Flow Measurement Conference
Technical Paper
Reynolds number based proving -
Reducing Coriolis duty meter uncertainty
Ton Leenhoven, Endress+Hauser
Erwin H. Doorenspleet, Endress+Hauser
Reinhold Bietzker, Endress+Hauser
1. Abstract
EMA, Enhanced Metering Accuracy, is a new proving procedure for Coriolis mass flowmeters which
reduces the overall metering uncertainty. By determining the Reynolds number of the flowing liquid by
using the Coriolis it becomes possible to dynamically compensate for the metering effect caused by
variations in fluid properties. This correction is achieved by applying a Reynolds number dependent meter
factor.
2. Background - Reynolds number
In fluid mechanics the Reynolds number (Re) is a dimensionless quantity that is used to help predict
similar flow patterns in different fluid flow situations.
Almost every flow meter principle is sensitive to fluid mechanics and fluid properties. Some principles are
more sensitive than others to these changes. Flow rate, line temperature, line pressure, fluid density, fluid
viscosity and the size of the flow meter are variables which contribute to the change of flow patterns.
The fluid properties are best described by the Reynolds number of the flowing fluid. The Reynolds
number incorporates all the above mentioned parameters in one dimensionless number (Re). The Re
number is calculated as:
Where V is the mean flow velocity, the fluid density, D denotes the internal diameter of the pipe,
ṁ= the mass flow rate, the kinematic and = the dynamic viscosity.
3. Coriolis technology
Compared to other metering technologies Coriolis meters have long been recognized to be largely
immune to variable process and installation conditions.
Flow Measurement Conference
Technical Paper
However, tests performed over the years, driven by the oil industry’s increasing demand for highly
accurate measurement of hydrocarbon fluids, proved that there is also an influence on the Coriolis
measurement uncertainty caused by changes in fluid properties.
A Coriolis meter, as a direct mass meter, is independent of the fluid flow profile. However in the Coriolis
meter’ measuring tubes an interaction between the oscillating Coriolis forces and the oscillating shear
forces (viscous effect) causes a negative shift of the indicated mass rate. The ratio of the two forces is
reflected in the Reynolds number.
This ratio becomes negligible at Reynolds numbers of 30,000 and up, where the oscillating shear forces
(viscous effect) no longer have a negative impact on the measurement. There is a strongly deterministic
effect depending on the Reynolds number. The lower the Reynolds number, the larger the negative
measuring error.
The measuring error, quantified by the Reynolds number, applicable to the Coriolis technology is just as
valid for ultrasonic and turbine flowmeters and in
fact for almost all flowmeter technologies.
However, here the flow profiles or patterns are
quantified by the Reynolds number, whereas with
Coriolis meters, the Reynolds number determines
the ratio of the oscillating Coriolis forces and shear
forces. This explains why in many hydrocarbon
flow measurement applications, there is Reynolds
number dependent measurement error.
Fig.1 raw and uncompensated data from a 4” Coriolis
without internal Reynolds correction
This error can become significant for highly viscous fluids flowing at relatively low flow rates.
4. Flowmeter correction
A common procedure to correct for fluid property related measuring errors is to determine meter factors
(MF) applicable for the various products applied at different flow rates. A flow computer can use these
MFs and interpolates between the known points based on the measured actual flow rate.
Figure 2 shows a typical behavior of a 4” turbine meter on three viscosities, ranging from 1cSt via 10cSt
to 100cSt over a range of 12 to 240m³/h.
In order to compensate for this non-linear behavior three independent meter factor curves, representing
three different products, have to be implemented into the flow computer. By selecting the applicable
product the appropriate curve will be automatically activated.
Flow Measurement Conference
Technical Paper
However, it must be understood that viscosity
changes, density changes or temperature
changes are not compensated for. This would
require a new proving of the meter.
Fig.2 Turbine meter error versus flow rates
As stated, viscosity changes, density changes and or
temperature changes are not part of the curve as
variables. Including those, next to the flow rate, will
create one single universal curve which covers all fluid
characteristics and is known as the Reynolds curve.
See Figure 3.
