The recommended practice covers the development of a leak detection program, far more than just the leak detection systems; company culture, control room actions, and more. However, this review is focused on the leak detection system aspect.

Examples of Metrics, KPIs and Performance Targets

Reliability KPIs

The following are examples of KPIs that may be used:

  • Number of non-leak alarms per unit time (alarms/month)
  • Number of missed leaks or percentage of missed leak events
  • Number of hours where the LDS capability is degraded by issues with component, electronics or software, etc.

Sensitivity KPIs

  • Average leak threshold. Track this separately for each leak observation time window
  • Minimum detectable leak size. Track this separately for each leak observation time window
  • Overall leak volume on which the LDS alarmed

Accuracy KPIs

  • Leak Flow Rate (Size) accuracy
  • Leak location accuracy
  • Leak volume accuracy

Diagnostic KPls

  • RTTMs - Deviations between measured and computed variable values indicate performance problems with the LDS.
  • RTTMs – Large deviations in adjustable (tuning) parameters related to pipe friction and heat transfer
  • Statistical Mass/Volume Balance: MCFD is able to learn full range of flow difference and continually learns through the day (same idea for static)
  • CPMs - Correctly identifies pumping scenario (switch LDUs or states depending on flow or no flow)
  • CPMs - System events reported such as bad data and known issues.
  • CPMs - Processing time on computer should be within a range for each installation

Robustness KPIs

How a LDS performs when some of the requirements of the LDS, such as measurements, are not available.

  • Loss of measurements, for instance, due to meter failure
  • Loss of communication
  • Unusual operating condition, such as draining the pipeline for maintenance, pigging, or operation with slack in the pipeline
  • LDS behavior during transient operating conditions
  • LDS performance when pipelines operations do not conform to the requirements of the LDS, for example:
  • Shutdown conditions when LDS is not intended to function
  • Column separation line conditions, when the LDS is not designed to deal with these

Performance Targets

Performance targets for a particular pipeline are a good way to select an LDS and to evaluate continual improvement possibilities for leak detection on that particular pipeline. Pipeline operators are encouraged to group performance targets according to the metrics of sensitivity, reliability, robustness, and accuracy. All of these metrics may not be equally appropriate for every LDS, so they can be assigned different importance depending on the pipeline.

  • The targets can be broadly stated (e.g. sensitivity of less than 20 %)
  • The attributes can be more focused or be given as ranges (e.g. sensitivity of less than 5 % to 10 %)
  • A LDS may have multiple performance targets that depend on shutdown, steady state, and transient operation. For example, the sensitivity target during different types of operations may be different.

Determination of Performance Targets

Performance targets should be based on sound engineering expertise and judgment. Performance targets may be determined by estimation or observation of the LDS performance. They will not provide an exact performance number but rather indicate what may be reasonably expected from a LDS.

API 1149 analysis provides a theoretical best case performance of a CPM system on a pipeline operating in nearly steady conditions. In the real world of transient conditions, CPMs will not achieve this theoretical performance. My preference is the other choice, observing the same system’s performance estimate the real performance, but that may not drive improvement. Together these numbers may provide a resample estimate of what is likely to be an attainable goal for a particular LDS.

Determination of Performance Targets by Estimation

Estimation uses detailed knowledge of how the inputs to the LDS and the operational environment affect the performance. API 1149 is an example of this approach.

Advantages of estimation:

  • May be performed before an LDS is implemented
  • Can compare the performance of different LDSs for a pipeline
  • Predicts the effects of changes to the configuration or operation of the asset or of the LDS

Disadvantages of estimation:

  • A theoretical exercise that is not accurate, and the accuracy of the estimation is not known
  • Comparing different LDSs, if the difference in accuracy of the estimations is of the same order as the difference in the estimated accuracy, it provides no basis for selection and may even be misleading
  • The configuration of the asset should be known in great detail, including items such as accuracy and precision of inputs that are difficult if not impossible to know or estimate
  • The physical principle of the technique used for the LDS should be known in detail however, this may not be available for proprietary technologies
  • Derivation of the uncertainty relations for an LDS require a thorough understanding of the mathematics and statistics of uncertainty analysis
  • Determination of Performance Targets by Observation
  • Analysis of historic performance of the LDS and/or testing to establish the performance

Advantages of observation:

  • Provides a definitive result for the performance
  • Accounts for as-built, real-world conditions

The disadvantages of observation:

  • Does not identify factors limiting the performance
  • Does not provide predictive information on how changing the configuration or operation of the pipeline system may affect performance.

These two methods (estimation and observation) of determining performance targets are not exclusive.

Testing

LDSs used in an LDP must be tested when implemented and on a regular basis not to exceed five (5) years or when there has been a significant change in the pipeline's operation or a physical change in the configuration.

The testing process should be based on the requirements of API 1130, the unique aspects of the LDS, and the specific pipeline. Actual leaks may be used in lieu of periodic testing, as outlined in API 1130.

API 1175 emphasizes the importance of creating a detailed testing plan per API 1130 as well as the cautionary aspects of testing also outlined in API 1130. The test plan, prepared prior to testing, should document the purpose of the test, the methods that will be employed, and the process and procedures that should be followed.

The pipeline operator may use the test results to improve the culture, procedures, and knowledge Levels Company-wide. Remember that knowledge or skills acquired in training should be tested as a part of the training program.

Read part 4

References

  1. American Petroleum Association. (2014). Recommended Practice 1175, Pipeline Leak Detection Program Management (2st ed.). Washington, DC: Author.

A Look Inside API Recommended Practice 1175 Series

A Look Inside API Recommended Practice 1175, Part 1
A Look Inside API Recommended Practice 1175, Part 2
A Look Inside API Recommended Practice 1175, Part 4
A Look Inside API Recommended Practice 1175, Part 5
A Look Inside API Recommended Practice 1175, Part 6
A Look Inside API Recommended Practice 1175, Part 7
A Look Inside API Recommended Practice 1175, Part 8

Categories: Best practice advice Industry update

By: Atmos International
Date: 12 April 2019