From the May-2017 issue
What’s Hot in Modeling Software
Developers of hydraulic and hydrologic modeling software have recently unveiled exciting new program updates and products to increase the efficiency and effectiveness. Dramatic decreases in model run times, thoughtful assembly of multiple information sources, and greater flexibility within hydrologic connectivity options demonstrate just a few ways these changes have improved the science of modeling. A complete list of every change recently introduced to the market would be overwhelming to the average modeler. Instead, a few case studies of some impressive changes are presented here.
Even small firms with limited budgets can make significant improvements to their overall productivity by updating as appropriate to their workload and budget. But any computing change demands caution. Conduct a full investigation into advantages and disadvantages of the new product, upgrade glitches and headaches experienced by others, and how changes within the new programming may affect modeling results, especially for ongoing projects or models periodically updated every few years. Not all software upgrades improve workflow or model results in every situation. Every engineer, company, and organization must consider their specific circumstances.
Pre-conference workshops Repairing Entrenched, Incised, and Degraded (Urbanized) Streams – Techniques and Case Studies Monday August 28, 2017 and BMP Selection to Improve Your Watershed Monday August 28, 2017. You may register for these without also registering for the annual conference. Download the StormCon Conference Program here.
Real-Time Flash Flood Prediction
Originally established in 1991, the city of Austin’s Watershed Protection Department (WPD) focuses on reducing the effects of flooding, erosion, and water pollution. The department’s role is especially important as Austin lies within the swath of Central Texas known as Flash Flood Alley, an area of steep terrain and shallow soil that also receives very high rainfall. The topographic transition between the state’s Hill Country and coastal plain regions means that frequently, atmospheric moisture sent aloft by the Gulf of Mexico delivers intense rainfall events across the transition zone, which runs roughly from Dallas in the northeast, through Waco and Austin, to San Antonio in the southeast.
The frequency of flash flooding and resulting destruction led WPD to create a Flood Early Warning System roughly 30 years ago to continually monitor rainfall, water levels, and low water crossings within Austin. “We have significant numbers of flood hazards, on the order of 10,000 properties with some level of flood risk, and around 400 roads that flood in a 100-year event,” says Kevin Shunk, floodplain administrator at WPD.
Add Stormwater Weekly and Water Efficiency Weekly to your Newsletter Preferences and keep up with the latest articles on water: green infrastructure, smart meters, stormwater drainage and management, water quality monitoring and water treatment.
The monitoring system utilizes 130 rain or creek gauges, gauge-adjusted radar rainfall data, 15 low water crossing signals, cameras at low water crossings, and predictive modeling and mapping. “When it rains, I have my methods of opening up various software, and those are my tools to watch things,” says Shunk. “We wanted to have software that can ingest these different sources of information and display it in one location, on a single site or map for us to look at.”
Unfortunately, the motivator to develop a consolidated system was a flood of record on October 31, 2013, on Onion Creek in neighborhoods between Interstate 35 and the Austin-Bergstrom International Airport. In 15 minutes, Onion Creek rose 11 feet and crested at a record 41 feet later in the day. The speed of inundation caught residents completely off guard and resulted in fourth deaths.
“We had about 850-plus homes flood, some with up to eight feet of water in them,” says Shunk. “It revealed some issues we had with the Flood Early Warning System and overall communication across different departments within the city.”
The need to create an integrated system led to Austin to bring in Vieux and Associates Inc., a firm based in Norman, OK, specializing in rainfall and runoff software. Together, WPD and Vieux are developing FloodVieux, a common operating picture of rainfall, water level, and runoff information, by connecting diverse sets of mission-critical data. By leveraging cloud computing through Amazon Web Services, FloodVieux facilitates situational awareness during emergency events. By harnessing the Web app, FloodVieux allows quick access to real-time information and easy retrieval of archival data. This software as a service (SaaS) mimics other cloud-based services, but specializes as an integrated system for display, data visualization, analysis, and notification used in flood warning and water management.
