SCADA-reported pressure and temperature data on typical RTU spacings along with inlet and delivery flows can be used to drive a real time model of a pipeline network. The model in turn can respond to detect, size, and locate leaks. The approach described in the paper has previously been applied successfully in the field to all - liquid systems. Studies presented show that the approach will work equally well for gas networks, dense phase ethylene systems and in liquid - filled portions of networks which are operating in slack-line flow. It is currently being implemented in the field for the latter case.


This paper is to describe a robust procedure for using typical SCADA data for detecting the existence, size, and location of leaks in pipelines carrying a variety of fluids. It is also to describe a related approach for of detectability of leaks in a given system before the procedure is actually implemented in the field. The latter permits an evaluation in advance of weaknesses, if any, in the SCADA data to be used. It also enables the determination of realistic expectations of leak detection performance during the design phase of a leak-detection project.


The leak detection procedure described here is an enhancement of an earlier system called TRIM which w a s described in 19801. To implement the procedure an accurate computer model of each element of a pipeline network is first built. This includes the pipe geometry and a representation of the fluid properties, which would generally include the bulk modulus, the viscosity, the thermal modulus, and the reference density of each batch of material which would be flowing in the pipeline. The required SCADA data are continual updates of temperatures and pressures at all RTU locations as well as inlet fluid properties, and all inlet and out let flow rates. Leak detection is accomplished by requiring that the concurrently running network model agree with the measured pressures at each SCADA scan at the locations corresponding to the measured values. The temperatures and other fluid properties entering the model are caused to match those of the fluids entering the pipeline at the corresponding locations. Figure 1 shows the layout of a typical pipeline and model system. This schematic shows a pipeline with 7 RTU locations. The typical spacing for RTU locations in commercial pipelines ranges from 25 to 100 miles. The measured pressures (and temperatures, which are not shown) are fed into the model. After running briefly, a properly constructed model, with temperature profiles and batches properly located, will be in "lock-step" with the pipeline. Being in lock-step means that all the modeled pressures, flows and temperatures agree with their counterparts in the pipeline. The initialization of temperature and batch locations will be discussed later. The time required for lock-step is usually in the range of 5 minutes for fluids with high bulk modulus, such as liquids, or 50 minutes for fluids with low bulk modulus, such as gases.

This content is only available via PDF.
You can access this article if you purchase or spend a download.