A Task Graph Parallel Computing Architecture for Distillation Column Simulation on Flash Granularity

• Published in: Computer Aided Chemical Engineering Vol. 53; pp. 373 - 378
• Main Authors: Qu, Shifeng, Yang, Shaoyi, Du, Wenli, Qian, Feng
• Format: Book Chapter
• Language: English
• Published: 2024
• Subjects: distillation column; parallel computing; process simulation; task graph computation; distillation column; process simulation; parallel computing; task graph computation
• ISBN: 9780443288241; 0443288240
• ISSN: 1570-7946
• DOI: 10.1016/B978-0-443-28824-1.50063-6

Simulations of distillation columns stand as the most time-consuming unit operation calculations, profoundly affecting both the efficiency and practical feasibility of process simulation, design and optimization. The complexity of distillation columns stems from the involvement of multiple chemical components as well as extensive energy and mass transfer stages. In this paper, a novel parallel computing architecture is presented specifically for the distillation process. The architecture revolutionizes distillation calculations by decomposing them into smaller, independently and concurrently computable subtasks, i.e., flash calculations and solving tridiagonal matrices for material balance. At the core of the architecture is the creation of an executable task graph, which maps out dependencies between subtasks and incorporates conditional tasking for efficient control-flow management during iterative material and energy balance calculations. The proposed architecture was applied to distillation cases with different computation intensities to ascertain its effectiveness. The results reveal that under optimal parallel granularity, parallelization substantially improves the efficiency of distillation simulation.