You Don’t Know What You Don’t Know: The Stormwater World is an Unpredictable Place

One of the StormSensor® company values: Design for a perfect world…and then add reality.

 

Data from a StormSensor Monitoring Point (Image © StormSensor)

Take a look at the chart above. The first two-thirds or so are a bit of a mess. There’s a clear trend, certainly, and the data go up and down (which correlates nicely with storm events; not shown). Yet some of the data are negative, which, when looking at depth in a stream—as we are here—seems pretty impossible. The latter third, however, is much cleaner. Flow is always a bit positive—as it “should” be—and there are fewer unexpected jumps. As if it the water was moving across a beautifully smooth plane.

The latter set of data is based on a model we developed based on the Manning’s Equation. It’s calibrated, of course, and the positive values in the model match the positive values of the unmodeled, measured data within about a +/- 10% accuracy. But the nice smooth bits are, frankly, a pipe dream based on a model with inherent perfection built in, and inherently messy reality left out.

Now here’s the fun part: just about everything that we know about stormwater and wastewater systems is generated through the type of modeling described above.

Our pipes are built based on modeled estimates of flow; they are sloped based on estimated velocities that allow the pipes to ‘self clean.’ Tree limbs, debris, corroded pits in once smooth concrete pipes- those messy bits of reality don’t exist in most models.

The required pipe size for a given system is generally based on the volume of water anticipated at to flow through any given point in time. Because we cannot be there all the time to keep tabs,  we build models and estimate flows during different storm events, and select a pipe size accordingly.

As we connect pipes to build a larger stormwater system, the capacity is modeled based on the likelihood of a repeating statistical storm event—and associated flows—within a certain time period. So, for example, one combined sewer system could be required to to hold enough stormwater/sewage generated during three 5-year storm events over the course of 2 weeks.

And treatment is the same; just about all of the Best Management Practices (BMPs) that we install and construct and utilize are based on modeled estimates of flow into and out of the systems. Many BMP models also include a treatment component, based on modeled concentrations and estimated pollutant removal rates.

And each of these models has myriad assumptions built into it: the area of adjoining pervious vs. impervious surfaces, the standards for statistical storms that vary by issuing agency, flow is present only during storms, empty treatment facilities just waiting to hold all the water generated.

Lab Conditions Vs. Real-World Conditions for flow can be very different (Image © StormSensor)

Which brings me back to one of the key values at StormSensor: assume a perfect world with “lab conditions” (All engineers do it! Ahem! Spherical cow anyone?), and then adjust for reality (this is the hard part).

Because we exist out here in the real word, in the sewers where things get *real*, just about every single StormSensor deployment has resulted in a surprise.

Sometimes it’s constant baseflow in what were assumed (and modeled) to be dry pipes whenever we had a sunny day.

Other times it’s simply because the roughness coefficient came out of a textbook and doesn’t reflect the actual pipe material, or the sediment buildup, or the presence of a backup farther downstream.

And at least one time it was 10’-diameter storm culverts thoughout an entire city running at 50% capacity because river water unexpectedly filled the storm system…which explained why the models didn’t come close to representing actual flow conditions AND why the city experienced such impressive flooding despite the size of their system.

It’s absolutely amazing the things that we see with real-world data! And it explains so much we don’t know, and why we don’t know it.

So when someone asks me why we need this data, it’s an easy answer: models don’t reflect reality, and reality is unpredictable at best.

I mentioned at the beginning of this post that we tend to rely on models because we simply cannot be everywhere at once to keep tabs on flow conditions. But the truth is, technology has progressed to the point where we actually CAN (at least our sensors can!). Keeping tabs on your system, finding those unexpected flows, back-ups, sediment deposits, and clogs that models cannot account for is what we’re all about. I can only imagine what our next surprise will be…any bets?

Contact us if you’re interested in learning more about model validation with StormSensor.

About the author

Erin Rothman

Stormwater Scientist

Talk stormwater with erin@stormsensor.io With more than 15 years of environmental consulting experience, Erin observed so many opportunities for innovation in the stormwater industry. With those in mind, she founded StormSensor to enthusiastically embrace new technology to help solve the problems of an age-old industry.