Fig.3 Turbine meter error versus Reynolds
number
What applies for a turbine meter also applies for a Coriolis meter as well, with a distinctly lesser effect, as
shown in the following two diagrams
Figure 4 shows the meter factor curves of three different
products plotted against flow rates.
Fig.4 Meter factor curves against flow rates
In Figure 5 the three curves are summarized into
one single meter factor curve against Reynolds.
Please note that also this example is with a Coriolis
without internal Reynolds correction.
Fig.5 Meter factor against Reynolds number
Incorporating an on-line viscosity measurement in
the metering system enables the flow computer to always apply the correct meter factor which is
applicable for turbine meters and Coriolis meters without internal Reynolds correction.
Flow Measurement Conference
Technical Paper
5. Reynolds number effect compensation
Extensive tests over more than a decade have been carried out, in order to understand and quantify the
relationship between fluid properties, Reynolds number and Coriolis measurement performance. Refined
products and crude oils with densities ranging from 690kg/m³ to 960kg/m³ and in viscosity from 0.6cSt to
over 1,000cSt were used in tests involving Coriolis mass flow meters in sizes from 3” to 12”. As a result of
the test program, a patented algorithm could be developed which accounts for the relationship between
the Reynolds number and its effect on Coriolis measurement uncertainty.
By implementing an algorithm in the Coriolis mass flowmeter it is possible to enable a dynamic
compensation for the ever changing Reynolds numbers. This Reynolds correction compensates for any
change in the fluid properties based on a continuously measurement by the Coriolis flowmeter.
This correction allows a standard factory (water) calibrated meter to measure any fluid within the error
limits of +/-0.2% according to OIML R117-1.
The NMi Certin B.V. Evaluation Certificate Number TC7149 certifies this. In other words, no additional
onsite expensive calibrations are required, while using an Endress+Hauser Coriolis mass flowmeters
having a capability to apply a built-in Reynolds number compensation. The maximum permissible error
MPE will not exceed +/-0.2%.
6. Measuring Reynolds number
For an online Reynolds number calculation is required, next to an accurate density measurement, an
online viscosity measurement. Fortunately, Endress+Hauser already holds a patent on the measurement
of this parameter by using Coriolis Technology.
The viscosity measurement is based on the measurable relationship between the oscillation force of the
Coriolis measuring tubes and the opposing shear force which a fluid
causes on the inside of the tubes.
The colored image shows the additional rotation of the fluid within the
tubes caused by the Coriolis force. See figure 6a.
It needs to be understood that this movement is not caused by
laminar or turbulent flow profiles, but only by the Coriolis forces.
The sketch in figure 6b shows the rotational movement of the media
at the beginning and at the end of the flow tube where it reverses.
The energy required for this motion caused by the shear forces is
Fig.6a Shear force proportional to the media. Such the dampening of the tube drive
current becomes a directly proportional measure for viscosity.
A Coriolis massflow meter naturally already measures massflow,
density and temperature. Together with the additional parameter of
viscosity the measurement uncertainty become largely independent
of changes of velocity, line temperature, line pressure, fluid density or
viscosity.
Fig.6b Shear force
Flow Measurement Conference
Technical Paper
7. EMA, Enhanced metering accuracy with Reynolds dependent MF
The Reynolds number of the flowing fluid incorporates the effects of velocity, density, viscosity,
temperature and pressure.
The Enhanced Metering Accuracy (EMA) procedure (see Figure 7) is based on the measured Reynolds
number by the Coriolis flowmeter. The patented Endress+Hauser Reynolds algorithm provides an output
signal that enables the EMA flow computer to apply a real-time MF for every small change in the fluid
properties.
This procedure ensures that after an initial proving and determination of MF curves and without any
further on-site proving the meter error will not exceed a maximum permissible error (MPE) of +/-0.05%.
Measured value = (Pulses / K-factor) x ReC x MFRe
Fig.7 Reynolds dependent measuring value
8. Proving a 4” meter “out-of-the-box” following MPMS and OIML R117-1
In a first step a 4” Coriolis mass flowmeter Promass 84F was installed in line with a 12” Brooks piston
prover (see Figure 8) and zeroed. No further adjustment was done.
