IMPLEMENTATION OF NEAR-INFRARED SYSTEMS FOR

MAIN FUELS BLENDING

DEP 32.31.56.30-Gen.
August 2009
(DEP Circular 10/10 has been incorporated)
DESIGN AND ENGINEERING PRACTICE
This document is restricted. Neither the whole nor any part of this document may be disclosed to any third party without the prior written consent of Shell Global
Solutions International B.V., The Netherlands. The copyright of this document is vested in this company. All rights reserved. Neither the whole nor any part of this
document may be reproduced, stored in any retrieval system or transmitted in any form or by any means (electronic, mechanical, reprographic, recording or otherwise)
without the prior written consent of the copyright owner.
 DEP 32.31.56.30-Gen.
August 2009
Page 2
PREFACE
DEPs (Design and Engineering Practice) publications reflect the views, at the time of publication, of:
Shell Global Solutions International B.V. (Shell GSI)
and/or
Shell International Exploration and Production B.V. (SIEP)
and/or
other Shell Service Companies.
They are based on the experience acquired during their involvement with the design, construction, operation and
maintenance of processing units and facilities, and they are supplemented with the experience of Shell Operating Units.
Where appropriate they are based on, or reference is made to, international, regional, national and industry standards.
The objective is to set the recommended standard for good design and engineering practice applied by Shell companies
operating an oil refinery, gas handling installation, chemical plant, oil and gas production facility, or any other such
facility, and thereby to achieve maximum technical and economic benefit from standardization.
The information set forth in these publications is provided to Shell companies for their consideration and decision to
implement. This is of particular importance where DEPs may not cover every requirement or diversity of condition at
each locality. The system of DEPs is expected to be sufficiently flexible to allow individual Operating Units to adapt the
information set forth in DEPs to their own environment and requirements.
When Contractors or Manufacturers/Suppliers use DEPs they shall be solely responsible for the quality of work and the
attainment of the required design and engineering standards. In particular, for those requirements not specifically
covered, the Principal will expect them to follow those design and engineering practices which will achieve the same
level of integrity as reflected in the DEPs. If in doubt, the Contractor or Manufacturer/Supplier shall, without detracting
from his own responsibility, consult the Principal or its technical advisor.
The right to use DEPs is granted by Shell GSI, in most cases under Service Agreements primarily with Shell companies
and other companies receiving technical advice and services from Shell GSI or another Shell Service Company.
Consequently, three categories of users of DEPs can be distinguished:
1) Operating Units having a Service Agreement with Shell GSI or other Shell Service Company. The use of DEPs
by these Operating Units is subject in all respects to the terms and conditions of the relevant Service
Agreement.
2) Other parties who are authorized to use DEPs subject to appropriate contractual arrangements (whether as part
of a Service Agreement or otherwise).
3) Contractors/subcontractors and Manufacturers/Suppliers under a contract with users referred to under 1) or 2)
which requires that tenders for projects, materials supplied or - generally - work performed on behalf of the said
users comply with the relevant standards.
Subject to any particular terms and conditions as may be set forth in specific agreements with users, Shell GSI
disclaims any liability of whatsoever nature for any damage (including injury or death) suffered by any company or
person whomsoever as a result of or in connection with the use, application or implementation of any DEP, combination
of DEPs or any part thereof, even if it is wholly or partly caused by negligence on the part of Shell GSI or other Shell
Service Company. The benefit of this disclaimer shall inure in all respects to Shell GSI and/or any Shell Service
Company, or companies affiliated to these companies, that may issue DEPs or require the use of DEPs.
Without prejudice to any specific terms in respect of confidentiality under relevant contractual arrangements, DEPs shall
not, without the prior written consent of Shell GSI, be disclosed by users to any company or person whomsoever and
the DEPs shall be used exclusively for the purpose for which they have been provided to the user. They shall be
returned after use, including any copies which shall only be made by users with the express prior written consent of
Shell GSI. The copyright of DEPs vests in Shell GSI. Users shall arrange for DEPs to be held in safe custody and Shell
GSI may at any time require information satisfactory to them in order to ascertain how users implement this
requirement.
All administrative queries should be directed to the DEP Administrator in Shell GSI.
 DEP 32.31.56.30-Gen.
August 2009
Page 3
TABLE OF CONTENTS
1. INTRODUCTION ........................................................................................................4
1.1 SCOPE........................................................................................................................4
1.2 DISTRIBUTION, INTENDED USE AND REGULATORY CONSIDERATIONS .........4
1.3 GENERAL DEFINITIONS ...........................................................................................4
1.4 SPECIFIC DEFINITIONS............................................................................................5
1.5 ABBREVIATIONS .......................................................................................................6
1.6 CROSS-REFERENCES .............................................................................................7
1.7 COMMENTS ON THIS DEP .......................................................................................7
2. NIR APPLICATIONS AND SYSTEM .........................................................................8
2.1 GENERAL ...................................................................................................................8
2.2 SCOPE OF NIR APPLICATIONS...............................................................................9
2.3 NIR SYSTEM COMPONENTS ...................................................................................9
2.4 NIR APPLICATION DEVELOPMENT SEQUENCE .................................................10
3. NIR IMPLEMENTATION STRATEGY......................................................................12
3.1 GENERAL .................................................................................................................12
3.2 KEY NIR IMPLEMENTATION REQUIREMENTS ....................................................12
3.3 NIR IMPLEMENTATION STEPS ..............................................................................13
3.4 REGULATION DOMAINS.........................................................................................13
3.5 PATENTS AND INTELLECTUAL PROPERTY (IP)..................................................13
3.6 NIR IMPLEMENTATION PROJECT TEAM..............................................................14
3.7 PROJECT RELATED ITEMS....................................................................................16
4. NIR ANALYSER DESIGN ........................................................................................19
4.1 GENERAL .................................................................................................................19
4.2 NIR INSTRUMENT ...................................................................................................19
4.3 OFF-LINE NIR ANALYSER ......................................................................................20
4.4 AT-LINE NIR ANALYSER.........................................................................................22
4.5 ON-LINE NIR ANALYSER ........................................................................................22
5. NIR SYSTEM QA/QC ...............................................................................................24
5.1 GENERAL .................................................................................................................24
5.2 QA PROCEDURES ..................................................................................................24
5.3 SPECTROMETER TESTS .......................................................................................24
5.4 SQC PROCEDURES................................................................................................24
6. REFERENCES .........................................................................................................29
APPENDICES
APPENDIX A FIGURES.........................................................................................................31
 DEP 32.31.56.30-Gen.
