W W W. TA P P I .

O R G

JOURNAL

THE PAP E R AN D PACK AGING IND UST R IE S TEC H NIC A L R ES OUR C E

AUGUST 2014 | VOL. 13 NO. 8

CORROSION

UE the
S
L
IA G IS from e
C
E N rch nc
SP ERI esea nfere
R o
NE ring RS C
I
G atu EE
EN Fe 3 P
1
20

Digester thinning:
Erosion-corrosion of internal
flow channel headers
Greg A. Busby and Peter W. Hart

RECOVERY BOILER

19

A laboratory study of recovery

boiler smelt shattering

Anton Taranenko, Markus Bussmann,

and Honghi Tran

RECOVERY BOILER

27

A novel method for determining

the internal recycled dust load
in kraft recovery boilers
Matheus Antunes Guimares,
Honghi Tran, and Marcelo Cardoso

CORROSION

37

Could biomass-fueled boilers

be operated at higher steam
temperatures? Part 1: Laboratory
evaluation of candidate
superheater alloys
Douglas L. Singbeil, Laurie Frederick,
James R. Keiser, and W.B.A. (Sandy)
Sharp

CORROSION

51

Could biomass-fueled boilers

be operated at higher steam
temperatures? Part 2: Field tests
of candidate superheater alloys

James R. Keiser, W.B.A. (Sandy) Sharp,

and Douglas L. Singbeil

INDUSTR
OF

VICE
SER

65

Could biomass-fueled boilers be

operated at higher steam
temperatures? Part 3: Initial
analysis of costs and benefits

100 YEARS

CORROSION

1 9 1 5 0 1 5
-2

W.B.A. (Sandy) Sharp, W.J. (Jim)

Frederick, James R. Keiser,
and Douglas L. Singbeil

PULPING

81

Combustion properties of
reduced-lignin black liquors
Niklas Vh-Savo, Nikolai Demartini,
Rufus Ziesig, Per Tomani, Hans
Theliander, Erkki Vlimki, and
Mikko Hupa

SUPERIOR SPRAY. SERIOUS RESULTS.

Automatic brush showers that run cleaning cycles automatically
Shower nozzles designed for long wear life
UltraStream nozzles provide precise, crisp edge-trimming. High-impact
performance allows nozzles to be placed further away from target to
reduce build-up
The industry's largest selection of nozzles for cleaning, cooling and deluge
Learn more at spray.com/papermaking download our new free
Paper Industry Catalog 66A and request a free paper machine
nozzle inspection.

NOZZLES FOR SHOWERS

Choose from at and
solid stream patterns,
wear-resistant orice
materials and more

NO-OBLIGATION PAPER MACHINE AUDITS

Free inspection service to document
all nozzles on machine; nozzle wear
rate assessment and more

TANKJET 360
is ideal for thorough and
sts
quick cleaning of stock chests
and pulp storage chests

Unmatched Global Engineering, Manufacturing and Technical Support

Nozzles | Control Systems | Headers & Injectors | Research & Testing
1.800.95.SPRAY

August 2014
TABLE OF CONTENTS

WWW.TAPPI.ORG

VOL. 13 NO. 8

5 EDITORIAL
Peer-Reviewed Engineering highlights from PEERS 2013

Peter W. Hart

9 CORROSION

Digester thinning: Erosion-corrosion of internal flow

channel headers

Greg A. Busby and Peter W. Hart

19 RECOVERY BOILER

laboratory study of recovery boiler smelt shattering
A

Anton Taranenko, Markus Bussmann, and Honghi Tran

27

RECOVERY BOILER

Matheus Antunes Guimares, Honghi Tran, and Marcelo Cardoso

37

CORROSION

Douglas L. Singbeil, Laurie Frederick, James R. Keiser,

and W.B.A. (Sandy) Sharp

51

CORROSION

James R. Keiser, W.B.A. (Sandy) Sharp, and Douglas L. Singbeil

A novel method for determining the internal recycled dust

load in kraft recovery boilers

Could biomass-fueled boilers be operated at higher

steam temperatures? Part 1: Laboratory evaluation of

candidate superheater alloys

65

Could biomass-fueled boilers be operated at higher

steam temperatures? Part 2: Field tests of candidate
superheater alloys

[email protected]

Associate Editor
Scott Rosencrance
+1 770 429-2753

[email protected]

Editorial Director
Monica Shaw
+1 770 367-9534

[email protected]

Vice President, Operations

Eric Fletty

[email protected]

PRESS Manager
Jeff Wells
+1 770 209-7228

[email protected]

Webmaster
Trina Heath
+1 770 209-7416

[email protected]

TJ EDITORIAL BOARD
Terry L. Bliss, Ashland Inc.
[email protected], +1 302 995-3523
Brian N. Brogdon, FutureBridge Consulting & Training LLC
[email protected], +1 678 581-9114
David A. Carlson, Carlson Consulting
[email protected], +1 847 323-2685
Jere W. Crouse, JWC Consulting
[email protected], +1 608 362-4485
Mahendra Doshi, Doshi and Associates
[email protected], +1 920 832-9101
Peter W. Hart, MeadWestvaco Corporation
[email protected], +1 409 276-3465
Carl J. Houtman, USDA Forest Products Laboratory
[email protected], + 1 608 231-9445
John A. Neun, Albany International
[email protected], +1 518 275-8139
Steven P. Ottone, SNP Inc.
[email protected], +1 919 641-7854
Arthur Ragauskas, Oak Ridge National Laboratory
[email protected]
John A. Roper III, Dow Chemical Company
[email protected], +1 989 636-5644
Scott Rosencrance, Kemira Chemicals
[email protected], +1 770 429-2753
Ricardo B. Santos, MeadWestvaco Corporation
[email protected], +1 540 969-2426
Paul Wiegand, NCASI
[email protected], +1 919 941-6417
Junyong Zhu, USDA Forest Products Laboratory
[email protected], +1 608 231-9520

Join TAPPI today!

CORROSION

Could biomass-fueled boilers be operated at higher

steam temperatures? Part 3: Initial analysis of costs

and benefits
W.B.A. (Sandy) Sharp, W.J. (Jim) Frederick, James R. Keiser,

and Douglas L. Singbeil

81 PULPING

Combustion properties of reduced-lignin black liquors

Editor-in-Chief
Peter W. Hart
+1 409 276-3465

JOURNAL

Niklas Vh-Savo, Nikolai Demartini, Rufus Ziesig, Per Tomani,

Hans Theliander, Erkki Vlimki, and Mikko Hupa

ON THE COVER: TAPPIs Engineering Division promotes the application of engineering principles to
the design, construction, operation, and maintenance of facilities for the manufacture of pulp, paper,
and related products. The Divisions Steam & Power/Energy Management and Corrosion & Materials
Engineering Committees actively explore topics like the digester, recovery boiler, and biomass-fueled
boiler research in this issue.

TAPPI JOURNAL is a free benefit of TAPPI membership,

and is only available to members. To join TAPPI, to renew your
TAPPI Membership, or to learn about other valuable benefits,
visit www.tappi.org.
TAPPI, 15 Technology Parkway S., Suite 115,
Peachtree Corners, GA 30092, publishes TAPPI JOURNAL monthly.
ATTENTION PROSPECTIVE AUTHORS: All papers published are subject to
TAPPI JOURNALs peer-review process. Not all papers accepted for review
will be published. Before submitting, check complete author guidelines at
http://www.tappi.org/s_tappi/doc.asp?CID=100&DID=552877.
Statements of fact and opinions expressed are those of individual authors.
TAPPI assumes no responsibility for such statements and opinions. TAPPI
does not intend such statements and opinions or construe them as a
solicitation of or suggestion for any agreed-upon course of conduct or
concerted action of any sort.
Copyright 2014 by TAPPI, all rights reserved. For copyright permission
to photocopy pages from this publication for internal or personal use,
contact Copyright Clearance Center, Inc. (CCC) via their website:
www.copyright.com. If you have questions about the copyright permission
request process, please contact CCC by phone at +1 978 750-8400.
To obtain copyright permission to use excerpts from this publication in
another published work, send your specific request in writing to Editor,
TAPPI JOURNAL, 15 Technology Parkway S., Suite 115,
Peachtree Corners, GA 30092, USA; or by fax to +1 770 446-6947.
Send address changes to TAPPI, 15 Technology Parkway S., Suite 115,
Peachtree Corners, GA 30092, USA; Telephone +1 770 446-1400,
or FAX +1 770 446-6947. www.tappi.org

| VOL.
| TAPPI JOURNAL
AUGUST
20142007
13 NO. 8 JOURNAL
MONTH
| TAPPI
3

How deinkers
save three ways
with one additive
Now you can save three ways with one
additive! DEKA can replace up to half of costly
surfactants, improve pulp brightness, and lower
ink residual numbers as well.
Sound too good to be true? In documented
mill tests, DEKA has replaced fully 50 percent of
expensive surfactant dosage without adversely
affecting deinking performance. And laboratory
studies prove DEKA usage can increase pulp
brightness up to 5 GE brightness points. Plus,
additional lab deinking tests have demonstrated
ERIC values lowered by up to 74 percent.
DEKA is also versatileflotation or wash
deinking processes both benefit from this
revolutionary deinking aid.
Contact Thiele today for details on how your
deinking process might benefit from using DEKA.
Well be glad to show you the test results and set
up a trial in your mill.

Thiele Kaolin Company Sandersville, GA

(877) 544-3322 www.thielekaolin.com

Editorial
PETER W. HART
Editor-in-chief

Peer-Reviewed Engineering
highlights from PEERS 2013
Editors Note: TAPPI would like to thank Brian Brogdon and Peter Hart of the TAPPI Journal (TJ) Editorial
Board for soliciting papers from the 2013 PEERS Conference and overseeing their review for this special
Engineering content, as well as additional past issues of TJ with PEERS content. Thanks also to the authors and
peer reviewers who participated in this process.
Monica Shaw, Editorial Director

he current issue of TAPPI Journal (TJ) represents the last in a series of special issues featuring highlights from the 2013 TAPPI PEERS
Conference held in Green Bay, WI, USA. The
first two 2014 issues with peer-reviewed PEERS content
focused on Pulp Manufacturing (March) and Nonwood
Pulping ( June).
The papers in this issue were obtained from PEERS
paper sessions sponsored by the Engineering Divisions
Corrosion & Materials Engineering and Steam & Power/Energy Management Committees. Several committee
chairs were asked to recommend papers appropriate
for an Engineering highlights issue of TJ. The authors
of the suggested manuscripts were solicited to submit
their work to the peer review process. The submission
and review process resulted in six papers available for
this special Engineering highlights issue.

A snapshot of Engineering issues

The papers in this issue represent a snapshot of some
of the ongoing research and development issues currently being addressed by the Engineering Division
committees in TAPPI. The topics include:
S
 melt Continued effort is being expended to
improve the industrys understanding of smelt
shattering and smelt runoff events. Both of these
items may be directly related to liquor quality and
consistency, as well as to the safe operation of
industry equipment.

C
 orrosion Ongoing corrosion studies relate to
equipment longevity and reduced maintenance
costs. Improved understanding of the type of and
mechanisms associated with corrosion may also
lead to reduced capital costs for new pieces of
equipment.
Other Engineering topics at PEERS 2013 included
efforts in understanding the impact of new processing
regimes, such as enhanced biomass boiler operation,
that may improve capital requirements, maintenance,
and efficiency for large scale equipment. Also, the
Engineering Division is actively following legislative
issues such as Boiler MACT and continuing to update
their constituents on the latest developments in both
proposed and currently active legislation that has the
potential to impact our industry.
An additional paper appears in this issue from
Niklas Vh-Savo, a research associate and doctoral
student at bo Akademi University in Finland. The
paper on p. 81 is part of his Ph.D. thesis on Behavior
of black liquor nitrogen in combustion Formation of
cyanate, to be defended later this year.

TAPPI PEERS 2014 in Tacoma

The Engineering Division will again sponsor several
sessions at the 2014 TAPPI PEERS Conference in
Tacoma, WA, this September 14-17. The Engineering
Program will consist of three distinct tracks: Corrosion
and Materials; Energy, Power and Recovery; and
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

Engineering and Maintenance. Additional tracks will

be sponsored by the Environmental Working Group
and the Pulp Manufacturing Division. Recycling
and Sustainability programs will also be offered. In
addition, committee meetings and steering committee

If you or your engineers want an excellent

networking opportunity, along with the
opportunity to shape the content of
future PEERS conferences, joining one of
the Engineering Divisions subject matter
committees (open to all TAPPI Members)
provides an excellent starting point.

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

meetings will be held during the conference. If you

or your engineers want an excellent networking
opportunity, along with the opportunity to shape the
content of future PEERS conferences, joining one of
the Engineering Divisions subject matter committees
(open to all TAPPI Members) provides an excellent
starting point. A list of committees is found at
www.tappi.org/Groups/Divisions/Engineering.
We hope you enjoy the PEERS Engineering highlights
from TAPPIs Engineering Division in this issue of TJ, and
we hope to see you in Tacoma. TJ
Peter W. Hart, Ph.D., editor-in-chief of TAPPI Journal,
is the director of Pulping, Bleaching, and Chemical
Process Technology for MeadWestvaco Corporation,
Atlanta, GA, USA.

Introducing

EnerVent
Patent Pending

For Maximum

TISSUE
MACHINE
Performance

Customers Report:
Increased post pressure roll consistencies
Reduced steam consumption
Lower hood temperatures

Learn More at Booth #424

Reduced energy consumption

Increased machine speeds

Advanced Tissue Roll Technology

XERIUM.COM

PEER-REVIEWED

CORROSION

Digester thinning: Erosion-corrosion

of internal flow channel headers
GREG A. BUSBY

and

PETER W. HART

ABSTRACT: A softwood continuous digester has experienced severe erosion-corrosion of the shell wall inside
the internal flow channel headers of the extraction and modified continuous cooking (MCC) zones. Although erosioncorrosion is a common form, it is not typically looked for inside digester flow channels. The worst damage was
located where high velocity liquor exits the screen orifices and enters the collection headers. With erosion-corrosion
rates as high as 200 m/year, the damage has effectively reduced the wall thickness almost by half in the worst areas.
Also affected were the horizontal backing rings that form the bottom of the flow channels. An API-579/ASME FFS-1
Part 5, Level 2 analysis was performed to allow the mill to continue operating the digester until the next scheduled
outage. Owner-users are encouraged to inspect these locations in their digesters to ensure that erosion-corrosion
has not caused accelerated and/or unexpected shell thinning. The internal flow channels represent locations that are
not readily accessible for internal visual inspection without removal of the flow channel header cover plates and
removal of the residual black liquor typically retained in these flow channels. External, on-stream ultrasonic
thickness testing (UT) at the proper collection header elevations has been determined to be an effective way to
detect the presence of this erosion-corrosion phenomena. Also discussed are considerations to make on-stream UT
as accurate as possible. Corrosion rates determined via electrochemical corrosion probe monitoring have correlated
very well with the corrosion rates determined by on-stream UT data.

Application: The detailed results from this study should enable the users of continuous digesters with internal

flow channels to improve their internal inspection techniques. Improved inspections will allow for safer operation of
these type digesters.

s part of fitness for service and general safety guidelines,

digesters are required to undergo routine shell thickness
inspections. The digester involved in this study was originally
designed in 1993 as a 1250 a.d. tons/day, 2-vessel hydraulic
continuous digester. Operation began in March, 1995. The
design consists of upper and lower internal flow channels for
the extraction and modified continuous cooking (MCC) zones
vs. external headers. The single-level of wash screens is
equipped with an external ring header. The design is a Kamyr
wet feed system consisting of a chip bin, chip meter, lowpressure feeder, steaming vessel, chip chute with a circulation
pump, in-line drainers, and a sand separator in the circulation
loop feeding the high pressure feeder. The digester has been
routinely monitored and well maintained since it was first
commissioned. Annual internal visual and ultrasonic thickness (UT) inspections through the years have always indicated that the general corrosion rates were low. Thickness readings were typically taken at three locations on each shell
course at eight sectors around the circumference. The typical
UT survey during the outages also included taking several
thickness readings of the shell that is directly accessible by
removing one of the flow channel clean-out windows. Usually, only one or two screens at each zone were opened for
visual and UT of the shell, as well as wet fluorescent magnetic particle testing (WFMT) of shell welds or penetrant testing

(PT) of stainless nozzle welds. Since the internal collection

headers were not equipped with drains that would remove
the residual heel of colored extraction liquor below the outlet
nozzles, there was no visual inspection of the lower portion
of the flow channels as viewed through the small clean-out
windows.
In 2012, the mill asked for a six-month extension to the annual inspection frequency. This necessitated a critical review
of the available thickness data. During this review, it was
noted that the UT data that had been taken inside the flow
channels through the clean-out windows varied by as much
as 0.300 inch around the circumference at the same digester
elevation. Even if the data were located on separate shell
plates, all shell plates at any given elevation should have the
same nominal thickness. This variation in thickness was the
first indication that the shell behind the flow channels had
experienced in-service thinning.
Based upon review of the literature, rapid thinning can be
a common problem in softwood continuous digesters running
modern cooking processes. Wensley reported in 1998 that
continuous digester extraction liquors may produce corrosion
rates up to nearly 0.25 in. per year [1]. Wensleys paper also
showed that corrosion rates increase rapidly with increasing
temperature. A corrosion rate of 0.079 in. per year in a softwood continuous digester extraction liquor at 327F increased
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

CORROSION

1. Roll-out reference sketch of digester.

more than 3.5 fold to 0.290 in. per year when the temperature
increased to 350F.
Although equations are available to predict corrosion rates
in white liquor from the liquor composition, no predictive
equations have been developed to predict corrosion rates in
extraction liquors. Predictive equations have not been correlated with the inorganic and organic constituents of extraction liquor. The available equations do not account for the
extraction liquor temperature, or the liquor flow velocity.
What is known is that extraction zone liquors are particularly
corrosive because of their high temperature, low caustic concentration, and high flow rate.
10

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

PROBLEM IDENTIFICATION
The current work describes a softwood digester that has recently experienced severe erosion-corrosion of the carbon
steel shell behind the internal flow channel headers. For the
softwood continuous digester in this case study, the combination of liquor heaters operating wide open, a faulty alkali sensor that resulted in caustic concentrations below intended
values, and a high flow velocity inside the flow channels resulted in serious wall thinning damage to the shell. The damage was greatest in the vicinity of the screen orifice holes, as
well as the bottom 4 or 5 in. of the flow channels. Based upon
observations such as the erosion patterns and location of dam-

CORROSION
age, it was concluded the local flow velocity was a key parameter affecting the rate of thinning damage. Descriptions of
how the damage was initially found and how it was monitored
with on-stream ultrasonic thickness and electrochemical corrosion probe testing are reviewed. Also discussed are details
specifically related to external on-stream thickness testing,
which ensure that the corrosion rate data are accurate. A rollout reference sketch of the digester in question, along with
the areas of high corrosion, is shown in Fig. 1.
ONSET OF DIGESTER THINNING
The morphology of the damage inside the digester flow channels clearly indicates the thinning is the result of erosion-corrosion. Erosion-corrosion is an acceleration of corrosive damage caused by the relative motion of a corrosive fluid against
a metal surface. When a certain velocity threshold is exceeded
for a given alloy, the relative motion is capable of removing an
otherwise protective (passive) film that forms on the metal
surface [2]. The thinning damage found in this digester exhibited the classic erosion-corrosion features such as grooves,
waves, comet-tails, teardrop-shaped pits, and horseshoeshaped undercutting of the metal surface. Figure 2 shows
some of the classic erosion-corrosion pictures obtained from
the digester flow channels. Erosion-corrosion can produce
dramatic rates of damage, which were also experienced in
this digester. Based on the locations and patterns of thinning,
the flow velocity appears to be the primary component of the
damage that occurred inside the digesters flow channel headers. Static corrosion rates have always remained relatively normal for this digester.
RESPONSE TO DIGESTER THINNING

On-stream thickness testing (pre-outage)

Once it was apparent that a hidden corrosion problem may
have developed inside the internal flow channels, it was necessary to do external on-stream UT testing at the flow channel elevations in order to verify the scope and magnitude of
damage. Originally, small 4 in. x 4 in. windows were cut in
the insulation for thickness monitoring locations (TMLs).
These TMLs were oriented 45 apart at the elevations that
corresponded to the mid-height of the flow channel areas.
The flow channels are actually 16 in. in vertical height, and
the true placement of the TMLs within the flow channel zone
was about midway. As the TML areas were scanned, the spotto-spot thickness data varied considerably. The first impression was that the UT readings were widely inaccurate; however, it was determined that the true wall thickness actually
did vary considerably from spot to spot in the TML. It was
also clear that the thickness decreased near the bottom of
each TML window. It was decided to remove the majority of
the insulation corresponding to the flow channel zones. The
new data collected indicated the true severity of thinning.
Extreme substrate loss was found near the bottom 5 or 6 in.
of the flow channels and also in separate vertical bands
around the circumference corresponding to the location of

shell

weld
corrosion

lower
extraction
nozzle

2. Wavy pattern of erosion-corrosion damage to the extraction

header plate directly beneath the screen orifice. In the lower
picture, notice the erosion-corrosion damage inside the flow
channel in the vicinity of the extraction nozzle.

the screen orifice holes. When plotted, the thinning had an

almost sinusoidal pattern around the circumference of the
digester.
With the shell thickness as much as 0.300 in. below the
calculated ASME t-min [3] in the worst areas, the mill contracted with an engineering consultant to perform an API-579/
ASME FFS-1 fitness for service evaluation [4-5]. Using Part 5,
a level 2 analysis was performed. Through the rigors of finite
element analysis (FEA), it was possible to demonstrate that the
digester was indeed safe to operate for a few more months
until the scheduled outage, which was months away.

On-stream ultrasonic thickness data

and accuracy
When trending UT data collected on a frequent basis, the data
needs to be as precise and accurate as possible so that the indicated corrosion rates are not exaggerated. When ultrasonic
thickness testing on operating digesters (on-stream inspection),
three considerations are necessary for accurate and precise
thickness data: 1) clearly identified and physically marked TML
locations directly on the digester surface; 2) temperature-velocity compensation of the UT calibration; and 3) elimination of
temperature drift effects of the transducer as it heats up.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

11

CORROSION
As the temperature of a metal increases, the velocity of
sound through the material slows down; thus, it takes longer
to go the same distance (thickness). This change in sound
velocity with temperature is an important variable for
accurate readings. If the high temperature velocity effect is
not accounted for when testing a hot component, a UT
technician will erroneously get a reading that indicates the
component is thicker than it actually is. For example, in the
wash zone of a typical continuous digester with a thickness
of 2 in. and an operating temperature of 320F, the thickness
error from operating temperature to ambient temperature
(85F) calibration will be about 0.050 in. [6]. ASME Section
V indicates a temperature correction factor of 1% per 100F
difference between the calibration and testing temperature.
The alternate way for high temperature calibration is to heat
a thickness step block to the operating temperature and then
store the calibration for this temperature corrected sound
velocity. Done with care, there should be excellent correlation
between these two high temperature ultrasonic thickness
calibration methods for the normalized SA 516-70 plate [7].
Another temperature induced error that can easily be
eliminated is one related to high temperature transducer
drift. As the plastic potting material surrounding the transducer heats up, the sound velocity passing through this material decreases even more than it would when passing through
a metal. Since the measured sound must travel through this
part of the probe, as well through the digester shell, the timing and gated thickness will be altered by this effect, referred
to as transducer drift. An easy correction for transducer
drift is to set the thickness gage in the echo-to-echo mode.
In this test mode, the machine is set to gate the thickness
between multiple echoes of the distance from the metal surface to the back wall. In this way, thickness readings in the
echo-echo mode only represent the metal path distance and
are not affected by the thickness of couplant, paint, or temperature drift. At typical digester operating temperatures and
testing techniques, transducer drift can create another 0.020
to 0.030 in. of error in the thickness reading.

Digester shell

Corrosion of bottom
plate directly beneath
screen orifice

Flow channel
bottom plate

3. Inside the lower extraction flow channel with cover removed

for access. Note the severe erosion-corrosion of the shell
near the bottom plate, as well as wash-out of the bottom plate
(holding liquor) that is directly beneath the screen orifice.

4. Inside the lower extraction flow channel with cover removed

for access.

Internal inspection of digester flow channels

The moment of truth
Consistent with TAPPI Technical Information Paper TIP
0402-27 Guidelines for inspection of continuous digesters
and impregnation vessels, the internal inspection consisted
of direct visual inspection, ultrasonic thickness testing, and
dye penetrant and wet fluorescent inspection of critical
welds. Additionally, the flow channel cover plates were removed for the first time in the operating history of the
digester. Once the flow channel cover plates were removed,
it was obvious that significant and very real damage had occurring within the flow channels. The damage was found to
be as bad as the on-stream UT data had indicated. Figures
3-5 are photos of the extraction flow channels showing the
extent of corrosion that was present within the digester at
the time of the inspection.
12

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

Bottom plate of
flow channel with
cover removed

5. Erosion-corrosion damage to the digester shell and bottom of

the flow channel.

CORROSION
blank plate
removed
(replaced)

screen (profile
bars)

localized wash-out
around stand-off pins
localized extraction flow channel
with covers removed

7. Upper MCC flow channel. (Nozzle height leaves 2.5 in. heel
of liquor in the bottom).

erosion/corrosion
entities inside lower
extraction f low channel
caused by f low eddies
around stainless
stand-off pins

6. Note the localized damage around standoff pins that are

attached to the flow channel covers that were removed. The
lower photo is a close up detailing the localized corrosion
caused by flow eddies around the standoff pins.

Extraction zone
The upper and lower extraction zone consisted of a checkerboard pattern of ten alternating carbon steel blank plates with
ten vertical stainless steel profile bar screens. The liquor collects through the screens and feeds through two flow orifices
into the flow channel header. The flow channel headers have
two outlet nozzles oriented 180 apart. The erosion-corrosion
damage was concentrated near the flow orifice locations and
affected the lower portion of the shell inside the flow channel. There was significant localized thinning of the bottom
plate (backing bar) of the collection header where liquor
flows through the screen orifices and impinges directly upon
the bottom plate (Fig. 5). At the lower extraction zone, severe
damage extended 5 or 6 in. up from the bottom of the collection header. Localized areas of shell thinning associated with
the stand-off pins were observed behind the screen plates
as well as inside the flow channel. Galvanic corrosion effects
were not seen in these areas as the standoff pins do not actually touch the impacted area of the digester shell. Rather, these
pins tend to collect pulp fibers and increase the flow rate of
liquor around these pins. Figure 6 shows the localized cor-

rosion around the stand-off pins. Flow eddies appear to form

around these pins, which causes accelerated erosion-corrosion with the classic comet tail and horse-shoe track patterns. The corrosion penetration into the shell at these standoff pins located inside the flow channel were much deeper
than that around the pins behind the screens. Even though
the process chemistry in these two areas is nearly identical,
higher velocities are present inside the flow channel than behind the screens, thus enhancing the erosion-corrosion rate.
MCC zone
The MCC zone consists of two levels of diagonal, mill-slotted
screens with no blank plates. There are 20 screens in the
upper zone and 20 screens in lower zone. At each screen,
there is only one open flow orifice, as one of the two orifice
holes was blanked in years past.
There was erosion-corrosion damage to the shell inside
both MCC flow channels, similar to that found at the extraction zone. Although this damage required weld buildup repair
to restore to ASME t-min thickness, the damage was much less
severe than the damage at the extraction level. The bottom
plates of the flow channel headers showed minor to moderate
damage underneath the open flow orifice. In general, the
damage to the shell in the MCC flow channels did not extend
more than 2 or 3 in. above the bottom plate of the collection
header. Figure 7 shows the MCC channel along with the
heel of liquor retained in the bottom of the flow channel
that impedes visual inspection of the internal flow channels.
WELD REPAIRS
All areas of the shell below the corresponding ASME t-min
value were repaired by weld overlay. Prior to welding, the area
was preheated to 300F using a rose-bud technique [8]. The
welding was performed by shielded metal arc welding
(SMAW) using 5/32 dia., E7018 electrodes in consecutive,
horizontal weld beads. The weld overlay areas were built-up
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

13

CORROSION
to the corresponding t-min value plus an additional 0.050 in.
of corrosion allowance. In a few of the worst areas of the lower
extraction zone, ER 309L SS electrodes were applied on top
of the carbon steel weld build-up.
POST OUTAGE ON-STREAM
CORROSION MONITORING
Because of the wide variance in local shell wall thicknesses
caused by corrosion, TMLs used for UT comparison were
marked directly on the digester shell. Thickness readings
were taken at these precise locations for optimum corrosion
rate monitoring accuracy. Due to the expected high erosioncorrosion rates inside the flow channels, the majority of TMLs
were concentrated at the flow channel elevations. Areas
below the flow channels were included to establish static
corrosion rates outside of flow areas.
Baseline thickness readings were taken prior to the
digesters return to service. On-stream thickness readings
were taken approximately every 30 days for the next six
months. Annualized corrosion rates were calculated for
each TML to determine a short-term corrosion rate (between
the last two readings) and a longer-term corrosion rate
(between the baseline and the last reading). These data
showed extremely high corrosion rates (100 to 200 m/year)
in the majority of TMLs located in the flow channels and low
corrosion rates (0-5 m/year) in static areas outside of the
flow channels. Also, during the six months of on-stream
thickness monitoring, the mill experimented with various
cooking schemes that showed a wide variance in corrosion
rates.
In addition to the on-stream UT thickness monitoring, six

ABOUT THE AUTHORS

We chose to study this aspect of digester thinning because exploration of variations in routine thickness
inspections led to the surprising discovery of extreme rates of erosion-corrosion occurring within the
digester flow channels. The need to understand the
cause, mitigation, and prevention of the rapid thinning led to the preparation of this work.
This work supports earlier efforts at identifying
various forms of rapid digester thinning experienced
in other continuous digesters. The current work compliments efforts in defining specific locations of
erosion-corrosion.
Determining the root cause of the digester thinning and understanding the UT thickness testing data
were the main difficulties in this work. The most interesting thing we encountered was discovering the
extent of digester thinning in areas that were deemed
to be relatively safe and rarely inspected.
We have identified specific causes and locations of
potential rapid digester thinning. Mills may use this

14

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

electrochemical corrosion monitoring probes installed by an

anodic protection contractor during the recent outage were
also used to trend the corrosion rates during the six month
post outage period. In general, there was very good correlation between the UT thickness data and the electrochemical
corrosion rate data.
CONCLUSIONS
Outside of the high velocity flow channel areas, the general
corrosion rates on the digester shell range between 5 and 10
mils/year. Inside the flow channels, directly below the screen
orifice holes and along the bottom of the flow channels, the
localized erosion-corrosion rate appears to be as high as 200
mils/year. No thinning was noted behind the wash zone
screens, as there is an external ring header instead of an internal flow header. Velocity appears to be the critical parameter affecting the erosion-corrosion damage found inside the
flow channels. Throughout the vessel, there has never been
any indication of stress corrosion cracking (SCC), and no indications of SCC were found during this outage. Because portions of the digester were well below the ASME t-min thickness, there was discussion about immediately shutting the
digester down for overlay weld repair. Instead, API 579/ASME
FFS-1 Fitness for Service Part 5 Level 2 analysis was performed
to verify that the localized corrosion did not require an immediate shutdown of the digester. Rather, the digester was
allowed to run until the next scheduled outage. Weld repairs
were made to restore damaged areas to ASME t-min plus a
small corrosion allowance.
Based upon on-stream ultrasonic thickness data collected
since the outage, high erosion-corrosion rates still exist. However, after experimenting with various cooking changes, the

work to enhance
their own digester inspection programs
to ensure safe
operation of the
continuous digester. The next
step includes rebuilding the lost
Busby
Hart
base metal and
overlaying the
digester flow
channels. Once this is done, routine monitoring of
the flow channel thickness will be performed.
Busby is senior metallurgist for MWV Corp. in
Richmond, VA, USA. Hart is director of Pulping,
Bleaching, and Chemical Process Technology for
MWV Corp. in Atlanta, GA, USA. Email Hart at
[email protected].

