Streamlining MEP Sustainability Analysis Through Power BI and the ACC Model Viewer

MARGARITA KRETERNOVA: Hello. Good afternoon, everybody. This is Streamlining MEP Sustainability Analysis Through Power BI and ACC Model Viewer. I'm Margarita Kreternova. I am the Director of Digital Delivery for Data Centers at Jacobs.

Since 2017, I've been focused on digital execution and delivery for data center projects, specializing in solving BIM challenges for hyperscale and co-location clients. My passion lies in sustainability and software interoperability, finding efficient ways to extract data from models to improve both environmental performance and project delivery.

After graduating with a dual major in architecture and sustainability in 2012, I wasn't sure I'd be able to apply my sustainability knowledge. And I feel very fortunate that after years of focusing on BIM, I'm now able to combine my passion with my professional work.

I'll begin by looking at data center Life Cycle Assessments, known as LCAs, from a high-level perspective and discuss how these assessments differ from those done for commercial and residential buildings. Next, I'll demonstrate the critical importance of accounting for MEP elements in data center LCAs, supported by evidence from several studies conducted by Jacobs.

Next, I'll explore a case study where Jacobs was challenged by a major confidential client to perform a comprehensive whole building LCA several years ago. This case study will lead into the technical substance of today's presentation, where I'll showcase our custom solution for effectively calculating the global warming impact of MEP systems. And, finally, I'll close with some lessons learned and discuss next steps for improving MEP LCA methodologies.

So, first, an introduction to data center LCAs and the missing piece to the puzzle. The global demand for data centers continues to rise, particularly following the surge in AI technologies, which has increased the need for cloud storage and processing power. Traditionally, LCAs have focused primarily on structural and architectural systems. However, we're now realizing that this approach overlooks substantial portions of global warming impact, especially for building types like data centers.

Over the past five years, our comprehensive whole building LCAs have grown into a significant market segment for us, providing us with the opportunity to refine and optimize our workflows, which I'll be sharing with you today.

LCA first gained popularity through LEED certifications, which at the time didn't focus very much on MEP systems. While this approach might make sense for commercial and residential buildings, when analyzing advanced facilities like data centers, a large portion of the building's components are being overlooked.

In our early work, we utilized some model auditing tools to extract item counts from data center Revit models and found that MEP services made up 74% of all items. While this didn't equate to 74% of global warming impact, it really underscored for us the necessity of calculating MEP impact to create a truly accurate whole building LCA.

As we began conducting whole building LCAs, we relied on case studies from other building types, such as health care facilities, to identify items that had a high, medium, and low global warming impact. During this analysis, we discovered that tools like Tally were missing large segments of high and medium impact MEP items, which you can see outlined here in dashed lines.

This gap in analysis led us to develop custom tools that would fill this void. I think it's worth noting that this happened incrementally. We actually began in Excel, then transitioned to Power BI, and ultimately developed the custom solution, which I'll demonstrate for you today. This progression reflects the organic development of a truly useful tool shaped by demand and collaboration with subject matter experts.

Through years of conducting the whole building LCAs, we've gathered substantial evidence that MEP systems contribute significantly to global warming, with MEP systems accounting for approximately 15% of total impact and equipment averaging 43%. Here we present data sets from different projects, clients, and regions, all showing a consistent trend in MEP contributions to global warming.

I was personally involved in the first two projects. And when I saw the result from the third study conducted by a completely different team, I felt very validated in our methodology as the percentages were strikingly similar. From the previous slide, I hope it's evident that MEP LCA analysis is crucial for data centers if we want to achieve an accurate representation of global warming impact.

Initially, our process was manual, involving exporting Revit schedules and then performing calculations in Excel. However, as client demand grew, we realized we needed a more efficient way to extract, analyze, and visualize MEP elements for life cycle analysis. This brings us to the technical portion of our presentation where I'll walk you through the custom solutions we developed to meet these needs.

So before zooming into MEP systems LCA an analysis, I am going to talk briefly about how we approached whole building LCA when we were first challenged to do it by our clients because I think walking through this process will allow you to understand our thinking and logic. Our process began by extracting quantities from Revit and assigning LCA factor using a combination of tools tailored to region and item type. These tools included Tally, SimaPro, and One Click, depending on where the LCA was located.