August 2009
Page 4
1. INTRODUCTION
1.1 SCOPE
This new DEP specifies requirements and gives recommendations for implementing Near-
Infrared (NIR) systems, focusing on:
- Implementation challenges and key requirements, in particular:
! the special nature of NIR as a model based analytical technique and
! the complexity of a NIR system, which is composed of analysers and a
dedicated organisation to deliver analytical applications;
- The way an NIR system implementation strategy should be rolled out and the site
organisation for accomplishing it, in particular the role of the individuals, the
application ownership, the calibrator and the site laboratory;
- How the NIR analyser shall be designed and installed for off-line, at-line and on-line
applications, highlighting the use of standard equipment and the setting of
performance targets;
- Which tools are used to validate the NIR analyser(s) and the frequency and
necessity of on-going quality assurance.
This DEP is applicable to:
- Laboratory (off-line or at-line) NIR analysers;
- Process (on-line) NIR analysers;
NIR is mostly applied for main fuel blending, in particular for gasoline blending, although its
scope of application is much larger. This DEP is dedicated to refinery main fuel blending
where the NIR measurements are applied to the blended fuel sampled after the blend
header or from a blended product tank or proportional compositor.
Excluded from the scope of this DEP are NIR calibration procedures and outlier detection –
for these, refer to SMS 2965 and ASTM E 1655. This DEP generally follows the validation
and QC requirements of ASTM D 6122, see (5).
1.2 DISTRIBUTION, INTENDED USE AND REGULATORY CONSIDERATIONS
Unless otherwise authorised by Shell GSI, the distribution of this DEP is confined to Shell
companies and, where necessary, to Contractors and Manufacturers/Suppliers nominated
by them.
This DEP is intended for use in main fuel blending, e.g. mogas and diesel, but not fuel oil.
When DEPs are applied, a Management of Change (MOC) process should be
implemented; this is of particular importance when existing facilities are to be modified.
If national and/or local regulations exist in which some of the requirements may be more
stringent than in this DEP, the Contractor shall determine by careful scrutiny which of the
requirements are the more stringent and which combination of requirements will be
acceptable with regard to the safety, environmental, economic and legal aspects. In all
cases, the Contractor shall inform the Principal of any deviation from the requirements of
this DEP which is considered to be necessary in order to comply with national and/or local
regulations. The Principal may then negotiate with the Authorities concerned, the objective
being to obtain agreement to follow this DEP as closely as possible.
1.3 GENERAL DEFINITIONS
The Contractor is the party that carries out all or part of the design, engineering,
procurement, construction, commissioning or management of a project or operation of a
facility. The Principal may undertake all or part of the duties of the Contractor.
 DEP 32.31.56.30-Gen.
August 2009
Page 5
The Manufacturer/Supplier is the party that manufactures or supplies materials,
equipment and services to perform the duties specified by the Contractor.
The Principal is the party that initiates the project and ultimately pays for its design and
construction. The Principal will generally specify the technical requirements. The Principal
may also include an agent or consultant authorised to act for, and on behalf of, the
Principal.
The word shall indicates a requirement.
The word should indicates a recommendation.
1.4 SPECIFIC DEFINITIONS
At-line The configuration of having the analyser near the process at a plant
laboratory, but not connected to the process.
NIR analyser or analyser All piping, hardware, computer, software, instrumentation and calibration
model(s) required to automatically perform analysis of a process or product
stream – ASTM D 6122 (either off-line, at-line or on-line (Figure 3)).
NIR analyser result(s) or The numerical estimate of a physical, chemical, or quality parameter
analyser result(s) produced by applying the calibration model to the spectral data collected by
the analyser – ASTM D 6122. Also called estimate or predicted primary test
method result(s) (PPTMR) (Figure 3).
NIR analysis The process of applying the calibration model to a spectrum, pre-processed
as required, so as to estimate a component concentration value or property
– ASTM D 6122.
NIR application A specific analysis by means of a combination of an NIR analyser and a
specific process/product sample fed, manually or automatically, to generate
one or more analyser result(s). There can be as many results as reference
methods but independent for each process, e.g. RON by NIR (Figure 2).
NIR calibration or A process for creating a model that relates component concentrations or
calibration properties to the absorbances of a set of known reference samples at more
than one wavelength or frequency – ASTM D 6122.
NIR Calibrator The party developing and maintaining the calibration models (Figure 1).
NIR instrument or Spectrophotometer, associated electronics and computer,
instrument spectrophotometer cell and, if utilized, transfer optics - ASTM D 6122
(Figure 3).
Calibration model or The mathematical expression or the set of mathematical operations that
Multivariate calibration relates spectra to component concentrations or properties (Figure 3).
model or Model
A model can be called:
! Local, if applied at a specific site using on-site data;
! Global, if applied at various sites using centralised data from various
sites.
NIR system Combination of NIR analyser, local staff/organisation, NIR Calibrator and
QA/QC procedures covering one or more NIR applications (see Figure 1).
Also called process analyser system when installed on-line.
Performance test A test that verifies that the performance of the instrument is consistent with
historical data and adequate to produce valid results – ASTM D 6122.
Outlier detection methods Statistical tests that are conducted to determine if a spectrum falls within the
validity range of the model.
 DEP 32.31.56.30-Gen.
August 2009
Page 6
Outlier detection results Outcome of the outlier detection method while applying appropriate outlier
or outlier results detection limits or statistics. The analyser results obtained for outliers have
unknown accuracy and therefore should be checked by the PTM (Figure 3).
Outlier sample or outlier A sample whose spectrum is declared to be an outlier.
Spectrometer cell or cell An apparatus that allows a liquid hydrocarbon to flow between two optical
surfaces which are separated by a fixed distance, the sample path length,
while simultaneously allowing light to pass through the liquid – ASTM D
6122 (Figure 3).
Off-line The configuration of having the analyser at the central laboratory.
On-line The configuration of having the analyser connected to the process through
a sampling system.
Validation test or A test performed on a validation sample that demonstrates that the result
validation produced by the analyser and the result produced by the primary test
method are equivalent within control limits (ASTM D 6122).
1.5 ABBREVIATIONS
AMADAS Analyser Maintenance And Data Acquisition System
APC Advanced Process Control
BOD Basis of Design project phase
COA Certificate of Analysis
DVPE Dry Vapour Pressure Equivalent
FAT Factory Acceptance Test
FT-NIR Fourier Transform Near-infrared spectrometer (ASTM E 1944)
LIMS Laboratory Information Management System
NIR Near-infrared region of light wave number: 4000 - 12500 cm-1
(780 to 2500 nm) (ASTM E 131)
PCA Principal Component Analysis
PCR Principal Component Regression
PLS Partial Least Squares
PQM Product Quality Management
PT Proficiency Testing Program/Scheme
PTM Primary Test Method
QA Quality Assurance, procedures applied for system acceptance
QC Quality Control, procedures applied for system performance monitoring
QE Quality Estimator
QMI Quality Measuring Instrument, also known as Process Stream Analyser
R Reference method precision according to standards (Method
Reproducibility)
R’ Site precision according to ASTM D 6299.
RMSEC (V) Root mean square error of calibration (cross validation)
RMSEP Root mean square error of prediction
RVP Reid Vapour Pressure
 DEP 32.31.56.30-Gen.