CORROSION
mill determined the corrosion rate can be significantly varied. For instance, comparing UT survey data collected at the
beginning of down-flow cooking (where no extraction flow
exists inside the extraction headers) with data 30 days after
the cooking change showed the corrosion loss went from an
annualized 0.190 in. per year to essentially zero. These changes in cooking conditions show that the liquor velocity is a
major component in the damage that has occurred within
the flow channels. Being able to trial and error various cooking schemes and correlate these cooking schemes with corrosion rates only underscores the critical need for precise and
accurate on-stream ultrasonic thickness testing techniques.
For digesters with internal flow channels, if no current
or reliable wall thickness data from the shell near the bottom
of the flow channel is available, owner/users are encouraged
to schedule on-stream UT thickness measurements at the
flow channel elevations or plan to remove flow channel
covers directly below screens at an upcoming outage for
inspection. TJ
LITERATURE CITED

3. ASME, ASME Boiler and Pressure Vessel Code: An International

Code, 2013. Available [Online] https://www.asme.org/
getmedia/1adfc3df-7dab-44bf-a078-8b1c7d60bf0d/ASME_
BPVC_2013-Brochure.aspx >[31July2014].
4. Antaki, G., Fitness-for-Service Evaluations for Piping and Pressure
Vessels: ASME Code Simplified, McGraw-Hill, New York, 2005.
5. Bennett, D.C. and Mixon, N., API 579-1/ASME FFS-1, Fitness for
service post-construction code for pressure equipment with flaws
and corrosion damage, TAPPI PEERS Conf., TAPPI PRESS, Atlanta,
GA, USA, 2011, p. 997.
6. Pezant, J.C., High temperature thickness monitoring using ultrasonic waves, M.S. thesis, Georgia Institute of Technology, Atlanta,
2008.
7. ASME Boiler and Pressure Vessel Committee on Nondestructive
Examination, ASME Boiler and Pressure Vessel Code, Section V:
Nondestructive Examination, American Society of Mechanical
Engineers, New York, 1 July 2011. Available [Online] https://law.
resource.org/pub/us/code/ibr/asme.bpvc.v.2010.pdf >[31July2014].
8. U.S. Department of Transportation Federal Highway Administration,
Guide for Heat-Straightening of Damaged Steel Bridge Members,
Chap. 2, Section 2.7. Available [Online] http://www.fhwa.dot.gov/
BRIDGE/steel/02.cfm >[31July2014].

1. Wensley, A., Corrosion of batch and continuous digesters, Proc. Int. Symp. Corros. Pulp Pap. Ind., 9th, CPPA, Montreal, 1998, p. 27.
2. Corrosion Technology Laboratory, Erosion Corrosion, Corrosion
Fundamentals, NASA. Available [Online] http://corrosion.ksc.nasa.
gov/eroscor.htm >[31July2014].

2015 TAPPI Centennial Celebration:

Georgia World Congress Center - Atlanta, Georgia

Honoring Our Past,

Inspiring Our Future

PAPERCON 2015
AT TAPPIS CENTENNIAL CELEBRATION

IF YOURE INTERESTED
IN ATTENDING,
SPEAKING OR
EXHIBITING, VISIT:

PaperCon.org
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

15

STRENGTH RESIN
TECHNOLOGY THAT
REALLY STACKS UP
Kemira legacy companies introduced the rst commercial
synthetic strength resins more than 50 years ago. Today,
over 100 researchers in our global R&D centers continue
to develop new generations of strength resins and other
innovative paper technologies.
Kemiras FennoBond dry strength and FennoRez
Temporary wet strength products help our customers
enhance sheet properties, reduce basis weight, optimize
furnish costs, and increase machine speeds. It is no
wonder Kemira is the North American market leader for
glyoxylated polyacrylamide (GPAM) technology. Kemiras
FennoStrength resins are the highest solids products
commercially available, delivering unbeatable cost
efciency for permanent wet strength applications.

TOGETHER, INNOVATING PAPERMAKING

We have unique expertise in applying chemicals and
supporting you in innovating and continuously improving
your operational efciency. For a partner in paper that
really stacks up, contact us or visit our website
www.kemira.com.

Complete control over microbiological control.

Measure, monitor and manage with advanced tools from Buckman.
The more control you have over your biocide program, the better you
can control plant efciency, productivity and costs. Thats why Buckman
offers powerful tools that keep you informed and your water systems in
perfect balance.

Buckman OnSite

MicroBio Advisor
The MicroBio Advisor precisely measures
key microbiological control parameters,
including total chlorine residual, dissolved
oxygen, pH, and ORP, and delivers the
data to you in real
time. It also provides
feedback directly to
your automated biocide
feed unit so it can
adjust dosage. Text
alarms keep you aware
of potential problems.
And data can be sent
to your DCS, accessed
by computer and smart
phone, or integrated
with Buckman OnSite.

With Buckmans web-based portal

and data dashboard, you and your
Buckman representative can access
vital system data, including automated
controller data, manually entered
updates, material safety data sheets and
more. Buckman OnSite enables users
to efciently manage documents and
quickly create comprehensive reports,
complete with graphical interpretations.
All of your systems all in one place!

Find out more.

Bring a new level of effectiveness to your microbicide program with MicroBio Advisor and
Buckman OnSite. Contact your Buckman representative for more information, or visit
buckman.com.
2013 Buckman Laboratories International, Inc. All rights reserved.

buckman.com

PEER-REVIEWED

RECOVERY BOILER

A laboratory study of recovery

boiler smelt shattering
ANTON TARANENKO, MARKUS BUSSMANN,

and

HONGHI TRAN

ABSTRACT: A scaled-down experimental apparatus was built to examine smelt shattering during typical recovery boiler operations. Water-glycerine solutions and air were used in place of smelt and steam. A high-speed camera
and image processing software were used to record and quantify liquid shattering in terms of droplet number and
size distributions, as a function of air velocity, air nozzle position, liquid flow rate, and liquid viscosity. The results
showed that increasing shatter jet velocity reduced average droplet size, increasing the liquid flow rate increased
droplet size, and placing the shatter jet nozzle closer to the liquid stream decreased droplet size. These results were
all as expected. The effect of liquid viscosity (1-50 cP) depended on the shatter jet velocity. At high air velocities,
even the viscous liquid was well shattered, but at lower velocities, the effect of viscosity on shattering was
significant.
Application: Understanding how a molten smelt stream is shattered by a steam jet will help operators and process engineers to optimize their recovery boiler smelt shattering efficiency and to improve dissolving tank safety.

n kraft recovery boiler operations, molten smelt (mainly

consisting of sodium carbonate and sodium sulfide, and
small amounts of sodium sulfate, sodium chloride, and
potassium salts) flows out of the boiler at 800C-850C
down a number of spouts, at a flow rate on the order of
1 L/s per spout, before falling into the dissolving tank below
(Fig. 1). In the tank, the hot smelt mixes with weak wash to
produce green liquor that is subsequently causticized with
lime to produce white liquor for reuse in the pulping process.
The interaction between the molten smelt and water in the
confined space of the dissolving tank is violent; the noise and

1. Typical smelt shattering practice.

vibration of the tank often can be heard and felt far from the
tank itself. To control the intensity of the smelt-water interaction, mills use steam shatter jets to break up the smelt stream
into a spray of droplets just below the ends of the spouts.
Intense smelt-water interaction may be necessary to effectively dissolve smelt in the dissolving tank. Smelt shattering
is important to distribute smelt evenly throughout the tank,
rather than have large amounts of smelt simply pour into the
tank from the spout. Inadequate smelt shattering increases
the violence of dissolving tank smelt-water interaction. Experienced boiler operators claim to be able to assess dissolving
tank operation by listening to the tank. At the extreme, inadequate smelt shattering can lead to a dissolving tank explosion that can cause equipment damage, an unscheduled shutdown, and even personnel injury [1,2]. Despite these
concerns, smelt shattering practices vary widely from mill to
mill, and the shattering behavior has not been studied before.
The safety implications and lack of standards for smelt shattering motivated the study presented here.
CURRENT PRACTICE
An informal survey of a number of mills led us to conclude
that smelt shattering practices vary widely. Shatter jet nozzles
come in a variety of designs. Figure 2 shows examples of a
few designs that we encountered during mill visits. Some
mills simply modify the end of a steam tube to create a single
nozzle (Fig. 2a), or attach a T to create multiple nozzles
(Fig. 2d). Shatter jet nozzles can be round (Figs. 2a, 2b, and
2d) or slit-shaped (Figs. 2c and 2e). Frozen smelt can build up
on shatter jet nozzles, which has led some mills to install the
nozzle within a guard (Fig. 2b) or to use hot water or weak
wash to clean the shatter jet nozzle (Fig. 2a).
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

19

RECOVERY BOILER

2. Examples of shatter jet nozzle designs.

Steam used for smelt shattering varies from mill to mill, but
typically measures 3-15 bar (45-220 psi) and 150C-250C.
Steam consumption also varies widely, from 180 kg/h to 2250
kg/h per nozzle, or an order of magnitude range.
Shatter jet nozzle placement and orientation and spout inclination can also affect shattering. Shatter jets are usually
aimed at the smelt flow just below the end of the spout, and
the nozzles are usually installed above the spout and point
downwards, or further from the boiler wall and point back
towards the smelt flow. The choice affects how well smelt can
be shattered when the flow rate is unusually high or low,
which along with spout inclination, affects the trajectory of
the smelt. When the smelt flow rate is low, and especially if
the spout inclination is shallow, smelt will simply drip off the
tip of the spout. When the smelt flow rate is high, and espe-

3. Shattering apparatus: a) schematic and b) illustration.

20

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

cially if the spout inclination is steep, smelt will shoot off the
spout. In either case, shatter jet placement will affect shattering effectiveness. To accommodate such situations, some mills
install two jets per spout and turn on both during periods of
high smelt flow. Some mills also install adjustable shatter jets
that an operator can point toward abnormal smelt flows.
EXPERIMENTAL SETUP AND METHODOLOGY
A laboratory-scale experimental setup (Fig. 3a) was constructed to study smelt shattering. A water-glycerine solution
(liquid) was used in place of molten smelt, and compressed
air in place of steam. Liquid shattering was characterized as
a function of air velocity, nozzle position, liquid flow rate, and
liquid viscosity by imaging the spray with a high speed camera (Fig. 3b), and then processing those images to extract

RECOVERY BOILER
droplet number and size distributions. Although the apparatus was built at a reduced scale, the various parameters were
chosen to reflect typical values of these parameters in recovery boilers.
As shown in Fig. 3a, the apparatus was operated as follows.
A large collection tank stored the water-glycerine solution.
The solution was pumped through a flow meter to an inclined
tank, with a spout mounted near the top (Fig. 3b). After that
tank filled, the liquid began to flow down the spout (the inclination was maintained at 15 for all of the experiments described here) at a known flow rate, and fell back into the collection tank.
The liquid stream was shattered into droplets by an impinging air jet (Fig. 3b). The air was drawn from the building supply. The air flow rate was controlled by a ball valve and measured by a flow meter. The air line was capped with a
Laval-type nozzle (11.9 mm outlet dia., 7.9 mm throat dia.).
The nozzle was positioned either 7.5 cm or 15 cm above the
liquid stream and pointed straight down. This nozzle proximity to the liquid stream, if scaled up, translates into 30-60 cm,
a typical distance between a shatter jet nozzle and the smelt
stream.
Shattering experiments were conducted at liquid flow
rates of 0.1 L/s and 0.2 L/s to yield liquid velocities at the end
of the spout similar to those in practice. The lower flow rate
(0.1 L/s) is representative of the smelt flow encountered under
normal recovery boiler conditions; 0.2 L/s is more representative of heavy smelt flow. Different water-glycerine solutions
were used to obtain liquid viscosities of 1, 2.5, 10, and 50 cP,
where 3-5 cP is typical of recovery boiler smelt at 800C-850C
[3]. (Liquid viscosity was limited to 50 cP because more viscous liquids could not be pumped at the requisite 0.2 L/s.) The
air velocity at the shatter jet nozzle exit was set to 100, 150,
200, 250, or 300 m/s, corresponding to inline air pressures of
10, 12.5, 15.5, 18.5, and 21.5 psig.

4. A view of the overall spray, and of the section that was

imaged (not to scale).

Image analysis
The liquid spray was backlit by a 300 watt light source and
imaged by a Mega Speed MS70K S2 digital high speed camera
(Canadian Photonics Lab, Inc.; Minnedosa, MB, Canada) fitted
with a 28-300 mm f/3.5-6.3 DG macro lens (Sigma Corp.;
Kanagawa, Japan). A translucent sheet behind the spray was
used to diffuse the light, which then shone through the spray
and into the camera. Figure 4 shows a characteristic spray
pattern from the end of the spout to 50 cm below, and the
position of the 3.5 3.5 cm section of the spray that was imaged. The depth of field of the images was about 2 cm.
After a set of images of a particular spray configuration had
been obtained, open source software (ImageJ) was used to
process the images and extract droplet size data. Figure 5
depicts the processing of a sample image. The original image
(Fig. 5a) was first converted into a binary (black and white)
format (Fig. 5b). ImageJ was then used to calculate individual
droplet areas (in pixels) from the binary image, which were
then converted into equivalent droplet diameters. Only circu-

5. Image processing: a) original image, b) binary of the original

image, and c) droplet outlines.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

21

RECOVERY BOILER
lar droplets were analyzed; larger liquid fragments, and droplets that were out of focus, were discounted. Figure 5c shows
the droplets that ImageJ identified from Fig. 5a.
Data from multiple images of any spray configuration was
then consolidated into droplet number and size distributions,
and the Sauter mean diameter (SMD or D32) was calculated.
The SMD is a common measure of the fineness of a spray; it
reflects the average ratio of drop volume to surface area [4,5]:

(1)
6. Shatter jet air velocity distribution; Nls = 7.5 cm.

where D is the droplet diameter and N is the number of droplets.

RESULTS AND DISCUSSION
Results are presented of the effect of four parameters on liquid
shattering: air velocity (uair), liquid flow rate (Ql), liquid viscosity (l), and the distance between the nozzle and liquid (Nls).
Images were processed for each of 80 experimental conditions, to allow for an analysis of the effect of each of the parameters while keeping the others constant.

Effect of air velocity

We considered five air flow rates, characterized by average
nozzle exit velocities of 100, 150, 200, 250, and 300 m/s, as

measured by the air flow meter. Axial velocities downstream

of the nozzle were then measured with a pitot tube to assess
the rate at which the air jets decayed. Velocities were measured horizontally, vertically, and diagonally across the jets.
The results confirmed that the jets were approximately axisymmetric. Figure 6 presents a sample measurement of
horizontal air velocity profiles measured 7.5 cm from the nozzle exit. The nozzle exit diameter was 11.9 mm. Figure 6 illustrates jet profiles six nozzle diameters downstream of the
exit. The jet center line velocities are only a half to a third of
the average nozzle exit velocity, and the jet is several times as
wide as at the nozzle exit.
Figure 7 displays a set of representative images that il-

7. Spray images depicting the effect of air velocity on shattering; Ql = 0.1 L/s, L = 2.5 cP, Nls = 7.5 cm.
22

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

RECOVERY BOILER

10. Liquid flow down a spout (front view).

8. The effect of velocity on droplet size distribution: a) number

density and b) volume density; Ql = 0.1 L/s, L = 2.5 cP, Nls =
7.5 cm.

11. Effect of liquid flow rate on spout exit velocity at different

inclination angles.

9. The effect of air velocity on Sauter mean diameter; Ql =

0.1 L/s, L = 2.5 cP, Nls = 7.5 cm.

lustrate the effect of air velocity on liquid shattering. In this

case, Ql = 0.1 L/s, l = 2.5 cP, and Nls = 7.5 cm. As expected,
the stronger the jet, the smaller the droplets, as a result of an
increase in the aerodynamic drag applied to the liquid stream.
The 100 m/s jet is not strong enough to adequately shatter
the liquid stream, as the image shows several large drops of
liquid. Increasing the velocity of the shatter jet by 50 m/s
largely eliminates those large drops, and further increments
of air velocity lead to a progressively finer spray.
Figure 8 shows two droplet size distributions as a function of air velocity. The number distribution (Fig. 8a) can be
misleading, as it appears to indicate that shattering is nearly
independent of air velocity. The problem is that even inade-

quate shattering yields many small droplets, but also a few

large drops that contain a large fraction of the total liquid
volume. The liquid volume distribution (Fig. 8b) conveys a
better sense of the effect of air velocity. The plot shows that
increasing air velocity shatters more of the liquid volume into
small droplets, and that large drops only appear at low air
velocities.
Figure 9 shows SMD versus air velocity and quantifies
the extent to which mean droplet size decreases with air velocity. In this case, the mean droplet size decreases by half, as
the air flow rate increases three-fold.

Effect of liquid flow rate

To examine the relationship between liquid flow rate and
average liquid velocity, the liquid cross-sectional area was
imaged at the end of the spout (Fig. 10) for various flow
rates and spout inclinations (0, 10, 15, and 20) and used
to calculate an average liquid velocity that is plotted in
Fig. 11. One can draw two conclusions from this plot. At a
given inclination, the velocity is almost independent of flow
rate. This means that the water level in the spout rises with
increasing flow rate. As long as the spout is inclined, the veAUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

23

RECOVERY BOILER

12. The effect of liquid flow rate on Sauter mean diameter;

l = 2.5 cP, Nls = 7.5 cm.

13. Liquid velocity at the spout exit versus viscosity; spout

inclination = 15.

locity is not a strong function of the inclination. It is only in

the horizontal spout that the liquid flows more slowly, and
thus the water level is higher.
To examine the effect of liquid flow rate on shattering, two
flow rates were considered: 0.1 L/s and 0.2 L/s. Figure 12
shows the SMD as a function of air velocity for each of the two
liquid flow rates. For these particular conditions, doubling the
liquid flow rate roughly increased the mean droplet diameter
by almost a factor of two.

Effect of liquid viscosity

The effects of liquid viscosity on the flow down the spout
and on shattering were studied. Figure 13 shows the average liquid velocity at the spout exit versus flow rate for four
different viscosities. As expected, as viscosity increased, the
liquid velocity decreased (and the liquid level in the spout
increased).
Figure 14 shows the effect of liquid viscosity on shattering. The 300 m/s air jet shattered all of the liquids well, as
24

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

14. The effect of liquid viscosity on Sauter mean diameter;

Ql = 0.1 L/s, Nls = 7.5 cm.

15. The effect of nozzle proximity on Sauter mean diameter;

Ql = 0.1 L/s, l = 2.5 cP.

demonstrated by the small variation in mean droplet diameter.

But when the air velocity was only 100 m/s, the effect of viscosity was significant; the SMD varied from 1.2 mm for the 1
cP liquid to 2.1 mm for the 50 cP liquid. Liquid viscosity obviously matters when the shatter jet is not strong enough to
simply overpower the liquid stream.

Effect of nozzle proximity

Figure 15 shows the effect of nozzle position on droplet size
(Ql = 0.1 L/s and l = 2.5 cP). As expected, placing the nozzle
closer to the liquid stream results in improved shattering.
Droplet mean diameter decreased as the shatter jet nozzle was
positioned closer to the liquid stream. Roughly speaking, the
SMD decreased by half when the nozzle is moved from 15 cm
to 7.5 cm from the liquid, for all air velocities.
PRACTICAL IMPLICATIONS
The results of this laboratory-scale study confirm what is

RECOVERY BOILER
common sense: shattering improves with increased shatter
jet flow rate, it improves as the shatter jet nozzle is moved
closer to the liquid stream, and it is easier to shatter the liquid
at lower flow rates than higher ones, as long as the nozzle is
properly positioned. For the range of liquid viscosities that
we studied (1-50 cP), the results show that adequate shattering requires a minimum shatter jet flow rate that must increase with liquid viscosity.
Although our results are preliminary, we can begin to
develop correlations that can be applied to actual scale. A
common parameter that is often used to characterize liquid
atomization (shattering) is the liquid-to-gas momentum ratio
q [6]:

(2)
where l and g are the liquid and gas densities, and ul and ug
are the liquid and gas velocities. Figure 16 shows median
drop diameter normalized by the diameter of the liquid
stream at the end of the spout, D l, plotted against the
momentum ratio q.
As expected, as the liquid-to-gas momentum ratio decreases, the mean drop size decreases as well. Despite considerable
scatter in the data, the results are generally consistent for the
two liquid flow rates that we considered. This is an example
of a result that we plan to build on to generalize shattering
behavior, and ultimately to reliably scale up the results.
Additional experiments are planned to examine the effectiveness of different nozzle geometries and positions, and to

16. Liquid-to-gas momentum ratio versus normalized droplet

mean diameter.

examine an even larger range of flow rates and viscosities

that might be encountered during boiler startup and upset
conditions. TJ
ACKNOWLEDGEMENTS
This work was conducted as part of the research program,
Increasing energy and chemical recovery efficiency in the
kraft process III, jointly supported by the Natural Sciences
and Engineering Research Council of Canada (NSERC) and a
consortium of the following companies: Andritz, AV Nackawic, Babcock & Wilcox, Boise, Carter Holt Harvey, Celulose
Nipo-Brasileira, Clyde-Bergemann, DMI Peace River Pulp,
Eldorado, ERCO Worldwide, Fibria, FP Innovations, Interna-

ABOUT THE AUTHORS

The way in which molten smelt is shattered by a
steam jet is believed to greatly affect dissolving tank
operation and safety. We chose this topic to study to
provide better insight into the shattering behavior of
molten smelt.
To our knowledge, no previous research has been
done on this topic. The most difficult aspect of this research was the scarcity or non-existence of literature
on smelt shattering. We addressed the problem by
visiting a number of pulp mills, consulting with mill
engineers and operators, and analyzing smelt shattering videos.
It was surprising to find that increasing the liquid
flow rate in an open channel did not increase its velocity, and that increasing the liquid viscosity from 1
cP to 50 cP had little effect on the liquid droplet size.
Understanding how a molten smelt stream is shattered by a steam jet might help mills to optimize their
smelt shattering efficiency and to improve their dissolving tank safety.
The next step is to continue to use the laboratory

Taranenko

Bussmann

Trann

apparatus to examine the effects of different jet nozzle designs and positions on droplet size and
distribution.
Taranenko is graduate student, Tran is professor,
Pulp and Paper Centre, and Department of Chemical
Engineering & Applied Chemistry, and Bussmann
is associate professor, Department of Mechanical &
Industrial Engineering, University of Toronto, Toronto,
ON, Canada. Email Tran at [email protected].

AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

25

RECOVERY BOILER
tional Paper, Irving Pulp & Paper, Kiln Flame Systems, Klabin,
MeadWestvaco, Metso Power, StoraEnso Research, Suzano,
Tembec, and Tolko Industries.
LITERATURE CITED
1. Grace, T.M. and Tran, H.N., Critical issues in dissolving tank operation, Int. Chem. Recovery Conf., TAPPI PRESS, Atlanta, GA, USA,
2010.
2. Lien, S. and DeMartini, N., Dissolving tank explosions: a review
of incidents between 1973 and 2008, unpublished report, Black
Liquor Recovery Boiler Advisory Committee and American Forest &
Paper Association, New York, 2008.
3. Tran, H.N., Sunil, A., and Jones, A.K., J. Pulp Pap. Sci. 32(3):
182(2006).
4. Yule, A.J. and Dunkley, J.J., in Atomization of Melts for Powder
Production and Spray Deposition, Clarendon Press, Oxford, NY, USA,
1994, Chap. 3, pp. 47-87.
5. Lefebvre, A.H., in Atomization and Sprays, Hemisphere Publishing,
New York, 1989, Chap. 3.
6. Mashayek A. and Ashgriz, N., in Handbook of Atomization and Sprays:
Theory and Applications (N. Ashgriz, Ed.), Springer, New York, 2011,
Chap. 29, pp. 657-684.

As of May 2014

2015 TAPPI CENTENNIAL CELEBRATION:

Honoring Our Past,

Inspiring Our Future

SPECIAL THANK YOU TO OUR SPONSORS*:

CENTURION SPONSOR

PLATINUM SPONSORS
MWV NewPage Nalco

GOLD SPONSOR
Domtar Sappi Valmet
SILVER SPONSOR
Dow Chemical GL&V Inc. Kadant
Packaging Corporation of America
Solenis

BRONZE SPONSORS
ABB FM Global Georgia-Pacific
Green Bay Packaging Hood Container Corp.
Imerys Ingredion MICA Corp.
Penford Products Co. Optest Poyry
Soundview Paper Sun Automation
Wausau Paper Corp Yates

Spring 2015 Atlanta, Georgia www.tappi100years.org

26

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

SUPPORTING SPONSORS
BrightKey Cascades Sonoco
Kruger Jedson Engineering
Shepard Exposition Services SunTrust
MEDIA SPONSORS
Board Converting News Corrugated Today
Flexible Packaging Magazine Ochre Media
PaperAge Paper360 PFFC
As of August 2014

PEER-REVIEWED

RECOVERY BOILER

A novel method for determining

the internal recycled dust load in
kraft recovery boilers
MATHEUS ANTUNES GUIMARES, HONGHI TRAN,
and MARCELO CARDOSO

ABSTRACT: In kraft recovery boiler operation, fly ash or dust generated from black liquor combustion is mixed
with the virgin black liquor in a mix tank and returned to the boiler with the as-fired black liquor. This internal recycled dust stream varies widely from boiler to boiler and from time to time and can have a great impact on the asfired black liquor flow and properties and, ultimately, on the boiler thermal performance. A new method has been
developed to quickly and accurately determine the amount of internal recycled dust in recovery boilers. The method
is based on the difference between the total organic carbon content of the virgin black liquor and that of the as-fired
black liquor. Tests using the method were performed on recovery boilers at three of Fibrias mills in Brazil. The
results show that while the specific virgin black liquor solids produced at these mills were about the same, the internal recycled dust load varied widely, from as low as 4 wt% of as-fired black liquor solids fired in the boiler at one mill
to as high as 15 wt% at another mill. Instead of total organic carbon values, heating values may also be used, but the
result is not as accurate.
Application: Mills can use a newly developed method to quickly and reliably determine the amount of internal
recycled dust in recovery boilers.

n kraft recovery boilers, the hot flue gas from black liquor
combustion gives off heat as it passes through a series of
heat transfer tube surfaces in the upper boiler to produce
steam. The cooling causes the vapors of alkali compounds in
the flue gas to condense and form fumes (Fig. 1). This, together with a small amount of physically entrained carryover,
forms a stream of fly ash or dust that is collected by ash hoppers and electrostatic precipitators (ESP) [1]. A small portion
of the dust stream ends up forming deposits on the tube surfaces, which are occasionally knocked off by sootblowers and
fall to the char bed or to ash hoppers. Only a small amount
(<0.5%) of the dust is lost with the stack gas as particulate
emissions.
The collected dust is mixed with the strong (or virgin)
black liquor, typically in a mix tank, and returned to the boiler with the as-fired black liquor. Because this dust stream
originates from the as-fired black liquor, and eventually returns to it, it is conventionally known as internal recycled dust
(IRD). The IRD typically accounts for about 5 wt% of the total
black liquor dry solids (BLDS) fired in the boiler; however, it
may vary from 1 wt% to 10 wt% depending on many factors,
including liquor firing load, bed temperature, boiler floor surface area, and whether a sootblower in the generating bank
or in the economizer is activated [27].
Tavares and Tran [4] conducted a systematic field study on
fume formation in a recovery boiler by (1) analyzing the thermal images of the char bed to obtain bed temperature distribution profiles, (2) using the profiles to calculate the amount

1. Dust particles and internal recycled dust stream.

AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

27

RECOVERY BOILER
of fume formed thermodynamically and equating the amount
with the IRD load, and (3) calibrating the calculated dust load
against the actual dust load determined independently
through a material balance around the mix tank over a
4-month period. The study showed that the increasing char
bed temperatures increased the internal IRD load, and that at
the same average bed temperature, a bed with a poor temperature distribution generated twice as much IRD as a bed
with a good temperature distribution.
In another study, Tamminen et al. [6,7] developed an online recovery boiler dust analyzer and used it to measure the
amount and composition of dust in the flue gas in two recovery boilers. The analyzer operated by continuously collecting
the flue gas dust sample, dissolving it in water, and analyzing
the flowing water potentiometrically for ion concentrations.
They found that at full liquor firing loads, the average dust
loads in these two recovery boilers were 6 wt% and 7 wt% asfired BLDS, respectively. Their data showed that the amount
of dust increased with an increase in liquor firing load and in
furnace floor surface area, and that the majority of the fume
(90%95%) was formed from in-flight burning of black liquor
droplets; only 5%10% came from the char bed.
As more and more kraft pulp mills have adopted high-solids firing practices in recent years, many recovery boilers operate at a high bed temperature. This inevitably results in an
increase in fume formation, and hence in IRD load. While
wide variations in IRD load can have a great impact on the asfired black liquor mass flow, composition, heating value,
steam production, and ultimately on the boiler thermal performance, there is no simple and reliable means of directly
measuring this dust stream. As a result, not many mills know
how large a recycled dust stream is circulating around their
recovery boilers.
In this paper, we first review the principles of several methods that have been used to estimate the amount of IRD. We
then discuss the results of a systematic study conducted on
recovery boilers at three kraft pulp mills within Fibria Celulose SA. The study was part of a larger effort to develop a
simple method that could be used to reliably determine IRD
load in recovery boilers, and then use it to investigate key
boiler operating variables that lead to high IRD dust loads.
COMMON METHODS

Mass balance method

where MAs-fired and MVirgin are the mass flow rates of as-fired
and virgin BLDS, respectively. Most kraft pulp mills monitor
the volume flow rates and solids content of the as-fired black
liquor online and use these values to calculate the mass flow
rate, MAs-fired, assuming a black liquor density value. Although
the liquor density changes with solids content [8], some mills
use only one fixed value for a wide range of their solids contents. This inherently results in an error in the MAs-fired calculation, particularly during the time when there is a wide variation in liquor solids content and IRD load. A more proper way
to calculate as-fired black liquor mass flow rates must take into
account the change in liquor density with solids content.
Furthermore, most mills do not continuously measure the
volume flow rate and/or solids content of virgin black liquor,
so MVirgin is usually not available. As a result, although Eq. (1)
is simple, it cannot be used readily.

