 |
Center for Nonlinear Studies |
This seminar will examine the potential for utilizing massive parallelization on General-purpose computing on graphics processing units (GPGPU) for pre-computing gas thermodynamic quantities by interpolation for effective evaluation of such parameters during pipeline simulation. Inverse transformations of these quantities can also be calculated similarly to significantly accelerate stationary and dynamic simulations. This concept, together with massive parallelization of the applied space-time numerical scheme, are used to implement a simulator suitable for modeling highly dynamic gas flow in pipelines. The simulator is successfully applied to reconstruct gas flow in a pipeline system after an accident and for solving the inverse problem of gas pipeline rupture localization. Moving beyond parallelization on a multi-core central processing unit (CPU), cluster of multi-core computers and/or cloud computing can be used to accelerate simulation-based steady state and transient pipeline optimization under uncertainty, where stochastic optimization algorithms are used to solve the inverse problem of finding unknown control parameters for a simulated scenario.
Igor Mračka is a postdoctoral research fellow at the Mathematical Institute of the Slovak Academy of Sciences, Bratislava, Slovakia. He has attained his PhD. degree in applied mathematics from the Comenius University, Bratislava, Slovakia. Currently, his research focuses on gas transport simulations, leak detection and location, signal processing and machine learning.
Host: Anatoly Zlotnik