Every time a valve opens or closes, a pressure wave is initiated. Thus, to optimize the sensitivity of a rupture detection system, it is necessary to be able to identify all detectable pressure waves due to natural causes so that any remaining pressure waves can be analyzed as possible rupture indications. Current microprocessor, instrumentation, and telecommunication technology have advanced to the point where much more information can be derived regarding pressure waves in the pipeline. The paper reviews what is available and what still needs to be done to optimize leak and rupture detection sensitivity achievable with state of the art sensors, microprocessors, and dynamic models. In the past, our efforts on dynamic models of a pipelines have tended to overwhelm available computational limitations. Predictive models still tend to be limited by computational capabilities. Thus, they are receiving the lion's share of current development attention. Computation power has advanced to the point where even the state of the art monitoring models are well within current limitations. The key questions are how much further should we go with monitoring models to improve their capabilities for leak and rupture detection. Their usefulness for operations improvement is probably receiving enough attention.
There is a pressure wave of magnitude, pw, moving upstream away from the cause of the change in velocity, and its mirror image of magnitude, pw, moving downstream from the source of the change in velocity. What happens to those pressure waves as they move through the pipeline systems is a very complicated subject. Some of the better dynamic models of pipeline behavior are capable of considering such effects (2, 3). Currently SCADA (Supervisory Control and Data Acquisition) systems discard essentially all information about pressure waves. Typically, SCADA systems operate on a report by exception basis (4). Data is sent only when the new value represents a significant change from the prior value. The net effect of this approach is that much potentially useful information is discarded at the data acquisition end of the SCADA system substantially limiting the performance potential of the whole system for leak detection. In the next section of this paper, we will review typical SCADA system processing at the RTU (Remote Terminal Unit) end and how an alternate processing technique based on available technology can gain much more information. In the following section, what the central processing is and should be doing with that information is addressed. That discussion concentrates on the Monitoring Hydraulic Model. Since ground temperatures do change and those temperatures do affect the behavior of the fluid flow in the pipeline, they should be considered also. For that purpose, we recommend a Monitoring Thermal Model as indicated in Figure 1. Monitoring models are intended to generate the best information feasible about what is happening in the pipeline from the measurements available. Thus, it runs in real time. On the other hand, prediction should be available before the predicted event occurs.