Talks with Terry (Engineering in Action): Stormwater Modelling

Welcome to the second of a series of ‘chats’ with Terrence Allen, one of our friendly engineers. To read the first chat all about BOD and De-Nitrification, click here.

Hey, I found that info last time pretty useful – how about you talk me through something else?

I suppose I can do that. After all, its all part of our friendly service. What do you want to talk about?

Well, it is summer, which means storm season, so let’s talk about that. What exactly  is stormwater?

Well, the short answer, obviously is that it’s water that comes from a storm, or, essentially, rainfall.

So, stormwater is natural, but big storms can be bad, they rip up trees, down power lines and cut off roads. And even little storms can cause problems in the right circumstances.

I know what you mean…but can you be more specific?

Other than the big acts of god we all hear about on the news, stormwater presents two major problems.

Firstly, as it falls and runs over the land, stormwater collects a wide variety of contaminants ranging from pesticides to fertilisers, soils (suspended solids), and litter, grease, oil, nutrients and metals and increasing water turbidity. Without proper management these contaminants end up in waterways,  which in turn leads to pollution.

I never thought about that, but it makes sense…all that water, washing away whatever is sitting around on the ground. What else?

Secondly, as you’d expect, stormwater causes an increase in water flows), in drains, creeks and across land.  This can lead to erosion of soil, damage to infrastructure (like roads) and, in more extreme cases, flooding, with infrastructure and houses getting inundated, or overwhelmed by water.

Sounds like a big problem. What do we do about it?

To avoid both pollution and flow rated problems it is important to properly manage stormwater flows, especially in urban areas where everything is closer together and there is less room for the water to spread out naturally.

How do you do that – I mean, specifically?

There are a variety of techniques we can use to manage stormwater management. They are often aimed at reducing pollutants and offsetting, storing, reducing or more appropriately channelling and controlling excessive flows.

Treatment and flow control devices are practical, hands on ways to manage stormwater, but planning policies, flood and inundation maps and municipal requirements do their part too.

Why would that matter?

Planning ahead helps to prevent construction or development within flood prone areas, where no amount of treatment or management infrastructure would stop a disaster.

So, that’s stormwater and stormwater management. How’s this stuff actually done?

Stormwater management systems for developments are now often designed and installed as part of the developments design phase. This design relies heavily upon stormwater modelling, of which there are two broad categories: Water quality and Hydrology.

Modelling….sounds like computer stuff to me.

Exactly! Planning and flood maps are usually generated using hydrology modelling, based upon observed flow data (what we’ve seen in the past) and predictive statistical algorithms (complex programs to predict what might happen). These can (and often are) created for each major catchment (and possibly even sub-catchment) within a development or council area.  These maps then inform councils decision making processes when it comes time to allocate land for development or approve applications for developments in flood prone areas.

Okay, I get that…but why would you have to look at water quality?

Water Quality modelling simulates the treatment of the potential pollution contained within stormwater that runs off the land (e.g. a development, road, farmland, etc), which allows the design of systems to maximise the protection of surrounding waterways or major water bodies (like rivers, creeks or wetlands).

And this “hydrology” thing?

Hydrology has to do with the movement of water across the landscape, and is normally employed to assess both erosion and flooding potential. Hydrology modelling can also include piped hydraulics and stormwater collection and distribution systems, allowing assessment of conveyance systems as well as land based flows.

Okay, got it….So you know about that stuff, but do you guys actually DO it?

Here at Simmonds & Bristow we can do both, and we use two different stormwater modelling software programs..

For Water Quality Modelling we utilise the MUSIC model.

As in….instruments and singing?

You don’t want to hear me sing. Seriously though, MUSIC – The Model for Urban Stormwater Improvement Conceptualisation (someone really wanted that acronym to spell MUSIC) was originally developed by the Catchment Hydrology CRC, and is now maintained by eWater.

Can you tell me more about how it works?

MUSIC is a stormwater quality model. The model contains a number of pre-configured stormwater quality improvement devices (SQID’s), which are combined with catchments (of which many are pre-configured) and outlet nodes to form a treatment train model.  MUSIC is relatively common with many of the larger municipalities publishing guidelines listing specific configuration parameters for runoff, catchment and SQID nodes.

Whoah…slow it down a bit…

The model uses 6minute time-step data to provide a continuous rainfall simulation over a set period, generally 30 years. Climate data can be obtained from the Bureau of Meteorology in a specific 6-minute timestep format for importing into the model.

Wow. Predicting every 6 minutes for 30 years….sounds exciting.

