Insights on Design Automation for Water Treatment Facilities

MIKE ETHERIDGE: Welcome, everyone, to Insights on Design Automation for Water Treatment Facilities. I'm Mike Etheridge, the process management lead for Black and Veatch. And joining me today will be Adam Tank, chief customer officer for Transcend. Just a brief agenda for today, I'm going to talk a little bit about just some general automation, artificial intelligence, machine learning trends in the water industry with a focus on the engineering space of that for AEC firms and EPC firms.

Adam is going to talk some on rules-based, generative design-based tools. And he'll also provide some specific project examples and talk a bit about the benefits for those advanced automation tools that he'll speak to.

So just from a high-level overview, as automation trends and AI/ML matures within the overall technology space, there are some places where AEC and EPC firms have started to take advantage of that automation and those tools to leverage design support. And one of the places where that has happened is the use of BOTs.

So in our experience, there's two types of BOTs that have been leveraged for that type of design support. One are the traditional Robotic Process Automation, RPA, BOTs that are programmed to perform repetitive tasks-- so generally, very task specific, tend to be fairly simple, and focused on performing just one activity with a limited number of inputs and a limited number of users.

And so in the design space, it's to maybe perform calculations. In the BIM space, it might be to automate title blocks and those sorts of activities, but again, very task focused and very focused on producing the typical design deliverables that an AEC or EPC firm would produce, just automating some of those tasks.

The other place where we've started to leverage this is with chatbots-- so chatbots that use natural language intelligence but also have some programmed intelligence primarily for finding information. So AEC and EPC firms tend to have a lot of standards, tools, processes, information, best practices that can be sometimes hard to find.

And we've actually been able to leverage BOTs to help users find that information more quickly in all the variety of places where it might be stored, whether that's document management systems, CAD and BIM systems, a variety of places where you'll grab that data in the context that the user is familiar with. So that's one of the places where we're starting to take advantage of better technology in the automation space.

The other place is a bit more traditional, and this goes back to the very beginning of AutoCAD, which is just automating CAD tasks within the CAD and BIM applications, so all the way back to LISP and VBA programming within AutoCAD, now extending to Dynamo within Revit and similar applications where, again, you're focused on production activities and specific tasks to improve efficiency of designers.

But we really haven't probably changed the overall workflow. We've just automated pieces of that to make it easier. And the focus has still tended to be on designers having to be in the design application directly producing models and drawings and the typical deliverables that they produce. We've just found ways to automate those activities both with the automation tools that are in those design applications or with third-party items as well, but again, very focused on tasks, not some of the broader stuff that Adam is going to talk about here in a little bit.

One other thing to just note real quickly is maybe a little bit tangential to automation but still very important is to move away from the activity-based automation and really be focused more on data and data integration because in the end, as you look at how all of these automations provide the most benefit to both the AEC firm and the end client, it's how well the data can flow from one phase of the project to the next phase, how it can be integrated with the design applications, operations applications, construction applications, which kind of leads to the concept of digital twins, which everyone hears more and more about in recent times.

From the perspective of automation, just briefly want to touch on digital twins themselves could have automation embedded in them. But from a perspective of looking at design and design tools, really see the automation play there being, again, data driven of we're generating all this data in the design space, whether that's in a tool like Revit or in other design applications that we're using to produce the design.

How can we grab that data in an automated fashion and move it forward to the construction phase and those tools that need that same data, then on to operations where the client can leverage some of that same data, which might be the graphical data for the model, might be asset data, operations data, move that data forward in an automated fashion so that it's not a heavy lift to transition from one phase of the project to the other?

But again, these activities that I have talked about and some of the current state of leveraging automation and AI and ML within the design space, again, is largely focused on automating the typical traditional deliverables and activities that we have done in this space for 100 years, even predating CAD tools.

And what Adam is going to talk about is transitioning to something that is more of a rules-based generative design tool that's allowing the engineers to take those rules that we use to do things manually and leverage automation and AI tools to take those rules, allow the engineer to focus on just the core inputs, the important engineering decisions, and then let the tool actually produce some of those deliverables that we were producing manually in the past.

So it's really a step change in using automation and better leveraging the engineering and designer resources that we have to focus on the important things and let the tool set do some of the generation of the output. So with that, I will turn it over to Adam and let him continue on from here.