Dust dissolution method

Several mills measure the amount of IRD on a daily basis by
directing the precipitator dust screw feeder to a small, dedicated tank filled with a fixed amount of water for a fixed period and analyzing the concentration and composition of the
dissolved dust in the tank solution. The analysis results are
then used to back-calculate the amount of IRD. The principle
of this method is similar to that of the online recovery boiler
dust analyzer used by Tamminen et al. [6,7] mentioned above,
except that it operates on a much larger scale. While the method is relatively accurate as it deals directly with the dust
stream, the procedure is complicated and the equipment involved is rather costly and requires frequent maintenance. As
a result, it has not been widely adopted.
In mills equipped with an ash treatment system to preferentially remove chloride (Cl) and potassium (K) from recovery
boiler precipitator ash [9], the amount of ash may be known
through the material balance around the ash treatment system. However, this method is limited only to mills where the
ash comes from one boiler and all is treated.
THEORETICAL CONSIDERATION
The amount of IRD can be determined by performing a material balance around the recovery boiler. Consider a system
involving three streams, in which stream 1 mixes with stream
2 to form stream 3, as shown in Fig. 2. M1, M2, and M3 are the
mass flows of streams 1, 2, and 3, respectively (Eq. [2]):

As-fired black liquor is the virgin black liquor that has been
mixed with dust collected from ESPs and ash hoppers. Therefore, the amount of IRD is essentially the mass difference between the virgin black liquor and the as-fired black liquor.
Customarily, IRD is expressed as the percentage, on a mass
basis, of the as-fired BLDS (Eq. [1]):

(1)
2. System with three streams.
28

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

RECOVERY BOILER
M1 + M2 = M3 (2)
In Eqs. (3) and (4), if a and b are two particular species in
stream 1, stream 2, and stream 3, then:
M1Ca1 + M2Ca2 = M3Ca3

(3)

and
M1Cb1 + M2Cb2 = M3Cb3

(4)

where Ca1, Ca2, and Ca3 are concentrations, in wt%, of a in

streams 1, 2, and 3, respectively, and Cb1, Cb2, and Cb3 are concentrations, in wt%, of b in streams 1, 2, and 3, respectively.
Rearranging the equations yields Eq. (5):

(5)

We now consider a case around a recovery boiler as shown in

Fig. 3, where MVirgin BL, MDust, M As-fired BL, MSmelt , and MStack,
respectively, represent the mass flows of virgin black liquor,
precipitator dust, as-fired black liquor, smelt, and stack gas.
Applying Eqs. (1)(5) to the black liquor mix tank, the
amount of IRD, expressed as % as-fired black liquor, can be
calculated (Eq. [6]):

(6)

where Ca As-fired BL, Ca Virgin BL, and Ca Dust, respectively, are concentrations in wt% of species a in as-fired black liquor, virgin
black liquor, and precipitator dust. Equation (6) suggests there
is no need to have the mass flow rates of various streams for
the calculation of IRD. In theory, only the concentrations of
one species or element (e.g., carbon, hydrogen, oxygen, sodium, potassium, sulfur, or chlorine) in as-fired black liquor,
virgin black liquor, and precipitator dust are needed.

Although this equation allows the use of any one component present in the as-fired black liquor, virgin black liquor
and precipitator dust to calculate the internal recycled dust,
the choice of such component, sampling locations, and sampling time can all have an impact on the accuracy of the calculated results. Both virgin and as-fired black liquors should
be collected at the liquor storage tank outlet where they have
supposedly been well mixed.
Collecting a representative dust sample is not as easy a task
as it may seem to be; dust and black liquor are often not well
mixed and the dust composition can vary widely with boiler
operation. For instance, boilers fired with high-solids liquors
tend to produce not only more dust, but also dust that contains
more carbonate, chloride, and potassium and less sulfate and
hence, more sodium than boilers fired with low-solids liquors.
Consequently, if sodium is used as the element of interest in
Eq. (6), the change in dust composition can lead to an error
in the calculation of IRD load.
Also, the amount and composition of dust collected after
a precipitator discharge cycle may be different from that collected during the discharge cycle. They may also vary greatly
depending on whether a sootblower in the generating bank
and/or the economizer region is activated during the time of
sampling. These variations can have a great effect on the asfired black liquor composition and hence, the accuracy of IRD
calculation.
PROPOSED METHOD
Choosing a component that is present in a large amount only
in the virgin and as-fired black liquors and not in the dust can
help minimize the system error and eliminate the need for
dust analysis. In this case, since Ca Dust = 0, Eq. (6) can be simplified as Eq. (7):

(7)

3. Flows around mix tank and recovery boiler

AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

29

RECOVERY BOILER

4. Total organic carbon analyzer.

One component that is abundant in both black liquor and dust

streams is carbon (C). In dust, C is present mostly as carbonate (i.e., sodium carbonate and potassium carbonate). In black
liquor, however, it is bound mostly with organic compounds,
which can be easily determined using a commercial total organic carbon (TOC) analyzer. Thus, by choosing TOC as the
component of interest, Eq. (7) becomes Eq. (8):

(8)

where TOCVirgin and TOCAs-fired are the TOC contents of virgin

and as-fired black liquors, respectively. Knowing the TOC
contents in these liquors, the amount of internal recycle dust
can be readily calculated.
METHOD VALIDATION

Reproducibility tests
To evaluate the validity and accuracy of the proposed method,
experiments were conducted to determine the reproducibility of the TOC analyzer used in this study. The analyzer was a
Model TOC-V CPN (Shimadzu Corp.; Kyoto, Japan) available
at mill 1 (Fig. 4). The analysis was performed on two sets of
virgin and as-fired black liquor samples collected from two
Fibria kraft pulp mills (mill 1 and mill 2). The virgin black liquor was collected from the weak black liquor tank to the
evaporation plant, while the as-fired liquor was collected from
the liquor feed tank to the boiler. Each liquor sample was diMill 1 Liquor

Mill 2 Liquor

Virgin

As-Fired

Virgin

As-Fired

Average of
10 samples,
g/kg BLDS

317

302

314

291

Standard
deviation,
g/kg BLDS

3.1

4.6

4.3

12.5

Variation
coefficient,
%

1.0

1.5

1.4

4.3

I. Total organic carbon contents in virgin and as-fired black

liquors from two kraft pulp mills.
30

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

5. Comparison between the amount of dust measured by the

proposed total organic carbon method and the amount of dust
initially added to the virgin black liquor samples.

vided into 10 subsamples that were dried at 105C for 12 h.

The subsamples were then analyzed for their TOC contents.
Table I summarizes the TOC analysis results for these 10
subsamples. The reproducibility was good for mill 1 liquors
and for mill 2 virgin liquor, with a small error (or variation
coefficient) of 1%1.5%. For mill 2 as-fired black liquor, however, the reproducibility seems to be poorer, with an error of
4.3%. The greater error for as-fired black liquor samples was
likely due to their more heterogeneous nature compared to
virgin black liquor samples

Calibration tests
Calibration tests were performed to determine if the TOC
analyzer could be used to determine the amount of precipitator dust in the as-fired black liquor. Two mill 1 virgin black liquor samples (with 17%25% solids) collected on different days
were used. Each sample was spiked with various amounts,
040 wt%, of mill 1 precipitator dust to produce well-mixed
liquor-dust mixtures. The mixtures were dried at 105C for 12
h and analyzed for TOC. The TOC values obtained for the virgin liquors and the mixtures were then used to calculate the
amounts of dust added to the samples using Eq. (8).
Figure 5 compares the amount of dust in the mixtures
determined by the proposed TOC method and the amount of
dust added to the mill 1 virgin liquor samples collected on different dates. The data points are located either exactly on or
close to the diagonal line of the graph, suggesting that the proposed method was accurate and the equipment was reliable.
IRD MEASUREMENTS
The TOC method was used to determine the amount of IRD
in recovery boilers at three Fibria mills (mills 1, 2, and 3). For
each mill, five sets of samples (virgin black liquor, as-fired
black liquor, and precipitator dust) were collected, and each
set was collected on a different day.
Figure 6 schematically shows the sampling locations. The

RECOVERY BOILER

6. Sampling locations.

virgin black liquor samples were collected at the weak black

liquor tank outlet, the as-fired black liquor samples were from
the concentrated black liquor tank, and the dust samples were
collected from the conveyor below the electrostatic precipitator. The samples were collected when the mills were at their
normal pulp production rates and the boiler operations were
stable. As in the reproducibility tests, the samples were dried
at 105C for 12 h before their TOC contents were determined.
The amount of IRD in each case was calculated using Eq. (8).
No organic carbon was detected in any of the dust samples. This, along with the relatively pure white appearance
of the samples, suggests that the recovery boilers at these
mills were running well with little or no unburned char entrained in the flue gas. Table II summarizes the results of
TOC analysis and the calculated internal recycled dust load
for these mills. The dust load varied widely from mill to mill,
averaging 12.7 wt% of as-fired BLDS for mill 1, 9.1 wt% for
mill 2, and 5.6 wt% for mill 3. Within each mill, the amount
also varied significantly from day to day, particularly at mill
2 where the value decreased from 15.2 wt% to 4.0 wt% BLDS
in one day (July 13) and then increased back to 8 wt% the
next day (July 14). Although it would have been beneficial to
know what caused the high variation in IRD load at these
mills, no attempt was made to investigate it because it was
not the main focus of this study.
The results clearly show that the recovery boiler at mill 1
operated with the highest IRD load, followed by the recovery
Mill 1
Sampling
Date

7. Average specific as-fired black liquor dry solids produced at

each mill.

8. Average specific black liquor dry solids versus recovery

boiler internal recycled dust load at mills 1, 2, and 3.

boiler at mill 2, with the lowest IRD load at mill 3. Mill operating data also showed that during the corresponding sampling period, mill 1 produced more specific as-fired black
liquor solids (tons of solids/ton of air-dry pulp) than mills 2
and 3 (Fig. 7).
Figure 8 plots the average specific black liquor solids
against the IRD load. There is a good correlation between the

Mill 2

TOC in Black Liquor

IRD

Virgin

As-Fired

g/kg
BLDS

g/kg
BLDS

wt%
BLDS

Sampling
Date

Mill 3

TOC in Black Liquor

Virgin

As-Fired

g/kg
BLDS

g/kg
BLDS

IRD

Sampling
Date

wt%
BLDS

TOC in Black Liquor

IRD

Virgin

As-Fired

g/kg
BLDS

g/kg
BLDS

wt%
BLDS

Jul 26

311

275

11.5

Jul-11

323

292

9.8

Sep-28

335

310

7.3

Aug 01

309

266

13.8

Jul-12

340

289

15.2

Sep-29

328

314

4.3

Aug 23

322

282

12.2

Jul-13

326

313

4.0

Oct-26

336

320

4.7

Sep 12

310

265

14.5

Jul-14

326

299

8.4

Oct-27

344

329

4.3

Sep 14

309

274

11.2

Jul-15

338

311

8.0

Oct-28

341

316

7.1

Average

312

273

12.7

Average

331

300

9.1

Average

337

318

5.6

II. Total organic carbon content in black liquors and calculated internal recycled dust load.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

31

RECOVERY BOILER

9. Effect of dust addition on the higher heating value of a virgin

black liquor sample from mill 4.

average specific as-fired black liquor solids at these mills and

the average IRD load in their recovery boilers.
Figure 8 also shows the specific virgin black liquor solids
produced at each mill, calculated based on the specific asfired black liquor solids and the internal recycled dust load,
using Eq. (1). Unlike the specific as-fired black liquor solids,
the specific virgin black liquor solids were the same for all
three mills: 1.41 tons of BLDS/a.d. ton of pulp. The results are
understandable. Because these mills employed the same
pulping technology, used the same type of wood species (eucalyptus), and used the same amount of chemicals (effective
alkali-to-wood ratio) to produce bleached eucalyptus pulp,
the specific virgin black liquor solids produced at each mill
are expected to be similar.
Extrapolating the data toward 0 wt% IRD load shows that
without the presence of internal recycled dust, the virgin and
as-fired black liquor solids would have been identical. The results, again, support the validity of the TOC method.
ALTERNATIVE METHOD
Precipitator dust contains a negligible amount (< 0.1 wt%) of
unburned carbon or char, even when it appears to be grayish
and contaminated with char. Thus, when mixed with virgin
black liquor in the mix tank and returned to the boiler with
the as-fired black liquor, the dust only increases the liquor
mass, not its heat content. The dilution and heat absorption
effects of the dust mass will lower the heating value of the asfired black liquor compared to that of the virgin black liquor.
As in the case of TOC, this difference in liquor heating value
can be used to estimate IRD load, using Eq. (9), a modified
form of Eq. (6):

(9)

where IRD is expressed as wt% in as-fired black liquor solids;

HV Virgin and HVAs-fired are the heating values of virgin liquor
32

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

10. Comparison between the amount of dust measured by the

proposed heating value method and the amount of dust initially
added to virgin black liquor.

and as-fired black liquor, respectively; and Dust is the heat absorbed by the dust in the bomb calorimeter during the heating
value measurements.
Laboratory tests were performed to determine the validity of Eq. (9). A sample of virgin black liquor from a kraft mill
(mill 4) was spiked with various amounts of precipitator dust
from the same mill to produce black liquor/dust mixtures.
These simulated as-fired black liquor samples were dried and
analyzed for their higher heating values (HHV) using a Model
6300 bomb calorimeter (Parr Instrument Co.; Moline, IL,
USA). As shown in Fig. 9, the HHV of the simulated as-fired
black liquor decreased linearly with an increase in dust content (R2 = 0.972). The linear relationship yields an HHV of
-1926 kJ/kg BLDS for a hypothetical liquor/dust mixture
at 100% dust (i.e., no liquor). This negative HHV value is
essentially the Dust itself (i.e., Dust = -165.39100 + 14,613 =
-1926 kJ/kg BLDS).
The tests were repeated on a different day. The obtained
Dust and HHV values were used to calculate the amount of
IRD for two sets of tests using Eq. (9). The results are shown
in Fig. 10. The measured dust amounts were in good agreement with the actual amounts added, confirming the validity
of the method.
Results of numerous tests using different ESP dusts
show that the term (HV Virgin - Dust) can be approximated as
HV Virgin/0.9. Eq. (9) thus becomes Eq. (10), which is simpler
because it does not need to deal with ESP dust to obtain the
internal recycled dust load:

(10)

LIMITATIONS
The proposed TOC method and its alternative heating value
method have their limitations. The methods can be used to
determine IRD load only at a specific time, not on a continu-

RECOVERY BOILER
ous basis. They work only for mills where the flows around
the black liquor mix tank can be clearly identified as shown
in Fig. 3.
For mills with two or more recovery boilers with one common black liquor mix tank, where a portion of the dust
stream is purged from the system or is treated via an ash treatment system and where makeup salt cake is also added to the
mix tank, the material balance becomes more complicated.
The equations developed in this work may not be applicable
without taking all important side streams into account.
As with other methods, the accuracy of the proposed
methods depends strongly on the liquor sampling timing and
location, and how well the precipitator dust has been mixed
with black liquor in the mix tank. The TOC method is expected to be more accurate than the heating value method due to
the simplicity of Eq. (8) compared to Eq. (9), which requires
an additional term Dust, or to Eq. (10), which requires an empirical correction factor of 0.9. Also, Dust is related to the heat
capacity of the dust and thus may change slightly with dust
composition, and the liquor heating value may change with
the state of oxidation of the liquor. Oxidized black liquor, for
example, has a heating value 3%5% lower than that of the
liquor before it is oxidized.
The choice of method is more likely dictated by the availability of a TOC analyzer and/or a bomb calorimeter at individual mills. While TOC analyzers are more expensive to
equip than bomb calorimeters, they are more available in
many mills as they are also used for monitoring TOCs in various wastewater treatment streams and mill effluents [10] and
for determining pulp yields [11].

CONCLUSIONS
A novel method has been developed that can be used to rapidly and accurately determine the amount of IRD in recovery
boilers. The method is based on the change in either TOC
content or the heating value of the black liquor after it has
been well mixed with precipitator dust.
Tests performed using the TOC method on recovery boilers at three different kraft pulp mills show that IRD varied
widely, from 4.0 wt% to 15.2 wt% of the total as-fired BLDS
fired in the boiler. While the specific virgin black liquor solids produced at these three mills was about the same, 1.41
tons/a.d. ton of pulp, the specific as-fired black liquor solids
burned in recovery boilers were different1.50, 1.55, and
1.59 tons dry solids/a.d. ton of pulp, respectively. The difference was directly related to the amount of IRD in the boilers.
Tests performed using the heating value method in the
laboratory show that the method is also valid, but is not as
accurate as the TOC method.
Now that the TOC method has been in place, our plan is
to use it to investigate how IRD load may be affected by black
liquor properties and by boiler operating conditions, and in
turn, how it may affect the boiler thermal efficiency. TJ
ACKNOWLEDGEMENTS
The authors acknowledge numerous recovery boiler operators and laboratory technicians at three Fibria mills for their
assistance in sample collection and analysis, as well as the
partial support received from the Energy and Chemical Recovery Research Consortium at the University of Toronto.

ABOUT THE AUTHORS

We chose this topic to study because it can potentially be useful for recovery boiler operation and control.
To our knowledge, no previous research has been
done on this topic.
Because of the lack of information regarding the
amount of internal recycled dust (IRD) in recovery
boilers, it was difficult for us to know if the results we
obtained were accurate. We addressed this problem
by carrying out a series of calibrating curves for each
individual boiler. Even a complex problem can be
solved with a basic mass balance and a simple tool
available at the mill. The most surprising finding was
that while all three mills in this study employed the
same pulping process, the amount of IRD in their recovery boilers was significantly different.
Mills now have a new method for determining IRD
load in their recovery boilers; knowing this can help
improve recovery boiler firing capacity and thermal
efficiency. The next step is to use the method to investigate how the IRD load may be affected by black
liquor properties and other boiler operating
parameters.

Guimares

Tran

Cardoso

Guimares is process and biorefinery specialist at the

Fibria Technology Center, Aracruz, Brazil. Tran is a
professor in the Department of Chemical Engineering
and Applied Chemistry, University of Toronto, Toronto,
ON, Canada. Cardoso is an associate professor in
the Department of Chemical Engineering, School of
Engineering, Federal University of Minas Gerais, Belo
Horizonte, Brazil. Email Tran at [email protected].

AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

33

RECOVERY BOILER
LITERATURE CITED
1. Tran, H.N., Tappi J. 69(11): 102(1986).
2. Wessel, R.A. and Verrill, C.L., Eng. Conf., [Proc.], TAPPI PRESS,
Atlanta, GA, USA, 1996, p. 775.
3. Janka, K., Walln, J., and Backman, R., Pulp Pap. Can. 105(1):
T15(2004).
4. Tavares, A. and Tran, H.N., TAPPI J. 80(12): 117(1997).
5. Tavares, A., Tran, H.N., and Reid, T., TAPPI J. 81(9): 134(1998).
6. Tamminen, T., Enroth, J., Landkammer, J., et al., Int. Chem. Recovery
Conf., TAPPI PRESS, Atlanta, 1998, p. 1047.
7. Tamminen, T., Kiuru, J., Kiuru, R., et al., TAPPI J. 1(5): 27(2002).
8. Frederick, W.J., in Kraft Recovery Boilers (T.N. Adams, Ed.), TAPPI
PRESS, Atlanta, 1997, Chap. 3.
9. Tran, H.N. and Earl, P.F., Chloride and potassium removal processes for kraft pulp mills: a technical review, Int. Chem. Recovery
Conf., TAPPI PRESS, Atlanta, 2004.
10. Ristolainen, M. and Aln, R., TAPPI J. 81(9): 195(1998).
11. Iwase, Y., Taneda, Y., and Inoue, T., Int. Pan Pacific Conf. Proc., TAPPI
PRESS, Atlanta, 1994, p. 15.

34

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

Nalco innovation delivers value

Engineered Filler
FillerTEK technology combines several aspects of wet-end chemistry
to achieve higher ller content while maintaining runnability and paper
quality. Increased ller content replaces high cost bers, reducing total
cost of operation.
X Reducing ber use by 3-10%
X Maintaining paper strength at higher ller loading
X Reducing energy cost for drying and rening
Value delivered
Uncoated fine paper
copy paper
X increased ash content
by 5%
X maintained strength
and bulk
X estimated annual savings:
US$1,500,000

Coated fine paper

X increased ash content
by 5%
X maintained strength
X reduced drying energy
by 18%
X delivered steam savings
of 23%
X estimated savings:
US$25/ton paper

Light weight coated paper

X increased ash content
by 5%
X maintained print quality
specications
X reduced drying energy
by 5%
X estimated annual savings:
US$2,600,000

Asia Pacic HQ
65 6505 6868
Europe HQ
41 44 877 2000
Latin America HQ
55 11 5644 6500
North America HQ
1 630 305 1000

FillerTEK, Ecolab, Nalco and the logos are trademarks

of Ecolab USA Inc.
2014 Ecolab USA Inc. All Rights Reserved

13TH TAPPI

ADVANCED

COATING
FUNDAMENTALS
SYMPOSIUM

Commons Hotel - University of Minnesota

Minneapolis, MN October 7-9, 2014
PAPER AND BOARD COATING
TOPICS INCLUDE:
Modeling New Products
Rheology Barrier Coating Technologies
Poster Session & More

Learn more at www.tappi.org/14acfs

Sponsorship opportunities available.

PEER-REVIEWED

CORROSION

Could biomass-fueled boilers be operated

at higher steam temperatures?
Part 1: Laboratory evaluation
of candidate superheater alloys
DOUGLAS L. SINGBEIL, LAURIE FREDERICK, JAMES R. KEISER,
and W.B.A. (SANDY) SHARP

ABSTRACT: A laboratory-based program was designed to evaluate candidate alloys for superheaters operating
at temperatures substantially higher than currently used in practice for biomass and chemical recovery boilers.
However, the data is also applicable to superheaters operating in very corrosive conditions at lower temperatures.
Alloys are ranked according to their performance in simulated environments.
Application: This paper provides information on the corrosion resistance of alloys that could potentially be
used to make superheaters in very advanced biomass and kraft recovery boilers with superheater steam temperatures much greater than 500C. Nearly all the alloys would be effective at preventing corrosion of superheater tubes
that operate in more normal conditions for existing boilers.

iomass is increasingly looked on as a cost-effective, environmentally friendly alternative to coal and petroleumbased fuels in the generation of energy. However, relative to
coal and fuel oils, biomass-based fuels are inhomogeneous,
and they also contain significant impurities, such as chlorides,
alkali metals and heavy metals, that deposit on heat transfer
surfaces. While these deposits may reduce the overall energy
efficiency of the boiler, their principal effect is to cause rapid
corrosion of heat transfer surfaces that approach, or exceed,
the first melting point of the deposits. As a consequence, boilers burning biomass have largely been limited to operating
with superheater steam temperatures of less than 510C
(950F) and steam pressures of less than 11 MPa (1600 psi).
The bulk of existing biomass boilers in North America operate at substantially lower steam temperatures and pressures,
particularly those utilized as waste utility boilers and kraft
recovery boilers. In contrast, the most modern utility boilers
burning coal are intended to operate at 26 MPa (3800 psi)
with 615C (1140F) superheated steam temperatures, and
research programs are in-place to help design and build even
more efficient boilers by the year 2015.
While small, incremental gains in energy efficiency can
be met by improving operation of existing biomass-fired boilers, the greatest overall gain comes from designing and building biomass boilers that operate at substantially higher pressures and superheater steam temperatures (HT/HP) than is
the current practice. To gain this enhanced energy efficiency
and retain operational reliability, challenging materials problems in the areas of high temperature strength and environmental degradation must be overcome, particularly at tem-

peratures near or above the first melting point of the

superheater tube deposits. These issues were addressed in a
multi-year research program that comprised several tasks to
explore and expand the operating envelope of future biomass-fueled steam generating systems. The programs tasks
included a literature review [1]; an assessment of European
best efforts to circumvent such problems [2]; an economic
assessment of the value of operating boilers with more aggressive superheater steam temperatures and pressures [3];
and laboratory studies in simulated service environments,
and in situ deposit and corrosion probes [4,5]. Further background about the program and discussion of issues related to
corrosion of superheater tube materials can be found in detail
in these earlier references, and will not be repeated here. Results from the laboratory studies in simulated service conditions in kraft recovery and biomass-fired boilers are presented in this paper.

EXPERIMENTAL PROCEDURE

Overview
A test matrix was developed to evaluate materials in simulated superheater deposit environments based on the analysis
of deposits from the three operating boilers used in the program one recovery boiler and two biomass-fired boilers. A
total of 25 tests were completed over the course of the program, ranging in length from 100 to 1000 h (Table I). Complete experimental details can be obtained from the final
program report [6]; what follows is a brief synopsis of the
important details.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

37

CORROSION
Simulated Deposits
Hours

RB

PB1

PB2

510C

530C

625C

650C

650C

100

T1

T5

T9

T13

200

T2

T6

T10

T14

500

T3

T7

T11

T15

1000

T4

T17

T8

T12

T16

Alloy Set 2

1000

T20

T24

T21

T22

T23

Alloy Set 3

1000

T27*

T28*

T29*

Alloy Set 1

* Cover gas for these tests included 50 ppm SO2.

I. The test matrix for the laboratory program. Cover gas for all experiments contained 5% O2, 10% CO2, 20% H2O, balance N2. Makeup
of alloy sets is shown in Table II.

Test temperatures
The principle established for selecting the test temperatures
was to set a target for metal temperature of 100C higher than
the current maximum operating superheated steam temperature for the type of boiler. This was anticipated to provide a
challenging, but realistic, stretch target for alloys and new
boiler design. Since tube surface temperatures are higher than
the steam temperature in service, this implies less than 100C
increase in steam temperature.
In the case of the recovery boiler, very few boilers worldwide operate with superheater steam temperatures of 525C,
and so the maximum test temperature was chosen to be
625C. Additional tests were conducted at 510C a temperature below the first melting temperature (FMT) of the synthetic deposit, and 530 C just above the FMT of the deposit. Some European wood-fired boilers operate with 550C
superheated steam [1,2], so the maximum test temperature
for those tests was chosen to be 650C. The FMT for the barkfired power boiler deposits was greater than 700C. This is a

much higher temperature than the boilers could reasonably

be expected to operate, even during upset conditions. Consequently, no tests were carried out at the FMT of the biomass
boiler deposits.

Deposit chemistry
The synthetic salts used for the bark-fired boiler tests were
based on analyses of deposits from air-cooled probes installed
in the model biomass-fired boilers [4]. One was considered to
be very aggressive, with high levels of chloride and potassium,
while the other was representative of a much cleaner fuel. The
recovery boiler salt composition was tailored to provide a salt
with an FMT of 525C. It is the same as that used in previous
work by FPInnovations and was based on analyses of superheater deposits in several boilers, including the recovery boiler from the pulp mill used as the host for the long-term corrosion probe work [4]. Figure 1 shows the composition of the
simulated deposits as an elemental mole fraction. Batches of
the salts were made in bulk using ACS reagent grade salts

1. Graphic representation of the three salt compositions used in this work.

38

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

CORROSION
Code

Alloy
Set 1

UNS No.

S21500

S31009

N08028

N08120

N06025

EP

N06025
pre-oxidized

N07214

FP

N07214
pre-oxidized

R20033

Alloy
Set 2

Alloy
Set 3

X
X

S33228

N12160

N06059

N06690

Cover gas composition

The gas mixture used for all but three tests was 5 vol% oxygen
(O2), 10 vol% carbon dioxide (CO2), and 20 vol% water (H2O).
The balance was N2. A small amount of sulfur dioxide (SO2),
50 ppm, was added to the cover gas for the final three tests.
The flow of each gas, except for water vapor, was controlled
by a mass flow controller (MFC). The total of 200 standard
cm3 produced a linear flow rate of 5 cm/min through the
furnace tube. The accuracy of the controllers was 3-5% of
the setpoint, depending on the gas and capacity of the MFC.
The flow of distilled water was controlled by a precision
peristaltic pump, and the water was added directly into the
gas inlet end of the furnace tubes. Tests showed that the
water vaporized before reaching the tube wall. The water
flow varied as the pump tubes aged and had to be monitored
daily and adjusted on occasion. Analysis of the daily water
usage data showed the volume % H2O normally ranged between 5 vol% of the setpoint.

Alloys

II. Alloys used in each alloy set. Due to limited space for data
in tables and graphs, codes are used throughout the report to
represent the alloys.

when available. The mixtures of compounds were placed into

a porcelain milling jar and tumbled with high purity alumina
balls for 45-60 min. The resultant powder was passed through
a 100 mesh screen and the FMT checked by differential thermal analysis (DTA) before being accepted for use.

The choice of alloys was based on consultations with boiler owners, manufacturers, and alloy suppliers participating in this project as industrial partners, as well as consideration of the literature and experience with these alloys in other types of service.
Not all of these alloys are currently available as code-certified
tube material; the first premise was to choose alloys for their
anticipated corrosion resistance in the test environments. The
alloys are listed in Table II by Alloy Set, where each set is the

CODE

Product
Name

UNS
No.