We then consolidated all the information into Power BI because we needed to create user friendly and accessible visuals for the client. Although we had to string together a bunch of different data sources and analysis methodologies in order to get a result, we didn't need the client to see that. For the client, the important thing was to see a combined report in Power BI and get a final combined number of what the global warming impact of their building was.

As we work through various elements in the Revit models, we noticed a pattern, how we calculated global warming impact for each Revit category. Architectural and structural elements were easily covered by existing plugins. For MEP systems, we extracted key dimensions and performed calculations to determine material mass and then applied LCA factors accordingly.

Equipment required us to extract quantities as well as model, make, and type information from the models and research Environmental Product Declarations, also known as EPDs, for each piece of equipment. Unfortunately, this is a highly manual task that was handled by our LCA experts.

From this work, three distinct workflows emerged, each dependent on the type of item from the Revit model. All three workflows converged in Power BI, allowing us to provide clients with a consolidated global warming impact figure and easy to use visuals. Among these workflows, the MEP system's workflow showed the most potential for packaging into a tool that could significantly increase the speed and accuracy of the calculation. This will be the focus of the technical part of today's presentation.

So let's take a deep dive into the MEP system tools we developed. To optimize MEP LCA analysis, we used two custom tools. One is called SIFT and the other one is called Alluvial. SIFT is a custom visualization tool that pulls an Autodesk Construction Cloud model view into Power BI, allowing for customization and manipulation.

If you're familiar with Power BI, you'll recognize this visualization pane. SIFT is a custom visualization that can be added to a Power BI dashboard. Alluvial allows for secure data access and processing via an Azure data lakehouse and facilitates importing data into Power BI as a web source. Again, Power BI users will recognize importing data as a web source in this image. Once these tools are integrated into Power BI, the dashboard design uses typical out-of-the-box Power BI functionality.

Our workflow can be broken down into three parts. First, we configure an ACC view and identify parameters to pull into Power BI for LCA calculations using Alluvial. Next, we move into Power BI for data cleanup, calculation of volumes, masses, material consolidation, and assigning LCA factors.

And, finally, we create the Power BI report, which visualizes total global warming impact alongside color-coded visuals. Before we dive into the details, here's a high-level roadmap of the ACC Alluvial portion of the workflow. In the upcoming slides, I'll walk through each step, but this is an overview. And it will give you an understanding of what's ahead.

So first of all, as I mentioned, it's key to create a publish view in ACC that contains the information that you're interested in analyzing. Then you create a new Alluvial SIFT workspace. After that, you allow the Alluvial platform access to your ACC data. Then you'll be prompted to select files from an ACC project.

And after selecting the file, you'll be prompted to select a specific view and a specific version of the view. At that point, a model ID and SIFT key will be generated, which it's important to note for use later in Power BI. After that, the next step that comes is importing all of the model data from ACC into Alluvial. And this takes a little bit of time.

After all of the data is imported, the user creates attribute sets, which are basically curated lists of attributes that are important to the analysis, which will then be brought into Power BI. After the curated lists are created, they are processed. After they're processed, unique URL is generated for each list, and that is what goes into Power BI.

So a well-developed ACC view is really the foundation for the success of this workflow. It's important to isolate content that you're interested in analyzing as this reduces the visual load times and simplifies attribute selection. For example, if doors were included in this view, you would have unnecessary width and height attributes for doors mixed in with information about ducts, making the selection process for useful attributes more time consuming.

Remember, the SVF file format used by ACC is structured differently than the native Revit file, so finding corresponding parameters might require some trial and error and just some getting used to. Also, it's important to understand that depending on the shape of the system, whether it's round, rectangular, or trough, you need to focus on different dimensions to perform calculations. Another important thing to keep in mind is that materials are also stored in different parameters depending on the system. And sometimes it takes a while to find where designers have put the material parameter.

Once the ACC view is configured, we move into Alluvial. After creating a new workspace, we begin by signing into the Autodesk Construction Cloud. This allows Alluvial to access the ACC data for the selected project. We navigate through the project folders to find a specific view.

And then Alluvial imports all the data associated with the geometry and the selected view into the Azure data lakehouse. Once this process is complete, we create attribute sets, which are very similar to Revit schedules and contain curated sets of attributes that will eventually import into Power BI.

Now, I'm going to do a short video demo and walk you through the process because I think it's much easier to understand when you see a visual. So on the left hand, we have the ACC model viewer open. And on the right, you can see the Alluvial interface.

