Automating the Design of Water and Wastewater Treatment Works: A Case Study with Brazil's Second Largest Water Utility

ADAM TANK: How many of you have been to Brazil? It is one of my favorite countries on this Earth. If you have been, maybe you are Brazilian, maybe you want to go, chances are, when you think about that country, you think about a couple of things.

You think about the fun people have dancing in the streets, which is a very real thing. You think about the beauty of the country. The Cristo Redentor, the statue on top of a mountain in Rio de Janeiro looking over some of the most gorgeous beaches on the planet. You think about the fervent support of the Brazilian soccer team and the amazing culture that they've built around.

Just unbelievable energy. The Brazilian people just bring a ton of energy into any room, any place that they are. And so I absolutely love the country. And I love working there. And I'm so excited to be able to talk to you about some of the work that some very innovative Brazilians with one of the water utilities are doing there relative to wastewater treatment.

One thing you may not know about Brazil is that 100 million households, or over 50% of the population, lack access to wastewater treatment. What that means is that the wastewater goes directly into rivers, into oceans, and it causes a lot of problems. It causes problems with health and sanitation of the population. It causes problems with ruining some of the natural beauty in Brazil. And it is a very significant challenge.

Enter wastewater treatment companies. Those like BRK Ambiental. BRK's mission is to promote public health and the preservation of the environment. It's a beautiful mission, and they're doing a tremendous job given that they serve a massive, massive area.

16 million people currently have service in water and wastewater treatment from BRK. They cover over 100 cities in Brazil. And you can see from the very northern tip, which is quite warm, to the very southern tip, which can get quite cold in the winters. And they own or operate over 171 wastewater treatment works.

So for those that are listening, if you are a consultant, you would love to work for a company like this. If you're someone in utility, you're probably thinking, oh, my god, how do they manage this number of projects. And if you're an equipment supplier, you're thinking, hey, I could sell a lot of my solutions or a lot of my products to them.

In either case, or all of those cases, your assumptions are correct. The consultants love it. The folks within BRK love their role and the work that they're doing. And the equipment suppliers do, in fact, sell a lot of equipment in and around Brazil and into BRK.

But BRK is on a path towards rapid growth. There's a couple very interesting things that I also wanted to set the stage for for BRK. So the first thing I want to talk about is the top three goals that they have as a company.

These three goals are quite simple. The first one is to grow profitably. There are a private company, which means that they need to grow profitably, and they're expected to generate a return for their shareholders. The second is regulatory compliance. So the Brazilian government has brought many mandates relative to water, wastewater treatment, sanitation that BRK has to comply with. And then the third is 100% BIM, which is a really interesting one that we'll talk about here in a bit.

So relative to profitable growth, I want you to think about a couple key questions here given your role in the water industry. Or even if you're not in water industry, think about these, because you probably deal with these from an engineering lens or in the work that you do with technology and design.

So the first one is how can BRK profitably bid on more treatment works, build new ones, and rehabilitate the ones that they already have? Key question number one.

Key question number two. How can BRK ensure that 99% of the population in their coverage area has water and sewage services by 2033? So I mentioned that 100 million households currently lack access to clean water and treated wastewater. But by 2033, the Brazilian government has mandated that 99% of the population has to have access to those. So BRK is one of the companies responsible for figuring out how to make that happen.

And finally, the last question. How can BRK implement BIM on 100% of projects by 2028? This as well is a very unique mandate from the government that was passed just a few years ago where all critical infrastructure projects have to incorporate BIM as part of the work. So BRK now has to adopt BIM across its entire asset base.

Of course, with these top three goals, there are some major challenges with achieving them. So I want to walk through those challenges. And then we're going to discuss a little bit about how you would think about solving them.

So the first key challenge is that in order to grow profitably, you want to be able to accomplish your projects profitably. But in current state, it takes over two months to evaluate and prioritize capital investments for a single wastewater treatment plant. And oftentimes it's fully dependent on costly engineering consultants. So it's very difficult to accomplish the amount of work that needs to be done and to do it affordably.

The second is that BRK has about a five person in-house engineering team, so the capacity is quite limited. And they're trained to evaluate conventional treatment technologies, meaning if a new regulation comes down the pipe that requires, let's say, a nitrogen limit as an example, a lower nitrogen limit, that conventional technologies can't meet, the team is not equipped to be able to evaluate those and implement them in the real world.

And finally, when it comes to 100% BIM adoption, they have no in-house Revit license or expertise, and none of the 171 treatment works have an existing BIM model. So it is extremely challenging for BRK to meet these three goals and to do them in a reasonable time frame.

So my question for you is if you were in their shoes, what would you do? How would you go about achieving these goals given those challenges? And by the way, you're owned by private equity, so you don't have the luxury of time. And also, by the way, you can't hire more talent. It either doesn't exist or you don't have the funds to do it. What options do you have to accomplish these goals?

