MODULE T14

TURBINE METER
K-FACTOR CONTROL CHARTS

Tyro T14.doc Rev 3a October 1999 mmk Page 1

 CONTENTS

1.0 TURBINE METER SENSITIVITY

1.1 CRITERIA
1.2 REPEATABILITY
1.3 ACCURACY OF CALIBRATION
1.4 SENSITIVITY TO FLOW RATE
1.5 SENSITIVITY TO FLUID
1.6 STABILITY OVER TIME

2.0 PERFORMANCE MONITORING

2.1 TECHNIQUES
2.2 PERFORMANCE CURVES
2.3 CONTROL CHARTS
2.4 NORMALISATION

3.0 SETTING UP THE CONTROL CHART

3.1 DATA
3.2 ANALYSIS

4.0 INTERPRETATION OF CONTROL CHARTS

4.1 INTERPRETATION
4.2 COMPARISON

5.0 OTHER INSTRUMENTS

Tyro T14.doc Rev 3a October 1999 mmk Page 2

1.0 TURBINE METER SENSITIVITY

1.1 CRITERIA

It can be demonstrated that the accuracy of a turbine flow meter is critically dependant upon
the accuracy of the meter K-factor.

The meter K-factor uncertainty can be classified by the following terms:

a) Repeatability
b) Accuracy of calibration
c) Sensitivity to flow rate
d) Sensitivity to fluid
e) Stability over time

1.2 REPEATABILITY

Repeatability, that is the quality of the flow meter to return a consistent K-factor at specified
conditions, is essential for a fiscal flow meter.

The ideal turbine flow meter would have a constant K-factor at all flow rates and at all fluid
conditions. The real turbine meter is subject to K-factor variation due to changes in:

a) The flowrate
b) The fluid conditions, especially viscosity
c) Wear and tear
d) Mechanical inconsistencies

Non-repeatability can be manifested in the following forms:

a) Random scatter of K-factors

b) Slow shift in K-factor with time
c) Sudden (step) change in K-factor

Item (a) above, a random scatter of K-factor is the normally accepted definition of
non-repeatability and cannot be accepted in a fiscal flow meter.

Tyro T14.doc Rev 3a October 1999 mmk Page 3

1.3 ACCURACY OF CALIBRATION

The accuracy of the turbine meter calibration is critical to the overall accuracy of
measurement.

Current methods of calibration are subject to uncertainty of measurement comparable to the

turbine meter uncertainty. Consequently statistical methods are required to monitor
performance.

1.4 SENSITIVITY TO FLOW RATE

A turbine meters K-factor will change with flow rate. The quantative shift will vary with
flow meter design and between the products of different manufacturers.

This change is fundamental and must be accommodated by operating procedure.

1.5 SENSITIVITY TO FLUID

Turbine flow meters are sensitive to fluid conditions. There is a degree of sensitivity to the
direct effect of the fluid temperature, density and pressure, but the most significant effect is
the sensitivity to fluid VISCOSITY.

The fluid viscosity cannot be readily measured but viscosity is a function to temperature,
density and pressure.

A focal problem with operating a turbine meter is in differentiating between random K-factor
shifts (non-repeatability) and viscosity related K-factor shifts.

1.6 STABILITY OVER TIME

Stability over a time period can be classified in a similar manner to repeatability (section 1.2
above):

a) Random scatter of K-factor values

b) Slow shift in K-factor with time
c) Sudden (step) change in K-factor

Item (a) exhibits non-repeatability and would result in the rejection of the turbine meter.

Item (b) could be considered as “bedding-in” of the meter and can be accommodated by
routine meter proving.

Item (c) must be investigated. There may be circumstances where the meter was rejected
and there may be circumstances where the meter could continue in use.

Tyro T14.doc Rev 3a October 1999 mmk Page 4

2.0 PERFORMANCE MONITORING

2.1 TECHNIQUES

There are two techniques commonly applied to monitoring the performance of turbine flow
meters:

a) Performance curves
b) Statistical control charts

2.2 PERFORMANCE CURVES

The performance curve is a chart of meter K-factor against flowrate.

When a new meter has been installed a series of meter proving runs must be carried out to
determine the meter K-factor over a range of flow rates (20% to 100% of maximum flow
rate).

If the fluid viscosity is subject to change then more than one performance curve will be
required for each meter. Performance curves at representative fluid conditions are required.

When the performance curve has been established action limits at + 0.1% and -0.1% must
be set up.

Each time the flow meter has been proved the operator should check the K-factor obtained
against the corresponding flowrate on the performance curve.

If the K-factor is within the action limit of the performance curve then the K-factor may be
accepted.

If the K-factor is outwith the limit then the meter proof should be repeated. If after three
full meter proofs have been carried out and the K-factor can be confirmed to have shifted
the operator must take action as specified for the particular meter station.

The action could be:

a) Change to another meter stream and investigate the fault

b) Establish that the shift is due to a fluid viscosity and continue normally
c) Accept the new K-factor but advise the Metering Engineer immediately

Tyro T14.doc Rev 3a October 1999 mmk Page 5

2.3 CONTROL CHARTS

The control chart is a chart of meter K-factor against consecutive meter proofs (time).

Statistical analysis is applied to the meter proof results to set up warning and action limits.