Karl McArthur, supervising engineer with WPD, explains that a large situational map view comprises the centerpiece of FloodVieux’s interface and can be surrounded by smaller pieces depicting source information, such as remote low water crossing pictures and real-time data from a particular rain or stream gauge watch point. In short, it combines all the source data in a single navigable program, focused around the overall map view.
“[FloodVieux] draws information from our other software models to help with predictions and operations. It’s a real-time flood inundation mapping application that can build on those other models that predict flows based on gauge-adjusted radar data or predicted rainfall data,” says McArthur. “It pulls together pieces we already had and puts them all more at our fingertips.”
The program can also monitor and report on social media activity, particularly Twitter, for potential flood conditions. “If it finds any tweets that have the word ‘flood’ in them, it populates them on a map,” says Shunk. “If you see 50 ‘flood’ tweets in one area, that gives you an idea something’s going on there.”
Because FloodVieux monitors the entire Austin metro area at one time, it can greatly reduce the burden on human monitoring of such extensive real-time conditions and spot where potential flood issues may arise. By writing conditional rules within the program, WPD can focus its attention on the most critical areas.
“We can write rules that trigger us to take appropriate action,” says Shunk. If specified conditions occur and relevant rules triggered, WPD can deploy public safety personnel. This rule-writing flexibility incorporates institutional knowledge into the system and facilitates better handling of future flood problems due to development-induced changes to drainage infrastructure.
Once the floodwaters subside, FloodVieux will also make post-event reporting easier. “It will summarize everything that went on during the storm for us, greatly facilitating that process and helping us identify places we could improve,” says McArthur, although this function has not been fully tested yet as FloodVieux is still under development.
“Some of it’s in place that we can use as an alpha version,” says Shunk. “All the pieces aren’t there yet. We’re building it on the fly, so something we learn today helps change the design of it next week.”
At present, WPD plans to use FloodVieux internally with Austin city government, providing flood-related information to other departments, but there has been initial discussion on making the program available outside WPD. “The emergency management department is very interested in having this available in the emergency operations center and maybe some expansion of it, similar to [FloodVieux], for their use as well.”
Vieux and Associates indicates that this is certainly possible—for other Austin departments, or totally unrelated groups—because the program builds on a common foundation and is extended to meet unique needs. “There are a lot of different services municipalities provide that might be able to benefit from such software that ingests a variety of information,” says Shunk. “It’s a way of taking in information and somewhat automating its management to alert the user to actions that need to be taken.”
Both Shunk and McArthur have found the software easy to use. “It’s not so much a learning curve on how to use it as it is a learning curve to figure out how to create it uniquely for us,” says Shunk.
However, McArthur points out that if a program like FloodVieux is intended for real-time flood conditions and a municipality does not already have a flood early warning system, it would be a big effort to generate all the contributing data pieces.
“If someone were to start from scratch, it’s a monumental effort just to have enough data to make the decisions,” says Shunk. “We were fortunate to already have the data in hand, and we’re using it to make decisions.”
The FloodVieux common operating system connects diverse internal and external datasets including US Geological Survey stream gauges and National Weather Service radar and forecast information, and can be expanded over time with development of additional data sources. As an Amazon cloud-based service, FloodVieux has a reliable and secure foundation that can be leveraged for application by other communities who have a need for some or all the components already developed. Customized permissions can be configured to allow limited data viewing by user groups within the community, including optional public viewing of generalized or less technical data.
As WPD and Vieux continue to develop the FloodVieux program, all are eager to make revisions and refine its use. “We have great hopes for it and we’re very optimistic that it’s going to be helpful for us,” says Shunk.
