Hypernet Labs and Computational Hydraulics International(CHI) collaborated to integrate Galileo’s remote deployment platform with water management software PCSWMM to make much needed scalable computing resources extremely easy to access for hydrology and hydraulics modeling.
While it’s clear that the complex modeling that PCSWMM is optimized to handle would greatly benefit from the recent explosion in cloud computing, current cloud solutions are not tailored for professional scientific computing applications and are therefore inaccessible and ineffective with regard to the needs of many engineers and researchers. CHI saw a path forward through Galileo, which now makes cloud and other remote machines directly available to users within the PCSWMM interface.
The collaborative effort exemplifies Hypernet Labs’ aim of breaking down barriers to cutting edge scientific computing resources. It also demonstrates the ease with which the Galileo platform can be integrated with other software tools. Of the integration work, CHI President and CEO Rob James says,“It’s been a dream, actually, really easy.”
Since PCSWMM encourages and facilitates more detailed modeling of larger systems for stormwater, wastewater, and watershed management, run times are significant. The issue is compounded when you consider that multiple runs are required for optimization and calibration, and project deadlines constrain the number of large runs that can be executed. The problem, then, becomes one of quality and not just raw productivity.
Galileo greatly helps to alleviate this pressure in that more runs can be completed in less time, through parallelization as well as access to powerful, compute-optimized machines. Engineers can work with more complex and precise models because they can do more calibration, or experiment with more“what if” scenarios to determine the best approach.
“In the past,” says James,“people had to simplify their models to achieve that. And now with services like Galileo, they can have a fairly high resolution, accurate picture, but do the runs in parallel and use cloud services. This really reduces the cost for doing that massively parallel approach to modeling.”
The kind of modeling enabled by PCSWMM allows engineers to see the big picture, both temporally and spatially. Instead of examining the potential local effects of building a single culvert, for example, engineers can model the entire watershed to get a more holistic sense of the impact of small local changes, which can have downstream and upstream effects.
PCSWMM also allows for continuous modeling, which means the modeler can examine a year or multiple years of rainfall, plus the initial conditions before each rain event and inter event periods. This contrasts with modeling work that focuses on a single storm or flood event that is meant to encapsulate a worst case scenario. Low impact developments or green infrastructure often have a larger effect on smaller, more frequent events, so this kind of modeling can be very important for the analysis of these kinds of projects.
Again, this wider range of visibility, in terms of both time and space, is accompanied by a greater need for computation.
In existence since 1978, CHI is a forward-looking firm in the hydrology and hydraulic modeling(H&H) world. For the most recent software release, they worked hard to develop a comprehensive framework for Python scripting within PCSWMM. The aim was to facilitate connections with and support for third party software, and the PCSWMM-Galileo integration was built on top of this framework as the first case study.
The scripting environment creates a very flexible system, in which the tool that runs SWMM models on Galileo can be unilaterally updated or enhanced by CHI, Hypernet Labs, or end users wishing to customize or add features. This being said, end users do not have to code in order to use Galileo through PCSWMM. It’s a tool that can be installed with a one-click download.
The flexibility of this system is consistent with the fact that fostering a healthy community-driven environment around PCSWMM is a central motivation at CHI. This is evident in their ample community support resources as well as the fact that the software is open source. In addition to a user forum, CHI hosts a free open access peer-reviewed journal centered on PCSWMM as well as an annual conference. Over 600 papers have been published in the journal, which has an independent academic editorial board.
CHI’s software and training clients include municipalities, multi-national corporations, universities, government agencies, and local consulting firms in over 85 countries. They host 40-50 workshops per year in the U.S., Canada, and overseas.
FiveThirtyEight: the models are always limited by the quality of data collection. Anastasios Angelopoulos and other Berkeley statisticians used Galileo while making this point earlier in the summer. https://t.co/veWYmTnIht
Supercomputer-driven genetic analysis supports new hypothesis to explain bizarre coronavirus symptoms. Hypernet Labs’ Galileo aims to connect more scientists to supercomputing infrastructure so they can carry out this kind of work!