Parallel computing in conceptual sewer simulations

摘要

Integrated urban drainage modelling is used to analyze how existing urban drainage systemsnrespond to particular conditions. Based on these integrated models, researchers and engineersnare able to e.g. estimate long-term pollution effects, optimize the behaviour of a system byncomparing impacts of different measures on the desired target value or get new insights onnsystems interactions. Although the use of simplified conceptual models reduces thencomputational time significantly, searching the enormous vector space that is given by comparingndifferent measures or that the input parameters span, leads to the fact, that computational timenis still a limiting factor. Owing to the stagnation of single thread performance in computers andnthe rising number of cores one needs to adapt algorithms to the parallel nature of the new CPUsnto fully utilize the available computing power. In this work a new developed software tool namednCD3 for parallel computing in integrated urban drainage systems is introduced. From threeninvestigated parallel strategies two showed promising results and one results in a speedup of upnto 4.2 on an eight-way hyperthreaded quad core CPU and shows even for all investigated sewernsystems significant run-time reductions.