August 2009
Page 7
SAT Site Acceptance Test
SEP Standard Error of Prediction
SQC Statistical Quality Control
VRM Validation Reference Material (ASTM D 3764, ASTM D 6122)
1.6 CROSS-REFERENCES
Where cross-references to other parts of this DEP are made, the referenced section
number is shown in brackets. Other documents referenced by this DEP are listed in (6.).
1.7 COMMENTS ON THIS DEP
Comments on this DEP may be sent to the DEP Administrator at [email protected].
Shell staff may also post comments on this DEP on the Surface Global Network (SGN)
under the Standards folder.
 DEP 32.31.56.30-Gen.
August 2009
Page 8
2. NIR APPLICATIONS AND SYSTEM
2.1 GENERAL
The principles to be recognised when implementing any NIR application are:
1. NIR is a model-based technique. A model is a mathematical expression or set of
mathematical operations that relates component concentrations or properties to a
pre-selected NIR wavelength domain.
An NIR analyser predicts or estimates a physical, chemical, or quality parameter
produced by applying a model to the spectral data (absorbance spectrum) collected
by the analyser.
An NIR analyser may embed one or more calibration models (Figure 3), depending
upon the type of application(s) (see Figure 2).
2. All NIR applications (2.2), whether off-line, at-line or on-line, contain a standard
instrument and software. One off-line analyser should be used to collect spectral
data for any on-line application. Analysers and software shall be sourced from
Manufacturers that are approved by the Principal.
3. NIR analysers not only deliver analyser results, but also generate outlier detection
results due to their “self-diagnostic” capability embedded in the model. Proper tuning
of outlier detection limits shall be performed to flag samples that are outside the
validity domain of the model and therefore have results that are of unknown
accuracy. Any sample whose spectrum is declared an outlier shall be analysed by
the reference method.
4. NIR is a secondary measurement technique. The implications of this include the
following:
! The application of NIR is intimately tied to the off-line laboratory measurements
by the PTM for model calibration, validation, SQC, and, in some instances,
correction. Benchmark data from recent implementations indicate that greater
attention to these measurements is often required for a successful NIR
implementation.
! Off-line laboratory measurements by the PTM may be reduced in number but
cannot be fully eliminated by implementing NIR.
! NIR is based on modelling several spectral variables which are tied to variations
in fuel compositions and regressed with PTM measurements that are often noisy.
This means that to develop accurate models spanning the blend compositions to
be produced, calibrations shall be built using a training set comprising (typically
hundreds of) samples to span and that composition range.
! Major compositional variations in blended fuels occur not only as a result of
variations in the ratios of blending components used but also from compositional
variations within those components. These latter variations are usually not
experienced over periods shorter than several months. As a result, it is important
that model training set data span all grades, all seasons, several months of
normal operation of the refinery and, where possible, include periods of
conversion unit turnarounds and outages.
! Lab-blends should not be used to prepare training sets since the blending
components used to prepare the handblends shall each span their normally
encountered composition ranges. This will usually mean using blending
components collected over a period of several months.
! Errors in NIR results cannot be eliminated simply by making many replicate NIR
measurements on the same composition. The averaging attained by making
many on-line measurements over the course of a blend will not eliminate NIR
model errors.
 DEP 32.31.56.30-Gen.
August 2009
Page 9
! For minor compositional variations over the course of a blend using blending
components of nearly constant composition, NIR model errors are nearly constant
for some properties.
! The use of NIR in making property measurements of blended fuels is, in part,
based on the assumption that all of the major compositional changes that affect a
given property are reflected in the NIR spectrum of the blended fuel. For some
properties, this is not the case (e.g., sulphur) and therefore not all properties can
be accurately measured by NIR. There are other properties for which the
compositional information available may be tenuous and model accuracies may
be insufficient to meet all measurement needs for those properties (e.g.,
distillation endpoint and Reid vapour pressure).
5. The need for good lab performance
NIR model errors can be dominated by PTM measurement precision errors in the
training set. The model is generally only as good as the PTM data allow. NIR
analyser results can be very precise and accurate and therefore can be used for
product certification, provided:
! appropriate calibration standards are applied;
! appropriate QA/QC procedures are applied (5);
! regulations allow its use (3.4);
! customers do not prohibit its use.
As a consequence, the Principal shall ensure that:
i. An NIR system (2.3) is set up around any NIR application or group of applications
(see Figure 1). NIR calibration uses multivariate calibration models requiring a
specific organisation (3.6.2) to develop and maintain the applications/models.
ii. An implementation strategy (3) is followed to ensure accurate, robust and
sustainable NIR applications, in which the proper NIR analyser(s) (4) are designed
and set up, and their performance (5) is monitored to ensure they are in-line with
their performance targets (3.7.5).
iii. The proper NIR application sequence is followed (2.4).
2.2 SCOPE OF NIR APPLICATIONS
Mogas and diesel blending and the corresponding blending streams are the most common
standard NIR applications.
An NIR application is a specific analysis done by means of a combination of a process
sample and an NIR analyser, which can deliver one or many results. There can be as many
results as reference methods, but independent for each process, e.g. RON by NIR (Figure
2).
Any NIR analyser can run many applications, depending on its capability to accommodate
different process samples (4). The specific cell design and instrument parameters are
specified in (4.2).
2.3 NIR SYSTEM COMPONENTS
For a specific site, the NIR system is a combination of all the NIR analysers, the local
staff/organisation, one or more NIR Calibrators, and QA/QC procedures covering all NIR
applications – see Figure 1.
Unless otherwise specified by the Principal, the composition of such a system shall include
the following elements in addition to the analysers:
1. An NIR focal point (3.6.2) who coordinates matters around NIR applications and who
is responsible for NIR performance.
 DEP 32.31.56.30-Gen.
August 2009
Page 10
2. A Calibrator, who shall be a person trained to build NIR models.
There are two alternatives:
a) The site shall contract out calibration to a certified expert; or
b) The site shall perform the calibration, in which case the calibrator shall:
! follow a specified expertise path (3.6.3);
! be approved by an authorized expert;
! have a back-up available.
If the Calibrator is not so approved, a certified expert shall review the models for
release.
3. An NIR team (3.6.1), including the staff operating the NIR analyser and the NIR
Calibrator, led by the NIR focal point, who interfaces with the PQM team and internal
customers.
During any project, tasks shall be allocated according to Tables 5 and 6.
The labour effort (Opex) required during the project and for on-going maintenance is
indicated in Table 6.
4. A set of QA procedures and SQC applications (5) shall be implemented and
maintained to monitor the NIR system's performance in terms of precision, accuracy
and robustness/availability.