ADAM TANK: Great, thanks, Mike. Thanks for the introduction and excited to be speaking to Autodesk University about some of these automation tools in the water and wastewater space. So the first thing I'll note is that the world at large is at an inflection point. And that inflection point is this, that a shift from manual tasks performed by humans like those that Mike was talking about relative to CAD and Revit, et cetera, to automated processes powered by machines is transforming entire industries right now.

And you can see that that's independent of the industry. It's happening all over the place. So whether it's in health care manufacturing or even HR, we're seeing these automation tools really start to take over those manual tasks that people traditionally did that took hours and hours and moving towards a more sophisticated automated system.

If you would have taken a look at Wall Street maybe 80 or 90 years ago, here's the sight you would have seen, hundreds if not thousands of people trading paper tickets screaming over one another in a pretty chaotic environment. And now if you walk in, it's largely been replaced by computers, automated trading, artificial intelligence, algorithms, et cetera. And the humans are now focused on analyzing the data, just as Mike said. So those decisions are now integrated, and it's more about data analysis, data transfer, and making smarter decisions and letting the computers handle, we'll say, the mundane work.

Something similar has happened in a lot of medical rooms or operating rooms. So if you take a look at this picture, 1920s, 1930s, you would have seen a room full of surgeons, techs, et cetera, working on a patient maybe not necessarily with the best outcomes. And of course, today, we now have robotics and automation, something like the da Vinci machine where you now only have a surgeon potentially not even in the room. You might have one technician, and you're delivering a better outcome for the patient through the use of sophisticated automation tools and robotics.

I love this video as a representation of what the shift in technology can do in order to help firms and teams be more competitive and the benefits of what automation can bring. So if you take a look at the top portion of the slide, F1 team in 1981 versus a team in 2019, there's a very distinct difference in how these teams are performing. The 1981 car we're still waiting on to get going.

If I ask you what the difference is between these two videos, you'll probably recognize things like maybe the technologies are different. Maybe the tools are different. The team members, of course, are wearing a lot more safety gear, which is also important. But something to note here is that you still have the human involved. They're just leveraging technology in a way that speeds up the entire process and delivers a better outcome for the group or for the challenge that they're working on. In this case, it's getting an F1 car back on the track.

So we come to the AEC space or EPC space, especially around design work. Just like you would have looked at Wall Street, you looked at an operating room, you looked at an old F1 team, a lot of people working manually on various tasks. If you've been in the industry long enough, this is a sight that would be familiar to you, a lot of folks working by hand drafting various documents.

Of course, now we've brought computers into the fray. But my question is, have we really automated the process? As Mike mentioned, we might have different ways of automating specific parts of tools we already use. So as an example, Mike mentioned Dynamo. We can use Revit families. We can look at some CAD automations. But are we really thinking about it from an end-to-end point of view and streamlining the entire design process?

We at Transcend believe that the answer is no and that we think AEC and EPC design work will be the next major industry that's transformed by these radical advances in automation. So as one example, and this is a short video from actually one of the things our software is producing, which is a P&ID automatically.

So a designer or a CAD technician no longer actually has to open AutoCAD to generate a P&ID. The software is generating a P&ID bespoke. This isn't based off of a copy-paste template. This isn't from a past library. This is actually a P&ID specific to the project based on project-specific parameters. And the software is taking care of the entire design process.

You can also take a look at something like processing mechanical outputs. So in the case of a wastewater treatment plant, we can actually take the process flow diagrams and the equipment and turn that into various spaces and rooms and buildings on a site. So what's happening now is you can see a software actually generating an abstract building model based off of those project-specific parameters.

So again, this isn't copy paste. There's not a human that's looking at past projects and trying to add a little bit of flavor on something they've already created. This is software creating a new design bespoke. And you can see that BIM file starting to come together on the left-hand side of the screen.

So another part of this, too, is that we can actually optimize site layouts with artificial intelligence. So in the past, very far past, where engineers used to cut out shapes of clarifiers or blower buildings or other assets on a treatment site and try to optimize that or arrange them on a piece of paper for the optimal layout, perfect problem for computers to solve, and computers can now do that.