It can be, if you’re in the business. The model then uses the input rainfall data, pan evaporation data and (generally) stochastically generated (a statistical method) pollutant data to simulate what would happen to stormwater that is directed through the treatment train. Results are assessed against council guidelines which may consist of load reduction targets, absolute concentration targets or a ‘no net worsening’ style scenario.  MUSIC is limited to 3 major pollutant parameters: Total Nitrogen (TN), Total Phosphorus (TP) and Total Suspended Solids (TSS).

Is it any good?

MUSIC is very good for simulating small developments and predicting stormwater treatment device performance. It does have a shortfall though, because it lacks hydraulic modelling elements. As such, MUSIC cannot predict the efficacy of a collection system, or predict whether excessive flow will cause soil erosion.

To include the hydraulics and hydrology we need to go to the next level.

Hey, before you move on, this thing sounds neat. Where can I learn more about it?

Sure – just click here and you can learn more.

Okay, go ahead. You were saying?

For hydrology and hydraulic modelling we use XP Solutions XPSWMM model. As in “Swim”.

Wow. These models really like acronyms.

They sure seem to! XPSWMM is a 1D/2D hydrology model that also allows for sanitary and water quality calculations in a similar fashion to MUSIC (albeit without the degree of pre-configuration). The model is generally much more freely configurable, and much more complicated, than the MUSIC model.

I’m trying to stick with you here, but can you make it a bit clearer why this matters?

It’s capable of calculating relatively complicated hydraulics and hydrology, in both 1 and 2 dimensions, which allows the model to simulate flood plains, stormwater distribution and catchment systems.

Dimensions? What? Does it do 4D modelling, 4D is time right?

Well 4D isn’t time, Back to the Future got it wrong, time is a much harder concept to actually define, as it’s pretty abstract, but anyway… Actually the models do use time, as they perform time-step calculations over a period of rainfall data…

The 2D model uses a terrain map of some description, which is split into a 2 dimensional grid of ‘cells’, the model then goes and calculates various parameters in these cells (water velocity, direction) based upon inputs.

You can then get (much) more complicated with some highly technical terms, setting ground types, setting imperviousness, manning’s ‘n’ numbers, soil conductivity, runoff model specifics, infiltration model specifics, etc.

And you’d want to do this because????

All of this complicated setup is used to generate runoff flows based on both incident rainfall, input from adjacent cells and losses through the soil profile. The model calculates a water balance across each cell in the model (so models with lots of cells take a very long time to calculate) and presents data on water depth (or height), flow, velocity, etc, generally on a time-step basis.

Great…..again, why would you want to do this?

This allows the modeller to predict whether excessive scouring might occur (and when in a storm it is most likely to occur), whether a new bridge will cause excessive flooding up-stream of the roadway, whether a culvert design is sufficient to pass a 1:100 year flood without major problems, or whether the new development going in will cause excessive flow into a neighbours land due to the change in area imperviousness.

Ah. I get it. Do you ever just do a 1D model or is that too old-fashioned?

Actually, not at all. A 1 dimensional model, models elements in a single dimension, with flow propagating linearly along the element (the element has depth and area, so “1D” isn’t quite accurate, it has more to do with the fact that the flow is linear along the element).

1D models are often used for pipework, open channel and network analysis, but a well-designed and set-up 1D model can be used to model runoff from an entire catchment in a similar manner to a 2D model (although good luck interpreting it without help, you essentially need to model each change in landform as a different element to gain useful velocity and flow data…).

A 1D model provides velocities and depths (the “hydraulic grade line”), which provides flow, in the 1-D element.

Okay…so where does your swimming model factor into it?

It’s actually XPSWMM and it combines 1D and 2D models.  It uses 2 separate engines (WBM BMT’s TuFlow for 2D flow, and XP’s own 1D hydraulic engine) for 1D and 2D flow, with the program passing flow data back-and-fourth between the two models during iterative solving.

This combination of models allows for an improvement in model design and efficiency, as the 1-D engine is better at calculating smaller elements (i.e. pipework, and on occasion small rivers or open channels and creeks that are too small to provide meaningful data when compared to the 2D cell size), and is generally (much) more rapid than the 2-D engine. The 2D engine then is left to handle overland flows, rainfall and generation capture etc.  The combination can also help keep model iteration and convergence error small, which improves overall accuracy.

XP SWMM focuses on stormwater hydrology, soil infiltration and goes one step further to also include water quality. This model also allows us to model stormwater impacts on sewage systems.


That’s not all. XPSWMM can accept a variety of inputs, from pre-set hydrographs, IFD (intensity-frequency-distribution) data for an area (allowing it to generate it’s own storm data), to continuous rainfall records. The model can run a Design storm event (generally a single event lasting for a set time at a set rainfall recurrence/ARI, e.g. a 2hr ARI100 storm) to continuous rainfall (which is often much more computationally intensive but can be very useful if sufficient data is available).