2.4 NIR APPLICATION DEVELOPMENT SEQUENCE
The sequence of development shall follow the following path (unless otherwise specified by
the Principal) (see Figure 4):
1. Application building, i.e. model development and generation.
a) On-line applications are normally built using an off-line analyser, whose spectra
are used to develop the models to be transferred to the on-line analyser. The
corresponding instrument shall be designed according to (4.2). Models shall be
built following the principles of Chemometrics, i.e. the application of
mathematics to build a relationship between (specific parts of) NIR spectra and
analytical/chemical data (properties of products).
b) The application design/specification [e.g. scope, range and performance targets
(3.7.5)] shall be defined by a project team and/or PQM team.
The model development shall begin with a training set. This training set contains
accurately analysed samples (3.2.1), which should include the entire range of
variation for which the application is designed.
These samples shall be obtained from either:
! Process rundowns, preferably from line samples so as to maximise the
capture of process variability; or
! Artificial blends that may be used if the process variability is not sufficiently
covered by process samples. They require proper design to capture all
blending component variabilities.
The calibration itself shall be carried out according to ASTM E 1655 and
SMS 2965. The calibration should include a standardisation step to ensure that
the model(s) can be transferred to another NIR instrument of the same brand.
c) Most models will be for “local” application, i.e. adapted to the individual refinery’s
type of processes, crude diet and product requirements. To accelerate the
process of building applications, a “global” library (i.e. a database with spectra
and reference data) may be used to develop “starter models”, which will be
refined and eventually replaced by means of local data.
 DEP 32.31.56.30-Gen.
August 2009
Page 11
2. Implementation of that application to another at-line or on-line analyser. Its
spectrometer shall be similar to that of the off-line analyser.
All analyser components (hardware and software) shall be of a type approved by the
Principal (3.2-2). The analyzer/software shall be souced from a Manufacturer
approved by the Principal.
For the implementation of an on-line analyser, an off-line analyser may be purchased
as part of the project if not already available.
3. A validation or QC check shall be performed according to ASTM D 6122, SMS 2966
and ASTM D 6708 before an application is released for use. If the validation is not
conducted prior to the release of the application, the validation shall be performed
soon afterwards.
To ensure consistency during the validation phase, the site laboratory shall deliver
the reference data used for model validation and shall perform the QC checks.
4. Most of the resources required for the maintenance of an NIR application are related
to model maintenance, not analyser maintenance (Table 6). The required resources
shall be taken into consideration during the project phase, as well as those required
for reference measurement and validation (5.) during the application phase.
 DEP 32.31.56.30-Gen.
August 2009
Page 12
3. NIR IMPLEMENTATION STRATEGY
3.1 GENERAL
In implementing any NIR application, the following shall be taken into account:
1. Ready-for-use (plug and play) applications are not available from equipment
Vendors/Manufacturers,
2. Successful implementations require a site strategy that takes into account the
complete NIR system (2.3), key implementation issues (3.2) and a very specific order
for rolling out an NIR implementation project (3.3) and the application itself (2.4).
3. Implementation issues are the same as those for any model-based analyser or
quality estimator used for process control, but these are particularly relevant to NIR
applications because:
" They are different from all other conventional work in a laboratory;
" They are relatively expensive in terms of both CAPEX (Capital Expenditure) and
OPEX (Operational expenditure – Table 6) compared to conventional analyser
applications, and they require specific model design and maintenance expertise;
NOTE: This is especially true if NIR is applied to measure a single property and/or for a single
stream.
" Time is required to develop and maintain location-specific models;
" They shall comply with appropriate QA/QC procedures.
3.2 KEY NIR IMPLEMENTATION REQUIREMENTS
The Principal shall consider the following items:
1. Laboratory performance is of prime importance to ensure the accuracy of the NIR
applications. Poor laboratory data precision/accuracy will directly result in poor NIR
performance. The laboratory shall comply with ISO 17025. The corresponding
performance rate, monitored through a recognised fuel proficiency scheme, should
be at least 90 %
NIR project are not intended to improve the lab PTM measurements (although it may
induce this to a certain extend).
2. The NIR analyser shall be a standard type of analyser:
a. For both hardware and software (2.1.2)
b. In its data pre-processing techniques (4.2.5),
These two key requirements enable models to be transferred from an on-line
analyser to any other analyser.
These standards shall be selected so as to enable exchange between sites.
3. Refinery (PQ) Management shall have a long-term strategy regarding the design,
maintenance and support of NIR technology, which in particular shall include:
a. The set-up of a proper NIR organisation (2.3, 3.6.2) as part of the Product
Quality Management organisation, comprising the site laboratory and normally
including on-line instrumentation or analyser departments.
b. Allocation of sufficient resources during both implementation and maintenance
(3.7.4).
c. Sufficient training of all staff involved in implementing and running the
application (3.6.3); and ensuring that sufficient expertise is gained from the
Calibrator.
d. The integrity of the NIR application is ensured by:
 DEP 32.31.56.30-Gen.
August 2009
Page 13
! embedding the NIR applications in the quality system for all the critical NIR
applications, particularly those used for product certification/release.
! implementing SQC procedures according to ASTM D 6299, ASTM D 6708
and SMS 2966.
3.3 NIR IMPLEMENTATION STEPS
Sustainable implementation shall be carried out in the following order unless otherwise
specified by the Principal:
Step 1: Ensure NIR awareness. If the site does not have significant experience with NIR,
awareness should be gained by consulting an NIR expert. This will provide the site
with an understanding of the potential benefits and the limitations of NIR technology
for the intended application(s). It will also acquaint the site with the key
requirements for successfully applying the technology. The site may address this
awareness requirement by having site personnel attend a formal awareness
course.
Step 2: Develop a plan:
a) A Project Execution Plan shall be compiled as part of or after the initial project
identification and assessment phase. It is required for the implementation of the
first on-line analyser.
and/or
b) For large sites, a long-term plan should be developed as part of a dedicated
analytical master plan covering the whole site.
This plan shall include a training plan in accordance with the NIR expertise path
specified in (3.6.3).
Step 3: Create an NIR team (3.6) involving both the Laboratory (technical focal point)
and Engineering (instrumentation/QMI) personnel, with clearly defined
responsibilities).
Step 4: Implement the application through good project management (3.7), and plan
maintenance resources at an early stage (3.7.4).
3.4 REGULATION DOMAINS
NIR can be, and is being, used for product certification/release.
Contractual arrangements and regulatory requirements may prevent NIR from being used.
NOTE: Before using NIR for this purpose, it shall be verified with the local authorities that NIR is allowed for
product certification for specific main fuel properties. Some authorities, e.g. USA, under EPA rules,
and Japan, do not allow the use of secondary techniques for certification of some product properties.
3.5 PATENTS AND INTELLECTUAL PROPERTY (IP)
Although chemometrics is widely applied to NIR modelling, some NIR applications have
been patented.
The Principal shall ensure that any patents are not infringed, either by:
(1) consulting the IP focal point when doing self-calibration, and acquiring a licence
where required.
(2) acquiring indemnity from the supplier who delivers the application, the calibrator
and/or analyser.
 DEP 32.31.56.30-Gen.