Fe

Ni

Cr

Mo

Esshete
1250*

S21500

67

10

15

31OH

S31009

Bal

19.4

24.3

0.088

Sanicro
28

N08028

Bal

30.4

26.7

3.33

HR120

N08120

Bal

37.3

24.92

0.26

602CA**

N06025

9.6

62.2

25.3

HR214**

N07214

3.86

~75

16.1

<0.1

A33

R20033

32.4

30.85

33

1.49

AC66

S33228

Bal

31.8

27.09

HR160

N12160

0.63

~37

28

0.27

30.2

A59

N06059

0.6

60.19

22.9

15.5

0.1

A690

N06690

8.7

62.5

27.85

Co

0.2

0.15

0.01

Al

0.028

Mn

Si

6.3

0.2 1.2

1.57

0.41

Ti

Others
V=0.15-0.4

<0.01

0.059

Cu=0.09

0.01

Cu=1.06,
N=0.081

1.74

0.43

0.07

0.72

0.45

<0.01

0.044

Nb=0.61,
N=0.22,
Cu=0.07,
B=0.002

2.3

0.02

0.03

0.1

0.17

P=0.05,
Z=0.09,
Y=0.07

4.34

0.21

0.06

<0.001

0.01

B=0.002,
Zr=0.011,
Y=0.004

0.61

0.24

0.015

Cu=0.61,
N=0.43

0.5

0.21

0.06

Nb=0.8,
Ce=0.08

0.48

2.75

0.01

0.3

0.48

0.056

0.17

0.02

0.004

Cu=0.01

0.2

0.15

0.02

Cu=0.01

* Nominal values. ** Alumina former.

III. Nominal compositions of alloys evaluated in the test program.

AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

39

CORROSION

2. Alloy Set 1 laid out in preparation for insertion to a furnace, showing the reproducibility of the initial salt packing and surface
height. The pre-oxidized coupons in the third and last rows from the right are darker than the remaining as ground coupons. Blank
guard pots placed at the gas inlet are also visible.

alloys exposed together in a single furnace test. Nominal composition of the alloys is shown in Table III, along with the
single letter code adopted for use in all documentation. The
baseline alloy for both laboratory and in situ field tests was
S31009, although the field trials also included S347009, which is
a commonly used comparable alloy in North America [4].
Peg coupons were produced from each alloy with dimensions 25 mm long x 5 mm dia., with a 0.4 to 1.6 m Ra ground
finish. A notch was cut in one or both ends of each peg coupon to act as an indicator of coupon orientation relative to the
direction of gas flow in the furnace. The coupons were
cleaned and dried with blowing hot air. The diameter of each
peg coupon was measured with a micrometer (+2.5 m) at
three points around either end and in the middle of the cylinder. These measurements were averaged for each location and
used as the before exposure diameters during post exposure analysis. The peg coupons were cleaned and stored in
plastic containers until use.

Exposures
Peg coupons were placed in alumina crucibles and synthetic
salts packed around the circumference (Fig. 2). The pegs were
taller than the crucibles, and hence permitted exposure of each
specimen to both the gas phase within the furnace (including
corrosive vapors) and to coverage by a bed of synthetic salt.
Fourteen crucibles at a time (with at least two of each alloy) were
placed onto an alumina D support platform, and placed in the
center of a horizontal tube furnace. Coupon orientation on the
D support platform was noted and remained consistent from
test to test. More alloys were evaluated than space allowed in a
single test, so in some cases, sets of coupons were independently exposed to the same test conditions. Once the coupons were
sealed in the furnace under flowing nitrogen, the temperature
was ramped slowly to the set point. When it was reached, the
test gases and water were introduced into the furnace. At the
conclusion, the coupons were allowed to cool in the furnace
with flowing nitrogen as a cover gas.

Analysis
Coupons were photographed immediately on removal from
the furnace, and then carefully pried from the crucible with
40

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

3. Six coupons embedded in epoxy with their support pins. The

mount on the right has not been machined square yet. The mount
on the left has been dry sectioned.

the salt layer as intact as possible. Six coupons at a time were

fitted to a jig, and embedded in an epoxy metallographic
mount for sectioning and cross-sectional polishing (Fig. 3).
Each mount was sectioned at a point above and below the
level of the salts in the crucible. Sections were then dry-polished to a 3 m diamond finish and the surface of each coupon photographed using a 10X objective in an automated
Zeiss AxioVision microscope (Carl Zeiss Microscopy; Jena,
Germany) to create a mosaic image. In total, more than 350
coupons were exposed in the test program, leaving some 700
surfaces to analyze. For coupons with more than 25 m of
corrosion, a custom image analysis program was developed,
employing Otsus Method and a circle fitting function to determine the original coupon center prior to segmenting the
coupon into 24 arcs. Measurements were made for both the
maximum metal loss and total-affected-metal (TAM) in each
arc (Fig. 4). Metal loss was determined at the interface between the scale and sound metal, while the TAM for the coupon was defined as the average of the deepest visible penetration of corrosion into the interior of the coupon in each of
the 24 arcs. The accuracy of the measurement was estimated
to be 3 m.

CORROSION

4. A mosaic image of a cross-section after analysis. Inset

shows top right corner in more detail. The outer red circle is the
original circumference of the coupon. The green lines indicate
the depth measured for metal loss, and the blue lines indicate
the total affected metal measurement for each of the 24 arcs
around the circumference. The average of the 24 total affected
metal measurements is defined as the TAM of the coupon.

RESULTS AND DISCUSSION

Overview
An example of the general appearance of the coupons after
exposure is shown in Fig. 5. These coupons were exposed
to the recovery boiler salts at 625C. Yellowing at the surface
of the salts was observed in many tests, but to a far greater
extent in the recovery boiler salts. EDX analysis of yellow and
white salts confirmed that the yellow color was associated

with chromium. Sodium chromate

(Na2CrO4) is an intense yellow salt that
appears the most likely to have been
formed. On several occasions, and for
more than one alloy, it was noted that
when one of a pair of coupons within a
given test had corroded significantly
more than the other, the salts had remained white for the corroded coupon and had yellowed for
the more resistant.
The complex test environment presented many experimental challenges, both in the setup and operation, as well as
for the analyses of the coupons afterwards. Throughout the
test program, steps and checks were put in place to determine
how consistent conditions were from test to test, and to enable data to be compared and normalized across tests. This
was achieved by putting two baseline coupons of S31009
(alloy B) in the same locations of the furnace in every test,
exposing duplicate coupons of every alloy, and exposing some
alloys (S31009, N06025, S21500) in all three alloy sets.
It is clear from an examination of the data that conditions
within the furnaces were not always identical, and neither
were conditions in duplicate tests. Nonetheless, the data
proved to be reasonably consistent within, as well as between,
tests. A few tests produced highly variable data that was almost certainly caused by artifacts in the test environment. The
stochastic nature of corrosion was also demonstrated as greater variability was found in the shorter-term tests due to the
randomness of initiation events on the coupons. Results in the
longer-term tests were more uniform.
A total of fifty S31009 (alloy B) coupons were exposed over
the entire program, giving 100 sections to be analyzed (half in
the vapor phase and half under-deposit). Data was analyzed
from the 50 individual S31009 sections for which the TAM was
greater than 25 m. The standard deviation of the 24 TAM
measurements around the circumference as a percent of their

5. Appearance of coupons at the conclusion of tests in RB salts at 625C. Red circles indicate coupons of pre-oxidized HR214
(Alloy FP). Note that the only Alloy FP coupon not to show substantial spalling is the one with the yellow-shaded salt.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

41

CORROSION
Standard Deviation as % of Average TAM
Code

Alloy

Vapor Phase

Under-Deposit

Average

S33228

14%

17%

16%

S31009

18%

14%

16%

N08028

18%

20%

19%

S21500

24%

20%

22%

N06059

22%

22%

N08120

24%

24%

24%

N06690

26%

23%

25%

EP

N06025 pre-oxidized

24%

26%

25%

N06025

34%

32%

32%

N07214

38%

28%

33%

FP

N07214 pre-oxidized

42%

29%

33%

N12160

29%

38%

34%

Average Standard
Deviation %

26%

24%

IV. Standard deviation in average TAM for alloys for all the test environments. The data is separated into vapor phase and underdeposit results and sorted by the average of those results.

average was calculated for each coupon. It ranged from 7% to

62% above the salts with an average of 15%, and from 5% to
30% below the salts with an average of 18%. The overall
average of the above for S31009 is 16%. For the balance of the
alloys, the standard deviation for any given set of measurements
in duplicate tests fell between 15% and 25% of the average TAM
for the alloy, consistent with the S31009 result. Only three
materials fell outside of this range those with the most severe
corrosion and those with the least corrosion (Table IV).

Power boiler 1 (PB1) tests

The PB1 salt was representative of deposits produced by clean,
uncontaminated fuel. There was minimal corrosion on all

coupons exposed to this salt, even for tests that lasted 1000
h. The average TAM was far less than 25 m for all coupons.

Power boiler 2 (PB2) tests

The PB2 salt was representative of deposits produced by fuel
that is highly contaminated with chloride. Even so, the depths
of corrosive attack on all coupons were much less than 25 m
after 100 h. Only coupons from test exposures 200 h and longer were subjected to the full standard analysis. Corrosion as
a function of time is shown in Fig. 6, while results for the
1000-h tests are shown in Fig. 7, plotted in order of increasing nickel content in the alloys. The most extensive corrosion
in these tests occurred underneath the layer of salt; there was

6. Variation of average TAM with time in PB2 salts at 650C (T14, T15, T16), in m.
42

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

CORROSION

7. Test data from alloys exposed to PB2 salts for 1000 h at 650C. Alloys are listed in order of increasing nickel content left to right,
with the last four on the right being the alumina formers (E, EP, F, FP). Metal loss is blue portion of bar, TAM is green portion of bar.

8. Test data from alloys exposed to RB salts for 1000 h at 625C. Metal loss is blue portion of bar, TAM is red portion of bar. The
alloys are listed in order of increasing nickel content, left to right, with the alumina formers at the right (E, EP, F, FP).

much less corrosion above the salt layer. There is also a clear
correlation between increased resistance to corrosion and
greater nickel content of the alloy. Conversely, there was no
correlation when the same data was plotted as a function of
chromium content. The results from the test conducted for
500 h appear anomalous and a review of the data showed
higher than average variation in TAM on several coupons. No
reason for this variation was discovered.

Recovery boiler tests

Negligible corrosion was observed on the alloys in tests conducted at 510C; the greatest TAM measured was only 17 m
for N06025 (alloy E). This temperature is below the FMT of
the salt deposits. More substantial corrosion was observed
at both higher temperatures, and two alloys completely disintegrated in the 1000 h test at 625C. These were N07214
(alloy F) and N06059 (alloy I). Data for the 1000 h tests at

625C is shown in Fig. 8. The correlation between corrosion

and nickel content is not apparent in these tests; nor does the
corrosion correlate with chromium content of the alloys. Of
significance in these tests was the observation that corrosion
in the gas phase was nearly as severe as underneath the salt
deposits. Especially in the vapor phase, internal penetration
accounted for a larger proportion of the TAM. A measureable
increase in corrosion was noted for many of the alloys when
SO2 was added to the cover gas (Fig. 9 and Fig. 10).

Alloy performance and ranking

Evaluation of alloy performance was based on the results of
1000 h tests, and compared the average TAM from all coupons
of the same alloy in each environment, as well as visual and
microscopic observation of the coupons. An arbitrary scale
was used to rank the alloys. An average TAM of less than
25 m was considered acceptable, between 25 and 50 m was
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

43

CORROSION

9. Corrosion results for Alloy Set 3 after exposure to recovery boiler salts at (a) 530C and (b) 625C for 1000 h in the standard
atmosphere plus 50 ppm SO2.

intermediate, and greater than 50 m was considered unacceptable performance. For reference, linear corrosion at
25 m/1000 h would extrapolate to 2.2 mm over 10 years and
50 m/1000 h would represent a loss of nearly 4.4 mm over a
10 year life. The ranking of alloys was different for each environment, and also varied by temperature in the same environment.
In the PB1 environment at 650C, all alloys performed well
enough to fit into the acceptable category.
In the PB2 environment at 650 C, only three alloys fell into
the acceptable performance range above the salt deposits
(Fig. 11). A number of other alloys performed moderately
well in the vapor space above the salts, but did much more
poorly underneath the deposits. The best three alloys overall
in the PB2 salts were N06025 (alloy E), N06059 (alloy J), and
the pre-oxidized N07214 (alloy FP). As already mentioned,
there was a general correlation between increasing nickel
content in the alloy and better performance in this environment. This correlation was less apparent in the vapor space
44

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

than beneath the salts. Two of the alloys that performed moderately well in the vapor space were R20033 (alloy G) and
S33228 (alloy H). These two alloys are among the group of
four alloys with the lowest nickel content, and did very poorly beneath the salts. The other low nickel alloys were S31009
(alloy B) and S21500 (alloy A), both of which fared quite poorly above and below the salt deposit in these tests.
In the recovery boiler environment at 510C, all alloys provided acceptable corrosion resistance. At 530C, the only alloy
to provide acceptable service both above and below the salt
level was N06059 (alloy J) (Fig. 12). A few other alloys
notably N06025 (alloy E & EP) and N07214 (alloy F) were
acceptable in the vapor space, but not beneath the salts.
N06690 (alloy K) and N12160 (alloy I) fell into the intermediate category above the salts, and N06025 (alloy E) was the only
alloy to fit into the intermediate category beneath the salts.
The worst alloys in this service were S31009 (alloy B), S21500
(alloy A) and S33228 (alloy H).
As expected, the recovery boiler environment at 625C was

CORROSION

10. Comparison of average TAM for the alloys that were in recovery boiler 1000 h tests, without and with the addition of 50 ppm SO2.
(a) 530C; (b) 625C.
Section
Above
salts
Below
salts

PB2 650C
E

EP

FP

18

18

23

31

31

35

36

41

44

47

59

65

71

EP

FP

10

12

20

38

63

89

151

154

271

283

303

316

385

TAM (m)/1000hr
10

< 25

30

25 < 50

271

50

11. The average TAM in microns for each alloy exposed to the PB2 environment at 650C. Alloy performance is color coded as
acceptable (green), intermediate (yellow), and unacceptable (red).

extremely aggressive and no alloy fit into either the acceptable

or intermediate categories (Fig. 13). Corrosion rates above
and below the salt level were similar. N06025 (alloy E) was the
best of the alloys, along with S21500 (alloy A) and N06690
(alloy K). Two alloys failed catastrophically in the recovery
boiler environment at 625C N06059 (alloy J) and N07214
(alloy F). The failure of N06059 (alloy J) was unexpected, given
its extremely good performance in the same environment at
the lower temperature. This alloy stands out from the others

due to the very high molybdenum content, and it is possible

that this is a contributing factor to the failure at the higher temperature. A boundary between hexagonal MoS2 and monoclinic Mo2S3 exists at 664 50C [7], but whether such a
change could contribute to a dramatic increase in corrosion is
unknown. N07214 (alloy F), in contrast, contains little molybdenum, but contains the second least amount of chromium
(next to S21500 [alloy A]). Chromium is well known to enhance corrosion resistance in kraft recovery environments, so
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

45

CORROSION
Section
Above
salts
Below
salts

RB 530C

RB 530C +SO2

EP

FP

12

13

15

15

25

42

44

54

62

127

133

140

272

47

47

83

154

168

198

260

TAM (m)/1000hr

EP

FP

14

27

39

59

61

81

87

99

115

159

169

185

194

41

91

118

151

160

199

301

10

< 25

30

25 < 50

271

50

12. The average TAM in microns for each alloy exposed to the recovery boiler environment at 530C. Alloy performance is color
coded as acceptable (green), intermediate (yellow), and unacceptable (red).

Section
Above
salts
Below
salts

RB 625C

RB 625C +SO2

EP

55

76

137

201

209

210

211

242

282

EP

65

92

133

134

211

217

219

230

278

FP

355 2000 2000 2000

G

FP

347 2000 2000 2000

12

144

165

182

195

201

295

46

103

179

184

215

241

255

TAM (m)/1000hr
10

< 25

30

25 < 50

271

50

13. The average TAM in microns for each alloy exposed to the recovery boiler environment at 625C. Alloy performance is color
coded as acceptable (green), intermediate (yellow), and unacceptable (red).

the failure of N07214 (F) was not unexpected. In contrast, the

relatively good performance of the S21500 (alloy A) coupons
in this environment was surprising, given the low chromium
content of this alloy. While it might be argued that the manganese content of this alloy imparted some level of protection,
there remains the possibility that it too might have failed catastrophically had the test been prolonged.
The ranking of alloys in the SO2-containing cover gas were
similar to the equivalent results in tests without SO2. Only 7
of the alloys were exposed in this environment, and they were
chosen as the best of the previous tests. 602CA (alloy E)
displayed by far the best corrosion resistance in this test.
There appeared to be little difference in corrosion above or
below the salt level. S21500 (alloy A) and S31009 (alloy B) were
the worst of the limited set of alloys exposed to the SO2containing environment. Given that the standard deviation of
these experiments varies between 15%-25%, the performance
of the balance of the other alloys could be described as similar.

Comparison to field data

Laboratory experiments are typically well-controlled and provide more reproducible results than field exposures, where
temperature, deposit flux and composition, and other critical
parameters vary dramatically as a function of time. The intent
of laboratory experiments is to provide comparable data in
less time, with useful results. Often, they also fail to replicate
critical aspects of the corrosive environment, and consequently under- or over-estimate the real rate of corrosion in-service.
One unique aspect of this particular program is the strong
linkage between field and laboratory tests [4,5]. The salt de46

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

posits chosen for the laboratory tests were meant to mimic

the real deposits collected from the one recovery boiler and
two power boilers involved in the program. The equivalent
field tests, in this case, typically ran for about 2000 h. As anticipated, exposure of the in situ coupons spanned a wide
range of temperatures along the length of the probe, as well
as very cyclic temperatures at each coupon location [4].
Overall, the general trends between field and laboratory
trials matched well in the case of the in situ recovery boiler
trials. N07214 (alloy F), N08028 (alloy C), and S31009 (alloy
B) all performed poorly in the field. These alloys were also
among the worst performers in the laboratory trials, particularly the N07214 (alloy F) at higher temperatures. The alloys
that performed best in the in situ probes were N12160 (alloy
I) and S21500 (alloy A), both of which also fared reasonably
well in the laboratory tests. Although N06025 (alloy E) was
clearly the best alloy in the laboratory tests, it did not fare as
well at the higher temperatures in the field probe.
The correlations between laboratory and field results were
less clear for the power boilers, because of the temperature
differences [4]. The laboratory trials were conducted at significantly higher temperatures than could be reached in the
locations where the field probes were installed, and consequently, there is no directly comparable data. In the case of
the PB1 tests (Port Mellon boiler in the field trials), the maximum exposure temperature did not exceed 480C in the field
probe. Overall, the conditions in this boiler were much less
corrosive than was the case in the PB2 boiler (Crofton). This
matches the laboratory experience. However, an interesting
note is that the most rapid corrosion in the field trial in the

CORROSION
Port Mellon boiler occurred at temperatures between 400C
and 450C. This was attributed to the presence of Zn and Pb
in the deposits, which form low-melting eutectic compounds
at those temperatures. This is a good example of not replicating the critical corrosive environment in a laboratory test.
These elements were not included in the synthetic salts due
to their low concentrations in mill deposit samples (<0.2
wt%), nor was the test temperature in the appropriate range
for this corrosion mechanism.
For the PB2 boiler, the maximum exposure temperature in
the field was about 550C. The best performing alloys in the field
were N08120 (alloy D) and S21500 (alloy A), neither of which
did particularly well at the higher temperature laboratory tests.
N06025 (alloy E) was also not a good performing alloy in the
field trials, despite doing well in the laboratory tests.
Across all three boilers, corrosion in the field trials was
roughly of the same order of magnitude, but somewhat higher than those measured in the laboratory experiments. This
is a good correlation, given the longer exposure times in the
field trials. Cyclic temperatures and heat flux are also known
to be strong accelerators for corrosion in high temperature
conditions, and neither of these was present in the laboratory
experiments. The difference in corrosion behavior for N06025
(alloy E) between the laboratory and field trials was surprising
and merits further investigation, given the good performance
of this alloy in the laboratory.

Implications for existing boilers

While the objective of the research program was to explore
materials options for advanced biomass-fired boilers operating
under extreme process conditions, some findings and observations from the laboratory test program can be applied directly to existing boilers. In general, the results from this test
program compare reasonably with other studies of corrosion
of superheater alloys in wood-fired and chemical recovery
boilers [8-10]. Although not all of the alloys used in this project are currently available as code-certified materials for use
as boiler tubes, the results suggest some might serve well
under more realistic operating conditions (surface temperatures < 525C) in kraft recovery boilers. However, caution is
required in extrapolating these test results to long-term performance in a boiler. A corrosion rate of 0.22 mm/year or less
was arbitrarily chosen as acceptable for the purposes of these
tests; such a rate of corrosion might still be too high for practical application in a boiler. One consideration touched upon
in another part of this program is whether the financial gain
from producing more steam (and hence more electricity)
could offset the cost of more frequent repairs/replacement of
superheater elements due to increased corrosion from hotter,
more challenging, environmental conditions [3].
Measured corrosion rates in the laboratory tests conducted
below the FMT of the deposits were low, and the corresponding in situ probe results show relatively low corrosion rates,
particularly for N06025 and S21500 [3,4]. Overall, N06025 did
relatively well in laboratory tests. The laboratory data was not

so positive for S21500, but for its alloy content, it did very well
compared to alloys with much greater nickel and chromium
content. The baseline alloys used for these studies (S31009
and S34709) are occasionally selected for use in higher temperature steam loops in recovery boiler superheaters [1], so
alloys in this program that performed the same as or better
than S31009 should do equally well in similar service.
On the hog fuel-fired boiler side, corrosion rates were significantly greater for all alloys in the salts representative of a
highly chloride-contaminated fuel. In the more benign fuel,
corrosion rates of all of the tested alloys were quite low. These
results were mirrored by the in situ probes [3]. One interesting observation is that corrosion in the power boiler environments was driven strongly by direct contact with the deposits.
This information could be valuable for mills that employ riskbased inspection protocols.
SUMMARY AND CONCLUSIONS
A laboratory test was developed that allows for assessment and comparison of corrosion that occurs above and
below deposits on the same coupon in simulated fireside
superheater environments.
In these tests, the rate of corrosion in the gas phase of
simulated power boiler environments was substantially
lower than for corrosion that occurred underneath deposits. This observation may have strong practical implications for companies that practice risk-based inspection
in biomass-fired boilers.
Under the test conditions employed, the vapor phase in
the simulated kraft recovery boiler environment was
found to be as aggressive to the alloys as the environment
underneath salt deposits.
I n terms of best corrosion resistance, Alloy N06025 was
ranked either first or second out of all the alloys evaluated in all laboratory test environments. The order of
ranking of other alloys depended on the test environment, whether the corrosion was measured above or
underneath the synthetic salt, and temperature.
In an environment simulating a chloride-contaminated
biomass-fired boiler, the alloy ranking at 650C from
most corrosion resistant to least is listed left to right
below. (Roughly equivalent performance alloys are
ranked together):
R20033
N07214
(pre-oxidized)
N06025
Above
salts:

N06025
(pre-oxidized)
N06059

N12160
<

S33228
N07214

N08120
<

S21500
S31009

N06690
N08028

AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

47

CORROSION
N08120

most corrosion resistant to least was (left to right, roughly equivalent performance alloys ranked together):

N12160

Below
salts:

N06025

N06059

S21500

N06025
(pre-oxidized)

N0714

N08028

<

N06059
(pre-oxidized)

<

N06690

S31009
S33228

N12160

Above
salts:
(no SO2)

N06025

N06690

N06059

N06025
(pre-oxidized)

N08120

N07214

S21500

<

S33228

<

N07214

(pre-oxidized)

S31009

R20033

N08028
R20033

In an environment simulating a kraft recovery boiler, the

alloy ranking at 530C from most corrosion resistant to
least is listed left to right below (roughly equivalent performance alloys are ranked together):

Above
salts
(no SO2):

N06059
N07214

<

N12160

N08028

N07214
(pre-oxidized)

S33228

N06690

N06025

N08120

N06025
(pre-oxidized)

R20033

<

N06690
N06025

<

N08120

N12160

N06025
(pre-oxidized)

N06690

N06059
(pre-oxidized)

<

S33228

S08120
N06025
(pre-oxidized)

S33228

Below
salts:
(no SO2)

S21500

N06025

N12160
<

S33228

S21500

N08028

N06025

S21500

S31009
S33228

R20033
N08120
N12160
Below
salts
(with SO2):

N06025

<

N06690
N08120

<

S21500
S33228
S31009

In an environment simulating a kraft recovery boiler

with 625C superheater tube temperature, all alloys
were ranked as providing unacceptable performance.
Based on these results, it is unlikely that a recovery boiler could be designed to operate at such a high temperature in the superheater. The alloy ranking at 625C from
48

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

N07214

(pre-oxidized)

R20033

S31009

N08028
<

N07214

N08120

N07214

N06690

<

(pre-oxidized)

N06059
<

N06059

N07214

N12160

N06025
(pre-oxidized)

S31009

N06690

S31009

Below
salts
(no SO2):

<

S31009

S31009

<

N08120

S21500

S21500

N12160
Above
salts
(with SO2):

Above
salts
(with SO2):

N06025

Below
salts
(with SO2):

N06025

<

N06690

<

S33228
N12160

S21500

ACKNOWLEDGMENTS
Jan van Heiningen, Greig Hull, Aleksey Bykov, Peggy Hung,
Shannon Hoekstra, and Jonas Ritter are acknowledged for
their contributions to specimen preparation, metallography,
and analysis. Material for the corrosion samples was donated
by Haynes International and Rolled Alloys. Funding was provided in support of this research by the member companies
of FPInnovations, with support from the U.S. Department of
Energy, Office of Energy Efficiency and Renewable Energy,
Advanced Manufacturing Office under contract DE-AC0500OR22725 with UT-Battelle, LLC, administered by James R.
Keiser at the Oak Ridge National Laboratory and supported
by in-kind contributions from the following project partners:

CORROSION
bo Akademi University, Andritz Oy, Babcock & Wilcox,
Catalyst Paper, Chalmers University of Technology, Domtar
Corporation, FM Global, FPInnovations, Foster Wheeler,
Georgia Institute of Technology, Haynes International, Howe
Sound Pulp & Paper, International Paper, MeadWestvaco,
Metso Power, Outokumpu Stainless, Rolled Alloys, Sandvik
Materials Technology, SharpConsultant, Southern Company,
Special Metals, ThyssenKrupp VDM, University of Toronto
Pulp and Paper Center, Vattenfall Power, and the Weyerhaeuser Company. TJ
LITERATURE CITED
1. Sharp, W.B.A., Singbeil, D.L., and Keiser, J.R., Superheater
corrosion produced by biomass fuels, CORROSION/2012 Conf.,
NACE International, Houston, 2012, Product No. 51312-01308-SG.

6. Singbeil, D.L. and Frederick, L., Improving heat recovery in

biomass-fired boilers Corrosion of superheater tube materials for
high temperature black liquor and bark-fired boilers, Final Report
ORNL Subcontract 4000089243, December 2012.
7. Mo-S System, Bulletin of Alloy Phase Diagrams, Vol. 1, #2,
American Society of Metals, Metals Park, OH, USA, 1980.
8. Tuiremo, J. and Salmenoja, K., Control of superheater corrosion in
black liquor recovery boilers, Proc. - Int. Symp. Corros. Pulp Pap. Ind.,
10th, VTT, Espoo, Finland, 2001, Vol. 1, p. 181.
9. Mkip, M., Kauppinen, E., Lind, T., et al., Superheater tube
corrosion in recovery boilers, Proc. - Int. Symp. Corros. Pulp Pap. Ind.,
10th, VTT, Espoo, Finland, 2001, Vol. 1, p. 157.
10. Salmenoja, K. and Mkela, K., Prevention of superheater corrosion in the combustion of biofuels, CORROSION/2000 Conf., NACE
International, Houston, 2000, Paper No. 00238.

2. Sharp, W.B.A, Singbeil, D.L, and Keiser, J.R., Energy from biomass Lessons from European boilers, TAPPI PEERS Conf., TAPPI
PRESS, Atlanta, GA, USA, 2011.
3. Sharp, W.B.A., Frederick, W.J., Keiser, J.R., et al., TAPPI J. 13(8):
65(2014).
4. Keiser, J.R., Sharp, W.B.A., and Singbeil, D.L., TAPPI J. 13(8):
51(2014).
5. Keiser, J.R., Sharp, W.B.A., Singbeil, D.L., et al., TAPPI J. 12(7):
45(2013).

