Unleashing the Power of ICM to Understand Risk of Environmental Impact

BEN DUNN: Hello, Unleashing the Power of ICM, Understanding Risk of Environmental Impact. My name is Ben Dunn. I'm an associate at Aurecon. And today I'm presenting with Alex Grist, a senior technical support engineer from Autodesk. And the safe harbor statement.

So I'm a wastewater planner specializing in wet weather performance for separate sewer systems. I have a strong focus on bringing digital solutions to these projects. And I enjoy the value they can bring. I live in Brisbane, Australia.

ALEX GRIST: Welcome, everyone. I'm Alex Grist a senior technical support engineer with a decade of expertise in InfoWorks ICM. This has been primarily focused on using integrated catchment models for flood analysis. I'm passionate about helping our customers solve important environmental challenges using our technology just like the project we're talking about today. I'm located near Oxford, UK right on the River Thames. So this is an issue that is close to home.

BEN DUNN: So there are wastewater networks across the world. In these wastewater networks, they discharge the natural environment. And understanding the impact of this is complex. Two years ago, we sat down and worked out ways to improve this understanding.

And we identified three key areas that really could be explored further. One was the hydraulic interaction between the sewer and stormwater system. It is quite dynamic. And at the moment, we just generally assume it's freely discharging. But this was a fairly false assumption.

The dynamic pollutant profiles across events, these vary considerably, small events through to large events from small overflows through to large overflows. And we really wanted to understand what these actually look like. And then the pollutant movement and its impact on the value in the receiving environment. Just because the overflows discharge doesn't mean to say that pollutant stays immediately next to that point and really understanding how it moves through that environment.

So we use ICM InfoWorks, ICM Ultimate. And this was supported by virtual machines Azure, databricks, and Esri. So Alex is going to take us back to basics.

ALEX GRIST: So combined sewer systems were developed in response to urbanization and industrialization of the 19th century in growing cities. As populations increased, so did the need for proper sanitation and drainage systems. These systems captured both the sanitary wastewater and stormwater runoff into a single piped network proving cost effective and efficient during the implementation.

While effective for their time, these systems will lead to environmental issues when heavy rainfall exceeds the system capacity as diluted sewage will be discharged to the receiving environment. Given their age, these systems are often well beyond their design life and can now be close to capacity simply from sanitary flows and infiltration alone. This gives them very little capacity to cope with runoff into the system.

In contrast, newer cities often adopted separate sewer systems recognizing the benefits of improved treatment of waste and better stormwater control. The sanitary sewer takes the domestic and non-domestic wastewater to a facility where it is treated before being discharged to the receiving environment. Runoff generated by rainfall is piped into the stormwater sewer. And should this be exceeded due to rainfall, no sewage is discharged to the environment.

Now, the project we're talking about is a separate system. But as I'm sure you will appreciate every design has a limit imposed on it through cost space or practicality, the design for these systems has varied through time and around the world. But at best, they may have been designed to cope with a 50 to 100 year design life.

So many of these systems are coming towards the end of their operational design life. And the capacity of these systems is additionally impacted by operational issues such as pump failures, blockages, and infiltration. So when these systems do reach capacity, which they do ever more frequently, the focus is usually to prevent flooding. Now, the solution to this historically has been a very simple one.

BEN DUNN: Overflows are placed strategically throughout the sewer network. Now, we often think of these assets as being in the combined systems. But they are also exist throughout the separate systems. They are a very important asset. Then they stop uncontrolled flooding happening throughout the network, so whether it floods in properties or through the maintenance hole covers.

Now, in the separate sewer systems, their performance is quite variable. In new areas, typically, where we might have very low levels of wet weather flow getting into the sewer network, their operation is extremely infrequent. But when we look at the older systems, the bigger systems, these overflows operate quite regularly.

And one of the problems with the separate system is that the flows are getting into the system in a very uncontrolled and inconsistent way. So their performance is extremely variable, which is where the hydraulic models become really important. But we need to understand what the volume and frequency of these overflows are. So we can understand the impacts of that environment.

ALEX GRIST: So is this a very local or a global issue? Since 2013, the UK government instructed water companies to install event duration monitors on all storm overflows to be completed by the end of 2023. Environment agency data for 2022 shows untreated sewage has discharged across England for a total of 1.75 million hours in around 300,000 spill events. So to combat this, a 56-billion pound investment plan is proposed to reduce the discharge from overflows through to 2050. This is a mammoth undertaking.