Well, if you guessed automate your processes and turn to technology, you're right. Great work. And I hope that the title of this presentation gave it away a little bit and the reason that you came was to learn more about the specific use cases around automating the design of water and wastewater treatment works to achieve goals like the ones that BRK has. And that's exactly what we're going to talk about today.

But first, I want to set the table, and I want to set the table relative to software. And the reason that this is important is because there's a lot of terminology being thrown around today about generative AI, generative design, automation. And there are some key differences between the two that are going to be important to learn before we get into the real-world applications of software like generative design.

So as simply as I can put it, generative design and generative AI are two completely different technologies. Here's a quick example. If you were to ask a generative design solution, please create an apple, you might get something like this. It's the apple that we all know and love. The fruit. What we would expect. Something that we can trust. This is a reasonable output.

Yeah, we might debate about the type of apple it is or maybe the fact that you wanted it sliced versus whole. But ultimately, generative design gives you an output that you understand. Hey, I get it. This is what I was looking for. Let me then validate how it got to the answer. But it seems reasonable that I would expect a fruit.

On the other hand, if you were to ask generative AI to please create an apple, you might get a Macintosh. Something you didn't expect. And you don't know how or why the software came up with the output. And that is the more important distinction between the two.

Generative design is not generative AI. The way that I think about it is that I can trust generative design and the outputs it provides, whereas generative AI might look really cool, but I have no clue how it came up with the answer.

So to get into a tiny bit more detail, generative design is a quantitative approach to creating outputs. It defines and execute procedures. It's deterministic, meaning that it comes up with an answer and you can determine how it got there based off of rules and real-world calculations or rules of thumb that you have provided it. And it's predictable. If I ask it to generate an apple, it will generate an apple, and it won't generate something that I wasn't expecting.

Generative AI, on the other hand, especially for any of you that use ChatGPT or similar tools, this is a qualitative approach. It trains, learns, and creates out of nowhere. It's indeterminate. It has statistical modeling weights, and it's unpredictable. Again, you might get a Macintosh where you were expecting a fruit.

So let's think about this in the context of wastewater. If I were to ask generative AI, please design a wastewater treatment plant, which I did, it gave me this. Now, for those of you watching this, you're going to look at this chart or this graphic, and you're going to see a lot that makes absolutely no sense. Even if you're not in the water industry, you can look at this and say, hey, something here is not quite right.

As one example, take a look at the staircase in about the middle right hand of the page. It goes to nowhere. Clearly unsafe. Makes absolutely no sense. Look at the bottom left. You have some sort of torpedo-looking structure crossing a road. Completely impedes the flow of traffic. Clearly is not useful.

You have pipes going to nowhere. You have buildings in places they shouldn't be. You have a massive set of offices for thousands of people to work from, when in reality you probably only need a couple dozen on site. Again, this isn't what I was expecting. And quite frankly, it leaves me with a feeling of W-T-F.

I would never trust a generative AI solution to design a water or wastewater treatment plant and most certainly wouldn't trust to drink the water from something that was created or treated by something that was generated by generative AI.

On the other hand, if we were to ask generative design, please design a wastewater treatment plant, you might get an output that looks something more like this. It's based in reality. It accurately reflects the existing site. You can see whether it's in 2D or 3D. It has the right models and characteristics and features. And it's something that you would expect to see.

Of course, you're going to want to do your diligence to figure out is all of this legitimate or not. Can I trust the output? But you can always tie it back to real engineering principles, real rules of thumb. And you can figure out exactly why it chose to place certain buildings where it did, certain assets where it did.

And in this case, it leaves me and maybe you all amazed. You're thinking, wait a second. I can ask a computer to generate a wastewater treatment plant and it will actually come up with a legitimate output that's quantifiable, that I can trust? Some of you might be saying, I don't believe it. And that's completely valid. When I first started this company, I didn't believe it was necessarily possible either.

So let me go a bit deeper. I want to show you exactly how a solution like this might actually work. And for those of you who are engineers in the room, you're probably thinking, there's no way. There's simply no way. But let's talk through the process of how a software could actually design a wastewater treatment plant automatically without any human intervention.

Like any project, you start with basic inputs, things like your water quality data, your site information, whether that's a greenfield or potentially brownfield site with existing assets already there. You're looking at things like the effluent that you're trying to create. You're looking at the treatment technology that you might want to select to treat that wastewater. And you're looking at things like specific preferences for the client.

So in Brazil, you may say, hey, temperature is really hot. We need to think about that when it comes to designing these treatment systems. Or you might say, hey, this is in a really arid part of Brazil or really humid part of Brazil, which also has impacts on what this treatment plant might look like or might operate.