ABOUT THE AUTHORS

Pulp and paper mills have the potential to contribute significant amounts
of electricity to North American
power grids. However, corrosion of
superheater tubes in our biomass
and chemical recovery boilers limits
their efficiency relative to coal-fired
boilers. We wanted to find alloys that
would perform well under extreme
conditions at superheater tube sur- Singbeil
face temperatures typical of coalfired boilers. Identifying better materials for these tubes would increase the amount of
steam available for conversion into electrical energy.
Our previous research has focused primarily on
corrosion of water wall tubes in recovery boilers. A
unique aspect of this work was the collaboration and
breadth of the research program, including the supporting field trials conducted by Oak Ridge National
Laboratory and economic evaluations by
SharpConsultants reported in other papers of the
same series [3,4].
Characterizing high temperature corrosion of small
alloy samples in a meaningful way is extremely difficult in a laboratory setting. We used a statistical approach to measure damage and looked at effects of
corrosion in the gas phase, under solid deposits, and
in the presence of a melt all in the same experiment.
We found a few alloys that proved surprisingly resis-

Frederick

Keiser

Sharp

tant to corrosion in these environments, given their

alloy content.
The test conditions used in these experiments
were extreme for most existing boilers, but we found
a few relatively cheap alloys that performed adequately and that would likely do very well in less extreme conditions. The next step in this research is another round of testing with less expensive alloys to
identify best replacement superheater alloys for existing boilers that are looking for smaller gains in energy
efficiency.
Singbeil is principal scientist and Frederick is
principal technologist with FPInnovations, Vancouver,
BC, Canada. Keiser is distinguished research and
development staff member, Materials Science and
Technology Div., at Oak Ridge National Laboratory,
Oak Ridge, TN, USA. Sharp is principal at
SharpConsultant in Columbia, MD, USA. Email
Singbeil at [email protected].
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

49

PEER-REVIEWED

CORROSION

Could biomass-fueled boilers be

operated at higher steam temperatures?
Part 2: Field tests of candidate
superheater alloys
JAMES R. KEISER, W.B.A. (SANDY) SHARP, AND DOUGLAS L. SINGBEIL

ABSTRACT: Operating superheater tubes of biomass-fired boilers at considerably higher temperatures than can
be tolerated by commonly used structural materials could improve boiler efficiency. However, corrosion of the
superheater tubes promoted by interaction with the relatively low melting point deposits that accumulate on the
tubes becomes a major issue. The objective of this study was to use field exposures to determine if there are materials acceptable for use as superheater tubes that can operate at temperatures at least 100 Celsius degrees above the
current maximum superheater temperature.
Corrosion probes containing multiple specimens of nine different alloys were exposed for at least 2000 h in the
superheater area of three biomass boilers where the deposits were determined to be enriched in potassium or
chlorine. Similar specimens were also exposed in a boiler co-firing coal and wood. For the probes, specimen
temperatures ranged from a low of less than 400C to temperatures above 600C for all but one case. Following
exposure, a section was taken from each specimen and examined using light microscopy and scanning electron
microscopy. Results of the examination of these specimens showed some alloys performed considerably better than
others, but the corrosion resistance could not be related to chromium or molybdenum content of the alloys.
Application: The results of this study provide guidance on the selection of superheater tube alloys for applications where it is desired to operate in hostile environment at temperatures as much as 100 Celsius degrees higher
than normally used in that environment.

t is generally recognized that more efficient utilization of

biomass or biomass-derived fuels would provide a means to
reduce consumption of imported petroleum and to decrease
the production of greenhouse gases, because the carbon dioxide produced during burning of biomass is roughly offset
in a fairly short time by the carbon dioxide consumed during
regrowth of biomass. However, the relatively low melting
point alkali metal salts and chlorine compounds released during burning of biomass can cause accelerated corrosion and
fouling of superheater tubes at fairly moderate temperatures.
Consequently, increased superheater tube corrosion may be
a result of efforts to increase boiler efficiency by increasing
the maximum steam temperature. The detrimental effects of
the alkali metals and chlorine have been well documented, as
have the efforts to identify a means to limit the severity of tube
degradation [1-10]. Extensive reference lists that provide additional sources of information on biomass-related corrosion
and corrosion prevention are included in the latter two of
these documents.
In a project competitively selected and subsequently funded by the United States Department of Energy, studies have
been conducted with four goals: 1) characterizing the super-

heater system designs and construction materials used outside

North America; 2) determining the economic benefits or penalties of operating boilers at a significantly higher temperature
(50-100 Celsius degrees) than currently used; 3) conducting
laboratory corrosion studies with candidate tube materials
using salt mixtures simulating those in selected boilers; and
4) conducting field studies with corrosion probes to measure
the corrosion rates of alternate superheater tube alloys in four
particularly aggressive biomass-fired boiler environments at
temperatures as much as 100 Celsius degrees above the normal maximum superheater tube temperature. Results from
studies conducted for the first three of these tasks have been
or will be reported elsewhere [11-14].
Before exposing the corrosion probes, in order to get an
indication of the composition of the superheater deposits, aircooled sampling probes were exposed in three of the four
environments. These sampling probes were exposed for a
relatively short time generally 7-10 days and care was taken
in recovering the deposits from the top and bottom surfaces
of the probes. Results obtained from analysis of these deposits helped determine the test compositions for the laboratory
studies.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

51

CORROSION
In order to assess the performance of the selected alloys
under the most severe environments that might be encountered during combustion of biomass, four boilers were selected as examples of what would be expected to be particularly
hostile biomass boiler environments. These environments
include a recovery boiler in Covington, VA, USA, that processes black liquor derived primarily from hardwood, thus yielding deposits particularly high in potassium content. A hogged
fuel boiler in Crofton, BC, Canada, that primarily burns bark
obtained from seawater-floated logs was expected to have
deposits that are high in chlorine content. Another hogged
fuel boiler in Port Mellon, BC, Canada, burns fuel consisting
primarily of wood from beetle-killed pine trees, but also uses
bark from seawater-floated logs and construction and demolition (C&D) waste, which together could result in superheater
deposits containing some heavy metal as well as some chlorine. The fourth environment was a power boiler in Gadsden,
AL, USA, that normally burns coal but supplemented the coal
with wood for the period of these studies. This resulted in an
environment containing sulfur from the coal, as well as the
alkali metal salts from the wood. This study summarizes the
results from the four corrosion probes that were exposed in
the boiler superheater areas of particularly hostile, but significantly different, boiler environments.
EXPERIMENTAL PROCEDURE

Deposit sampling
Deposit sampling probes were exposed in the recovery boiler
and the two hogged fuel boilers. The sampling probes in the
two hogged fuel boilers were prepared, exposed, and analyzed by the FPInnovations collaborators. The sampling probe
exposed in the Covington boiler was air cooled and had ten
thermocouples spaced every 15 cm (6 in.) along the length of
the probe. These thermocouples alternated between the top

and bottom of the probe, so temperatures measured on the

bottom of the probe reflected the effects of radiant heating
while temperatures measured on the top are more representative of the gas temperature. At the end of the approximately
one week exposure period, the deposit sampling probe was
removed from the boiler, allowed to cool, and then wrapped
with plastic wrap to keep the deposits in place on the surface
of the probe. Once the probe was safely returned to Oak Ridge
National Laboratory (ORNL) in Oak Ridge, TN, USA, samples
of the deposits were removed from both the top and bottom
of the probe for analysis.

Corrosion probe design

Each air-cooled corrosion probe was designed to contain 30
samples each at 5.08 cm (2 in.) long fabricated from 1.5 in.
schedule 40 pipe or the equivalent. The samples were fabricated so that they were interlocking and each had a slot machined in the surface to accommodate a 0.0625 in. dia.
sheathed thermocouple. Temperatures were measured by
twenty-four thermocouples so that four of every five samples
had a thermocouple attached. These thermocouples were fed
through the inner diameter (ID) of the corrosion probe. The
corrosion sample design, which is shown in Fig. 1, provides
for about 1 cm of the thermocouple to be in contact with the
outer surface of the sample. This design was intended to minimize the cooling effect on the thermocouples of the air flowing through the probe, thus making the measured temperature
representative of the temperature of the outer surface of the
samples. To assure that each thermocouple maintained full
contact with the sample surface, a piece of 0.78 mm (0.031 in.)
stainless steel shim stock was welded to the sample so that the
shim stock held the thermocouple tightly against the sample.
Pictures of the thermocouple arrangement without and with
the shim stock are also included in Fig. 1. Besides showing the

1. Schematic of the corrosion samples (left); photo of samples tack welded with thermocouples protruding from the surface of the
samples (center); and photo of samples welded together and with shim stock welded in place to hold thermocouples on the surface
of the samples (right).
52

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

CORROSION
Product Name

UNS No.

Fe

Ni

Cr

Mo

310H SS

S31009

Bal

19.10

24.30

0.25

Haynes 214

N07214

3.56

Bal

16.08

Sanicro 28*

N08028

35

31

27

602CA

N06025

9.60

62.20

25.30

HR160

N12160

0.63

Bal

28.00

Inconel 690*

N06690

62

29

HR120

N08120

36.26

36.68

24.74

0.07

Esshete 1250*

S21500

72

10

15

347H SS

S34709

Bal

9.02

17.21

Co

Mn
1.17

0.03
3.5

0.21

4.22

1.8
0.02

0.27

Al

30.20

2.30

0.48

Si

Ti

0.55

0.045

0.12

0.04

0.2

0.01

0.03

0.10

0.170

2.75

0.48

0.056

Others

Y=0.004,
Zr=0.011

Y=0.07,
Zr=0.09

0.02
0.11

0.68

0.16

0.73

0.057

N=0.21

6.3

0.5

0.1

Nb=1, V=0.25,
B=0.006

1.31

0.57

0.048

* Nominal values.

I. Composition of alloys selected for exposure in the corrosion probes.

shim stock that was welded in place over each thermocouple,

this figure also shows the tack welds that were used to position
and align the samples before application of the circumferential
welds that rigidly connect adjacent samples. To provide additional support for the probe, which extended at least 2.1 m (7
ft) from the boiler wall, a 0.5 in. dia. stainless rod ran the length
of the probe inside the corrosion samples (along with the thermocouples). A stiff spring was placed between the rod and the
outside base of the corrosion probe in order to allow for thermal expansion that was expected to differ between the array
of welded samples and the stainless steel support rod that was
cooled by the air flowing through the probe.
For the laboratory corrosion study, eleven alloys were exposed to synthetic environments similar to those encountered
in the biomass-fired boilers. The three co-investigators and the
project participants selected eight of these alloys plus 347H
stainless steel for exposure in the corrosion probe. These alloys and their compositions are listed in Table I. Nominal
compositions are given for three of the alloys, while the other
compositions are from actual heat analyses. The selected alloys
included two alumina-forming alloys, as well as seven more
traditional chromia-forming alloys. Alloy S31009 was chosen
because it is currently used in the hottest areas of the Covington recovery boilers superheater. Two other alloys (S21500 and
S34709) were chosen because of good experience from similar
superheater studies conducted in Europe. The two aluminaforming alloys were preoxidized at 1000C to promote formation of the alumina layer, while the chromia-forming alloys
were not given any pretreatment. Prior to assembly into the
corrosion probe, the wall thickness of each sample was measured in several locations around the circumference.

The thirty samples on each corrosion probe were arranged

in a repeating pattern so that samples of all but two of the materials (N06690 and N12160) were exposed to a range of temperatures. Four samples of each of the other seven alloys were
positioned so that one sample of each material was near the
hotter end of the probe, one sample was near the cooler end,
and the other two samples of each material experienced intermediate temperatures. In addition to the effect resulting
from a samples position on the corrosion probe, the temperature of a sample varied from top to bottom in that one
side was heated by the incident gas, and in some cases radiant
heat from the lower boiler; so, in one case, it was observed
that one side was as much as 80 Celsius degrees hotter than
the other side. The arrangement of samples on the corrosion
probe and the identification of thermocouple locations are
shown in Table II, and Fig. 2 shows a typical corrosion
probe with alloy locations identified. Instrumentation for data
collection and process control along with a computer were
provided with the corrosion probe. A variable flow valve controlled air flow to the corrosion probe, and the setting of the
variable flow valve was determined by computer software
that adjusted the flow to maintain the temperature of a particular thermocouple in a selected range. Generally, a thermocouple experiencing one of the highest temperatures was
selected for this control function, and this was usually a thermocouple on the bottom side near the hot end of the probe.
In addition to enabling the remotely-located operator to select
the reference thermocouple, the computers display panel also
permitted the operator to select the control range for this
thermocouple and the data collection rate.
Although the computer software permitted the operator to
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

53

CORROSION
Position

UNS No./Alloy

S31009/310H SS

N07214/Haynes 214

N08028/Sanicro 28

N06025/602CA

N12160/HR160

N06690/Inconel 690

N08120/HR120

S21500/Esshete 1250

S34709/347H SS

10

S31009/310H SS

11

N07214/Haynes 214

12

N08028/Sanicro 28

13

N06025/602CA

14

N08120/HR120

15

S21500/Esshete 1250

16

S34709/347H SS

17

S31009/310H SS

18

N07214/Haynes 214

19

N08028/Sanicro 28

20

N06025/602CA

21

N08120/HR120

22

S21500/Esshete 1250

23

S34709/347H SS

24

S31009/310H SS

25

N07214/Haynes 214

26

N08028/Sanicro 28

27

N06025/602CA

28

N08120/HR120

29

S21500/Esshete 1250

30

S31009/310H SS

Thermocouple No.
(on top of sample)

Thermocouple No.
(on bottom of sample)

24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1

II. Arrangement of samples on the corrosion probe and location of thermocouples.

select the data collection rate, all four computers were set to record the temperature of each thermocouple once a minute during their entire operating periods. On a daily basis, the temperatures and valve setting data were transferred via a telephone
connection from the on-site computer to a computer at ORNL.
Plots of each days temperature data were made and evaluated
to be certain there were no operational issues. About once a
week, the temperature plots were sent to the boiler operators.
At the completion of the exposure, the probe was removed
from the boiler with as little loss of the deposits as was possible. Once the probe had cooled, the portion of the probe
with the thirty samples was wrapped tightly with plastic to
retain as much of the deposits as possible. When the probe
was received at the laboratory, the orientation was carefully
noted, then samples of the deposits were collected and each
metal sample was cut off the probe. An example of the thirty
54

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

samples is shown in Fig. 3. For the metallographic examination, a complete ring was cut from the approximate center of
each sample and its orientation during exposure was noted.
So that the exposed cross-section would be representative of
the true thickness of the sample, every effort was made to
make the cuts perpendicular to the axis of each sample. Any
error introduced if cuts were not perpendicular would result
in the remaining thickness being overestimated; thus, any
error would be on the conservative side. Using the micrographs from the metallographic examinations, the thickness
of remaining unaffected metal was determined for the top and
bottom of each sample. Scanning electron microscopy was
used to determine the extent of subsurface degradation and
to collect compositional information for the reaction products
on and near the surface, particularly for the samples exposed
at the highest temperatures.

CORROSION

Deposit Sampling Probe

Covington Virginia

2. Typical corrosion probe showing the arrangement of samples as they were assembled on the probe.

RESULTS
Results from examination of samples from the various corrosion probes are summarized in the following sections. The
order of presentation is the order in which the probes were
inserted in the respective boilers.

Covington recovery boiler deposit probe

Figure 4 shows the average daily temperature of the ten thermocouples over the approximately eight day exposure period
of the deposit sampling probe in the Covington recovery boiler.
The measured temperatures ranged from about 300C to about
600C [11]. Analysis of the deposits using X-ray diffraction provided results that indicated the potassium-rich compound,
K3Na(SO4)2, was among the components of the deposit [11].

Covington recovery boiler corrosion probe

The deposit sampling probe and the corrosion probe were
inserted through a port on the rear wall (the side opposite the
bull nose and the smelt spouts) on the 14th floor of the boiler,
which was about 20 ft (6 m) below the roof of the boiler. This
location was high enough in the boiler that the temperature
could be controlled in the desired range with a reasonable air
flow, and, because this location was well away from the superheater tubes, there were no sootblowers in the vicinity to

3. Photo of the 30 samples removed from one of the corrosion

probes. Samples #6 and #10 still contain the spacers used to
position the support rod in the center of the probe. Several
other samples still contain the remnants of the thermocouple
that was attached to the outer surface of most samples.

affect or disrupt the accumulation of deposits on the probe.

As noted previously, deposits from the sampling probe exposed in the Covington boiler confirmed the presence of significant amounts of potassium in the deposits.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

55

CORROSION

4. Average daily temperatures for each of the ten thermocouples

on the deposit sampling probe.

5. Typical example of thermocouple temperatures from the

corrosion probe exposed in the Covington recovery boiler.
Note that the orange line shows the temperature of
thermocouple #23 which was being used to control the flow
of cooling air and thus the probe temperature.

The corrosion probe was exposed for 2000 h (12 weeks),

and the boiler operated under reasonably normal conditions
for the great majority of that time. An example of the data collected for one day early in the exposure period is shown in
Fig. 5. Unfortunately, several thermocouples failed during
the latter part of the exposure period, so temperatures shown
for some thermocouples do not cover the full 2000 h. At the
conclusion of the probe exposure, the arithmetic mean temperature for each thermocouple on each day was calculated,
and these results are shown in Fig. 6. From a plot of these
average temperatures, an estimate was made for the average
top and bottom temperature for each sample. During the 2000
h exposure, the temperature of every sample clearly varied
considerably from the average value, so the meaning of exposure temperature is much different between the field exposures and the well-controlled laboratory tests.
In order to determine the performance of each alloy as a
56

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

6. Average daily temperatures of thermocouples in the Covington

recovery boiler

7. Total affected material as a function of exposure temperature

for samples exposed for 2000 h on the corrosion probe in the
Covington recovery boiler

function of average exposure temperature and to compare

the performance of all the alloys, the depth of affected material (a sum of thickness loss and extent of subsurface attack)
was plotted versus the estimated average temperature. Results
for all the alloys exposed in the Covington boiler are shown
in Fig. 7, and results for selected individual alloys are shown
in Fig. 8. These results show there are several alloys that did
not perform very well in this environment at and above 570C
(N07214, N08028, and S31009), while N12160 and S21500
were among the better performers. Alloy N06025 showed a
very low rate of material affected around 590C, but considerably higher rates at 560C and 615C.
As a means to compare the results of the corrosion probe
exposure with actual boiler experience, it is important to note
that S31009 has performed well under the current boiler operating conditions presumably at a tube temperature of
about 470C. If it is assumed that the degradation rate observed at 470C for S31009 exposed on the probe is acceptable, then it would seem reasonable that any alloy that demonstrates approximately the same rate at 570C would be

CORROSION

8. Total affected material versus exposure temperature for individual alloys exposed for 2000 h on the corrosion probe in the
Covington recovery boiler

suitable for use at that temperature. On that basis, N12160 and

S21500 would be expected to perform satisfactorily if used in
a high potassium boiler operating with superheater tube
temperatures around 570C.

Crofton hogged fuel boiler

As previously noted, the hogged fuel boiler in Crofton, BC, was
selected because bark from the seawater floated logs would
result in a fuel with high chloride content. X-ray diffraction
analysis of deposits collected from the surface of several corrosion probe samples showed a high concentration of NaCl.

Figure 9 shows the results of examination of several samples.

The corrosion probe in the Crofton boiler was inserted
through a manway door in the superheater area less than an
estimated 2 m below a sootblower. Consequently, some cleaning of the tube surface possibly occurred on a fairly regular
basis. The probe was exposed for a total of about 2160 h, but
the exposure was interrupted and the probe removed from
the boiler because of a planned shutdown and water washing
of the boiler about midway through the exposure period.
Figure 10 shows the average daily temperatures for the
24 thermocouples on the corrosion probe that was exposed
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

57

CORROSION

9. Results of X-ray diffraction measurements of deposits

collected from the bottom of 7 samples on the corrosion probe
in the Crofton hogged fuel boiler.

11. Total affected material as a function of exposure temperature

for samples exposed for 2160 h on the corrosion probe in the
Crofton hogged fuel boiler.

alloys displayed in Fig. 12. The two alumina forming alloys,

N07214 and N06025, along with N08028 showed this type of
behavior while S34709, N08120, and S21500 didnt appear to
show such rapid increases when exposed to average temperature at and above 525C.

Gadsden power boiler

10. Daily average temperature of thermocouples exposed about

2160 h on the corrosion probe in the Crofton hogged fuel boiler.

in the Crofton boiler. The computer software was set to control the flow of cooling air on the basis of the temperature of
thermocouple #24, but it is clear from Fig. 10 that the temperature of thermocouple #24 frequently dropped below the
desired range. In these cases, the air flow was already reduced
to a minimal level, so this resulted in the significant fluctuations in temperature shown in Fig. 10. An average value for
the temperature of each sample was calculated, but it should
be recognized that each sample experienced significant variations around the calculated value.
Figure 11 shows the measured thickness of affected material versus exposure temperature for the samples exposed
on the Crofton boiler probe. With the current maximum superheater tube temperature at about 415C, the goal was to
identify the materials with the lowest degradation rate at
515C. This figure shows that most of the alloys had a fairly
low degradation rate at 500C, but many of the alloys showed
rapid increases in degradation at slightly higher temperatures.
This pattern is more clearly shown in the plots for individual
58

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

One of the boiler manufacturers participating in this project

suggested that the coal fired boiler at Southern Companys
power plant in Gadsden, AL, be considered for inclusion in
the project because the boiler had previously been used for
co-firing coal and biomass both wood and switchgrass. After
a description of the project was presented to power plant and
administrative personnel, the power plant operators agreed
to participate in the project and to operate the boiler continuously for an extended time with a blend of coal and wood. The
original goal was to operate with up to 15% biomass, but that
goal could not be achieved. The actual amount of biomass included in the coal-wood blend ranged from 8% to 15%.
This corrosion probe was inserted in a sootblower lane
between superheater tube platens and is shown in Fig. 13
just before it was removed. Deposits on the probe were a reddish color and were fairly limited in terms of the amount accumulated. Although there were a couple of interruptions in
boiler operation during the early stages of the exposure, as
shown in Fig. 14, the temperatures were far more constant
than observed in the other three boilers that were studied.
The total exposure period lasted for about 123 days, but the
first several days were during a startup period when temperatures were low, and there were two short periods, clearly
evident in Fig. 14, where the boiler briefly cooled, in one case
to near room temperature. The probe was actually exposed
to normal operating conditions for about 106 days or 2540 h.
The average temperatures for the top and bottom of each
sample were calculated from the daily average temperatures.
From these calculations, it was evident that the tops of the
samples were 60-80 Celsius degrees hotter than the bottom

CORROSION

12. Total affected material versus exposure temperature for individual alloys exposed for 2160 h on the corrosion probe in the
Crofton hogged fuel boiler.

of the samples. This is not the pattern generally seen in superheater tubes, but the boiler operators confirmed that this was
the pattern to be expected for this location.
After the probe and the supporting structure were returned to ORNL, preparations were made to dispose of the
portion of the probe that was just inside the boiler wall but
did not include any samples. In making the routine but required check for evidence of radioactivity on the structure, it
was found that low levels of lead and bismuth isotopes were
present. These isotopes are decay products of uranium, and
it is reasonable to assume that the low concentration of ura-

nium in the coal resulted in deposition of these isotopes on

the cooler portion of the corrosion probe. As a result, the nine
samples on the cooler end of the probe could not be removed
and examined. Deposits were removed from the samples on
the hotter end of the corrosion probe, and the analyses identified mullite, quartz, cristobalite, hematite, and magnetite as
the major components.
The 21 samples that could be examined were sectioned in
the customary way, and the reduction in thickness was determined, as well as the subsurface attack. Results of those measurements are plotted as a function of exposure temperature
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

59

CORROSION

13. Photo of the corrosion probe in the Gadsden power boiler

just prior to its removal (photo courtesy of Billy Zemo, Southern
Company).

14. Average daily temperatures of thermocouples on the

corrosion probe exposed about 2540 h in the coal and wood
co-fired Gadsden power boiler.

in Fig. 15. These results show that the degradation of the

alloys in the coal-wood-fired environment was significantly
less than seen for most of the alloys in the boilers fired exclusively with biomass. As can be seen from the figure, several
alloys had losses in the 0.20 mm to 0.25 mm range for this
exposure period.

Port Mellon hogged fuel boiler

The fourth probe was inserted in a hogged fuel boiler in Port
Mellon, BC, with the expectation that the boiler would be,
like the Crofton boiler, firing bark from seawater floated logs.
However, the infestation of pine beetles resulted in a ready
availability of beetle-killed pine trees. Consequently, at the
time of the corrosion probe exposure, the mill was using
60

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

15. Total affected material as a function of exposure temperature

for samples exposed for 2540 h on the corrosion probe in the
Gadsden power boiler.

wood from beetle-killed pine trees, bark from seawater floated logs, and occasional barge loads of C&D waste from Vancouver. The latter of these three sources of biomass may have
ultimately contributed some unexpected, but very interesting, results.
This boiler was converted to a fluidized bed system, which
delayed insertion of the corrosion probe, so this was the last
of the exposures to be started. Once the probe was inserted,
there was a problem with the variable flow valve that made it
impossible to reduce the flow of cooling air to less than the
fully open position. This made it impossible to control the
probe temperature, and consequently, these temperatures
stayed well below the desired level. Eventually, the variable
flow valve was replaced, and this permitted better control of
the temperature. However, the site available for insertion of
the probe was less than 30 cm (12 in.) above the bullnose, and
this proximity to a water cooled surface apparently prevented
the probe temperatures from reaching the desired levels. The
average daily temperatures are shown in Fig. 16, and it is
obvious that the temperatures stayed low until about the 70th
day when the variable flow valve was replaced. Even after the
valve replacement, for many days the highest temperatures
were not much above 500C. This resulted in not having sufficient data to predict degradation rates at and above 500C.
The exposure time at the higher temperature levels was about
118 days or about 2830 h.
Figure 17 shows the depth of affected material versus
temperature for the various samples exposed in the Port
Mellon boiler. In addition to the fact that the highest
temperatures were lower than desired, it was of particular
interest to note that most of the alloys showed a higher than
expected degree of degradation in the 400C-430C
temperature range. Note the data points inside the red oval
in Fig. 17. The deviation from the expected degradation
pattern is more evident in the plots in Fig. 18, which show

CORROSION

16. Average daily temperatures of thermocouples on the

corrosion probe exposed about 2830 h in the Port Mellon
hogged fuel boiler.

17. Total affected material as a function of exposure temperature

for samples exposed about 2830 h on the corrosion probe in the
Port Mellon hogged fuel boiler.

18. Total affected material versus exposure temperature for individual alloys exposed for 2830 h on the corrosion probe in the Port
Mellon hogged fuel boiler. Note the higher amount of material lost in the 400C-430C temperature range for most alloys.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

61

CORROSION
the variation in affected material versus temperature for six
of the alloys exposed in this boiler.
In an effort to determine why higher rates of degradation
were seen at relatively low temperatures, more thorough analysis was undertaken on deposits collected from the samples
exposed in the particular temperature range. Both X-ray diffraction and Raman scattering were used to try to identify
components of the deposits that were potentially responsible
for low melting point compounds that might cause accelerated
degradation in a fairly low temperature range. Results of the
X-ray and Raman studies indicate the presence of zinc (Zn [II])
and/or lead (Pb [II]), either one of which could form relatively
low melting point compounds. A more detailed review of the
analysis results of the deposits collected from the sampling
probe exposed by FPInnovations revealed zinc concentrations
on the order of 0.2% and lead concentrations of about a third
that level [15]. Studies conducted in Finland at bo Akademi
have shown that heavy metal salts or oxides containing zinc
and/or lead can cause accelerated corrosion of superheater
tube alloys like 2Cr-1Mo steel and 347H stainless steel [16].
Based on this information, it appears likely that the higher degree of degradation seen in the 400C -430C range on the corrosion probe samples is due to the presence of these heavy
metals that almost certainly were introduced into the boiler in
the C&D waste. It will be of interest to learn if the lower temperature superheater tubes and the tubes in the bull nose of
the Port Mellon boiler experience accelerated corrosion.
SUMMARY
Four corrosion probes, each containing a total of 30 metallic
samples of 9 different alloys, were each exposed in a boiler firing biomass. Three of the boilers were selected because of the
extreme environment that was expected in each, and the fourth
was chosen because it burned a mixture of coal and biomass.
The temperatures measured by the 24 thermocouples on
each probe showed very significant variations during the exposures, which had durations of at least 2000 h.
In the potassium-rich environment of the Covington probe,
two alloys (N06690 and S21500) seemed to demonstrate corrosion resistance that would suggest they would likely perform
satisfactorily in a potassium-rich environment at a temperature
100C above current operating temperatures.
For the chloride-rich environment of the Crofton boiler,
essentially all alloys showed fairly low degradation rates when
exposed at temperatures below about 500C, but the six alloys that had samples exposed above about 525C all showed
greatly accelerated rates above that temperature. Both alumina forming alloys (N06025 and N07214) performed very
well below about 510C, but, as noted, both alloys had much
higher rates above 525C.
The environment of the Gadsden probe likely showed
small variations as the amount of wood varied, but that variation was probably considerably less than that of the environment in the other boilers studied because of the expectation
of consistent steam production from this power boiler. All
62

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

samples exposed in the Gadsden boiler showed relatively low

degradation rates, possibly because of the absence of significant amounts of alkali metals and chlorides.
The presence of zinc and/or lead in the environment of the
Port Mellon boiler was not expected when the exposure sites
were selected, but this provided some very interesting results.
Essentially all the alloys (except S31009) showed higher degradation rates in the 400C -430C range, and this very likely
can be attributed to the presence of Zn and/or Pb in the deposits. Even though the average sample temperatures on this corrosion probe were well below the desired range, enough information was collected to note that degradation rates for
almost all alloys tended toward higher values even at 40 Celsius
degrees or more below the temperature goal.
It is generally expected that the corrosion resistance of an
alloy will improve with increasing chromium content. However, that was not a pattern that fits the results of the corrosion
probe studies. Alumina forming alloys are sometimes an alternative to chromia forming alloys because of an improved resistance in some aggressive environments, but the alumina
forming alloys did not consistently perform well on these corrosion probes. It is worth noting that the alumina forming alloys often perform better at considerably higher temperatures, and the temperatures to which the probes were
exposed may not have been high enough to get the best features of the alumina forming alloys. TJ
ACKNOWLEDGEMENTS
Dane F. Wilsons efforts in reviewing this document are recognized and appreciated. Adam Willoughby led the effort to
assemble the probes and the data collection system; Tyson
Jordan and Hu Longmire carried out the metallographic sample preparation and light microscopy examinations; Tracie
Lowe conducted the scanning electron microscope examinations; Robbie Meisner performed the X-ray diffraction and
Maggie Connatser the Raman examinations of the deposits.
Employees of all four of the facilities where probes were exposed assisted with installation and removal of the probes,
as well as helped in troubleshooting several problems that
arose. In particular, the help provided by Curtis Clemmons
at Covington, Bob Erickson at Crofton, Billy Zemo at Gadsden, and Peter Hildering at Port Mellon is appreciated. Neville
Stead of FPInnovations provided considerable help in installation, maintenance, and removal of the Crofton and Port
Mellon corrosion probes. Material for the corrosion samples
was provided by Haynes International, Rolled Alloys, ThyssenKrupp VDM, and Sandvik Materials Technology, and
other in-kind contributions were provided by bo Akademi
University, Andritz Oy, Babcock & Wilcox, Catalyst Paper,
Chalmers University of Technology, Domtar Corp., FM Global, FPInnovations, Foster Wheeler, Georgia Institute of Technology, Howe Sound Pulp & Paper, International Paper, MeadWest vaco, Metso Power, Outok u mpu Stai n less,
SharpConsultant, Southern Company, Special Metals, University of Toronto Pulp and Paper Centre, Vattenfall Power, and

CORROSION
the Weyerhaeuser Co. Research was sponsored by the U.S.
Department of Energy, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office, under contract DE-AC05-00OR22725 with UT-Battelle LLC.
LITERATURE CITED
1. Sharp, W.B.A., Corrosion and cracking in chemical recovery equipment, Kraft Recovery Short Course, TAPPI PRESS, Atlanta, GA, USA,
2011.
2. Crowe, D.C. and Youngblood, W.C., Recovery boiler superheater corrosion, Proc. - Int. Symp. Corros. Pulp Pap. Ind., 9th, CPPA,
Montreal, 1998, p. 25.
3. Adams, T.N., Frederick, W.J., Grace, T.M., et al., Kraft Recovery
Boilers, TAPPI PRESS, Atlanta, 1997.
4. Reeve, D.W., Pryke, D.C., and Tran, H.N., Corrosion in the closed
cycle mill, Pulp Pap. Ind. Corros. Probl., Proc. Int. Symp. Corros. Pulp Pap.
Ind., 4th, Swedish Corrosion Institute, Stockholm, Sweden, Vol. 4,
1983.

induced corrosion of 304-type austenitic stainless steel at 600C:

Comparison between KCl, K2CO3 and K2SO4, Mater. Sci. Forum,
2008, pp. 367-375, 595-598.
9. Sharp, W.B.A., Singbeil, D.L., and Keiser, J.R., Superheater corrosion produced by biomass fuels, CORROSION/2012 Conf., NACE
International, Houston, 2012, Product No. 51312-01308-SG.
10. Keiser, J.R., Kish, J.R., Singh, P.M., et al., Superheater corrosion and alloy performance in current generation high
pressure recovery boilers, Final Technical Report: Materials for
Industrial Heat Recovery Systems, 31 December 2007, pp. 53-59.
Available [Online] http://web.ornl.gov/sci/ees/itp/documents/
MidFurnaceSuperheaterFinalReportFnl.pdf >[6Aug2014].
11. Keiser, J.R., Sharp, W.B.A., Singbeil, D.L., et al., TAPPI J. 12(7):
45(2013).
12. Sharp, W.B.A., Singbeil, D.L., and Keiser, J.R., Energy from biomass Lessons from European boilers, TAPPI PEERS Conf., TAPPI
PRESS, Atlanta, 2011.
13. Sharp, W.B.A., Frederick, W.J., Keiser, J.R., et al., TAPPI J. 13(8):
65(2014).