First, I'll focus on the electrical model and proceed to create new attribute set specifically for cable tray. You can think of attribute sets much like schedules in Revit, where we define which parameters will be displayed and tailored to our analysis.

The initial field that needs to appear in all the schedules is the name, and the name is essentially what's at the top of the properties in the ACC model viewer. For models coming from Revit, this is going to be a combination of the Revit category with the element ID. From here, the process becomes a little bit more intuitive, as you simply scroll down and identify which information you'd like to bring in.

You can see that I'm going through and selecting anything that is relevant to the LCA. When I get to dimensions, things get a little bit more complicated. As you can see, length appears multiple times.

This happens because there are several components in the view, cable tray, conduit, and wires, that all have length parameters. I'll go ahead and include all of them. And they will become consolidated during the attribute processing stage. I'm continuing with the selection of different dimensions and including anything that matches the criteria.

Finally, I'm scrolling through and looking for any additional parameters that may hold valuable material or shape information. Specifically for cable tray, there is information related to perforation. Whether a cable tray is wire mesh or solid, that can play a critical role. So I'm identifying any parameters that might contain this type of information. Finally, I am saving the set and confirming that everything has been saved correctly.

So after the sets are created, the attribute sets need to be processed. The processing extracts the curated data set from the model and places it into the Azure data lakehouse and then generates a URL for the data set that can be input into Power BI as a web source. Note, when there's a change in the model, attributes need to be either reprocessed, or a daily reprocess needs to be configured at the beginning of the project. But this is incredibly valuable because it means that as the designers are working on the Revit file, the information gets updated in Power BI.

Now that we've completed the data extraction process in Alluvial, we're ready to move into Power BI. Similar to the ACC Alluvial workflow, here's a high-level roadmap of the Power BI workflow. At first, we'll import the data as a web source. Then we'll paste the Alluvial SIFT URL in.

Afterwards, we'll manipulate the data and do some vital cleanup. Then I will talk about importing LCA factors into the Power BI dashboard, how to do the LCA calculations, and then, finally, how to combine all of the MEP system results together to get a combined global warming number, as well as how to create and colorize the visuals with relevant information.

Again, I'm going to do a video demo and walk through the process with you. Here I'm retrieving the Alluvial SIFT URL for a specific attribute set. In this case, it's the conduit from the electrical model and then using the Power BI web source data connection to bring all of this information in.

The data will take a second to import. I'm going to jump straight into transforming the data in Power Query to perform some initial cleanup and filtering. One of the first things I'll do is rename the data source to something more intuitive, like conduit for clarity.

Then as we saw in the ACC model viewer, the name of each item imported from Revit consists of two parts, the Revit category and the element ID. In this step, I'm splitting this information into two separate columns since I'll need the Revit category column to filter out just the conduit data. I'm then cleaning up the element ID data a little bit more. Now, I'm filtering for just conduit before adding this data set into the Power BI dashboard so that it performs faster.

Finally, I'm loading it in, which reloads all of my data sets. And as you can see, conduit now appears in the upper right-hand side as a data source.

One of the first steps in Power BI is cleaning up how the data comes through. We did a little bit of this in the last step, so breaking up the name column is a vital step. Then it's essential to make sure that all of the dimensions have consistent units.

For example, in our calculations, we chose to standardize everything to feet. But while length comes in feet, the other dimensions like height, width, and diameter actually come in inches. And so a conversion was required in order to make sure that all of the units were aligned and our calculations were correct.

The next step is essentially consolidating materials. So rather than enforcing a strict naming convention in Revit, we find it easier to consolidate material names in Power BI. Often, we're analyzing models created by other firms or by clients. So enforcing a standard naming convention isn't always an option.

At this point, our LCA experts will parse through the data to identify what different terms refer to the same material as far as life cycle analysis is concerned. Here, I'm showing you an example where rigid polyvinyl chloride conduit, PVC, polyvinyl, and plastic are all used as different material names in the model. However, they actually refer to the same material for the purposes of assigning global warming multipliers. So we use a combination of if and search functions to identify all of these items as PVC.

Sometimes materials are undefined in the Revit models. In this case, there are two options, either return to the designer for clarification or decide on an assumed material when materials are undefined. We do prefer to have a default assumption built in for any undefined material in our dashboard so that a preliminary MEP LCA can be performed.