Once you do that, the software goes through a set of processes that automate individual components of what engineers today do manually. One of those would be document generation. So software can now accurately put different variables and calculations inside of a Word document and generate a preliminary engineering report or something like a technical description.

This data-- you're probably wondering, well, wait a second. How does it actually come up, though, with those numbers? Where does it get those from? Well, one, it would be user input. So it's literally just a copy paste exercise. But the more interesting one is that the software in generative design can actually automate a biological process simulation. And not only do that, but optimize a building layout and a civil and site arrangement.

So software will actually do the calculations to figure out what needs to happen in order for the wastewater to be treated effectively, take that data and those calculations, and then generate documentation based off of that.

It can also generate a P&ID. So in this case, the software has determined you have the process calculations, you have the data around each asset that's required for treating wastewater. Now draw it for me. And with a simple integration into AutoCAD, you can actually generate a P&ID.

Then finally, all of this data, you now have data around each asset type, the equipment that's required. You can actually have it in a process flow. The software then can generate a Revit file automatically based off of the APIs that Revit provides.

So by the end of it, with a simple set of inputs, you then have a complete conceptual design package effectively instantly. Now, again, some of you may be saying, I still don't believe it. I don't think it's possible. It doesn't make sense. I need to see it. I need to visualize it. Well, let me try to do you one better.

So I mentioned you have the dynamic biological process simulation, and you automate that component. Just basic math. Once the software understands the process units required, it then actually generates rooms for all of the equipment that's part of that process.

So you can actually watch a computer generate an abstract building model. It's placing equipment inside of rooms and then putting those rooms inside of buildings, because that's where rooms tend to go. So you can see, you zoom in and out you have, in this case, the wastewater treatment plant. You have the clarifiers. You have the membrane tanks. You might have a blower building. And that's all created based off of real calculations and real engineering principles.

Once you have those buildings, then the software can try to optimally arrange them on a site. And this one I'm going to come back to in a second because it's really interesting. I've heard many comments from civil engineers in particular over the years about the optimization of the site arrangement.

Once then you have the buildings and the assets, you can then generate that BIM model. So it's the same procedure engineering teams follow today, the same calculations that they use, the same technologies that they use. It's just all done automatically with a generative design software. We can trace it all back to real engineering rules of thumb, real engineering calculations.

And finally, when everything is done, because you have all of this embedded data, you turn it over to your document generator, and you're generating things like a civil bill of quantities, an equipment list, a load list, doing CapEx and OpEx calculations. You're figuring out rapidly what is this treatment plant going to cost me, how is it going to operate, and you're coming up with answers very quickly where traditionally it would take quite a bit of time.

I mentioned before that generative design is not generative AI. However, occasionally you will find that generative AI is used within generative design applications. In this case, I mentioned the site arrangement. Here you can see software automatically finding the best fit for buildings on a wastewater treatment site.

So there are existing assets that are highlighted in yellow. The new assets that the software has designed are in blue. And now you actually watch the algorithm figure out where is the best place to put these new assets.

Civil engineers have told me over the years that this used to be done on paper. You would cut out physical shapes of buildings, of clarifiers, whatever assets you have, and a human would manually try to arrange them in the best fit. But with generative design software and a tiny bit of AI, this is a perfect problem for a computer to solve. In fact, over 16 million possible combinations of these assets all done in seconds.

So let's go back to BRK. We're sort of taking the future of what wastewater treatment plant design looks like and bringing it to now. You may have heard the quote, "the best time to plant a tree was 20 years ago. The second best time is now." Well, someone else wise, may or may not be me, once said, "the best time to adopt generative design was five years ago. The second best time is now."

So despite being a technologist and someone who's really excited about the software, buying the software is the easy part. Adopting the tool, changing your workflows, getting value from it is a process. And that is the hard part.

So let's talk about that journey that BRK has been on for the last five years using generative design. It all started in 2020, where they ran a pay-per-use pilot to validate the ROI of the software. They ran two designs for a municipal wastewater treatment project. And based off of the data and the quality of the designs that they produced as part of their bid package, they won a $2 billion concessions contract for a wastewater treatment plant.

That was enough then to convert to an unlimited use subscription where anyone inside the organization can use the software on any project that they'd like. And they started to upskill their internal teams with not only generative design technology, but Revit in particular.

Fast forward to 2022 where they then actually mandated that people had to use this software in order to get any resourcing on a project. They created a centralized generative design team who were responsible for creating all of the designs across the company. And they actually issued an order, if you will, where it said all bids, preliminary designs, costing, BIM, et cetera, must first be completed using a generative design software.

This is a radical shift for an organization to make. But again, it's a process. Had they tried to do that right at the beginning, guaranteed would have failed, without a doubt.

Over the last few years, they've started to expand the use of their platform, not only in terms of use cases, but features. So moving to smart P&IDs and integrating those into Revit, looking at natural capital systems and natural treatment systems, so things like reed beds, low carbon, low OpEx treatment technologies.