5. Bruno, F., Thermochemical aspects on chloride corrosion in kraft

recovery boilers, NACE CORROSION/2001 Conf., NACE International,
Houston, TX, USA, 2001, Paper No. 01426.

14. Singbeil, D.L., Frederick, L., Keiser, J.R., et al., TAPPI J. 13(8):
37(2014).

6. Mkip, M., Kauppinen, E., Lind, T., et al., Superheater tube

corrosion in recovery boilers, Proc. - Int. Symp. Corros. Pulp Pap. Ind.,
10th, VTT, Espoo, Finland, 2001, Vol. 1, p. 157.

16. Bankiewicz, D., Enestam, S., Yrjas, P., et al., Fuel Process. Technol.
105: 89(2013).

15. Frederick, L., personal communication, FPInnovations, July 2013.

7. Tuiremo, J. and Salmenoja, K., Control of superheater corrosion

in black liquor recovery boilers, Proc. - Int. Symp. Corros. Pulp Pap.
Ind., 10th, VTT, Espoo, Finland, 2001, Vol. 1, p. 181.
8. Pettersson, J., Svensson, J-E., and Johansson, L-G., Alkali-

ABOUT THE AUTHORS

The efficiency of current biomass-fired boilers is severely limited because of corrosion associated with
deposits that accumulate on superheater tubes.
Identification of more corrosion resistant alloys and/
or process changes that would permit boiler operation at considerably higher temperature would result
in a significant improvement in efficiency.
Members of this research team have previously
studied other boiler corrosion issues floor and wall
tube corrosion and cracking and superheater tube
cracking and corrosion. Superheater corrosion studies conducted in Europe, primarily Finland, have produced results complementary to this latest work.
A major issue in this study was identifying biomass
boilers with particularly aggressive superheater environments and then obtaining permission and suitable
exposure sites in these boilers.
The severity of corrosion of some alloys in these
environments was somewhat of a surprise, and we
had not anticipated the apparent effect of Zn on superheater tube corrosion in the boiler burning a limited amount of construction/demolition waste.
If boiler operators want to make limited increases in
superheater temperature, these results provide some
guidance on selection of superheater tube materials.

Keiser

Sharp

Singbeil

Also, if a new boiler is being designed, these results

provide guidance on what materials could be used for
superheater tubes that could operate 100C hotter
than would conventionally be done.
We would like to continue our work with a more indepth study of the more promising alloys in operating
biomass-fired boilers.
Keiser is distinguished research and development staff
member, Materials Science and Technology Div., at
Oak Ridge National Laboratory, Oak Ridge, TN, USA.
Sharp is principal at SharpConsultant in Columbia, MD,
USA. Singbeil is principal scientist and Frederick is
principal technologist with FPInnovations, Vancouver,
BC, Canada. Email Keiser at [email protected].

AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

63

The science of wowing

your customers.
Paper with ColorLok Technology helps your business stand out.
The paper you choose makes a difference. The science behind ColorLok Technology
gives you faster drying times, more vivid colors, and bolder blacks. It all adds up to
the ultimate stopping power. Because every detail matters.
To stand out, be sure to buy paper with the ColorLok logo. Learn more at colorlok.com

2014 Hewlett-Packard Development Company, L.P.

The information contained herein is subject to change without notice.
HP shall not be liable for technical or editorial errors or omissions contained herein.

PEER-REVIEWED

CORROSION

Could biomass-fueled boilers be operated

at higher steam temperatures?
Part 3: Initial analysis of costs and benefits
W.B.A. (SANDY) SHARP, W.J. (JIM) FREDERICK, JAMES R. KEISER,
and DOUGLAS L. SINGBEIL

ABSTRACT: The efficiencies of biomass-fueled power plants are much lower than those of coal-fueled plants
because they restrict their exit steam temperatures to inhibit fireside corrosion of superheater tubes. However,
restricting the temperature of a given mass of steam produced by a biomass boiler decreases the amount of power
that can be generated from this steam in the turbine generator.
This paper examines the relationship between the temperature of superheated steam produced by a boiler and
the quantity of power that it can generate. The thermodynamic basis for this relationship is presented, and the
value of the additional power that could be generated by operating with higher superheated steam temperatures is
estimated. Calculations are presented for five plants that produce both steam and power. Two are powered by black
liquor recovery boilers and three by wood-fired boilers. Steam generation parameters for these plants were
supplied by industrial partners. Calculations using thermodynamics-based plant simulation software show that the
value of the increased power that could be generated in these units by increasing superheated steam temperatures
100C above current operating conditions ranges between US$2,410,000 and US$11,180,000 per year. The costs and
benefits of achieving higher superheated steam conditions in an individual boiler depend on local plant conditions
and the price of power. However, the magnitude of the increased power that can be generated by increasing
superheated steam temperatures is so great that it appears to justify the cost of corrosion-mitigation methods such
as installing corrosion-resistant materials costing far more than current superheater alloys; redesigning biomassfueled boilers to remove the superheater from the flue gas path; or adding chemicals to remove corrosive
constituents from the flue gas. The most economic pathways to higher steam temperatures will very likely involve
combinations of these methods. Particularly attractive approaches include installing more corrosion-resistant alloys
in the hottest superheater locations, and relocating the superheater from the flue gas path to an externally-fired
location or to the loop seal of a circulating fluidized bed boiler.
Application: The value of additional power that could be generated by raising biomass boiler steam
temperatures by 100 Celsius degrees appears to justify substantial expenditures to avoid superheater corrosion at
the higher temperature, including the use of much more expensive corrosion-resistant alloys or substantial design
and operational changes, particularly if the high temperature superheaters were designed for rapid replacement.

he 1997 Kyoto Protocol set the goal of stabilizing greenhouse gas concentrations in the atmosphere at a level that
would avoid dangerous climate change by obliging industrialized countries to reduce such emissions. European countries
that ratified this protocol have adopted systems of incentives
and penalties to reduce greenhouse gas emissions. Many have
developed and implemented technologies that improve the
efficiency with which steam and power can be generated
from biomass fuels.
Biomass fuels are attractive for power generation because
they are renewable and almost carbon-neutral. However, the
combustion of biomass produces superheater ashes that contain corrosive elements like chlorine, potassium and sodium.
When climate change and renewable fuels were of less con-

cern, outlet steam temperatures of biomass-fueled boilers

were kept low to avoid fireside corrosion of superheater tubes
and/or pluggage in the convective sections. Such temperature
limitations restricted the power that could be generated from
biomass fuels.
The United States (U.S.) Department of Energy (DOE) has
funded a project to evaluate corrosion mitigation strategies that
would enable biomass-fueled boilers to operate with higher superheater steam temperatures. Raising superheated steam temperatures would increase the efficiency of biomass-fired Combined Heat and Power (CHP) plants. The use of U.S.-grown
renewable biomass fuels would reduce the U.S. dependence on
imported fuels and fossil fuels. This DOE-funded project includes a review of the science and technology of superheater
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

65

CORROSION

1. Process mass and energy streams included in the software model.

corrosion in biomass boilers; calculations of the economic benefits of increased steam temperatures; and laboratory and field
evaluations of alternative tube alloys at temperatures near and
beyond current operating temperatures.
Previous results from this project have been published in a
series of three papers. The first reviewed European approaches to inhibiting superheater corrosion in biomass boilers by
using corrosion-resistant alloys, modifying boiler designs, and
using process additives [1]. The second reviewed the performance of corrosion-resistant alloys that may be suitable for
biomass boiler superheaters operating at higher temperatures
[2]. The third paper described tests to evaluate the corrosion
resistance of candidate alloys at higher temperatures in a black
liquor recovery boiler that operates with unusually corrosive
superheater conditions [3]. Two other papers report laboratory tests of the corrosion resistance of candidate superheater
alloys for higher temperature operation [4] and field tests of
candidate alloys in operating boilers [5].
The work on which the current paper is based analyzed
the economic impacts of higher superheater temperatures in
biomass-fueled boilers. Boiler performance and process conditions were estimated using proprietary thermodynamicsbased plant simulation software. We present financial calculations of the benefits of increased superheated steam
temperatures based on the analysis of steam and power production in five example boilers burning biomass. These financial benefits are compared with the additional costs of
alloys that could resist corrosion at the higher superheater
temperatures. Where the financial benefits exceed these
costs, using more corrosion-resistant alloys to enable higher
steam temperatures offers a net financial benefit to the boiler
operator. Second, we compare the financial benefits of high66

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

er steam temperatures with the cost of fuel treatments or additives [1] applied to reduce the concentration of corrosives
in the superheater ash. Third, we compare the financial benefits with the cost of boiler redesign, e.g., of constructing a
circulating fluidized bed (CFB) boiler with the superheater
in the loop seal [1] instead of in the convective pass.
CALCULATION METHODS
The primary tool used for these calculations is a thermodynamic software program called StmPowr (Table Mountain
Consulting; Golden, CO, USA), which models CHP systems
that supply process steam to an adjacent plant (or sell it for
district heating) and simultaneously generate electricity for
plant use or external sale. For example, in a pulp and paper
mill, a typical steam turbine would supply high-, medium- or
low-pressure steam to dry paper, evaporate water from black
liquor, and heat process units such as pulp digesters. It would
also generate electrical power.
The StmPowr software calculates the net electrical power
and process steam generated from a Rankine cycle power
plant. Almost all power plants use the Rankine cycle as their
heat engine. This cycle has four components. First, the working fluid (boiler feedwater) is pressurized by a pump. The
high pressure water is heated at constant pressure to become
a dry saturated vapor (steam). This dry steam expands
through a steam turbine, generating power and lower pressure process steam. The remaining turbine exhaust steam is
condensed immediately or later to a saturated liquid. The
working fluid (condensate) is recirculated and heated to become boiler feedwater, continuing through a closed condensate boiler feedwater steam loop.
Required inputs to the model include the rate of fuel input,

CORROSION
Specified Input Conditions

Specified Exit Conditions

Calculated Parameters

Fuel input rate and temperature, gross heating

value of dry fuel, fuel moisture content, and
elemental composition

Temperature and pressure of the

superheated steam produced

Thermal efficiency of boiler

Combustion air input rate, temperature, and

moisture content

Blowdown rate

Rate of generation of high pressure,

superheated steam

Boiler feedwater temperature and pressure

Boiler feedwater input rate

Internal steam used for air heating and sootblowing (pressures, temperatures, and flows)

Flue gas exit temperature

Correlation between flue gas exit temperature

and fuel input rate

Ash output rate

I. Components of boiler sub-module.

Specified Input Conditions
Flow rate of high pressure superheated
steam to the turbine inlet

Specified Exit Conditions

Calculated Parameters

Pressures and flow rates of process

steam extracted from the turbine

Temperature and pressure of steam delivered

to the turbine

Power generated for the steam

conditions specified
Temperatures of process steam
extracted from the turbine

Isentropic efficiency of the turbine

II. Components of the steam turbine sub-module.

the fuel characteristics (elemental composition, gross heating

value, and temperature), and process and equipment
parameters for various unit operations that will be described
later in this paper. The software uses the flow rate,
temperature and pressure of the inlet steam and turbine
efficiency parameters to predict the temperature, pressure,
and flow rates of the steam extracted from the turbine and
the power generated. Figure 1 shows the configuration and
the process mass and energy streams included in the model.
The program calculates the flow rates and conditions of the
process streams that cross the process boundary (red dashed
line) in this figure.
COMPONENTS OF THE STEAM POWER
CYCLE CALCULATION MODULE
The Rankine cycle analyzed by the StmPwr software program
consists of seven unit processes, which are discussed in the
following sections:
1. Boiler
2. Steam turbine
3. Exhaust steam surface condenser
4. Condensate transfer pump
5. Deaerator
6. Boiler feedwater pump
7. Combustion air preheater

Boiler
The boiler sub-module calculates the net rate of high-pressure
steam generation at specified steam conditions (T, P) and
other specified parameters from the fuel input rate, as shown
in Table I. The steam rate calculation is based on the heat
loss method [6,7].

Steam turbine
The steam turbine sub-module calculates the power generated
by the calculated flow of superheated steam generated by the
boiler. The turbine operates at its own specified isentropic efficiency. The power generated is calculated from the temperature, pressure, and flow rate of the steam supplied to the turbine;
the temperatures, pressures, and flow rates of process steam
extracted from the turbine; and the temperature, pressure, and
flow rate at the turbine exhaust, as shown in Table II.

Exhaust steam condenser

For condensing turbines, this sub-module calculates the cooling water flow required to condense the exhausted steam at
its specified temperature, pressure, and flow rate, as summarized in Table III. The condensate returns to the deaerator
for reuse as boiler feed water.

Condensate transfer pump

The condensate transfer pump moves the condensate from
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

67

CORROSION
Specified Input Conditions
Temperature, pressure and flow rate
of turbine exhaust steam

Specified Exit Conditions

Calculated Parameters
Flow of cooling water required to
condense exhausted steam

Temperature and pressure

of condensate
Inlet temperature of cooling water

Feed water make-up rate and temperature

Outlet temperature of cooling water

III. Components of exhaust steam condenser sub-module.

Specified Input Conditions

Temperature, pressure and flow rate of
condensate from the condensate sump

Specified Exit Conditions

Deaerator pressure

Calculated Parameters
Power required by condensate
transfer pump

Pump and motor efficiency

IV. Components of condensate transfer sub-module.

Specified Input Conditions

Temperature, pressure and flow rate of
condensate return to deaerator

Specified Exit Conditions

Calculated Parameters

Boiler feedwater temperature

Flow rate of steam extracted from the

turbine to the deaerator at the specified
temperature and pressure

Specified Exit Conditions

Calculated Parameters

Temperature and flow rate of boiler

feedwater make-up
Temperature, pressure, and flow rate of
steam used for deaerator heating

V. Components of the deaerator sub-module.

Specified Input Conditions

Temperature, pressure and flow rate of
boiler feedwater flow from deaerator

Pressure of feedwater at
boiler inlet

Power required by boiler feedwater pump

Efficiency of the feedwater pump and motor

VI. Components of the boiler feedwater pump sub-module.

Specified Input Conditions

Temperature and pressure of steam extracted
from turbine to heat combustion air
Temperature of ambient air entering preheater

VII. Components of the combustion air sub-module.

68

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

Specified Exit Conditions

Temperature of combustion
air entering boiler

Calculated Parameters
Required flow rate of steam extracted
from turbine to heat combustion air

CORROSION

2. Types of steam turbines used in calculations of costs and benefits of increasing superheater steam temperature.

the turbine steam condenser sump to the deaerator. Components of the sub-module are shown in Table IV.

Deaerator
Boiler makeup feedwater is added to the condensate in the
deaerator. This combined feedwater is heated to the saturation
temperature at the deaerator pressure by steam extracted
from the turbine. The components of the deaerator sub-module are listed in Table V.

Boiler feedwater pump

The boiler feedwater pump raises the pressure of the boiler
feedwater from the pressure in the deaerator up to the boiler
inlet pressure, as shown in Table VI.

Combustion air preheater

The combustion air preheater is a heat exchanger that uses
turbine extraction steam to heat the outside air from ambient
temperature to the temperature required for fuel combustion,
as summarized in Table VII.
BOILER SYSTEMS ANALYZED
To determine the magnitude of the economic costs and benefits of higher superheater steam conditions, we asked industrial collaborators in the project to supply data from typical
biomass-fueled boilers.
Calculations were made with data from five boilers: two
recovery boilers and three boilers burning hogged wood. Hog
fuel is produced from waste bark or wood that has been processed by a disintegrator, or hog, that processes the wood into
chips or shreds of practically uniform size.
The two boilers used for the calculations, designated recovery boiler B and recovery boiler M, were based on models.
Data for recovery boiler B were supplied by a boiler manufacturer as part of a software model of a recovery boiler in a
typical pulp mill where process steam is extracted from the
turbine. Steam data for recovery boiler M came from a Win-

GEMS (Metso Automation; Helsinki, Finland) software model

of a 1000 o.d. tons/day unbleached pulp mill with a recovery
boiler operating at 1450 psi. These two recovery boilers produced power outputs of 52 MW and 53.6 MW, respectively,
with conventional steam conditions.
All three biomass boilers were actual units. Biomass boiler
C operates on the Pacific Coast and is fired by hogged fuel
produced from sea-floated logs. Biomass boiler H also operates
on the Pacific Coast, but fires about 2/3 beetle-killed pine,
about 1/3 sea-floated logs and a small amount of wood from
construction waste. Biomass boiler M is a biomass boiler recently converted to bubbling fluidized bed operation by a
boiler manufacturer. It burns hogged wood.
Figure 2 illustrates the difference between the two types
of steam turbine that are used in the calculations that follow.
Condensing CHP turbines are shown schematically in Fig. 2a.
Steam used to turn the rotor is extracted at high, medium, and
low pressures (or sometimes at only two pressures). Any remaining steam is then condensed in water-cooled tubes in a
surface condenser. Non-condensing (back-pressure) turbines,
shown schematically in Fig. 2b, extract all the steam used to
turn the turbine rotor as process steam to be used in an adjacent plant.
Both of the example recovery boilers deliver their steam to
condensing turbines while all three wood-fired boilers deliver
their steam to non-condensing turbines. Condensing turbines
use all the steam supplied by the boiler to turn the turbo-generator rotor and generate power. Steam that has passed through
the full length of the turbine is condensed on water-cooled
tubes in a surface condenser. This condensation decreases the
volume of the steam by about 1800 times and creates a vacuum
in the condenser. Surface condensers can produce pressures
of 90 to 100 kilopascals (13 to 14.5 psi) below atmospheric
pressure. Creating this vacuum allows the steam supplied by
the boiler to expand much more, producing a larger enthalpy
change through the turbine and increasing the turbine efficiency. For this reason, utility boilers, where efficiency is of
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

69

CORROSION
primary importance, always use condensing turbines. Note
that the vacuum pressure developed by the surface condenser
depends on the temperature of the cooling water, which could
be subject to seasonal changes, and on the partial pressures
exerted by non-condensable gases (NCGs) accumulated in the
feedwater. Because the turbine efficiency increases as the condenser vacuum increases, accumulated NCGs should be
stripped out and the condensing temperature should be kept
as low as possible. The steam condensate, stripped of NCGs,
is returned to the boiler as feed water.
The simplest non-condensing (back-pressure) steam turbine would extract all the steam used to drive the turbo-generator rotor as process steam at a series of decreasing pressures. Back-pressure simply indicates that the steam is
extracted at pressures above atmospheric. The process steam
flows are extracted at the pressure(s) required by the process
operations, regardless of the load on the turbine. This is
achieved by controlling the steam flows at the extraction
points. The extraction pressures are chosen to suit the particular process steam requirements of the plant where the
boiler is located. The extracted steam can be used in an adjacent process plant (e.g., a pulp and paper mill or chemical
plant) or for district heating. The steam condensate is later
augmented by make-up boiler water, deaerated, and returned
to the boiler.
Once saturated steam has been generated in a boiler, relatively little additional fuel energy is required to raise its temperature and pressure to the superheat required to provide
more efficient power generation. Therefore, if a process plant
already needs a substantial amount of process steam, adding
a non-condensing turbine offers a simultaneous supply of
relatively inexpensive power without the need for the substantial volumes of cooling water that would be required to
operate a surface condenser in a condensing turbine. However, the capacity of a turbine to generate power falls substantially when process steam is extracted at appreciable
pressure instead of being expanded into a vacuum in a surface condenser.
The biomass-fueled boilers studied in the current project
all produce a combination of heat and power (CHP). All of
them provide process steam to an adjacent plant and all of
them generate power. All are once-through turbines that do
not take advantage of the efficiencies of reheating the steam
in the boiler between various turbine stages.
THERMODYNAMIC BASIS FOR
THE INCREASING VALUE OF STEAM
WITH INCREASING TEMPERATURE
The principle that underlies the work described in this paper
is that the availability of energy in steam increases as the temperature of the steam increases. Availability, also called available energy, is a thermodynamic property like enthalpy. Gibbs
introduced this concept in 1873 [8] by defining capacity for
entropy as the amount by which the entropy of a body can
be increased without changing the energy of the body or in70

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

3. Increase of specific enthalpy of steam with temperature at

50 and 100 bars pressure (725 and 1450 psi).

4. Increase of specific entropy with temperature at 50 and 100

bars pressure (725 and 1450 psi).

5. Increase of specific available energy with increasing steam

temperature at 50 and 100 bars pressure (725 and 1450 psi).

creasing its volume. For steam, the available energy is the

maximum amount of mechanical work that can be obtained
from it in an ideal engine by bringing it to the temperature
and pressure of a reference atmosphere without allowing heat
to pass into it or out of it, i.e.:
Available energy = Enthalpy of steam (Entropy of steam)
x (Temperature of surroundings) (1)

CORROSION
Recovery Boiler

Boiler Type and Identifier Code

Biomass (wood)-Fired Boiler

Steam production, kg/sec

91.8

101.68

119.7

59.98

60.48

Steam pressure, bar

61.22

99.66

41.84

89.63

89.63

Steam temperature, C

465.6

471.1

399.0

482.2

480.0

HP steam pressure, bar

N/A

N/A

N/A

N/A

28

HP steam temperature, C

N/A

N/A

N/A

N/A

328.2

HP steam enthalpy, kJ/kg

2933.74

2838.42

3132.56

3146.11

3069.18

HP steam extraction flow, kg/sec

MP steam pressure, bar

15.12

12.75

12.75

11.22

12.75

12.75

MP steam temperature, C

247.8

190.86

335.2

330.2

287.4

MP steam enthalpy, kJ/kg

2923.63

2814.46

3124.03

3110.13

3016.76

MP steam extraction flow, kg/sec

8.82

39.31

15.12

15.00

6.05

LP steam pressure, bar

5.10

5.10

4.15

4.15

4.15

LP steam temperature, C

152.60

152.60

177.1

175.9

167.9

LP steam enthalpy, kJ/kg

2768.81

2756.43

2811.52

2808.91

2791.51

LP steam extraction flow, kg/sec

63.76

48.26

104.58

44.99

39.31

Exhaust steam pressure, bar

0.14

0.14

N/A

N/A

N/A

Exhaust steam temperature, C

100

100

N/A

N/A

N/A

Exhaust steam enthalpy, kJ/kg

2715.63

2711.29

Exhaust steam flow, kg/s

19.22

14.11

Condensate flow, kg/s

72.58

87.57

119.7

59.98

60.48

18

18

1.24

1.24

1.24

108.5

108.5

100

100

100

High pressure stage

0.890

0.890

Medium pressure stage

0.459

0.459

Low pressure stage

0.887

0.887

0.813

0.813

0.813

Exhaust stage

0.650

0.650

Condensate pressure, bar

Adj. condensate temperature, C
Intrinsic Turbine Stage Efficiencies

VIII. Steam system parameters for the five boilers studied.

More rigorously:
a = h T0s

(2)

where:
a = specific available energy, kJ/kg
h = specific enthalpy, kJ/kg
T0 = absolute temperature of the surroundings, K
s = specific entropy, kJ/kg
Consider the effect of increasing temperature on the steam
parameters in this equation. Figure 3 shows that, as the temperature of a unit mass of steam at constant pressure increases, its specific enthalpy increases. Figure 4 shows that the
specific entropy of the steam also increases with the steam
temperature at constant pressure.
The availability of energy from the steam increases with
increasing temperature because the specific enthalpy of

steam increases proportionally more with temperature than

does the specific entropy of steam. This conclusion is demonstrated in Fig. 5, which shows the temperature dependence
of the specific available energy of steam at boiler pressures of
50 bars (725 psi) and 100 bars (1450 psi) within the temperature range of interest in this work. These data lead to the practical conclusion that, for a turbine to derive the most energy
from the steam produced by a boiler fuel, the boiler should be
designed to produce steam at the highest temperature that its
superheater metallurgy allows.
It is important to note that the volume of a given mass of
steam heated by burning a given amount of fuel will fall as the
final temperature of the steam increases. In the data presented
in Fig. 4, raising the steam temperature from 500C to 600C
without changing the fuel input to the CHP plant reduces the
volume of steam produced by 6.6% at a pressure of 50 bars
(725 psi) and by 7.2% at a pressure of 100 bars (1450 psi).
Whether or not this reduction in steam volume presents a
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

71

CORROSION
Recovery Boiler
Type of Boiler and Identifier
B

Condition/data
Current power generation

52.0 MW

53.6 MW

Power generated with steam temperature 50C higher

56.4 MW

60.8 MW

+4.4 MW

+7.2 MW

+8.4%

+13.4%

Additional power generated

Value of this additional power @ $40/MWh

+$1.54 million/year

+$2.52 million/year

Value of this additional power @ $80/MWh

+$3.08 million/year

+$5.05 million/year

Power generated with steam temperature 100C higher

Additional power generated

59.0 MW

66.3 MW

+7.0 MW

+12.7 MW

+13.4%

+23.7%

Value of this additional power @ $40/MWh

+$2.45 million/year

+$4.45 million/year

Value of this additional power @ $80/MWh

+$4.91 million/year

+$8.90 million/year

IX. Value of increased energy production in the two example recovery boilers. (All currency in USD.)

problem in plant operation will depend on the process steam

requirements and the turbine design. In a condensing turbine,
the smaller volume of steam produced at a higher temperature
could, in principle, be redistributed among the extraction
stages by controlling the steam flows at the extraction points.
However, in a back-pressure turbine, it may not be possible to
maintain the process steam flows if the steam volume is
reduced by heating the steam to a higher final temperature.
Table VIII lists the steam parameters for the two recovery boilers and the three wood-fired boilers studied in the
economic calculations. The boiler parameters were supplied
by project partners. Three of these example boilers used for
the calculations were also used for field tests to measure the
corrosion resistance of candidate alloys in higher temperature
superheater environments.
The steam extraction temperatures used in the calculations
were recalculated from the extraction pressures using standardized turbine stage efficiencies to determine the entropy
of the steam extracted. The condensate temperatures were
adjusted to those of saturated condensate at atmospheric temperature. These calculations used the procedures proposed by
Ulrich [9], although similar calculation procedures are presented in other engineering thermodynamics and process design texts. The turbine stage efficiencies used for the recovery
boiler calculations were provided by an industrial partner [10],
and the turbine stage efficiencies used for the biomass boiler
calculations were estimated from data presented by Ulrich [9].
ECONOMIC CALCULATIONS
Many different approaches could be taken for calculations of the
effects of increasing the outlet steam temperatures of a biomassfueled boiler. Different sets of constraints could be applied to
evaluate the effects of specific local factors, such as changing
needs for process steam, opportunities to sell high-priced
power, or a decision to install a new high-efficiency turbine.
72

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

The purpose of the current project was to estimate the

economic benefits that would accrue to a boiler operator if a
biomass-fueled boiler could be operated with substantially
higher exit steam temperatures. To make these estimates, calculations were made with the data from Table VIII. The
value of the increased steam and power production that
would be obtained by increasing the steam temperature by
50C and by 100C was calculated for each of the five example
boilers, using the following assumptions:
1. The existing turbine could operate with the higher temperature steam.
2. The heat produced by the combustion of the fuel would
be unchanged when the final steam temperature was
increased.
3. The product of specific steam enthalpy and steam flow
would be unchanged when the final steam temperature
was increased.
4. The amount of steam extracted from the turbine to meet
process heating requirements would remain unchanged
when the final steam temperature was increased.
We will first review the results of calculations made with
these assumptions and later discuss the uncertainties associated with these assumptions.
Table IX shows the value of the increased power that
would be produced in the two example recovery boilers that
would accrue to the owners of these boilers if they could increase the exit steam temperature by 50C or 100C above the
current operating temperature. Because the value of electrical power varies both with location and with time, the value
of the additional power generated was calculated for power
prices of both US$40/MWh and US$80/MWh. Current power
prices typically lie within this range.
In related work, three types of virtual turbine were studied

CORROSION
Wood-Fired Boiler
Type of Boiler and Identifier
C

Parameter

42.8 MW

27.5 MW

27.4 MW

Current power generation

42.8 MW

27.5 MW

27.4 MW

Power generated with a steam temperature 50C higher

51.2 MW

31.3 MW

31.0 MW

+8.4 MW

+3.8 MW

+3.6 MW

+19.6%

+13.8%

+13.1%

Value of this additional power @ $40/MWh

+$2.94 million/year

+$1.33 million/year

+$1.26 million/year

Value of this additional power @ $80/MWh

Additional power generated

+$5.87 million/year

+$2.66 million/year

+$2.52 million/year

 educed volume of steam generated with a steam

R
temperature 50C higher

3.22 kg/s

2.18 kg/s

2.2 kg/s

Power generated with a steam temperature 100C higher

58.8 MW

34.9 MW

34.3 MW

+16.0 MW

+7.4 MW

+6.9 MW

+37.3%

+26.9%

+25.1%

Additional power generated

Value of this additional power @ $40/MWh

+$5.59 million/year

+$2.59 million/year

+$2.4 million/year

Value of this additional power @ $80/MWh

+$11.18 million/year

+$5.17 million/year

+$4.82 million/year

 educed volume of steam generated a stem

R
temperature 100C higher

8.09 kg/s

4.15 kg/s

4.2 kg/s

X. Value of increased power that could be produced by raising exit steam temperatures by 50C and by 100C in the three example
wood-fueled boilers. (All currency in USD.)