In Australia, in preparation for the 2000 Sydney Olympics, a huge network of storage tunnels were built to reduce overflows and pollution in Sydney Harbor. The project comprised 22 kilometers of tunnel driven with a diameter of 3.8 to 6.6 meters. The lowest point in the tunnel sits at the terminal end where it is always 100 meters below sea level at a depth of 160 meters below ground level.

There are two pumping stations, each with over 16 megawatt pump capacity. The storage capacity of the tunnel is almost half a gigalitre. And it came at a cost of 460 million Australian. So this was a mega infrastructure project.

And with a year to go, to the Olympics, Paris is in the final phase of a historic clean up, which will soon see swimmers and divers back in the river Seine. At its heart, Paris has a combined sewer system dating back to the 19th century. Banned for a century because of the filthy water, city swimming is set to be one of the major legacies of the games thanks to a 1.4 billion Euro regeneration project. The final piece of the puzzle has again been to build a vast underground reservoir, which will store runoff in times of heavy rain in a huge cylindrical space 34 meters deep and 50 meters wide.

Now, for the US, there is no overflow data that I could easily find. But for the Great Lakes basin, there are 166 permits, which authorize discharge from combined sewer overflows. And according to the national coastal condition assessment in 2010, 18% of the nation's coastal and great lake waters were in poor biological condition. So this certainly seems to be a common global theme.

So the solution to remove polluting spills has often been through the use of huge storage tunnels. But this comes with a considerable price tag. And it often has a very large driver such as the Olympics to fund it. For the majority of cities, there needs to be a different approach, especially with combined and aging systems, which looks at long term strategic investment and separation.

BEN DUNN: So before we get into the project, let's just take five overflows as an example. And we're looking at volume here. And not every overflow is the same. They're all different. They all have different volumes. They have different volumes across different events. Some can be big. Some can be small.

And as we transition into the frequency conversation, that also changes quite considerably. Some can go off once a year, twice a year. But some can be a lot more regularly than that. And looking at these two matrices is the traditional approach. That's what we typically do across the world looking at the performance of wastewater systems.

We wanted to take the next step. We wanted to look at the pollutant load that was coming out of these wastewater systems. And a number of pollutants could be chosen. For this project, we're looking at Enterococci, nitrogen, phosphorus, and suspended solids.

And, again, the pollutant load varies considerably. So in some overflow structures, the loads can be huge but can be spread over quite a long time. And, similarly, there could be very small discharge volumes or frequencies and very polluted load going out very concentrated load going out to that receiving environment.

So those three variables are extremely complex. But if we take the next step into that conversation, which is, well, how is that impacting the receiving environment, which can be big in terms of a 2D model, it's quite dynamic. It moves around a lot. And understanding that movement becomes really, really important.

Now, when looking at this, we really wanted to try and do this in one single environment, one single model. What we didn't want was lots of models moving data around, running at risk of human error with data management. We just wanted one single environment. And that is where ICM came in.

ALEX GRIST: So InfoWorks ICM is Autodesk's hydraulic modeling solution for analysis of city and catchment-scale models. When it was first released 10 years ago-- and it still remains to be a real game changer in this space as it allows for the representation of almost everything in a single model. As Ben said, that was central to this project.

For those that aren't familiar with hydraulic modeling, pipes and channels are often represented as links. The direction of the flow is fixed. So it's referred to as a one-dimensional analysis. When you model flows over the surface, it can move in the x and y directions. So it is referred to as a two-dimensional analysis.

What ICM does is allows the user to model the interactions between these in the single simulation engine. The ICM engine employs a dynamic time step, which means it excels in simulating long time series events ensuring it is running at the optimum speed. The engine can smoothly transition between low flow dry weather, surcharged wet weather conditions.

This was very important for this project as the requirement was to run a 10-year continuous time series simulation. Alongside the hydraulic components, ICM also offers considerable water quality and sediment capabilities. This includes dissolved sediment and detached pollutants. This was a must for this project because it wanted to look beyond just the frequency and volume of spill and to understand the pollutant load as Ben has just explained.

ICM comes with a fully managed database and geospatial user interface, which separates it from many competitors. This has now been further enhanced with cloud database and simulation capabilities that are included from ICM 2024. Quality assurance is guaranteed through the database, which tracks all changes to the model down to the individual field level.

Data flagging allows the source and confidence of each model object field to be set and understood. Finally, users are able to build out custom workflows and remove users from repetitive tasks with exchange. So why is it that I'm here with Ben talking about this brilliant project?

