Scope
This guideline establishes the requirements for generation of meter
performance charts and the criteria by which acceptable performance is
determined.
Frequency Every meter proving.
METER PERFORMANCE MONITORING GUIDELINES & PROCEDURE
Statistical Methods
Statistical methods may then be employed to assist the operator in
deciding whether the result of a meter prove shall be accepted. API
MPMS Section Chapter 13.2 (‘Meter Proving’) shall be consulted for
guidance in this area.
The following points shall be considered when devising a strategy for
statistical proving:
o If it is not possible to achieve the repeatability criteria within 10
runs, then it may be the case that attempts to obtain a
representative m-factor shall be unsuccessful even if many more
runs are completed.
o Under these circumstances it is recommended that a simple
arithmetic average of a large number of runs is calculated. The
number of runs required shall depend on the specific
circumstances.
o The poor repeatability may be caused by process instability. If this
instability is in some way ‘cyclical’ then the number of runs shall be
sufficient to cover a complete ‘cycle’, even if this number is
relatively large. Alternatively, if the instability is not regular, a
smaller number of runs would probably be sufficient.
o A large number of runs may be impractical in certain
circumstances, especially those where, due to the large size of the
prover or the low prevailing flow rates, the time taken to complete
all the proves would be unacceptably high.
o Where the number of averaged prove runs is less than 20, a
statistical analysis shall be performed on them in order that
‘outliers’ can be rejected. This can also be performed when the
number of runs exceeds 20, although there the larger total provides
some insurance against the undue influence of individual outlier
proves.
o Any unexplained shift in the meter factor in excess of 0.1% shall be
reverified by a repeat run prior to its acceptance.
Meter Performance Curves
Meter Performance Curves plotting the history of meter factors versus flow
rate shall be maintained for each meter. This allows the Operator to
determine the variation in flow rate that would cause a shift in k-factor of
greater than the value of one standard deviation r which would essentially
be a trigger to reprove the meter.
These curves shall cover a range from 10% to 100% of maximum flow
rate, subject to any system restriction on flow rate. It is recommended that
a number of proves (typically 5) are completed at each of these nominal
flow rate points, between which intervals of 10% are suggested.
If the variation in k-factor of a turbine meter under normal operating
conditions is significantly greater than the value of 0.1 %referred to above
there shall be an increase in the requirement to re-prove the meter.