August 2009
Page 14
3.6 NIR IMPLEMENTATION PROJECT TEAM
3.6.1 General
A project team shall be formed for any new NIR application that involves product release or
certification.
Unless otherwise specified by the Principal, the team shall comprise at least:
1. A project lead, who shall be responsible for:
a. Application specification (range and precision targets) (3.7.5);
b. Delivery of the project (organisation, resources, skills and schedule);
c. Supporting the NIR team.
2. An NIR focal point (normally the application owner).
3. A member of the QMI discipline, if dealing with an on-line application.
4. A member of the process control team, if the application is included in a control loop.
5. An Operations person should be included, if dealing with a major application, e.g. an
Oil Movements person for on-line product certification of mogas or diesel.
3.6.2 NIR Organisation
The Laboratory is normally responsible for NIR technology and validation of all site
applications; however, the instrumentation/QMI discipline may be responsible for on-line
NIR analysers and perform hardware maintenance.
An NIR focal point is typically:
1. A chemist,
2. A control (APC) engineer,
3. Or a QMI engineer;
This person coordinates all tasks related to the applications and is responsible for approval
of the operational use of each application, and for maintaining all quality records, and is
normally in direct contact with the following:
! Product quality section;
! Supplier of the analyser;
! NIR Calibrator (Contractor or in-house) (2.3.2), with whom data is exchanged for
model building and maintenance;
! Laboratory staff dealing with routine off-line NIR measurements and NIR QA/QC;
! QMI technician responsible for the hardware part of the on-line and at-line
applications.
 DEP 32.31.56.30-Gen.
August 2009
Page 15
3.6.3 NIR Expertise Path / Training
The training sequence shall be as shown in Table 1 unless otherwise specified by the Principal.
Table 1 NIR expertise path
Attended by
Training Given by Given when NIR focal point Calibrator Analyst in charge QMI technician
(2.3.2) of off-line NIR
NIR awareness NIR expert Before plan or during BOD Yes Yes, from site Recommended Recommended
course
NIR spectroscopy During SAT of off-line NIR Yes Yes Recommended
analyser
Spectra Calibrator After SAT Recommended
pre-processing
procedure
NIR analyser Manufacturer During FAT of on-line analyser Yes Yes
operation by
supplier
NIR analyser Manufacturer During SAT of on-line analyser Yes
maintenance
Basic multivariate Calibration Before Chemometrics level B Recommended Yes
data analysis software course
supplier
Chemometrics Certified After some practical Recommended Yes, only if done
level B (basic) chemometrician experience by site
Chemometrics Certified Before use of NIR for product Yes, only if done
level A chemometrician certification by site
(advanced)
Knowledge transfer to the future site calibrator shall be scheduled by a certified Chemometrician, so that training is mixed with practical modelling
and regular QA/QC checks of the applications by the calibrator.
This process should lead to the certification of the calibrator (2.3.2).
 DEP 32.31.56.30-Gen.
August 2009
Page 16
3.7 PROJECT RELATED ITEMS
3.7.1 General
For on-line NIR applications of main fuel blending, the key implementation phases typically
contain the following items.
NOTE: For a first off-line NIR application, the steps in Table 2 shall also apply except for the on-line execution
phase.
Table 2 Project phase requirements
Project phases Items/actions
Identify and assess Investigate local regulations with the governing authorities.
At that stage, the laboratory performance data shall be
captured. If the results are not good enough (< 90 % PT
scheme performance), this stage shall be postponed until
a performance improvement scheme has been
successfully implemented.
Select The Basis of Design (BOD) shall contain the site NIR
strategy (analyser standardisation and who does the
calibration) and the performance targets. The analyser
redundancy/back-up requirements shall be defined.The
Project Execution Strategy (PES) shall include all required
project items, in particular:
! A task allocation matrix (Tables 4 and 5)
! Project /Contractor structure/organisation
! Project and maintenance Opex (Table 6)
A Risk Management workshop shall be used to generate
the Risk Management Matrix.
Start collecting relevant process samples.
NOTE: Collecting and storing the relevant samples should commence
at the initial stages of the project in order to capture as much
process variability as possible.
Define Order the off-line NIR analyser (2.4.1, 4.3)
SAT of the off-line analyser including standardisation
tests.
Develop models.
Execute Order the on-line NIR analyser.
FAT including standardisation tests.
Integration test (iFAT) using models.
SAT of the on-line analyser.
Running Post-implementation review.
In the case of a new application, a feasibility study shall be carried out prior to any Front
End Design phase.
3.7.2 NIR Project Timing
An NIR project is constrained by three factors:
! The availability or the delivery time of the off-line analyser (which usually takes three
months);
! The delivery time of the on-line analyser (which usually takes six months).
! The availability of sufficient training/calibration set variability and sample number
according to SMS 2965. For models that will be based principally on samples from a
 DEP 32.31.56.30-Gen.
August 2009
Page 17
single site, the time required to collect an adequate sample set will be at least the better
part of a year.
Benchmark data show that the implementation of an NIR on-line product certification
project takes two years after the BOD kick-off meeting.
3.7.3 Project Task Allocation
During the project, the tasks shall be allocated according to the following table unless
otherwise specified by the Principal:
Table 4 Task allocation table - Generic
Specialist Tasks Deliverables
Supplier Off-line, at-line, on-line NIR analyser
Spare parts
Supplier's regional Installation, testing and
support training
Calibrator Kick off meeting Models, QA/QC check reports,
Training Calibration QA documents (5.5)
Training set design
Modelling
Engineering, Project management
Procurement and Integration between
Construction supplier, process and IT
Project leader Project management Risk Management Matrix, Project
status review.
Performance targets (3.7.5)
NIR team Reference data, spectral data
Off-line analyser SQC data
QMI engineer On-line analyser SQC data
Table 5 Task allocation table
NIR
Description Site Supplier Calibrator
expert
Off-line NIR systems manufacturing Responsible
Kick-off meeting & project review Responsible Participate Participate
Purchase of hardware and software Responsible Participate Participate
Collect samples Responsible
Collect spectra Responsible
Primary reference method
laboratory analysis. Responsible
Development and acceptance of Responsible
calibration models Participate
Installation of the final models Participate Responsible
QC on all models Responsible Informed
NIR analyser accessory installation Participate
and commissioning (FAT, SAT) Informed Responsible
 DEP 32.31.56.30-Gen.
August 2009
Page 18
NIR analyser operations and
maintenance training Informed Responsible
NIR analyser documentation Informed Responsible
3.7.4 NIR OPEX
The OPEX data given in Table 6 should typically be taken into account for both project and
maintenance resourcing.
NOTE: These values are for guidance – actual values can vary significantly.
Table 6 OPEX data
Project Maintenance
(one time only) (ongoing)
On-line NIR At-line and On-line NIR At-/off-line NIR
(months) off-line NIR (number of full- (number of full-
(months) time time
employees) employees)
Project leader 3 2 0.05 0.05
NIR focal point 5 5 0.05 0.05
Laboratory a 3 3 0.1 0.1
Calibrator 5 0.1 to 0.2 b
QMI 5 0.1
a
Spectra collection and analyser validation.
b
In case of changes of product specification.