And then PFAS, of course, is an emerging issue. So looking at how to automate the design of PFAS treatment systems and be able to do that cost effectively and in a timely manner.

So for the engineers who are listening, you're probably saying, OK, great, well, this all sounds fine and dandy, but I need to see a real example. Give me a real-world thing that I can actually chew on. So let me give you one.

BRK was trying to-- this is one simple example of the hundreds, if not thousands at this point that they have completed. But this one is relatively straightforward. What's the best treatment technology to use for a 300 liter per second wastewater treatment plant? Calculating CapEx and OpEx and given a range of how we might operate the facility.

And in this case, they wanted to look at three different technology types-- conventional, or activated sludge, Upflow Anaerobic Sludge Blanket, UASB with a combination of activated sludge, and then MBBR.

This is something that would have taken six months using traditional manual processes and outsourced design consultants. And instead, it took one week and was done in house.

So let me get into a little more nitty gritty. I'm not a process engineer, so forgive me if it's not technical enough. But of course, happy to have the conversation after the fact.

They started with the inputs. So how are they planning to operate this treatment plant in regards to the influent and effluent that they're looking at treating and generating? And of course, all of this is based off of, again, standard engineering rules of thumb. So in this case, Metcalf and Eddy, the wastewater engineering Bible, if you will. There's a beautiful little 3D model of the plant that you can see down below that was generated.

But ultimately, the important part is, OK, what decisions are we going to make based off of this data. Well, here's a beautiful chart for those of you that like those. But most importantly, here are the conclusions. So UASB and activated sludge had a higher CapEx, MBBR with a higher OpEx. Activated sludge offered the best relationship between the two, so that's the one they actually ended up choosing.

And they learned that as they operate this facility, they have to pay super close attention to the total nitrogen limit or else things could get thrown out of whack.

So now you might be left saying, so what? Seems cool. Cool technology. I kind of see the benefits, right? Like, save some people some time. Eh, whatever. OK, let me make this as clear as I can for you. Using generative design, BRK has been able to accomplish a couple different things.

Number one, they slashed their conceptual wastewater design costs by 80% and the time it takes to evaluate site and site options for wastewater treatment plants from two months to less than one week. They generated eight to 10 more conceptual design options per facility to inform investment prioritization.

And this includes BIM modeling for every single one of those scenarios, because again, a BIM model is automatically generated. But not only existing facilities, but any new facilities or new proposed builds or expansions.

This is one that a lot of people don't talk about but I think is very important. Using new technologies not only encourages more innovative thinking and honestly, retention of existing teams, but also increases their confidence. They start to get more curious about the new possibilities of software, the possibilities of doing cool things, of really starting to play around with the options that might exist to deliver water and wastewater treatment services to the population of Brazil.

And then finally, accelerating innovative treatment technology and integration into greenfield and brownfield projects. So just like in that example we evaluated, they traditionally would have only evaluated activated sludge, maybe one other option. In this case, they looked at three options and were able to make a smart decision, not only in terms of CapEx and OpEx, but also operations for what that facility needed to look like and needed to do.

So I am done with my piece, but I would like to share something directly from BRK themselves. In case all of this is lost and you don't remember anything, I just want you to hear it directly from BRK themselves. So here is what [INAUDIBLE] from BRK has to say about generative design technology and the solutions that we and Autodesk can bring to companies like BRK.

[VIDEO PLAYBACK]

- I believe it's very difficult to change the world. However, you can affect the place where you are. Being in the water sector, I can improve people's quality of life, water to drink, that they have their sewage being collected and treated. This is basic human rights and needs.

BRK adopted technologies like Transcend and Autodesk in our company because when we were dealing with conceptual design, we faced a lot of trouble with the third parties. The cause of it was very expensive. Transcend allows us to do unlimited scenarios. We can run designs and have the answers in eight hours. And we can do the business cases, the conceptual studies of it in one week. Like a whole project.

We also have the opportunity to use the data provided to get approvals on permits for environmental agencies and for bank loans. Transcend and Autodesk have done such amazing work with their partnerships. Now we can take the outputs of the Transcend platform and run on Revit, on AutoCAD, and we are able to change things, we're able to edit things.

When we want to change the building, like the thickness of the wall, the length of it, change that in a click and the tables will be automatically optimized. You will improve your capabilities. You will have more control over your process, over your design, and you will be able to make decisions and calls faster. It's a great partnership.

[END PLAYBACK]

ADAM TANK: Thanks so much for listening. Of course, if you have any questions, if you're curious about how generative design can impact your organization, whether you are a water utility, an engineering services provider, or an equipment vendor, I would love to talk to you. And I hope you've got a chance to see a glimpse of the future today. Thank you.