Fig.8 Proving 4’’ Coriolis
A Reynolds enabled flow computer, which we now call EMA-FC,
was connected to the Coriolis transmitter. The Coriolis
transmitter continuously provides the EMA-FC, among the mass
flow rate pulses, the fluid property data over Modbus.
See figure 9.
Fig.9 EMA Flow computer
Flow Measurement Conference
Technical Paper
The EMA-FC has the same Endress+Hauser patented algorithm embedded. This allows it to calculate
Reynolds number in the same way as the Coriolis transmitter.
This enables the EMA-FC itself to determine and apply the
Reynolds dependable meter factors, MF.
The meter was then proved over five different flow rates
(Reynolds numbers) applying five consecutive runs as per API
Fig.10 Correction factors to be programmed
MPMS. During proving, the set of metering errors, in
conjunction with their associated Reynolds numbers, were recorded. The correction factors (meter
factors, MF) were programmed in the EMA-FC.
The tested meter performed very well in terms of
uncertainty and repeatability. The maximal errors
recorded by the piston prover were +0.162% and -
0.0137%.
The red dots in the graph show the meter error points
in %. The red lines at +/-0.2% are the limits of the
most stringent MPE as per OIML R117-1 custody
transfer requirement for pipelines. The blue lines
indicate the maximum error limits of the meter proved.
Fig.11 proving results with runs, flowrates, Reynolds and errors. (1- (ME%/100) = MF)
Since the correction factors applied by the Coriolis transmitter are based on Reynolds number, the meter
is insensitive to any changes in fluid properties.
For example:
A fluid flowing through a meter at 5m/sec @ 100cSt will have the same Reynolds number as a fluid
flowing of 1m/sec @ 20cSt and consequently the same meter factor is applicable.
The meter error curve in figure 12 shows each point
and its repeatability band of +/-0.025%.
The green belt shows the area of interpolated MF
line calculated by the EMA-FC.
Consequently, one single Reynolds based correction
curve replaces the multiple products curves as we
used to apply before.
Fig.12 Meter error curve repeatability band
Flow Measurement Conference
Technical Paper
9. Viscosity influence on Meter Factors
Let’s assume that all these MF were taken at exactly the same flow rate but at different fluid conditions,
which result in different Reynolds numbers. This is possible by varying e.g. the viscosity or density.
At a stable flow rate the graph in figure 13 shows a wide spread of meter factors.
It becomes obvious that traditional proving versus flow rates
incorporates a potential for a significant range of MFs
despite a constant flow rate. If fluid properties like viscosity
and density are not accurately measured, the applied MF
might be far off the correct value.
Fig.13 Meter error @different fluid conditions
10. EMA-Flow Computer
Once the EMA-Flow Computer is programmed with MF versus Reynolds numbers, it will interpolate
between the proven points, in order to automatically calculate and implement the meter factors.
Fig.14 Interpolated Meter Factors
In the example shown in figure 14 the EMA-FC will recognize at a Re 2,000 an error of + 0.08% for which
it will implement a Meter Factor of 0.9992.
11. Meter error and applied linearization, 4” Coriolis
Knowing the meter error shown in red points and
line plotted against Reynolds, the flow computer
will apply a meter factor proportional to the
inversed meter errors (figure 15; shown in blue
points and line) in order to theoretically
compensate towards a zero error meter reading
(figure 15; shown in green points and line).
Fig.15 Applied linearization 4’’ Coriolis
Flow Measurement Conference
Technical Paper
Verifying the EMA performance of the 4” Coriolis mass flowmeter after proving.
Following the proving, the flowmeter was left untouched in its installation and the MF table established
during proving was implemented in the EMA-FC. Unlike in normal operations, we only implemented the
MF and its corresponding Re, not related to any flow rates.
Consequently, the EMA-FC adjusts the MF when the Reynolds number changes.
It receives real time the Reynolds information from the Coriolis meter, enabling it to make the appropriate
final adjustments on top of the corrections of the flowmeter at the same time.
Fig.16 Reynolds number versus EMA systems error
The improvement by this procedure on the overall metering
accuracy was demonstrated in a second step, the
verification.
The objective was to run the verification at four different
Reynolds numbers (flow rates) at five repeats each.