3.7.5 Performance targets
The following performance targets should be set.
! Acceptance during FAT and SAT (5.5) according to the Principal's test list.
! The long-term precision (expressed as a standard deviation) of the NIR analyser
shall be smaller than or equal to the long-term precision of the reference method.
Collection of the corresponding data requires periodic QC sampling or VRM
validation whilst monitoring the application (once per day or once per blend).
! The accuracy expressed in terms of RMSEP depends upon the type of application:
o For process control, RMSEP < R’
o For product certification, RMSEP =< R’/2 or R/2
The collection of the corresponding data requires periodic:
o line sample validation whilst monitoring the application (once per day or once per
blend) else during the running part (weekly) (5.4).
o tank sample validation whilst monitoring the application (once per day or once per
blend), otherwise whilst running (weekly) (5.4).
 DEP 32.31.56.30-Gen.
August 2009
Page 19
4. NIR ANALYSER DESIGN
4.1 GENERAL
An NIR analyser is a model-based analyser incorporating a multipurpose
spectrometer-based instrument (Figure 3). It comprises the following components:
1. A spectrometer;
2. A cell/probe;
3. Transfer optics to link the cell with the spectrometer; optical pipes are required
4. Specific software for:
! Generating the NIR spectrum/spectra acquisition and carrying out instrument
standardisation;
! Pre-processing the spectrum and/or running the models;
! User interfacing/communication;
! Model building and outlier detection.
All analysers shall integrate the same spectrometer and the same software.
For any particular application, any on-line analyser shall have a similar instrument design to
that of the off-line analyser on which the “same” models have been built:
1. The spectrometer setting (4.2.1.1) shall be the same;
2. The cell material, path length and sample temperature shall be the same (4.2.2);
3. The instrument standardisation and spectra pre-processing parameters shall be the
same (4.2.5).
To eliminate spectroscopic disturbances, the key design parameters for sample handling
are:
1. Sample temperature shall be controlled within +/- 1 °C;
2. Sample integrity shall be secured;
3. Removal of particles and free water, which can lead to light scatter.
4.2 NIR INSTRUMENT
The key design parameters for any NIR instrument are as follows:
4.2.1 SPECTROMETER
4.2.1.1 Type
The spectrometer shall be:
1. Type: Fourier Transform type (FT-NIR)
2. Scanning range from 3500 cm-1 to 9500 cm-1 to cover from the 1st CH combination to
the 2nd CH overtone
3. Number of scans (interferometer averaging): up to 512
4. Spectral apodisation: cosine
4.2.1.2 Light Source
The light source shall be of the quartz halogen type and shall be pre-aligned by the
Manufacturer.
 DEP 32.31.56.30-Gen.
August 2009
Page 20
4.2.1.3 Detector
! The detector shall be of the cooled "InAs" type.
4.2.2 Cell
The design parameters are:
1. Cell path length = 0.5 mm
2. Type: CaF2 or Infrasil
3. Temperature = (25 " 1) °C
The cell window shall be wedged for an at-line and an on-line analyser, and should be
wedged on an off-line analyser.
The sample temperature shall be controlled:
! For the off-line and at-line instrument by controlling the temperature of the cell
! For the on-line analyser by controlling both the cell temperature and the incoming
sample.
4.2.3 Cell Referencing
Cell referencing (background reading) shall be carried out regularly; it shall be done at least
every day, using an empty cell for an on-line analyser and an open beam on an off-line
analyser.
4.2.4 Software
The spectra collection and interface/communication software are specific to the instrument.
The software used for model building and prediction:
! shall run PCA, PCR and PLS models;
! shall detect outliers and inliers in prediction based upon preset confidence levels.
These software programs shall be sourced from Manufacturers that are approved by the
Principal.
4.2.5 Standardisation Procedures
Standardisation procedures are required for model transferability between an off-line and
an at-line or on-line analyser instrument. This consists of two steps:
1. Instrument standardisation, which shall be carried out by the analyser interface
software.
This should be performed by digitalisation and/or truncation and/or toluene spectral
shift.
2. Spectra pre-processing, which should be carried out in two sequential steps by the
software running the models:
i. Offset (baseline) correction by means of:
– baseline correction for white samples;
– second derivative for dark samples.
ii. Full Multiplicative Signal Correction (MSC) or Area Normalisation.
4.3 OFF-LINE NIR ANALYSER
4.3.1 General
The off-line analyser shall be located in the main laboratory.
 DEP 32.31.56.30-Gen.
August 2009
Page 21
This analyser should be used for both model development and application.
4.3.2 Implementation
Any off-line or at-line analysers shall be implemented as specified by the Principal.
4.3.3 Sampling System
The sampling shall be done in such a way that sample weathering is avoided.
The sample shall be stored in a rubber-sealed container, with a low nitrogen pressure
applied to the sample (see example of set-up in Figure 5).
4.3.4 Transfer Optics
No fibre optics shall be used.
4.3.5 Analyser Set-up
A standard list of analyser components is provided in the following table:
Table 7 Instrument components
Off-line NIR analyser On-line NIR analyser
Spectrometer + transfer optics same
Detector type same
Source (+ spare) same
Cell type same
Window type same
Path length same
Temperature controller same
Interface/prediction software same
Spectra acquisition software same
Calibration software same
Prediction software (on demand) Prediction software (real time)
4.3.6 Communication with LIMS
Especially when the analyser is used for product certification, the off-line analyser should
be interfaced with LIMS in order to accelerate the transfer of analyser results and outlier
warnings and also to avoid human/typing errors.
4.3.7 Spare Parts
The required spare parts shall be in line with the recommendations of the analyser Vendor.
4.3.8 Preventive maintenance
The following items should be replaced at the stated intervals:
! Source – every six months;
! Laser – every five years.
 DEP 32.31.56.30-Gen.
August 2009
Page 22
4.4 AT-LINE NIR ANALYSER
The main feature of an at-line analyser is its operability by operators, in particular for
sampling.
The use of automatic devices/functions should be maximises in order to handle:
! Cell referencing in order to avoid removing the cell
! Cell cleaning
! Sample injection.
4.5 ON-LINE NIR ANALYSER
4.5.1 General
Amended per
Circular 10/10
The on-line analyser system shall be selected, specified, designed, installed, inspected and
tested in accordance with DEP 32.31.50.10-Gen. and DEP 32.31.50.13-Gen.
It shall embed an instrument spectrometer that is of the same brand as that of the off-line
analyser (4.3).
Figure 6 provides a scheme of existing DEPs applied to process NIR analyser design.
4.5.2 Analyser redundancy/back-up
The Principal shall specify the required analyser back-up configuration during the initial
project identification and assessment phase, depending on the type of application and/or
the degree of reliability/availability needed. The following options are available, in
increasing degrees of redundancy:
1) Reliable analyser but no redundancy; single NIR analyser with single cell.