Productivity Gains in Stormwater Design
For some engineering consultants, traditional stormwater pond design can be one of the most frequent projects within their services portfolio. An engineer with prior experience can size and produce plans with assistance from modeling software, AutoCAD, and supporting tools with relative ease. But even very routine projects can benefit from modeling software improvements. McBade Engineers and Consultants, based in Youngsville, LA, recently deployed XPDRAINAGE 2017 and have already experienced remarkable improvements in project efficiency.
This latest project involved designing the right stormwater pond for a subdivision to avoid stormwater flooding,” says Lucas Hudspeth, an engineering intern III at McBade. “We wanted to see how an active stormwater event runs in the model, checking if there are any unanticipated overflows or shortages and making sure the pond and its outlets are sized correctly.”
Hudspeth has found that XPDRAINAGE 2017 is very user friendly and designed intuitively, including the interface. His prior experience with US Army Corps of Engineers’ HEC-RAS and Louisiana Department of Transportation and Development’s HYDR drainage programs provided him basic initial familiarity with XPDRAINAGE 2017, although this was his first experience using new software.
“It’s easier to navigate on XPDRAINAGE than on other programs,” says Hudspeth. “When putting in elevations, pipe sizes, lengths of pipe, and so on, it’s all in a chart. It’s easy to follow and you don’t have to hunt around for the information.”
The XPDRAINAGE 2017 interface employs standard ribbon and toolbox options organized around a typical workflow involved in designing stormwater facilities. As the user inputs information and completes actions, the interface responds accordingly and adapts its options, helping guide the user through the process. XPDRAINAGE 2017 also seamlessly integrates with AutoCAD, making import and export of layout and facility details efficient and eliminating possible errors caused by manually transferring data between applications.
The large profile and plan views within the user interface facilitate visualization of design input, depicting spatial and connection properties of the proposed facilities. “Design is definitely easier on XPDRAINAGE, and it’s more foolproof,” says Hudspeth. “If there is a mistake, it’s easier to spot. It’s also easier to check the design information if you’re a senior engineer, making sure everything is fluid and running correctly.”
This can be especially helpful when designing linked stormwater facilities. For this purpose, XPDRAINAGE 2017 incorporates a treatment train approach to properly account for downstream water levels and their potential effects at each individual control.
Hudspeth has been very pleased with the new program and looks forward to employing it on future projects. “It’s a very capable and excellent drainage modeling program and a big time-saver for subdivision or residential developments. If there are stringent drainage requirements, it’s a lifesaver.”
Consistent Low Impact Development Modeling
District of Columbia’s Water and Sewer Authority (DC Water) plans to reduce combined sewer overflows (CSOs) by constructing a massive underground storage tunnel system and implementing low impact development (LID) retrofits to intercept stormwater runoff in areas of the district. The combined improvements, collectively named the Clean Rivers Project, aim to reduce overflows to the Potomac River and Rock Creek. As a member of DC Water’s program management contractor team, LimnoTech has provided extensive modeling services to estimate both quantity reductions and quality improvements—including the deployment of EPA’s SWMM model to predict effects of LID implementation scenarios.
“It was about seven years ago when EPA implemented LID controls and introduced flexibility in how we could model green infrastructure in Washington DC,” says Steve Skripnik, senior project engineer with LimnoTech. “We decided to use SWMM for this project because we could envision growing alongside it, especially related to modeling green infrastructure in an urban area.”
SWMM was one of the first software packages to combine hydraulic and hydrologic modeling, and is still one of few that incorporate modeling of LID features, including permeable pavement, rain gardens, and green roofs. As compared to modeling software produced by private companies, many years can pass between releases of new SWMM versions, but minor bug fixes and program updates are released quite regularly. For instance, EPA unveiled SWMM 5 in 2005, but many updates have become available since then, including the most recent string of 5.1.011 updates in August and September 2016.
“In general, part of the allure of SWMM is its consistency,” says Skripnik. “There isn’t a flashy new feature that is added when they do a release; it’s pretty consistent with small improvements or upgrades, and that’s actually a pretty big benefit for our use.”