The runs were performed at different Reynolds numbers
compared to the ones implemented in the EMA-FC forcing it
to interpolate between those meter factors.
Fig.17 Reynolds number versus measuring error
Except for the fourth run at Re 7,400 all runs were at a new
interpolated Reynolds number.
Although we are aiming for the theoretical zero error measurement it must be understood that multiple
variables will influence the measurement and contribute to slight deviations which will always occur.
Variables like viscosity measurement, density measurement, flow computer interpolation and last but not
least, flow measurement error will contribute to the overall system error.
The four verification runs realized by the FC show the following final overall systems error of +0.04% to -
0.018%, coming from +0.162% to -0.137%. This reduced the error band width from 0.299% to 0.058%.
An improvement of the overall error bandwidth by factor 5 was achieved.
Flow Measurement Conference
Technical Paper
This makes a noticeable difference on custody transfer metering results. We recall: The red dots are the
data points of the initial proving based on the flowmeters factory water calibration and internal Reynolds
correction.
All data points are within required OIML limits of
+/-0.2% shown by the red lines.
The green dots represent the overall system
error of enhanced metering accuracy. The green
lines show the drastically improved measurement
accuracy, using EMA, Enhanced Metering
Accuracy.
Fig.18 Promass F 4’’ with overall system uncertainty
12. References
Research on the measuring performance of Coriolis with automatically applied Reynolds correction in
fluids with different viscosities has been carried out at accredited calibration facilities, including TUV/NEL
in Scotland and SPSE in France.
A comprehensive data base on a wide range of oil products, like for example Naphta, Kerosene, Ural
Crude, Arabian Heavy, Fuel, Bunker Fuel and Aztec has been gathered.
The scope of these tests included standalone Coriolis mass flowmeters in sizes of 3", 4”, 6”, 10” and 14"
without a flow computer. The Reynolds number ranged from about 10,000,000 down to almost 100,
covering a viscosity range of 0.7cSt to high viscosity liquids of 1,000cSt!
The figures 19 and 20 present all this data plotted together and show that the metering performance is
virtually independent of the fluid properties. This can only be achieved with an active Reynolds number
correction within the Coriolis transmitter.
Fig.19 Coriolis mass flowmeter error of 5 different sizes plotted
against Reynolds number. Measurements performed at TüV/NEL (3’’
– 6’’) and SPSE (10’’ and 14’’)
Flow Measurement Conference
Technical Paper
Fig.20 Coriolis mass flow meter error of 5 different sizes plotted
kinematic viscosity. Different hydrocarbons were measured
ranging from naphtha and kerosene over crude to heavy bunker
fuels. Each meter was measured with at least two different
liquids with strongly disctinct viscosities. The range of the
viscosity reached from 0.7 up to more than 1000cSt
These test results with the automatic Reynolds correction are impressive. Over the entire range of
viscosities all Coriolis mass flowmeters tested maintained their metering error within the +/-0.2%
bandwidth without any external correction factor applied.
All this information is described in detail in a paper prepared by Endress+Hauser Flowtec for the 2nd
European Flow Measurement Workshop 2014.
13. Conclusion
It has been demonstrated that Promass F and X Coriolis mass flowmeters can be used for accurate
hydrocarbon measurement, varying severely in both density and viscosity, without exceeding +/-0.2%
uncertainty. Without onsite proving, a standard out-of-the-box Promass F and X Coriolis mass flowmeter
will meet the international measurement guidelines of API and OIML R117-1.
Applying the EMA methodology and system approach as demonstrated at the NMi EuroLoop enables an
improvement of the error band from +/-0.2% to better than +/-0.05%.
Flow Measurement Conference
Technical Paper
14. Notation
V = mean flow velocity
= fluid density
D = internal diameter of the pipe
ṁ= mass flow rate
= kinematic viscosity
= dynamic viscosity
MF = Meter Factor
Re = Reynolds number
ReC = Reynolds correction
MFRe = Reynolds number dependent Meter Factor
EMA= Enhanced Metering Accuracy
15. Acknowledgement
We appreciate and thank for the support: NMi EuroLoop, SPSE, CEESI, Trapil and TUV/NEL.
