Need Speed? Get Model Simulation Results, When You Need Them

Posted on September 6, 2018 by Matthew Anderson


There’s no shortage of modeling packages available to calculate flooding depths. But what if you need to model 1D infrastructure and 2D surface flooding? Freely available 2D analysis tools can take hours, or days, to run complex rainfall scenarios and don’t have the capability to accurately portray constructed flood catchments. And in an age when engineers have to do more within the same amount of time to stay ahead, analysis time is critical.

The solution here is in making the most from your existing hardware to quickly build and simulate 1D as well as 2D overland flooding components.

If you need to conduct holistic modeling that includes hydrology, hydraulics, water quality, 2D overland and sanitary flows, running simulations on these complex components can take hours. Long simulation run times may be okay if you are using your model for annual or even quinquennial analysis for master planning. However, if you use a storm, sanitary and overland flood model for operational action and flood risk assessment then saved time allows for quicker determination for mitigation strategies.

Optimizing existing hardware can save you hours

Typically, Central Processing Units (CPUs) might leverage a single processor to run complete 1D / 2D model simulations or the Graphics Processing Units (GPUs) would perform a simulation while not including the underground infrastructure. Now, significantly faster simulations are achieved by multi-threading core processors and making use of GPU cards in highly parallelized computational (HPC) schemes for 1D / 2D analysis.  This change is highly impactful to modelers’ productivity and can output valuable information from your model in a fraction of the time compared to traditional platforms.

Leveraging HPCs suited to run on GPU devices or as a multi-threaded CPU opens the technology to answer flooding questions with drastically shortened model simulation run times. GPUs are commonly used to visualize graphics, but with an upcoming XPSWMM 2018.2 release, the new XP2D Extreme engine – powered by TUFLOW – provides an engine that accelerates 1D infrastructure and 2D surface flood combined model simulations. Making the most from this tech leads to a 5x – 100x simulation improvement compared to single-thread CPU computing.

Improved productivity = faster results for risk assessment

Currently, many people approach flood modeling by picking up the HEC-RAS 2D simulations.  These simulations – especially in the urban environment – are a highly complex mixture of surface and underground systems, of which underground system is never included.  A relatively simple comparison 2D simulation produced a 2.5x faster run speed on the single core of the CPU, and 8xfaster speed on the Nvidia Quadro GPU with XPSWMM/XPSTORM 2018.1.

The XP2D Extreme engine produces results 3x faster than the latest HEC-RAS release, which itself boasted of a 2x reduction in time in the latest release.  The XP2D Extreme engine provided by the upcoming 2018.2 release runs the simulation 13.5x faster than HEC-RAS originally does.  Imagine what can be done with this saved time. It opens the opportunity to evaluate multiple scenarios if you achieve the results significantly faster, and to provide faster project times to your clients and project managers.

Even more, the simulation took up ALL the computational power available when run on HEC-RAS, limiting users’ ability to multi-task while the simulation was running. In the updated XPSWMM, users are potentially able to continue uninterrupted while sending emails, responding to calls, and running a different model in the background. All of which vastly improves productivity.

When it comes to flood modeling, time may not equal money, but it does equal action. As technology continues to evolve, predictive analysis is becoming the norm. Use of hydraulic/hydrologic models for rapid simulation and “live” modeling is an essential tool for extensive rainfall and flooding analysis.

Share this post!

We want to hear what you think!