2) Software redundancies through QE or APC modules. NIR can be used as back-up
for some conventional analysers, e.g. density or, in some circumstances,
RVP/DVPE.
3) Hardware redundancies through duplicate analysers. Two NIR systems equipped
with two cells are only justifiable for multiple product blending or multiple blenders for
products of a single type.
4.5.3 Sampling system
Amended per
Circular 10/10
Sample take-off, transportation and sample conditioning shall be in accordance with DEP
32.31.50.10-Gen.
Spectrometric disturbances (4.1) shall be eliminated by:
! Removing particles – preferably by means of self-cleaning filters, e.g. automatic
tangential filters;
! Removing water from the sample by coalescing after pre-cooling; and
! Controlling the temperature of the sample at (25 " 1) °C
NOTE: If the temperature of the cell is controlled with a vortex cooler, an appropriate air treatment
facility shall be installed.
The sampling system shall allow for the introduction of:
1. The QC sample and / or reference sample (VRM)
2. Nitrogen to dry the cell;
 DEP 32.31.56.30-Gen.
August 2009
Page 23
3. Cleaning material (isopropyl alcohol should be used if there is a risk of cell fouling
with polar materials or water); and
4. Instrument check material (typically toluene)
NOTE: The cleaning material and instrument check material should be stored in pre-filled containers with at
least two months’ supply. The contents of these containers should be monitored by means of mass
flow meters.
4.5.4 Light transfer optics
Optical light pipes shall be used; fibre optics shall not be used.
4.5.5 Spare parts
The required spare parts shall be in line with the recommendations of the analyser Vendor.
A personal computer (PC) may be used as back-up and also for off-line configuration
testing.
4.5.6 Preventative maintenance
See 4.3.8.
The laser should be replaced every two years to prevent an analyser failure due to low
laser power.
4.5.7 Communication with IT systems
The on-line analyser shall be linked with the DCS, through Ethernet or Modbus
communication, but it may be hardwired.
A remote access facility using Ethernet should be provided to give laboratory staff access
to the on-line NIR analyser.
4.5.8 Software configuration
The configuration of the on-line analyser should be tested on an off-line analyser prior to
downloading its configuration.
4.5.9 Analyser housing
Amended per
Circular 10/10
The analyser housing shall be in accordance with DEP 32.31.50.13-Gen.
The project team shall specify the requirements for the analyser housing, including its
temperature.
Depending on the site location and the criticality of the application, e.g. if used for on-line
certification, the analyser housing should incorporate a redundant air conditioning system.!
 DEP 32.31.56.30-Gen.
August 2009
Page 24
5. NIR SYSTEM QA/QC
5.1 GENERAL
QA/QC shall be specified by the Principal in terms of type and frequency in accordance
with appropriate QA/QC procedures.
Because NIR is model based, these models shall be documented through dedicated
calibration QA documents (5.5).
5.2 QA PROCEDURES
NIR analysers used for product certification shall comply with the Principal's QA procedures
and should be compliant with ISO 17025.
The QA procedures shall cover:
1) the spectrometer tests;
2) the type of SQC procedures;
3) their frequency;
4) the host IT system.
5.3 SPECTROMETER TESTS
The spectrometer shall comply with the Manufacturer’s specifications for diagnostic testing
(e.g. signal to noise, linearity, light intensity, alignment, wavelength/frequency accuracy,
etc.).
The spectrometer tests shall be conducted on a regular basis. The recommended
frequency is at least every month and after any change of a light source or laser.
5.4 SQC PROCEDURES
5.4.1 General
NIR validation/QC shall be in accordance with ASTM D 6122, ASTM D 6299, ASTM D 6708
and SMS 2966.
The types of test and their frequency should be in accordance with Table 8, in addition to
the monthly spectrometer tests (5.3):
 DEP 32.31.56.30-Gen.
August 2009
Page 25
Table 8 SQC requirements
Type of QC Off-line IT tool On-line NIR IT tool
NIR frequency
frequency
System
performance test
Level 0 Daily Test flag Before and after Test flag
recorded in each blend recorded in
analyser log AMADAS or
or
book
as specified
every day,
by the
whichever is Principal
more frequent
Or Level A Daily Outlier recorded Before and after Outlier value
in analyser log each blend recorded in
book value AMADAS
or
or
every day,
whichever is as specified
more frequent by the
Principal
Level B – Never Before and after AMADAS,
VRM/QC each blend
or as
sample
or specified by
the Principal
every day,
(reference: ASTM
or
D 6122 – sections
9.1 and 10.1) at operator
request
Line sample Required LIMS SQC Required weekly, LIMS SQC,
validation weekly but
or or AMADAS,
recommended at
(reference: ASTM
as specified by least every two or as
D 6122 and ASTM
the Principal blends in case of specified by
D 6299)
product the Principal
certification
Tank sample and Required LIMS SQC Required weekly LIMS SQC,
composite sample – recommended
weekly or or
validation – in case at least every two
of product as specified by blends in case of AMADAS,
certification the Principal product
or as
certification
(reference: ASTM specified by
D 6299 and ASTM the Principal
D 6624)
Assessment of According Using specific According to the Using
agreement with to the ASTM D 6708 Principal's QA Specific
PTM Principal's software procedures ASTM D
QA 6708
According to ASTM
procedures software
D 6708
 DEP 32.31.56.30-Gen.
August 2009
Page 26
5.4.2 NIR system performance test
This test shall be carried out automatically by the NIR analyser itself or upon operator or
DCS request (on-line analyser). This is a supervisory test to monitor the long-term
performance of the analyser using:
1. a check sample (so-called levels 0 and A); or
2. a QC sample (level B), which is similar to the VRM validation (ASTM D 6122).
Level A validation is carried out using a check sample through a PCA model; the
corresponding outlier value should preferably be recorded in AMADAS or by the IT system
specified by the Principal.
Level B validation is carried out using a QC/VRM sample. It is used to monitor the long
term precision of the analyser and shall be performed for product certification purposes.
The QC sample, captured from the process line, should be stored in a two piston cylinders,
as follows:
! each cylinder shall be equipped with an internal mixer;
! one cylinder shall contain one of the main process grades and the other cylinder
shall contain the other main process grade;
! the cylinders shall be large enough to ensure that the sample has an autonomy of at
least three months.
The VRM is shared with the other conventional analysers/QMI e.g. DVPE, sulphur.
5.4.3 Model applicability
The model's applicability or outlier test results are diagnosed automatically at every
prediction by the analyser. It shall be tuned for both:
! Mahalanobis distance based upon a certain confidence level; and
! Spectral residual.
If there are distinct clusters or grades, and there are no separate models for the clusters or
grades, inlier detection shall be applied.
