This paper presents a method for assessing the leak detection performance of commercial real-time pipeline monitoring software. The method has two key advantages:
it is applicable to all types of leak detection software that accept ASCII input data, and
no prior knowledge of the software calculation method is required.
Real Time Transient Model (RTTM) performance has been assessed for the Statoil-operated Troll oil pipeline (TOR1). Leak scenario boundary data sets (pressure, flow and temperature) are calculated by transient simulation of leaks using production data as input. This was accomplished using an Olga pipeline model tuned against production data. RTTM simulations indicate that for the TOR1 system with current leak detection thresholds, leaks representing approximately 5 % of the pipeline flow rate are detectable within about two hours of the leak onset. A theoretical assessment based on the API publication 1149 indicates that leaks of this size should be detectable within about 15 minutes. The approach outlined in this paper enables Statoil to carry out targeted alterations to improve the leak detection performance of these software systems.
Liquid transport pipelines on the Norwegian Continental Shelf (NCS) show excellent performance. In the last decade, approximately 740 miles of pipelines conveyed an estimated 900 Mbbl of liquid hydrocarbons . In the same period, a single leak event caused a total leakage of 777 bbl of condensate . The reasons for this pipeline performance include robust design, and careful operation and maintenance. Nevertheless, pipeline leaks and ruptures can occur, for several reasons : Pipeline corrosion and wear; Operation outside design limits; Unintentional third party damage; Intentional damage. Analogous leak events in the (much larger) gas transportation system include the 2004 Jotun gas export pipeline rupture  and the 2008 Kvitebjørn gas export pipeline leak . Both of these incidents were preceded by large mechanical stresses on the pipelines, caused by unintentional third party interference. These leak causes are also relevant to liquid pipelines. With respect to health, safety, the environment and economics, the purpose of leak detection is to reduce the consequences of the leak. Full pipeline rupture will, with few exceptions, be discovered by control room operators within a short time, through abnormal declining pressures or flow rates. However, sizeable leaks can give only small sensor deviations and thus pass undetected unless sensitive, dedicated leak detection systems are installed. Time-continuous methods covering entire pipelines include simple or complex mass balance (inventory tracking) methods, acoustic detection of leaks, plus other forms of distributed sensing such as oil permeable tubes. As with all leak detection methods, the goal is to detect the leak-causing event and the resulting fluid emanating from the pipeline. These technologies require:
Coverage of the entire pipeline length;
A robust and unambiguous correlation between the leak and the measured quantity (loss of mass, acoustic signal, detected hydrocarbon), and;
A means of communication and evaluation between the sensor and operational personnel.