A Topological Approach to Partitioning Flow Networks for Parallel Simulation

System partitioning for effective simulation of civil infrastructure flow networks on parallel processors is a nontrivial problem. Arbitrary focused only balancing processor workload can lead to large interprocessor communication burden that limits speedup. Thus, there need intelligent algorithms balance the estimated computational load while minimizing number connections between partitions. Graph theory provides widely used methods, but these are applicable with power-law connectivity and where proportional system nodes—conditions do not hold finite-volume solution water drainage (e.g., river systems, stormwater systems). This paper presents novel BIPquick algorithm, which shown be an approach identifying network partitions reduced systems directed acyclic graphs (DAGs) have physical limit per node. Novel developments include (1) node-cut allows function exactly balanced in work link length nodes, (2) finite-pass ensures known time, (3) new scaling metric simple evaluation comparison different results. The model tested up 10,000