The Clean Rivers Project has a high-profile status in the region’s public and private communities, which means any significant modeling changes can result in significant changes to the results and then prompt questions and explanation among DC Water’s community. The overall consistency of SWMM across its long history dominated by numerous minor upgrades makes it reliable for high-profile projects. Periodically, though, more significant updates are unveiled to fix major bugs or make substantial functionality or computational engine improvements, like an update in April 2015 that yielded benefits to LID-modeling companies like LimnoTech.
“LID practices were modified to allow underdrains to be routed to a different node or subcatchment than overflow from an LID practice,” says Skripnik. “This has allowed us more flexibility in using SWMM 5 to represent a variety of real-world situations, including downspout disconnection, where disconnected flow often travels along a different flow path than connected rooftop flow to enter the sewer system.”
In addition, the April 2015 update incorporated multi-threading capability to greatly improve processing speed and resulting model run times. Faster processing makes the current SWMM 5 version more nimble for experimentation.
“We use SWMM for everything from just getting an idea of what runoff is in an area, to the more stringent regulatory side of things,” says Skripnik.
The frequency of updates is partially driven by its open-source nature. Users around the globe find bugs and program glitches, and EPA responds by fixing the problems or improving the programming, then issuing an update to the community. One of the challenges with modeling, via SWMM or other software, can be understanding how an available upgrade may affect modeling results and deciding the desirability of those results relative to the project goals.
“If you have a model that you’ve been running on a release from two years ago and you know that they’ve made an improvement or fixed the bug, it’s on the modeler or the modeling team to decide whether to switch to a newer release,” says Skripnik. “You have to investigate what the expected differences in the two different model runs will be just based on the bug fix or improvement.”
One way that SWMM users can determine the appropriateness of an upgrade is by following or engaging in discussions on the community list server established by the University of Guelph. At minimum, the frequency of SWMM upgrades means users need to stay informed about software activity. That’s easy to do via the list server’s email-based forum where SWMM users spanning every experience level, including the program developers, discuss every facet of the software. Skripnik says he commonly receives five to 10 list server emails each day, as the community can be quite active.
“When there is a release or change, people talk about it on the list server community, and if there are issues, they’re typically exposed quickly because people are open about it,” says Skripnik. “I think SWMM 5 is probably one of the most used H&H models out there because it’s free and has a very vibrant user community.”
Clearly these characteristics overcome any disadvantages that its minimalist user interface presents. Those already familiar with SWMM know its user interface lacks flash, prompting the generation of what Skripnik calls “wrappers,” like PCSWMM, XPSWMM, and others. Wrappers provide more aesthetic and feature-rich interfaces based on the open-source EPA SWMM computational engine.
“The PCSWMM interface is more user friendly, and there are additional tools like calculating subcatchment area, drainage length, and slope. If you don’t have something like PCSWMM, then you need to do those things off-line and then input into the more bare-bones SWMM 5 user interface,” he notes.
Simulations at Warp Speed
Situated on the British Columbian coastline at the mouth of the Fraser River, the city of Richmond has a relatively large watershed thanks to its river delta topography. “The city is 130 square kilometers and it’s very flat,” says Lloyd Bie, manager of the city’s Engineering Planning Department.
The topography and climate make for a beautiful place to live but result in rather lengthy stormwater modeling run times. “A model could take 24 to 30 hours to run based on older, single-core processing,” says Bie.
To help minimize run times and increase modeling efficiency, the Engineering Planning Department invested in a combined software and equipment upgrade. On new computers, department staff deployed DHI’s 2016 release of MIKE URBAN, which includes the latest computational engine for sanitary and stormwater collection systems, MIKE 1D. The upgrade, which Bie attributes mostly to MIKE 1D, results in dramatically reduced run times—down from roughly 24 hours to an hour and half.