Aurecon were commissioned to undertake a pilot project to demonstrate a new approach assigning risk of impact. And having explored different possibilities, they chose InfoWorks ICM due to the power of the simulation engine. Aurecon have plenty of experience of modeling in house but not specifically with InfoWorks ICM and water quality modeling.

Engagement was set up through consultancy hours using our global team of regional and international specialists. We initially sat down every two weeks to talk through the approach and where additional support was going to be required. I was also on hand to answer any questions directly with the project team as well as act as a conduit with product and development as needed on any specific questions.

So this model is really pushing the boundaries of what has been done and the capabilities of the software. Our users run 10-year time series. They run water quality and sediment models. They run 1D, 2D models. But no one has done everything in a single model, especially on this scale. So we always anticipated some challenges.

With that being said, the project has run fairly smoothly. And that really speaks for the power of the InfoWorks simulation engine. Yes, there's been a few occasional road bumps on the way. We had some challenges with some of the functionality that hadn't until this point ever been used in combination.

But because of the collaboration, we were able to quickly highlight the problematic workflow and enhance the software as a result. And, overall, it's just been a great project that I've been fortunate enough to be part of. So it's difficult to show you the model in detail such as its size. But I promise that there is an image of the whole thing later on.

So here we have a simplified snapshot to explain what's been included. Firstly, the model contains 256 kilometers squared of subcatchments of which 120 kilometers squared cover the urban environment. These drains were a combination of 269 kilometers of sanitary sewers, 36 kilometers of storm sewers. And there's also 66 kilometers of simplified river network.

All major ancillaries, such as pumping stations, force mains, and of course, the very important overflows and cross-connection structures, are modeled in detail. The receiving environment, in this case, the tidal estuary, is represented using a 13-kilometer square 2D zone. The 2D zone is discretized into mesh elements, which range from 250 to 10,000 meters squared. The 1D, 2D tidal systems are coupled via outfall 2D nodes, allowing the direct exchange of flow and pollutants. In total, there are 178 1D, 2D connection points in the model.

Hydraulically, there are four primary inputs into the model. For the sanitary flows, there are daily profiles for domestic and non-domestic sources of inflow. And these are applied through what we call the wastewater and trade profiles. Infiltration is represented via fixed base flows.

And the rainfall runoff response uses the [INAUDIBLE] methodology, which is a very common one for Australia, with approximately 40% of the area attributed to the sewers. Lastly, the very important time varying tidal boundary level is applied to the 2D zone via a level file. The simulation engine itself takes care of the hydraulic interactions and exchanges the flows between all of the different systems.

For the water quality, three dissolved determinants, including coliforms, sediment, and two attached pollutants were simulated. The inputs are driven by three main mechanisms the domestic and non-domestic flows have the pollutant concentrations associated via the wastewater and trade profiles. Background levels of pollutants have been added to the tidal zone via a polluted graph.

And, finally, the InfoWorks native wash off and gully pop flushing model has been used to determine the buildup and inflow of both dissolved and attached pollutants into the network from the runoff surfaces. The example graph in the top right hand side shows the sediment wash off and why this methodology is so important getting the pollutant loads correct. High concentrations, especially of sediment wash off, often occur at what are commonly termed first flush events, when you have rainfall after a dry period when the sediment is built upon the surface. And with ICM, this process can be represented over the full 10-year continuous simulation.

BEN DUNN: So, hopefully, you're now starting to appreciate how big this model is. And to take it to the next step, we needed to get a load of data, particularly around how the networks, the sewer networks, stormwater network, receiving environment were all behaving. So we installed 28 depth velocity gauges throughout the sewer network, 18 level monitors against overflow structures. And we chose strategically important ones that had high frequencies or high volumes.

We had 10 depth velocity gauges throughout the stormwater system. And, again, we chose quite strategically important ones, ones where we thought there was a lot of interaction going on between the sewer system and the stormwater system. And we chose 20 sites for water quality sampling.

Now, across the water quality sampling, we used a variety of tactics to gather the data. We used autosamplers, buoys out in the bay. We used people in boats going out during wet weather to gather manual samples. And we also had a variety of regimes in there.

So we were collecting data on an hourly basis, a daily basis, a weekly basis, and a monthly basis. So there's a lot of information gathered through the water quality. And, in addition to that, we gathered it across the sewer network, the stormwater network, and the 2D area.

