Inverse laplace transform for transient-state fluid line network simulation

Abstract

Inverse Laplace transform methods have a long history in the development of time-domain fluid line models. This paper presents a study combining the new Laplace-domain input/output (I/O) model derived from the network admittance matrix with the Fourier series expansion numerical inverse Laplace transform (NILT) to serve as a time-domain simulation model. A series of theorems are presented demonstrating the stability of the I/O model, which is important for the construction of the NILT method. In the previous work by the first author, the Fourier series expansion algorithm was studied, where qualitative relationships between the parameters and numerical errors were analyzed, and reliable parameter heuristics were developed. These heuristics are used for a series of numerical examples dealing with networks of 11, 35, 51, and 94 pipes by using five different pipe models. The examples are used as the basis from which the accuracy and numerical efficiency of the proposed NILT are compared to the standard method of characteristics (MOCs) model for transient pipeline networks. Findings show that, for all case studies considered, the proposed NILT is numerically efficient for the pipe types involving convolution operations, and it is accurate for networks composed of both linear and nonlinear pipe types.