Meter K-factor control charts provide a convenient graphical means for monitoring the
performance of turbine meters and the meter system as a whole. The control chart can be
used to assist in deciding whether an observed change in meter factor is acceptable, based on
the statistical analysis of past performance data, or whether there is some "assignable cause"
influencing the change.

In the latter case the control chart cannot indicate where the problem lies. If the change falls
outside acceptable limit of random variation then it may have been the result of damage or
wear to the meter; leaking valves; incorrect data in the flow computer etc. An investigation
will be necessary.

A meter K-factor control chart should therefore be maintained for each meter in service and
should be used to check the validity of new K-factors determined after each reproof.

2.4 NORMALISATION

Ideally the meter K-factor should be normalised to counteract the effect of flowrate and
viscosity sensitivity.

Without normalisation the K-factor will exhibit changes which could be due to any of the
following:

a) Flowrate variation
b) Viscosity variation
c) Damage to the meter
d) Meter wear and deterioration

Normalisation would remove the effects of items (a) and (b), which are normal, and
highlight the fault conditions (c) and (d). Graphical techniques and regression analysis have
been proposed.

3.0 SETTING UP THE CONTROL CHART

3.1 DATA

Meter factor control charts should be set up as soon as possible after the meter system has
been started up or a new meter has been installed. Repeated meter proving sequences
should be carried out until sufficient data has been gathered to calculate meaningful control
limits. Ideally at least 15 proving sequences should be carried out but the feasibility of this
will depend on operating conditions. In any case the results of not less than 10 proving
sequences should be acquired.

Tyro T14.doc Rev 3a October 1999 mmk Page 6

 All turbine meters are to some extent sensitive to temperature (viscosity), density, pressure
and flowrate changes. The initial K-factor data should therefore be gathered under normal
stable operating conditions.

If the system is known to operate under different operating condition regimes it may be
necessary to construct a number of control charts to accommodate the different operating
conditions. Where operating conditions are reasonably constant a control chart reflecting
average operating conditions can be used to test the validity of subsequent K-factor
determinations.

3.2 ANALYSIS

Having gathered the initial K-factor data it can be processed statistically in the following
way.

3.2.1 The data can be used to calculate a mean meter factor

K=
(K1 + K 2 + K3 .... + K n )
n

Where:

K = is the mean meter factor

K1 = etc are the meter factors determined from each proving sequence.
n = is the number of proving sequences

3.2.2 The next step is to calculate the standard deviation of the K-factor
determinations K1, K2 etc. This is done using the following formula.

∑ (K - K)
2

σ = i

n -1

Where

σ = (sigma) is the K-factor sample standard deviation

The term

∑ (K - K)
2
i

means

(K 1 - K ) + (K 2 - K ) + (K 3 - K ) .... + (K n - K )
2 2 2 2

Tyro T14.doc Rev 3a October 1999 mmk Page 7

 3.2.3 Assuming that the K-factors are randomly distributed about the mean it can be
shown statistically that:

95.5% of all K factors will lie between K ± 2 × F

99.7% of all K factors will lie between : K ± 3 × F

It is usual to adopt control limits of ± 3 × F. Any subsequent determination of

K-factor which falls within the band K ± 3 × F is said to be in control. Any
determination which falls outside this band is said to be out of control and will
require investigation.

4.0 INTERPRETATION OF CONTROL CHARTS

4.1 INTERPRETATION

In order for the control chart to provide useful information it must be kept up to date at all
times.

If K-factors drift outside or repeatedly fall outside the control limits then the system must be
investigated to find an assignable cause. The control chart can give no indication of the
likely cause of the shift and a number of possibilities need to be considered before it is
concluded that the problem lies with the meter itself. A typical list of possible causes is
given below.

(i) The operating conditions and/or flowrate have changed

(ii) The viscosity has changed
(iii) Valve leakage affecting meter proving
(iv) Detector switch malfunction
(v) Incorrect sphere inflation
(vi) Temperature of pressure transmitter malfunction
(vii) Prover lining damage
(viii) Vapour in the system
(ix) Deposits on meter blades
(x) Damage to meter blades
(xi) Meter bearing wear

4.2 COMPARISON

Where the meter system comprises a number of parallel runs a check on the K-factors of
other meters may assist in determining whether the problem, if it is a meter system problem,
lies with the meter or with the prover.

4.3 DTI GUIDANCE

Note to Rev 3, October 1999: The DTI (OGO) are investigating the preparation of
guidelines to set common standards for monitoring and accepting meter K-factors

Tyro T14.doc Rev 3a October 1999 mmk Page 8

5.0 OTHER INSTRUMENTS

Control charts or trend charts can be used to monitor the performance of many different
types of instrument.

Turbine meter control charts can be applied to positive displacement meters and can be
adapted to apply to Coriolis mass flow meters.

Trend charts and deviation charts can be applied to monitor density meters.

FIG. 1

SAMPLE STANDARD DEVIATION CALCULATION

K-Factor
Determination K-Factor
Run Number Pulses/m³ K −K (K − K )2

1 6000.36 2.08 4.33

2 6001.77 0.67 0.45

3 6005.15 2.71 7.34

4 6002.88 0.44 0.19

5 6002.03 0.41 0.17

________ _____

K = 6002.44
∑ (K - K ) = 12.48
2
i

12.48
σ =
(5 - 1)
F = ± 1.77 pulses/m³
2F = ±3.54 pulses/m³
3F = ±5.31 pulses/m³

Tyro T14.doc Rev 3a October 1999 mmk Page 9