It is a very well established belief that a leak detection system based on real time transient modeling will perform independently of the pipeline control modes (pumps, regulators, etc.). These control modes are typically not very flexible and often cannot be changed to achieve the best leak detection performance. However, when this is possible, is there any rule of thumb to set operating control modes in order to achieve the best possible leak detection performance? A leak is essentially an increase of system uncertainty in a systematic way. Its detectability depends, among other things, on the leak detection system's sensitivity. Leaks can occur for any hydraulic system, but some locations are more likely than others. With this in mind, is it always true that a leak close to a flow measurement will increase system flow imbalances faster than a leak further away from the flow measurement? The present paper analyses some scenarios to provide answers to these questions. A conceptual approach follows various examples to illustrate the implication of field control modes in model performance. The given example is based on an A to B pipeline and a pressure/pressure leak detection model; however the paper focuses on providing generic conclusions.
The present study is based on the transients that a leak perturbation introduces into a pipeline system. The way that the pipeline system is controlled is considered relevant. Also examined are the relations between the pipeline control modes and the leak detection system's boundaries. This work has been written whilst considering the problem of how to detect product theft (a common activity in some countries) using real time transient models (RTTM). The idea of simulating the leak by a flow controlled delivery has been adopted. Even when the leak appearance is an obvious transient, it is useful to think in terms of initial and final steady states. Appendix A provides simple graphs that depict head and flow profiles for a number of cases and conditions. The aim is to provide a quick and simple reference that can assist in the analysis. In advance it can be seen from the Appendix A that a pressure-pressure (PP) controlled pipeline does not generate any transient in a pure PP leak detection system (LDS). These simplifications make the analysis easier but they should be reconsidered once a real project requires a decision. However, these simple graphs provide a good starting point for the analysis of the initial and final steady states for several control modes. The study also considers some common configuration errors and misleading values in order to analyze their effects.
The base case analysis consists of building two models. The first one is a hydraulic model acting as the real world: its responses are referred to as ‘field’ as they represent the physical behavior of the real hydraulic system. This model's inputs are the pressure/flow rate setpoints according to Figure 1. In this way the ‘field’ model outputs act as SCADA measurements.