So if we put all those pieces together and we automate that entire workflow, what we end up with is a complete BIM model with all of that data that Mike discussed embedded in the BIM model. Each building is generated dynamically. All of the equipment is sized and selected appropriately. And the engineers now have an output to work from and make smart decisions based off of to ideally deliver better outcomes for their clients, in this case, a water or wastewater utility.

Once you have that, BIM model, you also, of course, have all the data embedded, which can help you create things like a civil BOQ. You can have CapEx and OpEx estimations. You can have full equipment lists, load lists, the list goes on. And it's all done in the cloud by software, again, no human intervention during any of this process.

Regardless of the construction project methodology that's being used or project methodology that's being used, in either case, conventional approaches or more design build or collaborative approaches, this preliminary design phase is critically important, and it's really what kicks off the whole project and where things can really go astray.

So as Mike mentioned, if you have a good set of data that can flow through from the very beginning stages all the way to the end, it's going to be better for all the parties involved in a project. And what we see as Transcend is that EPC firms, our clients and some of the larger ones are spending more than 800,000 USD per preliminary design package because, of course, the internal cost of doing this is incredibly laborious, a ton of manual work, and oftentimes, it's work that depending on the project methodology can't be billed for. So it's literally just cost, pure cost.

And if we look at the traditional process of designing wastewater treatment plants, what we hear from engineers is oftentimes, it's a manual, cumbersome, and in many cases, boring process. I can't tell you how many engineers have said, I'm tired of feeling like a glorified copy-paster. So in this process, what you'll see is a typical project's launched.

You then have a process engineer that's focused on using their particular tools that they will assess various process, improvements, and equipment, et cetera, to do the treatment works. They'll hand that data over to a mechanical engineering team. The mechanical engineers will hand that data to a controls and instrumentation team and electrical engineers. And then, of course you'll, hand it to a civil and architectural team.

Again, each of these disciplines may have their own sets of automation tools, but no one is actually taking the rules that each of them are using, automating them, and then streamlining that in the cloud. So this process not only happens once, but oftentimes, happens many, many times. And there's multiple iterations based on client feedback, which, of course, increases rework, increases the reduce of QA/QC errors, and just, quite frankly, it's a very challenging process to manage.

So when you think about what software can do in this process, let's talk about some examples. So in this case, what we're going to be talking about is an example of one utility in the Southern US that had an RFP out for a master plan update to one of their wastewater treatment plants. And we'll talk about how automation was brought into the project to accelerate that process.

So in this case, TRA had a 24 MGD facility where they were requesting their engineering consulting firms that were proposing to work on this project to evaluate three upgrade scenarios. TRA wanted to look at what happens if there are future effluent requirements or more stringent regulatory requirements for our wastewater discharge. And how about, what if there's increased loading or different wastewater types that come into the treatment plant based on population growth?

And of course, Texas had severe flooding in recent memory, and so they were thinking, what type of expansion should we consider if we want to be able to handle wet weather events? But of course, in the traditional model, it's very difficult to evaluate a number of these scenarios and all the permutations of those scenarios unless you're using something like a design automation tool.

So what TRA ended up doing was actually issue an addendum to that RFP where they included language around automation and tools like what Transcend provides. So actually, in this addendum, what they did was that they requested that their designs, instead of three scenarios, they requested the EPC firms provide 30 scenarios.

And they specifically asked for conceptual design software that automates every one of those engineering disciplines, process engineering, mechanical engineering, electrical, and Civil. And they also wanted a full 3D BIM model in addition to a number of other documents, including those native files that Autodesk software produces, so DWG files, Revit files, et cetera.

So for anyone that's in the audience that's a consulting engineer, typically, your jaws are dropping because you're saying, oh, my God, how in the world could we possibly do 30 conceptual designs in the same time and budget where traditionally we can only do three? And the answer is simply automation.

So one of the interesting parts about this in addition to including the types of files that they're looking for and the use cases that they're assessing is why they selected 30. That's typically a question we hear. If you only wanted 3, why have 30? Well, the answer is pretty simple.

So you imagine those first three scenarios that they were looking at evaluating, but if you have the ability to automate some of the decision making around those scenarios, you can see very quickly, you can now have 3, 6, 9, 12 additional scenarios for each of the first set that they wanted evaluated. And then in addition to those, you can also have more options in each of those additional scenarios.