And where can I learn more about this one?

Just click right here.

That seems pretty in depth – is there a “quickie” option that is similar to this XP SWMM?

Actually there is. It’s called XP RAFTS and it focuses on stormwater hydrology  including soil infiltration. It can be used for a much quicker 1D Model only – but could be a good starting point.

Can you compare the models for me again?

Sure. So the MUSIC model might be used to model a stormwater treatment train in a small development to prove to council that the development will not result in excessive pollution in a waterway.

An XPSWMM model might be used to model an entire development to prove to council that it will not cause excessive stormwater discharge to waterways or surrounding properties.

Or an XPSWMM model might be used to estimate 1:100 year flood levels to allow a council to set flood lines for their development guidelines.

Or, if you wanted a quick 1D model focused on stormwater hydrology we could use XP RAFTS.

Generally speaking MUSIC is the simpler and, hence, cheaper model to run, followed by XP RAFTS.  A full XPSWMM model can provide a significant amount of data, to the point of fully modelling your entire stormwater system, but is often more expensive to set up and run.

So…how do I decide which one I need?

You’re not on your own. The choice of model comes down to your requirements and we’ll always work with you to sort through what those are and what we’d recommend for your specific situation. As always, we’ll use the most appropriate for your needs, saving you time and money.

Okay, so I suppose that’s everything I really need to know, right?

Not exactly! You might want to know a bit more about the input data for XPSWMM.

Input data? What’s that?

A lot of hydrology modelling is really about estimating stormwater flows.

Estimating stormwater flows can be difficult.

Sewage and Potable Water flows and demands are well known enough that decent estimates can be made using ‘standard’ figures, but stormwater is much more variable, and rely upon local climatic conditions, which can change significantly between regions.

So, how do you figure out what to put in the model?

The biggest guide for flow estimation is Australian Rainfall and Runoff (AR&R). AR&R was originally published by Engineers Australia in 1987 and contained flow estimation techniques for use in estimating both rainfall and the runoff generated by said rainfall (e.g., the Rational Method).

So, if it’s been around since the late eighties, why do I need to worry about it?

This guideline has been undergoing significant review in recent years, with a new version due to be released in the near future. Much of the guideline is available online in draft form (click here), but be warned there are 9 significant chapters to read through.

The new guideline features revised temporal patterns, which splits Australia up into more granular rainfall zones that govern rainfall parameters (intensity-frequency-duration), which have been based on a significantly longer data set. It also features new base flow and loss calculation methods and a new regional flood frequency estimation model.

I’m interested…but that sounds like a time I’d call you guys in.

Absolutely – we can definitely help you out! We’ll provide you with a much more in depth overview of the changes once the full documentation is released (and we’ve had time to digest it).

You also mentioned a design storm?

Yeah, a Design Storm is usually either set by municipalities or by experts in the field. Most stormwater collection systems are tested against a “Design” storm event.  The event combines “likelihood” factor with a timeframe, so it might look like “20min 1:200y ARI storm” or “2hr 0.5% AEP storm”.

The “Likelihood” (also known as the “return Interval”) has traditionally been presented as an ARI, the Annual Recurrence Interval; this is your “1 in 100 year storm”.

Once every century…doesn’t sound like such a big deal.

The nomenclature for the ARI can be a bit misleading, which is why much of the industry is moving to the Annual Exceedance Percentage (AEP) as a frequency indicator. Technically a 1:100 year storm has a 1:100 chance of occurring every year.  Long odds but nothing precludes it happening more often than once in every one hundred years.

As an AEP that’s then a 1% storm.

Right, ok, so timeframe?

The storm timeframe is essentially the intensity, and the storm ARI needs to have an intensity added to it. A 20min 1% AEP storm is going to be more intense than a 2 hour 1% AEP storm.

You also need to consider Distribution (you may have noticed the term IFD used previously in the AR&R section…), the “Distribution” is basically how rainfall is distributed over the time of the storm. The actual rainfall depth will remain the same, but the distribution changes how that rainfall depth occurs.

So one distribution might have an early “Burst” (a short, intense fall) followed by a period of more sedate rainfall, whereas in other areas may have a more evenly distributed fall, with gentle peak and a more consistent intensity throughout.

This can generate very different flows when modelled, so it’s important to pick the right factors for the right areas, and make sure you at least tick off the minimum requirements for analysis, whether it be set by Council or an expert.

I had no idea there was so much to think about when it came to stormwater. Thanks for the chat. Can we do this again soon?

Absolutely. Part of our job is to help our clients understand what we do to help them get good clean water. When you know more about what you need, we can do a better job walking you through how we can help. And if you want to learn more, don’t hesitate to give me or any of the team a call.