Whenever an outlier or an inlier is detected, a signal should be sent to the DCS to inform
the plant operators, and a grab sample shall be taken automatically, depending on certain
criteria (e.g. process criticality, if the outlier or inlier flags are raised for many properties at
the same time).
5.4.4 Line sample validation
Periodic line sample validation covering various process modes/blending ranges shall be
performed to demonstrate that the analyser results remain in statistical agreement with the
initial NIR performance.
Unless otherwise specified by the Principal, the samples shall be taken from a cylinder
continuously purged with the process fluid and stored in a rubber-sealed bottle. Such a
device is preferred to sampling with bombs.
5.4.5 Tank sample and composite sample validation
Periodic tank sample validation shall be performed in the case of on-line certification.
The sample shall be collected by:
1. TMB (Top, Middle, Bottom) tank sampling, in the case of in-line blending to tank; or
2. A proportional sampler, in the case of blending to pipeline or ship.
 DEP 32.31.56.30-Gen.
August 2009
Page 27
5.4.6 Test for assessment and agreement with the reference method
For product certification applications, ASTM D 6708 shall be applied.
5.5 CALIBRATION QA DOCUMENTS
Key modelling/calibration steps shall be recorded in quality documents:
o Laboratory notebook, and/or
o E-notebook, and/or
o Report
These shall be available for audit by an NIR expert, especially for product certification
purposes.
They shall contain the following items:
1. Application name and application type.
2. List of the property models developed or updated and the electronic locations of the
models and their associated files.
3. Descriptions of the data sets and their electronic locations:
a. QC/QA primary reference method data and NIR spectrometer data.
b. Calibration data
c. Validation data.
4. Conclusions based on QA/QC data. For example, identify data that were excluded
from calibration or validation data sets and the reasons for excluding them.
5. Spectrometers used.
6. Modeling software used.
7. Modeling methods used. For example:
a. PCA for examination of X and Y block data.
b. PLS for model building.
8. Data analysis results – reference data
9. Data analysis results – spectra
! Spectral domain selected
! Clusters
! Odd samples
! Interpretation
10. Spectral preprocessing procedure.
11. In addition to any data excluded in (5.54), list additional data (outliers) excluded
along with the reasons for excluding the data.
12. Reference plots of particular interest to someone reviewing or extending this work.
The plots themselves do not need to be included in the documentation.
13. Calibration results. For example, the RMSECV, number of latent variables and
property range of each model.
Set-up of the cross-validation (how many segments, random/systematic)
14. Validation results. For example, the RMSEP, SEP and bias of each model.
SQC settings
15. Calibration transfer results.
16. Application flag using colors to indicate the model service status
 DEP 32.31.56.30-Gen.
August 2009
Page 28
Green: Operational for certification
Orange: Operational for process monitoring only
Red: Not operational
17. Outlier limits (e.g., property ranges, spectral residual limits, Mahalanobis distances
limits).
 DEP 32.31.56.30-Gen.
August 2009
Page 29
6. REFERENCES
In this DEP, reference is made to the following publications:
NOTES: 1. Unless specifically designated by date, the latest edition of each publication shall be used,
together with any amendments/supplements/revisions thereto.
2. The DEPs and most referenced external standards are available to Shell staff on the SWW (Shell
Wide Web) at http://sww.shell.com/standards/.
Amended per
Circular 10/10
SHELL STANDARDS
On-line process analysers DEP 32.31.50.10-Gen
Analyser housing DEP 32.31.50.13-Gen
Shell Method Series: Development and validation of SMS 2965
multivariate calibration models for quantitative near-
infrared analysis
Shell Method Series: Quality control of quantitative SMS 2966
near-infrared analysis
AMERICAN STANDARDS
Standard practice for validation of the performance ASTM D 3764
of process stream analyzer systems
Standard practice for the validation of multivariate ASTM D 6122
process infrared spectrophotometers
Standard practice for applying statistical quality ASTM D 6299
assurance to evaluate analytical measurement
system performances
Standard practice for determining a flow- ASTM D 6624
proportional average property value for a collected
batch of process stream material using stream
analyser data
Standard practice for statistical assessment and ASTM D 6708
improvement of the expected agreement between
two test methods that purport to measure the same
property of a material
Terminology relating to molecular spectroscopy ASTM E 131
Practice for describing and measuring performance
of laboratory Fourier transform near-infrared (FT-NIR) ASTM E 1944
spectrometers
Standard practice for infrared multivariate ASTM E 1655
quantitative analysis
Issued by:
American Society for Testing and Materials
100 Barr Harbor Drive, West Conshohocken
PA 19428-2959
USA
 DEP 32.31.56.30-Gen.
August 2009
Page 30
INTERNATIONAL STANDARDS
General requirements for the competence of testing ISO 17025
and calibration laboratories
Issued by:
ISO Central Secretariat
1, ch. de la Voie-Creuse
Case postale 56
CH-1211 Genève 20
Switzerland
Copies can also be obtained from national standards
organizations.
 DEP 32.31.56.30-Gen.
August 2009
Page 31
APPENDIX A FIGURES
1. NIR calibrator
3. Local
4.
QA/ QC
NIR team
NIR
2. Local
Product Quality Application(s)
Management team
Figure 1 NIR system components
process
sample Samples
Sample
Analyser
Analyser
results
Figure 2 NIR application components
 DEP 32.31.56.30-Gen.
August 2009
Page 32
Process analyser system estimate(s)
or PPTMR
Analyser (Predicted Primary
Spectrometer cell Test Method Results)
Piping Hardware
Process line

Analyser models
Instrument Multivariate Models
Transfer Cell Analyser Result(s)
Optics
Optical fibres Calibration
Optical light pipes models Outlier Result(s)
FT-NIR
Spectrophotometer
Electronics
Software
Computer
Figure 3 NIR analyser components (on-line NIR)
1.1
1.1Specification
Specification Products
Product
Reference
Reference
Design
methods
methods At/off/on -line NIR
At/off/on-line NIR
Reference
data 1.3.
1.3.Standardisation
Standardisation
1.2.
1.2.Calibration
Calibration Models 2.1.
2.1.Prediction
Prediction
training
set
results
2.2.
2.2.Validation
Validation
#.
Off-
Off-line NIR
$.
Off-line NIR
Master
COA
3.
3.Maintenance
Maintenance
Figure 4 NIR application implementation sequences
 DEP 32.31.56.30-Gen.
August 2009
Page 33
Figure 5 Light product sample injection system
 DEP 32.31.56.30-Gen.
August 2009
Page 34
LIMS
Scope!
Analyser houses Off-line NIR
DEP 32.31.50.13-Gen
DCS
NIR analyser communication
Process line

Spare parts
AMADAS
Validation
Multiplexing System
• line sampling
Sample • Spectro QC compatibility
• VRM sample
Conditioning
DEP 32.31.50.11-Gen
Laboratory
Sample analysers
Take-off & transportation
DEP 32.31.50.10-Gen
Figure 6 On-line NIR analyser and other analyser DEPs
Last page of this DEP