So as an example, a technology evaluation, these are traditional wastewater treatment technologies. If you're looking at a conventional plant versus something more innovative like an MBBR plant or maybe an MBR plant, you not only have that scenario to evaluate, but you also have to evaluate that scenario in the context of a new temperature, a new effluent requirement, new flow limits, et cetera. And so the number of permutations gets very complex if you're not doing it with automation.

So we talked about this old process or the standard process, the traditional process of designing a wastewater treatment plant. So the future process is much different in that it's exactly what Mike spoke to earlier, which is you have the engineers, rather than opening up their individual programs, designing these files or these outputs, the software designs the outputs, combines all of that decision making, and then the engineers can look at the outputs and spend their time evaluating outcomes rather than inputs, spend their time evaluating what's best for the utility instead of spending time manually creating documents in their program of choice.

So one of the other cool parts about a program like Transcend or other automation tools is that, again, it's not just discipline-specific output. So it's not just a P&ID, or it's not just a Revit file. You're also generating things like Word documents, Excel documents. You're automating process simulations.

Of course, you are doing the P&IDs and the CAD work. You have site and civil layouts that are optimized based on artificial intelligence and rules for how to design a wastewater treatment plant. And of course, you have the full BIM model that comes at the back end of all of this. So it's quite powerful when you have all of these rules that are being used to generate the entire facility layout effectively instantly. And then the engineers can spend their time evaluating, again, outcomes.

So a particular part of an artificial intelligence to talk about, as I mentioned, the old engineering model of cutting out physical shapes and then laying them out on a paper drawing and trying to figure out the best arrangement. Artificial intelligence specifically in our case, genetic algorithms, phenomenal challenge for a computer to solve.

So the end outcome, you can actually design a treatment plant based on what you're looking for, be it carbon reduction, GHG, site footprint, CapEx or OpEx, and you can actually have the software come up with optimal layouts for each of those scenarios that you're trying to evaluate.

So let's take a look at what happened in TRA's case. So this was their existing wastewater treatment plant. There's a really nifty tool inside of our software where you can actually assess the site based on a Google Maps overlay. And we were basically trying to say, here's the site we have to work with. If we're going to expand this facility, what are the assets required, and how much space will they take up?

So AI generates the site layout for new assets. So here are the existing ones. We mark those within the tool. And then AI actually generated everything above the fold, so what just came in, so new biological treatment, new clarifiers, maybe watering building, et cetera. And again, it was done instantly, all the process engineering, mechanical, electrical, and civil.

So the 3D BIM model as well, so all the greyed out buildings were the existing buildings. All the buildings that are in color are the new buildings. And again, you have the fully editable Revit file for any engineers that want to do more detailed work. So what happened here is that the winning firm included Transcend as part of their design package. And I'm proud to say that Black and Veatch has been a phenomenal client with us in order to deliver better outcomes to TRA through the use of automation tools.

Another example of this in practice is in West Chicago. So in West Chicago, of course, site or land is at a premium. So being able to evaluate different scenarios for site footprint is really important. So as one example, you can see very rapidly, if we're going to build a traditional, a conventional plant versus an MBBR plant, you can take that to the utility and say, hey, I know you only wanted us to evaluate a conventional plant, but here is a new innovative treatment type. We think it might be a good fit given that there are some site constraints. And you can see the MBBR plant is far more compact, far more consolidated.

And again, we're only scratching the surface here. So if you look at in the case of the West Chicago example a CAS and MBBR plant on the top left-hand side that are highlighted in yellow, we're only looking at footprint, CAS and MBBR, and maybe an expansion due to population growth. But there's this whole other realm of possibilities that can be assessed basically instantaneously through the use of automation tools.

We also believe that having more options evaluated in the early stages leads to less risk and improved outcomes. So a lot of what we hear from our clients or even utilities is, well, this is risky. This is a new technology. Why should we invest or why should we adopt this? And as it turns out, when you have someone evaluating or a software evaluating thousands of potential options rather than an engineer manually evaluating maybe one or two, you're actually reducing risk or helping manage risks for a project because you're able to assess far more than you ever could before.

So I'll leave you with this. We cannot do this all manually. There's far too much infrastructure, far too many projects, far too much time that has to be completed with manual work that we cannot deliver the best outcomes for our infrastructure without automating. And so we at Transcend believe that design automation is the key to resilient water infrastructure. And I hope we're able to work on future projects together.