As the successor to DHI’s MIKE 11 and MOUSE computational engines, MIKE 1D is the next-generation engine for one-dimensional network modeling of open channel and closed pipe systems. Merging the two older engines utilizes functionality from both domains and optimizes performance. In addition, the MIKE 1D engine is parallelized, making it capable of utilizing multiple cores during simulations.
Upgrading to DHI’s latest product felt natural to Bie and his staff. “When we first started doing our drainage and sanitary models in Richmond, we felt DHI had a good solution,” says Bie. “Richmond is very flat and we have a lot of really big drainage canals, and the dynamic flow solution was important to us at the time.”
Although Engineering Planning briefly considered options outside DHI, those were quickly dismissed. “For our drainage models, I’m not positive anything is showing a huge competitive advantage at this point in time. It costs significant money to change packages beyond just the product, including consulting work to migrate the system over.”
Engineering Planning staff already have a wish list of projects on which they plan to unleash MIKE 1D. For starters, they began verifying previous modeling contained within Richmond’s 2041 Official Community Plan.
“We ran extensive simulations in MIKE URBAN to determine future updates for future growth,” says Bie. “However, in our Olympic gateway neighborhood, some upgrades hadn’t gone as planned. So we recently reviewed what had actually happened versus what was planned and then rejigged the plan to accommodate the current construction.”
Bie anticipates doing a similar update to the East Richmond Agricultural Water Supply Study. Every three or four years, the study evaluates the supply and demand of irrigation water in the regional ditch network servicing the city’s eastern agricultural areas. In addition to evaluating supply and demand, Engineering Planning updates the study’s model by calibrating it with storm data every time a significant event occurs.
Another wish list project for MIKE 1D includes experimenting with the spatial context of existing stormwater models. The size of Richmond’s watershed means accurate depiction of the catchment’s peak intensity can be nearly impossible. “For a small catchment, distributing the peak intensity across the entire catchment is a reasonable assumption,” says Bie.
But he points out that the typical area of a storm cell’s peak intensity is frequently smaller than Richmond’s large catchment, thus the amount of water indicated in the model can be inaccurate. “Richmond is increasing its number of rain gauges to better capture how storms behave and will be looking for ways to apply this information to the models. Shorter run times will be valuable in this effort.”
More accurate results will in turn save the city money—for example, by reevaluating whether a single box culvert in a road crossing really needs upgrading to mitigate flooding potential. “These box culverts are worth $20,000 per meter, so you don’t want to willy-nilly upgrade them,” says Bie. “Modeling software is critical to identifying appropriate upgrades and being cost effective or finding different paths to upgrades that maybe are more effective financially.”
The Engineering Planning staff say transitioning to 2016 MIKE URBAN and MIKE 1D was painless, which surprised them as they had experienced glitches with the previous system upgrade. If anything, the new program leans toward being less tolerant of bad information. Modeling completed so far with the upgrade has pointed out bad information not previously pointed out by older software.
Bie feels the differentiators between today’s software options are speed and approachability of the user interface. “I don’t know that large sanitary or drainage networks are truly manageable without software, and the better it is, the better results you get. [MIKE 1D] allows us to ask 24 more questions in a day than previously.”
Technological advances happen at an increasingly faster pace, and improvements to hydrologic and hydraulic software are no exception. The organizational and computational demands of modeling mean software developers constantly research and experiment with ways to improve their programs. The companies and organizations that use modeling software then must make difficult cost-benefit choices as new versions become available, including whether computing equipment should be upgraded as well.
“That’s the thing about modeling—it really does pay off to have fast, good equipment,” says Bie. “Otherwise, a large project could take months if you have a system failure or program crash, or you’re trying to debug a model. You could set it to run overnight, just to come back the next day and find out it didn’t work.”
With properly selected software and processing tools, computation times and personnel frustrations can be dramatically reduced. The updates discussed here only sample the options on today’s market. Staying informed on changes related to modeling software, especially those already in an organization’s toolbox, requires a commitment to consistent monitoring and periodic research of the market.