to evaluate the effects of higher steam conditions in recovery

boiler M. In the first case, the turbine generator was optimized for the original (conventional) steam conditions and
was therefore less efficient at the higher steam temperatures.
In the second case it was assumed that the turbine generator
used for higher steam conditions was optimized for these conditions, i.e. that the turbine efficiency remained constant as
the steam conditions were increased. In the third case a reheater was added between the high pressure and medium
pressure turbine extractions. These calculations showed that
increasing the isentropic efficiencies of the various turbine
stages towards the maximum attainable efficiency increases
the amount of power generated as much as increasing the
steam temperature. This shows the value of using a steam
turbine designed for higher steam conditions when redesigning a boiler to produce higher temperature steam.
Table X shows the value of the increased power that
would be produced in the three example wood-fired boilers
assuming that the isentropic efficiency of the steam turbine
was maintained at the higher input steam temperature.
Table XI summarizes the calculated economic benefits
of increased exit steam temperatures in all five example boilers under the previously-stated assumptions. These results
indicate that very substantial economic benefits would accrue
if biomass boilers could be designed to operate with higher
superheater steam temperatures.
The magnitude of the increased power generated at higher temperatures in these calculations is similar to that published by groups in Sweden [11] and Denmark [12]. Stlen-

heim and Henderson [11] have presented similar plots of the

increase of electrical efficiency with final steam temperature
at various steam pressures in advanced steam cycle units.
Kamuk [12] commented that the electrical efficiency of biomass-fueled boiler systems is typically limited to about 35%,
compared to about 49% for coal-fired boiler systems, because
of superheater corrosion. She also noted that achieving higher steam conditions is the key to raising electrical efficiency.
The most attractive route to achieving higher steam
temperatures appears to be to identify new superheater
alloys or composite tube materials that can resist corrosion
and creep in the flue gas and fuel ash environment around
superheater tubes in biomass boilers. If suitable tubing were
installed in the hottest portions of a biomass-fueled boilers
superheater, the boiler owner would make a net profit if the
additional cost of this tubing did not exceed the value of the
additional steam generated. For example, if 5 m of high alloy
tubing were installed in each of ten superheater loops in
45 platens (for a total of 2250 m) and served for 5 years before
being replaced, tubing costs of less than US$5378/m to
US$24911/m would more than pay for a 100C increase in
steam temperature. Current costs of superheater tubing are
about two orders of magnitude lower than the lowest of
these estimates of the justifiable cost. To gain the full
financial advantage of tubes designed for a five-year
replacement life, it would be useful to also design the
superheater for rapid replacement. We noted in an earlier
publication [1] that the superheater of a 67 MW waste-fired
boiler operated by the City of Amsterdam, which had been
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

73

CORROSION
Recovery Boilers

Biomass Boilers

Parameter
B

+$1.54
million/year

+$2.52
million/year

+$2.94
million/year

+$1.33
million/year

+$1.26
million/year

+$3.08
million/year

+$5.05
million/year

+$5.87
million/year

+$2.66
million/year

+$2.52
million/year

Steam temperature 50C higher

Increased value
@ $40/MWh
Increased value
@ $80/MWh

Steam temperature 100C higher

Increased value
@ $40/MWh

+$2.45
million/year

+$4.45
million/year

+$5.59
million/year

+$2.59
million/year

+$2.41
million/year

Increased value
@ $80/MWh

+$4.91
million/year

+$8.90
million/year

+$11.18
million/year

+$5.17
million/year

+$4.82
million/year

XI. Summary of calculated value of additional power generated by raising exit steam temperatures by 50C and 100C.
(All currency in USD.)

designed for rapid replacement, was recently replaced in 48

hours, from fire-to-fire.
A second route to enabling higher steam conditions would
be to chemically treat the biomass fuel to remove corrosive
components before or during combustion. This could involve
diluting a corrosive biomass fuel with a less-corrosive fuel,
leaching corrosives out of biofuels, or using additives (such as
ammonium sulfate or SO2) that convert alkali chlorides to less
corrosive sulfates, or additives such as kaolin (40-60 wt%
Al2O3.SiO2) that convert alkali salts to less-corrosive silicates.
More details about these approaches can be found in [1].
If the corrosivity of the biomass fuel could be reduced by pretreatment or by combustion additives, the fuel processing
would have to cost less than US$2,420,000 to US$11,210,000
per year to justify the increased value of the power generated
by producing steam 100C hotter than current values.
Other approaches to enabling higher steam conditions
that have been implemented in Europe [1] involve redesigning biomass-fueled boilers to maximize the service life
of superheater tubes by separating the superheater tubes
from the corrosive fly ash (either by adding an empty pass
between the furnace and the superheater or by condensing
the fly ash on low temperature tubes in a chlorine trap); by
increasing the superheater temperature above the dew point
temperature of corrosive components of the fly ash; by
moving the superheater of a circulating fluidized bed boiler
(CFB) out of the flue gas passage into the recirculated
fluidizing medium; and by using an external superheater
fired by a less corrosive fuel. With the current abundance and
relatively low cost of natural gas in the U.S., this last approach
appears particularly attractive. To justify the financial benefit
of the increased power production provided by a 100C
steam temperature increase within five years, the capital
costs of boiler design changes would have to be less than
74

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

US$12,000,000 to US$56,000,000. The magnitude of these

financial gains associated with increased steam temperatures
appears to justify substantial boiler redesign or fuel treatment.
ISSUES RAISED BY
SIMPLIFYING ASSUMPTIONS
We will now review the credibility of the simplifying assumptions that were used to calculate the potential benefits of increased steam conditions. The first assumption, that the existing turbine could handle the higher temperature steam, is
probably not true. Steam turbines have designs and materials
of construction optimized for particular steam conditions, and
the calculations described previously for recovery boiler M
indicate that the efficiency of an existing turbine would be
substantially lower if it operated with a steam temperature
100C higher than its design conditions. Substantial infusions
of capital could be required to simultaneously upgrade both
biomass-fueled boilers and their turbines to take advantage of
the increased power generation potential of hotter steam.
However, we have shown previously that the value of the additional steam generated would more than justify the cost of
any foreseeable corrosion-resistant alloy tubing and that the
increased cost of a turbine designed to handle higher temperature steam could probably be paid off by the value of the
additional steam generated in one or two years. New biomassfueled boilers are being built every year and will very likely
be designed for efficient combustion and efficient steam generation, following the approaches discussed here.
The second assumption in the calculations is that the heat
produced by the combustion of the fuel remains unchanged.
This assumption appears reasonable. The firing of the fuel in
the furnace and the volume of water supplied to the boiler
would not be changed, but the steam would be heated to a
higher final temperature in the superheater. The final steam

CORROSION
Recovery
Boiler

Steam Flow, kg/sec

Case
HP Extraction

MP Extraction

LP Extraction

Exhaust

Total

Current operation

8.82

63.76

19.22

91.80

Steam temperature 50C higher

8.82

63.76

16.04

88.62

Steam temperature 100C higher

8.82

63.76

13.12

85.70

Current operation

39.31

48.26

14.11

101.68

Steam temperature 50C higher

39.31

48.26

10.27

97.84

Steam temperature 100C higher

39.31

48.26

6.82

94.39

XII. Calculations of redirected steam flows in the steam turbines of recovery boilers B and M as the inlet steam temperature is
raised by 50C and by 100C.

Biomass
Boiler

Steam Flow, kg/sec

Case
HP Extraction

MP Extraction

LP Extraction

Exhaust

Total

Current operation

0.00

15.12

104.58

0.00

119.70

Steam temperature increased

by 50C

0.00

14.59

100.89

0.00

115.48

Steam temperature increased

by 100C

0.00

14.10

97.51

0.00

111.61

Current operation

0.00

15.00

44.99

0.00

59.99

Steam temperature increased

by 50C

0.00

14.45

43.35

0.00

57.80

Steam temperature increased

by 100C

0.00

13.96

41.87

0.00

55.83

Current operation

15.12

6.05

39.31

0.00

60.48

Steam temperature increased

by 50C

14.57

5.83

37.88

0.00

58.28

Steam temperature increased

by 100C

14.07

5.63

36.58

0.00

56.28

XIII. Calculations of steam flow reductions in the steam turbines of biomass-fueled boilers C, H, and M as the inlet steam
temperature is raised by 50C and 100C.

temperature would be raised by extracting more heat in the

superheater, most likely by redesigning the superheater to add
more heat transfer surface. Increasing the heat transfer surface in the superheater would require installing additional
tube platens or extending the existing platens. Although the
costs of this type of rebuild are highly plant-specific, the financial calculations suggest that these capital costs could be
paid for by two to five years of additional power production.
The third assumption, that the product of the steam enthalpy
and the steam flow would be unchanged, is closely related to
the second assumption. It assumes that the amount of heat transferred to the steam is unaffected by raising the final temperature
of the steam. This assumption also seems reasonable.
The fourth assumption in the calculations is that the
amount of steam extracted from the turbine for process heating would remain constant. The provision of process steam is

a critical function of CHP turbine generators. However, we

noted previously that raising the final temperature of steam
in a boiler reduces the total volume of the steam produced. In
condensing turbines, it may be possible to adjust the extraction flows so as to maintain the existing requirements for process steam while reducing the steam flow to the surface condenser. Table XII shows calculations of the steam flows in
the turbines used by recovery boilers B and M as the temperature of the steam delivered by the boiler is raised by 50C and
by 100C above current operating conditions. The total volume of steam produced falls by 6.6% in boiler B and by 7.2%
in boiler M as the temperature is raised by 100C. Fortunately, for both of these boilers, it would be possible to maintain
the flows of medium and low pressure steam from the turbine
by reducing the volume of steam exhausted to the surface
condenser (Table XII).
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

75

CORROSION
The three wood-fueled boilers in this study all use noncondensing, back-pressure turbines. These have no exhaust
flow to make up for the reduced volume of hotter steam that
would be produced if their final steam temperature were increased. The steam flow calculations for these turbines presented in Table XIII are similar to those for the condensing
recovery boiler turbines in Table XII. As with the recovery
boilers, the total volume of steam produced falls as the final
steam temperature is raised. This would reduce the total
steam flow by 6.76% in boiler C, by 6.93% in boiler H, and by
6.94% in boiler M. The steam flows from the various turbine
extraction points would be correspondingly reduced. Although a pulp and paper mill that has implemented substantial energy savings since its non-condensing turbine was installed might be able to operate with these reduced flows of
process steam, this will generally not be possible. However,
three other approaches could be used to maintain the volumes
of process steam while still obtaining the benefits of increased
power generation. Probably the most economic would be to
attemperate the steam, either by a water spray or by passing
some of it through a heat exchanger cooled by boiler feedwa-

ter. Other approaches would be to fire slightly more wood

fuel in the boiler or to purchase a small amount of make-up
steam from another source.
ENVIRONMENTAL BENEFITS
OF HIGHER STEAM CONDITIONS
Increasing the energy efficiency of biomass-fueled boilers
would directly improve energy efficiency in all industries that
use biomass fuels. In addition, improvements in energy efficiency would make renewable fuels more attractive as replacements for fossil fuels.
Table XIV and Table XV show the cumulative savings
resulting from the replacement of natural gas with biomass,
assuming there are at least 500 biomass-fired boilers in the
U.S. that could be converted within a ten-year period to operate with final steam temperatures 100C above current conditions and that the additional energy generated would replace
energy currently produced from fossil fuels. These calculations assume that power production would increase 25% with
a 100C increase in final steam temperature, based on calculations submitted by a project partner [13].

Year 2020

Year 2025

Total primary energy displaced

268 trillion Btus

1,406 trillion Btus

Cumulative economic benefit

$1.5 billion/year

$8.5 billion/year

XIV. Cumulative energy benefits from using additional energy produced in improved biomass-fired boilers to displace energy from
fossil (natural gas) fired boilers. (All currency in USD.)

CO2 emissions displaced (C equivalent)

NOx reductions

Year 2020

Year 2025

5.3 million tons

27.4 million tons

47,000 metric tons

243,000 metric tons

XV. Cumulative environmental benefits from using additional energy produced by advanced biomass-fired boilers to displace
energy from fossil (natural gas) fired boilers.

Total primary energy displaced

Cumulative economic benefit

Year 2020

Year 2025

160 trillion Btus

866 trillion Btus

$774 million/year

$4,300 million/year

XVI. Cumulative energy benefits from replacing natural gas fired boilers with biomass-fired boilers. (All currency in USD.)

CO2 emissions displaced (C equivalent)

NOx reductions

Year 2020

Year 2025

2.31 million tons

12.5 million tons

17,000 metric tons

92,000 metric tons

XVII. Cumulative environmental benefits from replacing fossil fuel (natural gas) fired boilers with biomass-fired boilers.
76

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

CORROSION
Replacing fossil fuel boilers with renewable fuel boilers
would also bring substantial economic benefits. Assuming
that state-of-the-art biomass-fired boilers could economically
replace fossil fuel units, then for 500 installed units with a 2%
growth rate per year, an ultimate potentially accessible
market of 90% and a likely technical market share of 90%,
there could be 21% market penetration in 2020 and 71%
penetration in 2025. The savings shown in Table XVI and
Table XVII are based on the replacement of fossil fuel
(natural gas) fired boilers.
DISCUSSION AND CONCLUSIONS
We noted earlier that biomass-fueled boilers typically operate
with electrical efficiencies of about 35%, substantially lower
than those of coal-fired units [12] because their final steam
conditions are limited in order to restrict fireside corrosion of
the superheater tubes. In Europe, where the development of
power from renewable fuels has been defined as a strategic
priority by many nations, innovative approaches have been
developed to improve the efficiency of biomass-fueled boilers,
not only by identifying alloys with greater corrosion resistance
[2], but also by using combinations of design modifications,
process modifications, and corrosion-resistant alloys to enable
higher steam conditions [1].
The calculations presented here using data from two recovery boilers and three wood-fired boilers suggest that the value
of the increased power that could be generated if the steam in
a biomass-fired boiler could be heated 50C above current operational temperatures would be worth between about
US$1,260,000 and US$5,890,000 per year. If the steam could
be heated 100C above current operational temperatures, the
increased value of the power generated would be between
about US$2,420,000 and US$11,210,000 per year. Gains for
other biomass-fueled boilers would depend on local plant conditions and the price of power. The calculations indicate that
the increased steam value could justify the use of corrosion
resistant alloys costing orders of magnitude more than current
superheater alloys if the use of more corrosion-resistant alloys
were coupled with the introduction of superheater designs
that allowed very rapid superheater replacement.
Most modern wood-fired boilers in Europe also incorporate design improvements to remove corrosive fly ash before
it reaches the superheater, or to remove the superheater from
the flue gas path altogether. Given the likelihood that U.S.
natural gas prices will remain relatively low for many years,
the option of installing an external superheater fired by this
less-corrosive fuel seems a particularly attractive route to
higher superheated steam temperatures. To achieve a fiveyear payback, a boiler redesign and rebuild to achieve a 100C
steam temperature increase would have to cost less than
US$12,000,000 to US$56,000,000. A wood-fired boiler at
Nssj near Jnkping in Sweden, which has its superheater
located in the loop seal, already operates with a steam temperature of 540C, and another boiler with a loop seal superheater at Avedre in the suburbs of Copenhagen, Denmark,

which fires coal with straw containing up to 1% chloride,

already operates with a steam temperature of 580C [1].
Another approach widely used in Europe to achieve higher
superheated steam temperatures in biomass-fueled boilers
without excessive corrosion is the use of combustion additives
to convert (corrosive) potassium chloride into less-corrosive
sulfates or silicates. The increased value of higher steam conditions could offset the cost of such corrosion control additives.
Although the costs and benefits of achieving higher superheated steam conditions in specific biomass-fueled boilers
depend on local conditions, the magnitude of the increased
power that can be generated by increasing superheated steam
temperatures are so substantial that ongoing studies of technical pathways to achieving substantial steam temperature increases (such as the 100C increase studied in the current
project) are likely to pay substantial dividends. The most economic pathways to higher steam temperatures will almost
certainly involve combinations of more corrosion-resistant
alloys and either boiler redesign or the removal of the superheater from the flue gas path. TJ
ACKNOWLEDGEMENTS
This research was sponsored by the U.S. Department of Energy,
Office of Energy Efficiency and Renewable Energy, Advanced
Manufacturing Office, under contract DE-AC05-00OR22725
with UT-Battelle, LLC, administered by James R. Keiser at the
Oak Ridge National Laboratory and supported by in-kind
contributions from the following project partners: bo Akademi
University, Andritz Oy, Babcock & Wilcox, Catalyst Paper,
Chalmers University of Technology, Domtar Corp., FM Global,
FPInnovations, Foster Wheeler, Georgia Institute of Technology,
Haynes International, Howe Sound Pulp & Paper, International
Paper, MeadWestvaco, Metso Power, Outokumpu Stainless,
Rolled Alloys, Sandvik Materials Technology, SharpConsultant,
Southern Company, Special Metals, ThyssenKrupp VDM,
University of Toronto Pulp and Paper Centre, Vattenfall Power,
and the Weyerhaeuser Co. Particular thanks are also due to John
D. Andrews, Jr., (MeadWestvaco) and to D. William Francis
(FPInnovations) for helpful comments.
LITERATURE CITED
1. Sharp, W.B.A., Singbeil, D.L., and Keiser, J.R., Energy from biomass Lessons from European boilers, TAPPI PEERS Conf., TAPPI
PRESS, Atlanta, GA, USA, 2011.
2. Sharp, W.B.A., Singbeil, D.L., and Keiser, J.R., Superheater corrosion produced by biomass fuels, CORROSION/2012 Conf., NACE
International, Houston, 2012, Product No. 51312-01308-SG.
3. Keiser, J.R., Sharp, W.B.A., Singbeil, D.L., et al., TAPPI J. 12(7):
45(2013).
4. Singbeil, D.L., Frederick, L., Keiser, J.R., et al., TAPPI J. 13(8):
37(2014).
5. Keiser, J.R., Sharp, W.B.A., and Singbeil, D.L., TAPPI J. 13(8):
51(2014).

AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

77

CORROSION
6. Heselton, K.E., Efficiency, Boiler Operators Handbook, Fairmont
Press, Lilburn, GA, USA, 2005, pp. 100-105. Available [Online]
http://books.google.com/books?id=8y0qRQl7XDEC&pg=PA101
&lpg=PA101&dq=heat+loss+method+boiler+efficiency&source=
bl&ots=Rdm1iVHNe1&sig=QQX9-ouWVLetjl9fQ8KQjuzaOj4&hl
=en&sa=X&ei=s_PVUPKeFMjxigL6lYD4DQ&sqi=2&ved=0CFIQ6
AEwAg#v=onepage&q=heat%20loss%20method%20boiler%20
efficiency&f=false >[7Aug2014].
7. Adams, T.N., Frederick, W.J., Grace, T.M., et al., Kraft Recovery
Boilers, TAPPI PRESS, Atlanta, 1997, p. 26, Tables 1-8.
8. Gibbs, J.W., A method of geometrical representation of the thermodynamic properties of substances by means of surfaces, Transactions
of the Connecticut Academy II, December 1873, pp. 382-404.
9. Ulrich, G.D., A Guide to Chemical Engineering Process Design and
Economics, John Wiley & Sons, New York, 1984, pp. 87, 90-91.
10. Andrews, J.D., MWV Corp., Charleston, SC, USA, personal communication, 12 Sept. 2011.
11. Stlenheim, A. and Henderson, P., Materials for higher steam
temperatures (up to 600C) in biomass and waste fired plant a
review of present knowledge, Report M08-833, Varmeforsk
Service AB, Stockholm, Sweden, 2011.
12. Kamuk, B., Technical, economical, operating consequences by
operating at extreme steam parameters, Waste-to-Energy Research
and Technology Council Annual Meeting, 2010. Available [Online]
http://wtert.eu/Default.asp?Menue=26&Bereich=&SubBereich=&K
W=&ArtikelPPV=18598&AnbieterID=578#>[7Aug2014].
13. Duo, W.L., FPInnovations, Vancouver, BC, Canada, personal communication, 27 March 2008.

ABOUT THE AUTHORS

Efforts to use renewable fuels have
been hindered by the inefficiency of
biomass combustion, which arises
because biomass fuels corrode hot
boiler tubes. As energy prices rise,
there will come a time when it
makes financial, as well as environmental, sense to learn how to burn
renewable biofuels efficiently.
Sharp
This third part of the research in
this issue of TAPPI Journal (see also
[4-5]) examined the costs and
benefits of operating biomass-fueled boilers more
efficiently. Our conclusions here are consistent with
limited data available from European sources (e.g.,
Kamuk [12]).
In this study, it was difficult to obtain real steam
system data, but a few industrial partners provided
data sets. In addition, the thermodynamic modeling
was quite complex.
It was surprising and gratifying to find that even
with todays low energy prices, high efficiency
burning of biomass fuels seems an economically
attractive option. Our results indicate that installing
high efficiency biomass boilers and turbines designed

78

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

Frederick

Keiser

Singbeil

for high temperatures is a viable option for mills that

are rebuilding their biomass boilers.
Our next step is to gather long-term data on the
corrosion resistance of attractive alloy materials
installed in operating biomass-fueled boilers.
Sharp is principal at SharpConsultant in Columbia, MD,
USA. Frederick (W.J.) is principal at Table Mountain
Consulting, Golden, CO, USA. Keiser is distinguished
research and development staff member, Materials
Science and Technology Div., at Oak Ridge National
Laboratory, Oak Ridge, TN, USA. Singbeil is principal
scientist with FPInnovations, Vancouver, BC, Canada.
Email Sharp at [email protected].

TThe
h New 8-Series Forklifts.
Our new 8-Series Forklifts are built tough for the paper industry with an engine protection
systtem, fully stamped and thicker steel side plates, and load capacities from 8,000 to 15,500 lbs.

toy
oyootaforklift.com/paper for more information.

toyotaforklift.com/paper
800-226-0009
The new engine results in fuel savings of up to

PEER-REVIEWED

PULPING

Combustion properties of
reduced-lignin black liquors
NIKLAS VH-SAVO, NIKOLAI DEMARTINI, RUFUS ZIESIG, PER TOMANI, HANS THELIANDER,
ERKKI VLIMKI, and MIKKO HUPA

ABSTRACT: The growing interest in production of green chemicals and biofuels from biomass provides an
incentive for pulp mills to identify new possibilities in recovering more wood components from the pulping process.
One possibility is to use lignin, separated from black liquor. We undertook this work to determine the combustion
properties of reduced-lignin black liquorstwo kraft liquors and one soda liquorin a laboratory-scale, singleparticle furnace. The combustion times, maximum swollen volume, nitric oxide formation, cyanate formation, and
sulfur release were measured for the original liquors, the filtrates, and intermediate levels of lignin reduction.
Combustion experiments were conducted at 900C in 10% oxygen. Cyanate formation experiments were carried out
by pyrolyzing the droplets at 800C in 100% nitrogen to form a char. The chars were then gasified at 800C in a 13%
carbon dioxide/87% nitrogen atmosphere to obtain the smelt. Sulfur release was studied by pyrolyzing the samples
at temperatures ranging from 300C to 900C. Liquors with the lowest lignin content had a smaller maximum
swollen volume than the original sample. The devolatilization time was not affected by the lignin removal to any
great extent, but lignin removal did have a clear effect on the char burning time. The amount of formed nitric oxide
(g N/kg black liquor solids) remained constant or decreased slightly with increasing lignin removal in the kraft liquor
samples, while for the soda samples the amount of nitric oxide formed increased. The amount of cyanate decreased
clearly when comparing the samples with lowest lignin content to the original liquor samples. The peak sulfur
release occurred at 500C for both kraft liquors. In almost all experiments, the share of sulfur released was highest
for the original samples and lowest for the sample with lowest lignin content. These results provide new data on
combustion properties for reduced-lignin black liquors and indicate that for lignin removal levels up to about 20%,
no significant changes are expected in the combustion behavior.

Application: This work will help mills identify the effect of lignin precipitation on combustion properties of

black liquor.

he pulp and paper industry is identifying possibilities in

new value-added products from biomass to increase revenues. One possibility is the use of lignin that could be applied
to various applications in the chemical industry [1-3]. The removal of lignin is also an option to increase the pulping capacity, if the recovery boiler is the bottleneck of the pulp mill.
The separated lignin can be used in the pulp mill to replace
fossil fuels used in the lime kiln [4,5] as a first step until new
markets open up for lignin as a biomaterial.
Up to 70% of lignin from low dry solids (<30% dry solids)
kraft black liquor can be separated by lowering the pH of the
black liquor with carbon dioxide (CO2) to pH 10 [6-8]. One
such process based on this concept is the LignoBoost process
(Fig. 1). Carbon dioxide is added to a stream of black liquor
(30-45 wt% dry solids) taken from the evaporation plant. The
precipitated lignin is then filtered off and the filtrate is pumped
back to the evaporation plant. The precipitated lignin is dispersed in a solution acidified with sulfuric acid (H2SO4) and
the lignin is filtered off a second time. The filter cake is washed
with warm water acidified with H2SO4; part of the filtrate is
pumped back to the early stages of evaporation and part is
used to disperse lignin from the first filtration. The amount of
lignin removed from the black liquor sent to the process is

1. General layout of the LignoBoost lignin removal process.

approximately 70%. By treating a larger stream of black liquor,

more filtrate is produced and then blended back in with the
liquor to the recovery boiler. This results in as-fired black liquor with lower lignin content.
Lignin is 30%-45% of the dry solids in black liquor [9].
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

81

PULPING
Hardwood and softwood kraft lignins have higher heating
values (i.e., 23-27 MJ/kg), while kraft black liquor has a much
lower high heating value (i.e., 13-15 MJ/kg) [10,11]. Therefore,
the removal of lignin from black liquor will decrease the high
heating value of black liquor. On the basis of the changes in
black liquor composition with varying levels of lignin removal, mathematical calculations estimate that up to 20% lignin
removal is possible without any considerable changes in recovery boiler operation. Also, the estimated minimum load at
which the recovery boiler can operate without support fuel
is 40%-50% lignin removal [11,12].
Recovery boiler NOx emissions come from nitrogen (N) in
the black liquor [13]. The N content of lignin is higher than
black liquor so black liquor N will be reduced with lignin reduction [10,11]. Thus, one of the questions is whether lignin
removal will affect nitric oxide (NO) emissions from black
liquor. Lignin is the primary source of organic N in the char
[14] and may affect char N more than pyrolysis N. Char N
forms the inorganic compound cyanate during char conversion [15,16]; for this reason cyanate formation was studied for
different levels of lignin removal.
Roughly 65% of the sulfur (S) in black liquor is present as
inorganic compounds (i.e., sulfide, thiosulfate, sulfite, and
sulfide) [9,17] and the remaining 35% as a wide variety of organic bound sulfur compounds [18]. With lignin removal, a
fraction of the organic bound sulfur is removed from black
liquor, meaning that the ratio between inorganic and organic
sulfur will be changed.
The S content of the reduced-lignin black liquor increases
slightly with increased lignin removal [11,12] because the removed lignin has lower S content than the original black liquor from which it is precipitated. The partial recycling of
used acidified solution to black liquor evaporation will further
increase the total S content of the black liquor and the share
of inorganic S in black liquor.
The S in black liquor is released during pyrolysis mainly
as methyl mercaptans, dimethly sulfide, dimethly disulfide,
or hydrogen sulfide (H2S) [19-22]. The main contributors to
volatile sulfur compound formation are sulfide, thiosulfate,
and organic sulfur compounds found in black liquor, while
sulfite and sulfate do not result in volatile sulfur formation
[23,24]. Lignin methoxyl groups reacting with HS- have been
hypothesized as a route to mercaptan formation [20,25].
Soda
Lignin Content
mg / g d.s.

Kjeldahl
N (wt-% d.s.)

Black liquor

391

0.071%

8.4% lower lignin

358

0.066%

30% lower lignin

274

0.063%

I. Lignin and nitrogen content of the soda liquor samples.

82

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

Thus, the impact of lignin removal on S release during the

devolatilization stage of black liquor combustion was of interest in this work.
A single-particle reactor was used to determine the impact
of lignin removal on the combustion properties of black liquor. The properties studied were maximum swollen volume,
combustion times, NO formation during combustion, amount
of cyanate in smelt, and release of S during pyrolysis. Intermediate levels of lignin reduction were obtained by mixing the
weak black liquors and filtrates.
EXPERIMENTAL
The lignin removal of the soda liquor from hardwood pulping
was conducted at Chalmers University of Technology, and the
removal of lignin from kraft black liquors from softwood and
eucalyptus pulping was conducted at Innventia AB. One original soda liquor sample and two samples with levels of lignin
reduction were obtained from Metso Power Oy. The soda liquor was generated in a laboratory cook. The received soda
samples had dry solids content of <41% and were dried to
>74% dry solids before the tests in the single-particle reactor.
The lignin and N contents of the soda liquor samples are given
in Table I. The lignin content of soda liquors was determined
by UV spectrophotometry, and the total N content of samples
was analyzed with a modified Kjeldahl method using the Devarda alloy (CAS No. 8049-11-4) based on Finnish Standards
Association (SFS) 5505 Wastewater inorganic and organic
nitrogen determination.
Both the softwood and hardwood (Eucalyptus globulus)
kraft black liquors used in this study were processed in
bench-scale equipment, similar to what has been used earlier
[6]. The process was operated as follows. Industrial kraft
black liquors (~40% DS) were used and the pH was lowered
to about 10 by addition of CO2 (g), causing the lignin to precipitate (Fig. 2). The precipitated lignin was separated by
filtration (chamber press filter), referred to as filtration 1. The
lignin filter cake was then resuspended at pH 2-3, using sulfuric acid solution. The lignin slurry was filtered once more
(filtration 2) and the separated lignin cake was washed with
acidified water at pH 2-3. After washing, the lignin cake was
dewatered by pressing and by blowing compressed air
through the filter cake. All filtrates were collected and mixed.
These mixtures were used to represent the filtrate mixture
that would be obtained in a full-scale LignoBoost plant.
Two kraft liquor samples (one softwood and one eucalyptus) before lignin removal and two filtrates were obtained.
The softwood filtrate had 54.65% lower lignin content than
the softwood black liquor and the eucalyptus filtrate had
43.65% lower lignin content than the eucalyptus black liquor. The received samples had dry solids content of <25%.
For both black liquors, the original sample and filtrate were
mixed to obtain two more samples with different levels of
lignin content. The lignin, S, and N contents of the obtained
and mixed kraft liquor samples are shown in Table II. The
samples were dried in an oven at 105C to a dry solid content

PULPING

2. Schematic figure of the bench-scale equipment used for lignin removal.

of approximately 85% before the tests in the single-particle

reactor.
The lignin content shown in Table II for the kraft liquors
was analyzed using TAPPI T 222 om-02 Gravimetric determination of acid-insoluble lignin and UV spectrophotometry
(acid soluble lignin). The total N content of selected samples
was analyzed with a modified Kjeldahl method using the Devarda alloy (CAS No. 8049-11-4) based on SFS 5505; the S content was analyzed using SCAN-N 38:10 Acid-soluble metals.
Combustion, cyanate formation, and S release experiments
were conducted in a laboratory-scale quartz glass reactor as
illustrated in Fig. 3 [26]. The temperature and the atmosphere in the quartz glass reactor can be adjusted. Six droplets
weighing 100.5 mg of each sample were suspended on platinum hooks and combusted at 900C in a 10% O2/90% N2 atmosphere. The formed NO emissions were measured with a
Model 200EM chemilumenescence analyzer (Teledyne Technologies; Thousand Oaks, CA, USA). The combustion was recorded with a video camera to determine the maximum swollen volume and the combustion times.
The maximum swollen volume of the droplet is estimated
by capturing an image from the recorded video. The captured
two-dimensional image is compared to an area of a circle or

Lignin Content
mg/ g d.s.
Black liquor

Softwood
Sulfur Content
wt-% d.s.