On top of all that information, we also went and did some surveys. So we got a bathymetric survey of the entire 2D area. And we also undertook topographical surveys of the key stormwater systems and stormwater asset levels. So we pulled all that data into a model, the 1D sewer network, the 1D stormwater network, and the 2D tidal system, what we refer to as the 1D, 2D model. And we calibrated it.

So we started with the hydraulics first. We calibrated all three systems concurrently. And that made sure we were getting the interactions between all three systems as we were building that calibration. And we had a really strong focus on the wet weather performance because we needed to make sure that the hydraulic parameters in the wet weather were getting the dilution effects that we needed.

Once they got the hydraulic calibration sorted for all three systems, we moved on to the water quality. And, again, we did all three systems concurrently. But this time we had a lot more focus on the dry weather because in particular within the sewer system, this is where the pollutant load exists in the dry weather flows. And as the wet weather flows kick in, that's where the dilution comes from.

So we spend a lot of time making sure that was right across all three systems. And to run those simulations, we ran the single events on the desktops. But as we ran the time series both for the 18-month period, the calibration period, and also the 10-year time series that we're running, we're running those on virtual machines.

And when the 2024 ICM cloud service became available, we switched over to that. And that's been enormously successful. So our 10-year time series have taken about 12 days to run. We're actually really happy with that. We thought it was going to be taking a lot longer.

And so with regards to the output from these files, they are huge. I mean, 12 days of data on a virtual machine, that chomps through some code. And so we end up with about a 1 terabyte file per 10-year time series. And we are running three 10-year time series concurrently as well. So we've got two time series with some additional variables in for sensitivity analysis.

So for each of those time series within the 2D mesh, we've got about 60 billion points of data with all the pollutant loads that we have. And whilst this is really important data and really valuable for the analysis of impact, it's also really complex to digest. There's a lot of information. And when you're using time-based data, it's really complex to absorb which events are important, which aren't important, and how do you analyze that.

And one of the other aspects of this became that this information was being given to environmental people to look at to actually assess what the impact was. And they weren't users of ICM. They didn't have access to the software. And they didn't understand how to use the software. So it was a limited access to this information. So we needed to change that. And we switched over to Esri.

So we pulled the data out of the model and put it onto portal for ArcGIS to enable everyone to access it. And this is also beyond the scientists as well. We brought it to the client so that everyone could see it and see what was happening with this information.

And bringing it into this GIS system also allowed us to do quite big analytics, particularly between the three 10-year time series and assessing the deltas between those models. We could also visualize the 1D statistics in a much more powerful way. And we could also visualize the 2D simulations as we have there.

Now, although these simulations were a little bit more simplified than they were from the ICM model software, the information was good enough that people could really understand how the pollutants were moving around and what exceeded certain thresholds. And, lastly, one of the really important aspects of this when we're pulling into Esri is we wanted it to be repeatable and easy to update the model when we did updates in the model. And so we got this process down to about 24 hours so that the teams could actually access the information really, really quickly after those runs had finished.

So with the project success going back to those three criteria you had at the very beginning, yes, we did see interactions between the sewer and the stormwater system. And those interactions were important. They pushed the flows into other parts of the system when they came out in different places that we weren't expecting.

We did witness dynamic pollutant profiles across multiple assets and different events. And it varied considerably throughout the system. And we did see a lot of pollutant movement through the system. And, particularly, with the tidal aspect, it really moved around and moved around in ways that we were not expecting. So ICM was great for this. The image on the map there is the export from the model. So you can see the size and the complexity of it with all the systems.

And, lastly, was this a valued exercise? Was it valuable? Did we get information out of it that was worth this effort? And so it's really helped us understand the impact and, particularly, the impact of social sites. We've got a lot of high value beaches in this project area.

A lot of people use these beaches. And so we could understand how the Enterococci moves around comes out of the overflow structures and what the statistical impact is on those beaches. We can also take this onto flora and fauna. We can see how the suspended solids, nitrogen and phosphorus, are moving around. We can see how they're moving into areas where we've got high value flora and fauna.

And we can also look at it from the acute and the chronic point of view, so acute being short term frames, particularly, discrete events, versus chronic, where it's the long term time frame, so that 10-year time series. We can actually see the difference between the two. And we can analyze it differently.

So all of this information enables us to prioritize capital investment to maximize benefit to the receiving environment. It's not always the highest frequency or the highest overflow structure that's actually having the biggest impact. It can be the smaller ones and those ones that might be discharging more dilute flows that you wouldn't typically expect. So it has been a huge success. Thank you.