One of the most significant issues for those who deal with math models and calculation tools is to "tune" the models to the actual operating conditions to the end of obtaining the best results. Obviously, in an existing network this "tuning" is carried out by comparing the measured variables and those obtained by making calculations under different operating conditions. As a matter of fact, in a pipeline system in operation different actions can be performed to check up the accuracy of the simulation model vs the system performance, e.g., run or stop compressor stations; put loops into or out operation; open or close block valves; etc. But all the above-mentioned maneuvers should be considered"standard conditions"and can be frequently performed in any pipeline network. Now then, the case we are focussing on in this paper is the analysis of a pipe rupture. Needless to say, this is a"non standard condition"and when this kind of incident takes place, it turns out to be an extraordinary and unique opportunity to compare simulation analysis results vs measured data. Therefore, the aim of this paper it to gain a better understanding of the model capabilities and accuracy to study this type of unusual phenomenon (non isothermical flow). In evaluating the rupture phenomenon and the gas mass flowing through a given pipe area, we might establish an analogy between the affected pipe area and the throat of an adiabatic nozzle, where the nozzle walls become parallel.
This case study is a 20-inch-wide and 650-mile-long pipeline that carries 85 MMSCFD of natural gas from Argentina to Chile, crossing the Andes mountain chain reaching a height of up to 15000 ft above sea level. On January 27, 2002, at 10:44 p.m. a rupture took place between the V2AR and V3AR valves on the Argentine side as shown above. Rupture data were measured by the SCADA system in V1AR,V3AR and V5AR. The objective of this paper is to compare these measured data with results obtained from a simulation analysis. Our mission will be to calculate theamount of gas lostas a result of the rupture, for which purpose we need to know the sequence of events reported that will be used as input data for the simulation. Also, we need to determine when V2AR was closed since there are no SCADA measurement available for that valve.
With the exception of the rupture and the closure of V2AR, the rest of the above-cited actions were taken by the shift operators from the control room in Buenos Aires. As exhibited in the network scheme, the rupture could have been completely isolated by closing only V1AR and V3AR, plus the subsequent automatic closure of V2AR as a result of the minimum pressure set point (355 psig).