3. Schematic drawing of the quartz glass reactor used in the

experiments.

Kjeldahl
N (wt-% d.s.)

Lignin Content
mg / g d.s.

Eucalyptus
Sulfur Content
wt-% d.s.

Kjeldahl
N (wt-% d.s.)

366

6.11%

0.062%

429

5.10%

0.086%

10% lower lignin

329.4*

6.56%*

0.060%*

386.1*

5.76%*

0.082%*

25% lower lignin

274.5*

7.25%*

0.057%*

312.75*

6.76%*

0.077%*

166

8.58%

0.052%

242

7.99%

0.070%

Filtrate

* Calculated based on the analysis of black liquor and filtrate.

II. Lignin and sulfur content of the kraft black liquor samples before evaporation.
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

83

PULPING
an ellipse, which is then used to calculate the volume of a corresponding sphere or ellipsoid. In this work, the maximum
swollen volume is represented as cm3/g of dry solids.
The combustion of black liquor droplets can be divided
into four stages: (1) drying, (2) devolatilization, (3) char burning, and (4) smelt coalescence [27]. The times for the different
combustion stages were determined from the recorded video.
The drying time was not determined in these experiments,
because the sample dried before it entered the combustion
chamber. Devolatilization time is defined as the time from the
appearance of the flame to the disappearance of the flame.
The droplets often ignited as they entered the combustion
zone; thus, pyrolysis times can be slightly longer (by about 0.5
s) than what is observed. The char burning time is defined as
the time from the point where the visible flame disappears to
the point where smelt coalescence occurs.
The cyanate formation in the samples was determined by
pyrolyzing six droplets (100.5 mg) of each sample in the
quartz glass reactor at 800C in 100% N2 for 10 s to form a char.
The chars were then gasified at 800C in 13% CO2/87% N2 to
obtain smelts. The formed smelts were dissolved in ion exchange water using a ratio of 1.25 mg smelt:1 mL ion exchange
water and filtered. The filtrates were analyzed for cyanate
with ion chromatography using a Metrosep Anion Dual 2 column and conductivity detector (Metrohm AG; Herisau, Switzerland) [28].
The S release for the kraft liquor samples was measured in
the quartz glass reactor by pyrolyzing six droplets (100.5 mg)
of each sample at 300C, 500C, 700C, and 900C in 100%
N2 atmosphere. After the quartz glass reactor a flow of pure
oxygen was added to the flue gas. The oxygen content in the
flue gas was around 45 vol% and was measured with a Servopro 4900 continuous emissions analyzer (Servomex; Crowborough, UK). The flue gas flow containing oxygen was then
introduced to an SCC-K catalytic converter (ABB; Zurich, Switzerland) to oxidize released sulfur species to sulfur dioxide
(SO2). The released S was measured as SO2 with an ABB
AO2020 infrared analyzer. The total amount of S released was
calculated based on the SO2 measurement and the total flue
gas flow before dilution with oxygen [26].
RESULTS AND DISCUSSION

Combustion characteristics of reduced-lignin

black liquors
The combustion times; maximum swollen volume; and distribution between volatile matter, fixed carbon, and smelt for
the softwood kraft liquor samples is shown in Fig. 4 and for
the eucalyptus samples in Fig. 5. The devolatilization time
was around 1 s for all of the samples, and no clear difference
between devolatilization time and the level of lignin content
in the samples was detected. This was consistent with the idea
that the dissolved hemicelluloses are the main source of volatile products in black liquor combustion [14]. For the softwood samples the char burning time and total combustion
time decreased with increasing lignin removal, while the op84

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

4. Combustion times (top) and maximum swollen volume

(middle) from experiments conducted at 900C at 10%
oxygen/90% nitrogen. Distribution of volatile matter, fixed
carbon, and smelt (bottom) calculated based in weighed
initial samples, chars, and smelts from the cyanate formation
experiments of softwood kraft liquor samples. Error bars
represent one standard deviation of six droplets.

PULPING

6. Maximum swollen volume images of the kraft liquor samples.

5. Combustion times (top) and maximum swollen volume

(middle) from experiments conducted at 900C at 10%
oxygen/90% nitrogen. Distribution of volatile matter, fixed
carbon, and smelt (bottom) calculated based in weighed
initial samples, chars, and smelts from the cyanate formation
experiments of eucalyptus kraft liquor samples. Error bars
represent one standard deviation of six droplets.

posite was observed for the eucalyptus sample. This reduced

char burning time was most likely due to the combined effects
of slightly increased swelling and the reduced organics in the
black liquor, specifically the lignin that will contribute to char
formation. Increased lignin removal decreased the share of
fixed carbon and increased the share of smelt as seen in Figs.
4 and 5. The eucalyptus black liquor had considerably larger
maximum swollen volume than the softwood black liquor.
The swelling for the eucalyptus black liquor samples decreased with increasing lignin reduction, which may be the
reason for slightly higher average char burning times despite
lower organic content. Aln et al. [29] showed that hardwood
lignin swelled more than softwood lignin because the presence of xylan increases swelling. The ratio between hemicelluloses and lignin affects the swelling, and at a certain ratio
maximum swelling occurs. This could be one reason for the
difference in the average maximum swollen volume pattern
for the eucalyptus and softwood samples. In both cases, the
filtrate swelled considerably less than the original sample. The
high standard deviations seen in the maximum swollen volume can be explained by nonuniform swelling of black liquor
and use of two-dimensional images for estimating volume.
Figure 6 shows captured maximum swollen volume images
for different kraft black liquor samples.
The combustion times; maximum swollen volume; and
distribution between volatile matter, fixed carbon, and smelt
of soda liquor with different lignin content are shown in
Fig. 7; the capture maximum swollen images are shown in
Fig. 8.
The devolatilization time and the share of volatile matter
slightly increased with increasing lignin removal. The char
burning time and total combustion time increased and the
maximum swollen volume decreased with increasing lignin
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

85

PULPING

8. Maximum swollen volume images of the soda liquor samples.

removal for the soda liquor samples. The amount of smelt increased with increasing lignin removal, while the amount of
fixed carbon decreased. The char burning times clearly depend on the maximum swollen volume and the amount of
fixed carbon left in the char.

NO formation

7. Combustion times (top) and maximum swollen volume

(middle) from experiments conducted at 900C at 10%
oxygen/90% nitrogen. Distribution of volatile matter, fixed
carbon, and smelt (bottom) calculated based in weighed
initial samples, chars, and smelts from the cyanate formation
experiments of soda liquor samples. Error bars represent one
standard deviation of six droplets.
86

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

The NO formation (as g N/kg dry black liquor) for the eucalyptus and softwood samples are shown in Fig. 9. For both
black liquors, the NO formation decreased with increasing
lignin removal. Part of the black liquor N is removed with lignin, which lowers the N content of the lignin-reduced black
liquor compared with the original black liquor samples shown
in Table II. The conversion of fuel N to NO was roughly 40%
for the eucalyptus samples and roughly 45% for the softwood
samples. This includes NO formed during pyrolysis and NO
from char burning. In a recovery boiler, much of the char N
will exit the recovery boiler as cyanate in the smelt.
The NO formation (as g N/kg dry black liquor) slightly increased for the soda liquor with 30% lower lignin content
compared to the original soda liquor, as shown in Fig. 10.
An equal amount of NO was formed from the 8.4% lower lignin sample and original sample. In the soda liquor samples, N
content also decreased with increasing lignin removal, while
the amount of formed NO increase remains unknown. The
conversion of fuel N to NO was roughly 45% for the original

PULPING

9. Average nitric oxide formation of softwood and eucalyptus

kraft liquor samples. Experiments conducted at 900C at
10% oxygen/90% nitrogen. Error bars represent one standard
deviation of six droplets.

10. Average nitric oxide formation of soda liquor samples.

Experiments conducted at 900C at 10% oxygen/90% nitrogen.
Error bars represent one standard deviation of six droplets.

soda sample and increased to approximately 55% for the 30%

lower lignin sample.

Cyanate of reduced-lignin black liquors

Cyanate formation (as mg N/kg smelt) for the softwood and
eucalyptus samples is shown in Fig. 11. The amount of
formed cyanate from the 10% lower lignin eucalyptus sample
slightly increased when compared with the original sample,
while the amount of cyanate formed from the 10% lower lignin softwood sample and original sample are equal. In both
kraft liquors, less cyanate was formed from the 25% lower
lignin and filtrate samples compared with the original black
liquor samples. The fuel N to cyanate conversion for the eucalyptus samples was between 40% and 50%, and the lowest

11. Cyanate formation after pyrolysis (10 s at 800C at 100%

nitrogen) and gasification (800C in 13% carbon dioxide/87%
nitrogen) to smelt for eucalyptus and softwood kraft liquor
samples. Error bars represent one standard deviation of six ion
chromatography analyses.

12. Cyanate formation after pyrolysis (10 s at 800C at 100%

nitrogen) and gasification (800C in 13% carbon dioxide/87%
nitrogen) to smelt for soda liquor samples. Error bars represent
one standard deviation of six ion chromatography analyses.

conversion occurred for the original black liquor sample.

Roughly 47% of the softwood filtrates N was converted to
cyanate. For the rest of the softwood samples, the fuel N to
cyanate conversion was between 57% and 62%. It has been
shown that reaction between alkali salts (i.e., sodium and potassium carbonate) and the char N results in cyanate formation [30]. The reason for the increase in cyanate formation
from the eucalyptus liquor sample with a 10% lower lignin
content is not clear. The ratio between inorganic and organic
compounds in the fuel and the amount of inorganic compounds in the char or simply a higher char N content may be
possible explanations, but further work would be needed to
clarify this question.
Figure 12 shows the cyanate formation (as mg N/kg
smelt) for the soda liquor samples. The amount of formed
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

87

PULPING

13. Sulfur release of softwood black liquor samples during

pyrolysis (100% nitrogen) at different temperatures. Error bars
represent one standard deviation of six droplets.

14. Sulfur release of eucalyptus black liquor samples during

pyrolysis (100% nitrogen) at different temperatures. Error bars
represent one standard deviation of six droplets.

cyanate slightly increased for the 8.4% lower lignin sample

when compared with the original sample. Almost no cyanate
was formed from the 30% lower lignin sample. One reason
behind the low cyanate formation is that soda liquor samples
did not form a smelt at these conditions. The remaining residue still looked partially swollen and thus the char was not
fully converted to smelt. Previous studies show that when
gasification is conducted at 800C, maximum conversion occurs when almost 100% of the char is converted [16]. The
levels of cyanate formed represent 14% of the fuel N converted to cyanate for the black liquor and the 8.4% lower lignin
sample and only 2% for the 30% lower lignin sample.

precipitated lignin, along with the recycling of H2SO4 acidified

washing solution back to the black liquor, affects the distribution of sulfur species found in black liquor. Previous studies
have shown that at most only a very small fraction of the SO42results in S release during black liquor pyrolysis [23,24].
Therefore, the share of S released from lignin-reduced liquors
was expected to be lower compared with the original black
liquor before lignin removal. While this was observed for the
eucalyptus liquors (Fig. 14), this was not the case for the softwood black liquor samples with 10% or 25% lower lignin
where S release on black liquor dry solids was the same or
increased slightly.

Sulfur release of reduced-lignin black liquors

CONCLUSIONS
This research gives new information regarding the combustion properties of reduced-lignin black liquors. The removal
of lignin decreased the N content of all of the liquor samples.
The use of acidified warm-water H2SO4 solution in washing
of lignin increased the sulfur content of kraft liquor samples.
Lignin removal also reduced the organic content of the black
liquor and can affect both swelling and char burning. In all
samples, the char burning time was affected by lignin removal, while no clear difference could be seen in the devolatilization time. These changes in combustion time were small and
will not result in any observable change in char burnout in a
recovery boiler. The changes in maximum swollen volume
were fairly significant and could slightly affect carryover, but
sheet breakup and droplet formation from the liquor spray
need to be studied before any predictions can be made about
the effect of swelling changes on carryover.
The NO formation was not greatly affected by lignin removal, and NO emissions from the recovery boiler on a dry
solids basis are not expected to be significantly affected by
lignin removal. However, on a per ton of pulp basis, NO emis-

Sulfur release during pyrolysis was measured at different temperatures for the softwood samples (Fig. 13) and the eucalyptus samples (Fig. 14). Softwood and eucalyptus samples
showed S release peaks at 500C and decreases with increasing temperature. This has also been observed in previous
studies [20,21]. For the softwood black liquor and the 10% and
25% lower lignin samples, the share of S released was quite
similar when the pyrolysis was conducted at 500C, 700C,
and 900C. However, at 300C, the share of S released was
clearly higher for the 10% and 25% lower lignin samples than
for the original black liquor sample. The share of S released
from the softwood filtrate sample was clearly lowest of all the
samples when the pyrolysis was conducted at 500C and
700C. For the eucalyptus samples, the share of S released was
highest for the black liquor sample in all of the pyrolysis experiments and lowest for the filtrate sample, except when the
pyrolysis was conducted at 900C. For the S release experiments conducted for the eucalyptus samples, the share of S
released was lower for the liquors with lower lignin content.
The removal of a portion of the organic sulfur with the
88

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

PULPING
sions should go down as the flow of as-fired black liquor solids/ton of pulp will decrease. The concentration of cyanate in
the smelt will likely be quite similar to the original black liquor, at least at 10% lignin reduction. The lower furnace temperature and air distribution are expected to influence cyanate decomposition; therefore, lignin reduction also may
indirectly influence the amount of cyanate exiting the boiler
with the smelt.
Assuming the temperature at which devolatilization happens in the recovery boiler is the same, S release from reduced
lignin black liquor is expected to be the same or slightly lower
than for the original black liquor. This clearly can be seen at
all temperatures for the eucalyptus black liquor. For the softwood liquor this is less clear, but when looking at S release at
900C for the softwood black liquor, there is little difference
with lignin reduction. Older boilers with lower temperatures
at the gun level and char bed will need to adjust firing with
reduced-lignin black liquor to keep up the temperature; otherwise, S emissions could increase due to a cooler lower furnace, which can result in higher S release from black liquor.
The influence of lignin removal on S release should be studied
in more detail to better understand what role, if any, lignin
has on the formation of volatile sulfur compounds.
On the basis of this work, the combustion of reduced-lignin
black liquors in the recovery boiler furnace is possible without
any major changes. Mill trials with reduced black liquor firing
would provide more details on variables such as sheet formation and droplet formation during spraying. TJ
ACKNOWLEDGEMENTS
The authors acknowledge the Graduate School in Chemical
Engineering (GSCE) for the scholarship supporting Niklas
Vh-Savo. This work has been carried out within the Future
fuels for Sustainable Energy Conversion (FUSEC) project
(20112014) as part of the activities of the bo Akademi Process Chemistry Centre. Support from the National Technology
Agency of Finland (Tekes), Andritz Oy, Metso Power Oy, Foster Wheeler Energia Oy, UPM-Kymmene Oyj, Clyde Bergemann GmbH, International Paper Inc., and Top Analytica Oy
Ab is gratefully acknowledged.
LITERATURE CITED
1. Kadla, J.F., Kubo, S., Venditti, R.A., et al., Carbon 40(15):
2913(2002).
2. Borges da Silva, E.A., Zabkova, M., Araujo, J.D., et al., Chem. Eng.
Res. Des. 87(9): 1276(2009).
3. Gosselink, R., Lignin as a renewable aromatic resource for
the chemical industry, Ph.D. thesis, Wageningen University,
Wageningen, Netherlands, 2011.
4. Richardson, B. and Uloth, V.C., Tappi J. 73(10): 191(1990).
5. Richardson, B., Watkinson, A.P., and Barr, P.V., Tappi J. 73(12):
133(1990).
6. hman, F., Wallmo, H., and Theliander, H., Nord. Pulp Pap. Res. J.
22(1): 9(2007).

7. Wallmo, H., Wimby, M., and Larsson, A., Increase production in

your recovery boiler with LignoBoostextract valuable lignin for
biorefinery production and replacement of fossil fuels, TAPPI Eng.,
Pulping Environ. Conf., TAPPI PRESS, Atlanta, GA, USA, 2009.
8. Tomani, P., Cellul. Chem. Technol. 44(13): 53(2010).
9. Adams, T.N., in Kraft Recovery Boilers, TAPPI PRESS, Atlanta, 1997,
p. 381.
10. Blunk, S.L., Jenkins, B.M., and Kadam, K.L., Combustion properties of lignin residue from lignocellulose fermentation, National
Renewable Energy Laboratory, Golden, CO, USA, 2000.
11. Vakkilainen, E. and Vlimki, E., Effect of lignin separation to
black liquor and recovery boiler operation, TAPPI Eng., Pulping
Environ. Conf., TAPPI PRESS, Atlanta, 2009.
12. Vlimki, E., Niemi, P., and Haaga, K., Int. Chem. Recovery Conf.,
TAPPI PRESS, Atlanta, 2010, p. 148.
13. Forssen, H.M., Hupa, M., and Hellstrom, P., TAPPI J. 82(3):
221(1999).
14. Alen, R., Bioresour. Technol. 49(2): 99(1994).
15. Kymalainen, M., Forssen, M., Jansson, M., et al., J. Pulp Pap. Sci.
28(5): 151(2002).
16. Vh-Savo, N., DeMartini, N., Hupa, M., Energy Fuels 25(11):
4944(2011).
17. Forssen, M., Frederick, W.J., Hupa, M., et al., Proc. For. Prod.
Symp., TAPPI PRESS, Atlanta, 1992, p. 11.
18. Niemel, K., Low-molecular-weight organic compounds in birch
kraft black liquor, Ph.D. thesis, Helsinki University of Technology,
Helskinki, Finland, 1990.
19. Feuerstein, D.L., Thomas, J.F., and Brink, D.L., Tappi 50(6):
258(1997).
20. McKeough, P., Kurkela, M., Arpiainen, V., et al., Int. Chem. Recovery
Conf., TAPPI PRESS, Atlanta, 1995, p. A217.
21. Sricharoenchaikul, V., Frederick, W.J., and Grace, T.M., J. Pulp Pap.
Sci. 23(8): J394(1997).
22. Vainio, E., Brink, A., Demartini, N., et al., J. Pulp Pap. Sci. 36(34):
135(2011).
23. Strohbeen, D.T. and Grace, T.M., Tappi J. 65(10): 125(1982).
24. Harper, F.D., Sulfur release during the pyrolysis of kraft black
liquor (combustion), Ph.D. dissertation, Georgia Institute of
Technology, Atlanta, Georgia, 1989.
25. McKeough, P., Kurkela, M., Kyllnen, H., et al., in LIEKKI 2 Technical
Review 19931998 (M. Hupa and J. Matinlinna, Eds.), bo Akademi
University, Turku, Finland, 1998, p. 855.
26. Bjrkvik, T., Measurement of sulfur release from black liquor
droplets in a single particle reactor, M.S. thesis, bo Akademi
University, Turku, Finland, 2010.
27. Hupa, M., Solin, P., and Hyoty, P., J. Pulp Pap. Sci. 13(2): J67(1987).
28. DeMartini, N., Determination of cyanate in kraft green liquor,
M.S. thesis, bo Akademi University, Turku, Finland, 2001.
29. Alen, R., Hupa, M., and Noopila, T., Holzforschung 46(4): 337(1992).
30. Vh-Savo, N., DeMartini, N., and Hupa, M., Energy Fuels 25(11):
7108(2013).
AUGUST 2014 | VOL. 13 NO. 8 | TAPPI JOURNAL

89

PULPING
ABOUT THE AUTHORS
Lignin removal from black liquor and lignin usage
are going to increase in the near future.
Understanding how reduced-lignin black liquor
burns is important for smooth implementation of
this technology.
We have a long history at bo Akademi
University of characterizing the combustion
properties of different black liquors. Our earlier work
provided a platform for characterizing reducedlignin black liquor. This work used established
techniques but raised some interesting questions,
such as why the nitrogen chemistry doesnt change
more with lignin reduction. The sulfur release was
also higher than expected for the reduced-lignin
black liquors.
Determining the sulfur release was the most
difficult part of this work. We were most surprised
by the comparatively high sulfur release with the
reduced-lignin black liquors. These results do not
conform to what was expected based on our current
understanding of sulfur release.
In implementing a new technology in the
chemical recovery cycle, mills need to know how the
process will affect the rest of the cycle. How the
black liquor burns is one important piece of
information. This work indicates that from a
combustion standpoint, reduced-lignin black liquors
with 10%20% removal should behave similarly to
normal black liquor. However, mills will need to think

Vh-Savo

DeMartini

Theliander

90

TAPPI JOURNAL | VOL. 13 NO. 8 | AUGUST 2014

about how they spray the black liquor and air

distribution. Reduced-lignin black liquor has less
char carbon, which will affect the char bed and
reduce efficiency and potentially sulfur dioxide
emissions if the lower furnace temperature drops
significantly.
Short of a mill trial, perhaps the most beneficial
next step would be computational fluid dynamics
modeling of the carbon/air distribution when firing
reduced-lignin black liquor. This would give mills
some indication of how the firing conditions should
be adjusted, including spraying and air distribution.
Sulfur release and additional experiments with
the nitrogen chemistry are interesting topics for
further work.
Vh-Savo is a research associate and doctoral student
and DeMartini is a senior researcher, Laboratory
of Inorganic ChemistryCombustion and Material
Chemistry, bo Akademi University, Turku, Finland.
Ziesig is a research engineer, biorefinery processes
and products, and Tomani is team leader, lignin and
carbon fibers, at Innventia AB, Stockholm, Sweden.
Theliander is a professor, Forest Products and Chemical
Engineering, at Chalmers University of Technology,
Gothenburg, Sweden. Vlimki is R&D project manager,
pulp and energy business line, Valmet Power Oy,
Tampere, Finland. Hupa is a professor, Laboratory
of Inorganic ChemistryCombustion and Material
Chemistry, bo Akademi University, Turku, Finland.
Email Vh-Savo at [email protected].

Ziesig

Vlimki

Tomani

Hupa

China Pulp & Paper

Contents
and associations around the
TAPPI works with organizations
Peer Reviewed Papers

China Pulp & Paper

11 1967/T

MONTHLY

2014
world to keep members informed regarding global research in the
Vol.33 No.7
Effect
of
Novel
filler
Fly
Ash
Based
Calcium
Silicate(FACS)
on
Sizing
pulp, paper and related industries. Through a cooperation with
July. 2014
Efficiency
and theanCorresponding
Mechanism
China Pulp & Paper, we
are providing
updated list of
published
Published
jointly
by:
HAO
Ning
ZHANG
Mei-yun
WU
Pan
et
al
1
papers from the associations journal to offer TAPPI members a
China
Technical
Association
of Paper
Application
of
Ground
Calcium
Carbonate
Covered
with
Starch
Sodium
Starch
perspective on research in this rapidly growing market.

Industry(CTAPI)

Sodium Stearate Complex as Paper Filler

FAN Hui-ming XU Bin-feng LIU Jian-an et al 7 China National Pulp and Paper
Study of Using Blended Calcium Carbonate and Nano-4A Zeolite Adsorbed Research Institute(CNPPRI)
C Oas
NFiller
T E HE
NTS
Ca2+
Yue-feng SUN Hai-li FANG Fei et al 12 Address: Building 2, 4 Yard, Qiyang
13

J U LY 2 01 4

Road, Chaoyang District,

A New Type of Light-induced Yellowing Prohibitor-Synthesis and Application
PEER REVIEWED PAPERS
Beijing, China, 100102
ZHANG Guang-hua XIANG Rui WANG Peng et al 17
Performance
Modified Collagen
Emulsion
Telephone:86-10-64778158
Effect of Novel Filler Fly Sizing
Ash Based
Calcium of
Silicate(FACS)
on Sizing
Efficiency and
the Corresponding Mechanism
HAO Ning ZHANG
Mei-yun WANG
WU Pan Qun
et al.........................................................
ZHANG
Su-feng
WANG Xue-chuan et al 122 Fax:86-10-64778174
Optimization
of Covered
Mulberrywith
Stalk
SCMP
Process
Using
Response Surface http://www.cppmp.com
Application of Ground Calcium
Carbonate
Starch
Sodium
Stearate
Complex

as Paper Filler FAN Hui-ming

XU Bin-fengZHAO
LIU Jian-an etGui-ling
al........................................................................................
7
Methodology
WEN Feng ZHAO Chen et al 26

AnalysisCarbonate
on Dynamic
PaperAdsorbed
Sheet Forming
of the
Study of Using Blended Calcium
andFiltering
Nano-4Aand
Zeolite
Ca2+ as Characteristics
Filler
HE Yue-feng SUN Hai-li FANG
Fei
et
al
...............................................................................................................................
12
Forming Fabric HU Zhi-jun HU Gui-lin LIU Bao-jian et al 31
Articles
A New Type of Light-induced
Yellowing Prohibitor-Synthesis and Application
ZHANG Guang-hua XIANG Development
Rui WANG Pengand
et al Application
............................................................................................................
17 Board
of Heavy Multi-layer Composite Pulp

E-mail:[email protected](Manager)
[email protected](Editor)
[email protected](Advertisement)
Editorial office:
Editor in Chief: Shi-jun Kuang

Felt
Bang-chun WANG Xin-wen WANG Xing-fei 36 Executive Editor in Chief: Ma Xin
Sizing Performance of Modified
CollagenHAN
Emulsion
ZHANG Su-feng WANG QunDesign
WANGand
Xue-chuan
et al.....................................................................................................
Manufacture
of New Type Fabric Using in Drying Section22
of Paper Advertisement:
Manager: Du Rong-rong
JIAO
Cheng-yun 42
Optimization of MulberryMachine
Stalk SCMP
Process Using Response Surface Methodology
ZHAO Gui-ling WEN Feng ZHAO
Chen
et al ......................................................................................................................
26
Subscription:
Repair
Technology
of Yankee Dryer Surface
ZHANG
Feng GAO
Analysis on Dynamic Filtering
and Paper Sheet Forming Characteristics
of Zhi-fu WEI Tao 45 Circulation Department
China Pulp & Paper
the Forming Fabric HU Zhi-jun
HU Gui-lin
LIU Bao-jian et al .................................................................................... 31
Forum

Solutions to the Problems of Air Pollution in Paper Industry

ARTICLES

TONG Xin

ZHANG Zhen-bin

YANG Heng-yu et al

Building 2, 4 Yard, Qiyang

Road, Chaoyang District,

49
Development and Application
of Heavy Multi-layer Composite Pulp Board Felt
Consideration
on the
Development of Highly Filled Paper
HAN Bang-chun WANG Xin-wen
WANG Xing-fei
...........................................................................................................
36
SONG
Shun-xi
ZHANG
Mei-yun
WANG
Jian
56
Design and Manufacture of New Type of Fabric Used in Drying Section of Paper Machine
Discussion on the Symbiosis Model of Forestry-paper Integration
JIAO Cheng-yun .........................................................................................................................................................................
42

Beijing,China,100102

YANG
Jia-meng
ZHANG
Zhi-guang 61
Repair Technology of Yankee
Dryer Surface ZHANG Feng GAO Zhi-fu
WEI Tao
........................................
45

ISSN 0254-508X
Distributed abroad by:

FORUM

China

Production Trials
An Example of Trouble Shooting of Ash Sensor Fault in QCS System

ZHU
Yong
Solutions to the Problems of
Air Pollution in Paper Industry

DAI Le-yu

LIANG Yue-ming 69

Linter Cleaner
Pulping Process XUE Run-lin49
72
TONG Xin ZHANG Zhen-binCotton
YANG Heng-yu
et al ............................................................................................................
Exchangeof Highly Filled Paper
Consideration on the Development
TABLE
OFJian
CONTENS,
TAPPI JOURNAL May 2014(76)
SONG Shun-xi ZHANG Mei-yun
WANG
....................................................................................................................
56

Others

Discussion on the Symbiosis Model of Forestry-paper Integration

News6,16,35,41,44,55,60,68,71,75
YANG Jia-meng ZHANG Zhi-guang
...................................................................................................................................... 61

Price Per Copy US $ 20.00(Air Mail

Charge Included)
CN11-1967/TS

Publication

Corporation

for

Foreign Trade
DK11070
No responsibility is accepted by China
Pulp & Paper for the facts and opinions
expressed in its papers. CTAPI and
CNPPRI reserve copyright for all papers
published in this journal. Published

PRODUCTION TRIALS
An Example of Trouble Shooting of Ash Sensor Fault in QCS System
ZHU Yong DAI Le-yu LIANG Yue-ming................................................................................................................................ 69

Cotton Linter Cleaner Pulping Process XUE Run-lin .............................................................................................. 72

OTHERS
News............................................................................................................................ 6,16,35,41,44,55,60,68,71,75

every month by CTAPI and CNPPRI.

For nearly 90 years, Ashland has been delivering the right people, products and application insights to its customers around
the world. Were the leading supplier of specialty chemicals and support services to the global pulp and paper industry,
y
and our technologgies have a direct imppact on almost 70 percent of the worlds papper tonnagge. See how our longg leggacyy of
meeting customer needs has positioned us to help you operate more eciently and protably.
Experien
encce the Ash
shlland Advantage today.

ashland.com

Registered trademark, Ashland or its subsidiaries, registered in various countries

T ademark, Ashland or its subsidiaries, registered in various countries
Tr
2013, Ashland
AD-12502
TM

With good chemistry great things happen.