The timely detection of leaks and the accurate tracking of product batches in pipelines carrying highly compressible liquids has long proven to be a difficult task. The implementation of a Real Time Transient Model on a major Alberta liquids pipeline network has reduced leak detection times on that network and has resulted in accurate batch tracking. The paper describes a real time transient computer model that is being used by Peace Pipe Line Ltd. to model for purposes of batch tracking and leak detection, a 1200 kilometre pipeline network carrying a wide range of hydrocarbon liquids ranging from ethane rich Natural Gas Liquids (NGL) to crude oi1.
The paper will present the basis of real time transient modelling, an overview of the mathematical solutions involved, the practical considerations of interfacing a numeric model to non-perfect data measurement and gathering systems and the methods whereby the computed hydraulic data is presented to the pipeline dispatcher in a user friendly way.
A discussion of actual leak detection tests and performance data will also be presented.
The use of computers to perform leak detection on liquid pipelines has until recent years been mainly restricted to simple material balance algorithms.
These material balance algorithms compare metered inlets into a pipeline system with metered outlets and generate an alarm whenever the difference between the two exceeds a preset alarm threshold. The selection of al arm thresholds is crucial since large thresholds allow relatively large leaks to go unnoticed and small thresholds will cause many false alarms to be generated.
Ideally, alarm thresholds for material balance algorithms should be selected to take into account metering errors and maximum normal changes in linefill (i.e. linepack) as the pipeline system pressures up and down due to station startups and shutdowns.
This threshold selection procedure is valid for small pipelines carrying relatively incompressible liquids but becomes questionable for pipelines where normal linepack variations are large.
Methods whereby the linepack variations are calculated and taken into account when performing the material balance calculation are referred to as Line Pack Compensation (LPC) algorithms. LPC allows for substantial reductions in the alarm thresholds but their accuracy is related to the number of pressure and temperature measurements on the pipeline system. In ideal circumstances, such measurements should be spaced every 5 to 10 kilometres on the pipeline. In practice, reasonable accuracy can be obtained with 10 to 20 kilometre spacings.
On pipeline systems where such measurements are only available every 50 to 100 kilometres (i.e. at pump stations). LPC algorithms lose their precision and other techniques must be considered. In particular, Real Time Transient Modelling (RTTM) techniques allow for the computation of linefill by using complex transient hydraulic equations driven by the few sparse pressure and temperature measurements that do exist on the pipeline system. RTTM methods have now been implemented on several pipeline systems worldwide and have met with varying degrees of success.
This paper describes the implementation of a Real Time Transient Model a major Alberta liquids pipeline network owned and operated by Peace Pipe Line Ltd.