Towards scalable parametric/stochastic analysis for 2D hydrodynamic modeling via process containerization

Computationally intensive two-dimensional hydrodynamic models typically result in long runtimes, and due to the complexity of their boundary conditions, many scenarios are required to ascertain risk profiles. Additionally, Monte Carlo analysis is becoming more common for the purpose of uncertainty quantification, requiring thousands to tens of thousands of realizations for numerical convergence. This presents a unique challenge to engineers who use desktop-based software packages designed for use on local computational assets. Desktops and laptops are limited to modest amounts of CPUs, and we can no longer rely on Moore’s Law for passive increases in simulation performance.

Modern container technology offers a powerful virtualization tool for scaling traditional desktop-based physics kernels to meet the growing demands of modeling and simulation engineers. However, “containerization” of legacy software is often a challenging technical undertaking with a steep learning curve. This work will demonstrate how containerization can be leveraged to deploy the open source SWMM engine to a heterogeneous computing environment for the purpose of facilitating large scale hydrodynamic modeling.

 Click here to download a static PDF version of the presentation.

 Click here to watch recorded presentation on YouTube.