Abstract:
Gas distribution networks are systems with large pipelines, storage units, compressors,
and many other devices such as regulators and valves. These networks cover broad geographical
area and their analysis require large human resources, equipment and time
which can lead to human/measurement device errors. To avoid these errors, one possibility
is to analyse the network through modeling and simulation. This will require a
mathematical model of the complete network and a computationally efficient simulation
of the large scale complex network. In this thesis we explore fast simulation techniques
for such complex models using model order reduction. The concept of model order reduction
is to approximate large-scale dynamical systems effectively and efficiently into
much smaller dimensions and produce nearly the same input/output characteristics. We
observe the applicability of proper orthogonal decomposition (POD) based model order
reduction on gas distribution network models as they are well used for nonlinear systems
in the literature. A comparison between the original and the reduced models is made
in terms of computational time and accuracy using different gas networks. Numerical
analysis show that reduced order model is highly accurate, stable and takes lesser time
to simulate as compared to the original model.