Our assumptions are made based on materials we typically see for systems in data centers. An example I can give you is for undefined pipe, we assume that it's steel. For undefined conduit, we assume that it's PVC. It's very essential to make these assumptions transparent and flag any of the undefined materials. So in our dashboard, we actually have a page that shows exactly what percentage of materials were undefined and where we had to make a material assumption.

The nice thing is that at any point the materials can be defined in the Revit model and then the data reloaded into Power BI. At this point, the calculation is automatically rerun based on the new information, and a new global warming number is calculated that's more accurate. This is a screenshot from a project where no duct material was defined and about half of the conduit material was undefined. In this case, it would be beneficial to go back to the designers and have them define the duct material because 100% undefined is a dangerous assumption to make.

The next step that I'll walk you through is how we do the LCA calculations. So an LCA calculation in general is composed of the following steps. First, you identify the cross-sectional area of each item, which varies based on shape. The calculation for pipe is going to be different than the calculation for duct is going to be different than the calculation for cable tray because all of those items have different cross-sections.

Once you have found the cross-section, you multiply that by length. And then you multiply that by material density. And then you multiply that by the LCA multiplier. And you're able to get a global warming impact number for any type of MEP system. We keep the material density and the LCA multipliers in a separate Excel table that's referenced into the Power BI dashboard so that the LCA experts can update those numbers more easily because they generally are more familiar with Excel.

A couple more things about the LCA multipliers. As I mentioned, they're maintained in a separate spreadsheet, allowing the LCA subject matter expert to easily update and modify it without needing to use Power BI. These multipliers are broken down into life cycle stages, A through D, so that they can be included or excluded based on the project requirements.

I don't know if you noticed, but when I showed the examples of the three different data center LCAs that we've done, the different clients actually wanted different life cycle stages reflected. And so by splitting out the global warming impact by life cycle stage, you're able to essentially add or remove certain things.

For this particular dashboard, we used the Ecoinvent database sourced from SimaPro. If you are using a different database, you could essentially update the underlying Excel LCA multipliers with different numbers and rerun the analysis.

This is an example of what that table looks like. So I'm going to show an example of what applies to different types of systems. So, for example, these two line items would be used as the multipliers if you were doing an A1 through A3 analysis of steel duct. And then these two multipliers would be used if you were doing an A1 through A3 analysis of steel pipe.

If you wanted to do an analysis that included stage C, you would also add these multipliers into the calculation as well. So splitting things out by life cycle stage allows for more flexibility of what is and is not included in the LCA, which we found different clients are interested in, including different life cycle stages. So this is critical.

Once all the calculations are complete, we aggregate the data using a union function, focusing only on high-level columns that are common across all the MEP systems. These columns typically include material, length, the global warming impact, and the system type, which is a little bit different depending on whether you're looking at pipe, conduit, duct, or cable tray.

After completing the analysis, we use the SIFT visualization tool to colorize the model based on global warming impact. The model applies a red, a green, to yellow, to orange, to red gradient, where green represents low impact and red indicates higher impact.

In this example, you can see that the chilled water return piping system, which contributes to 37% of the overall global warming impact, appears in a darker orange versus the sanitary system, which only accounts for 2% is highlighted in green. The SIFT visual allows you to pick any kind of parameter to colorize your model based on. And this can be dynamic. The interesting thing is if the underlying data set was updated and more MEP systems were added to this that were more significant, the color gradient would start to change.

So now I'm going to give a demo of the dashboard as it appears in the Alluvial interface. So in the first tab, I'm presenting the overall calculated global warming impacts. The total value is predominantly displayed on the right in a large, easy to read format. Next, I've included a series of graphs that offer a high-level breakdown across MEP system categories, specifically piped, duct, conduit, and cable tray.

To the right, we dive a little deeper by breaking down these broader categories into more specific systems. Because this is a dynamic Power BI report, I can interact with the visual by clicking on parts of them. And the values on the screen will automatically update.

Moving to the second tab, here's where we provide some additional informational insights. We've documented the assumptions made for any materials that were undefined in the Revit model. Fortunately, since this particular dashboard is based on a sample file, there aren't any undefined materials. However, as I showed earlier, if there are any missing material definitions, they would show up colorized and with percentages.

And the third tab is where we list the LCA multipliers used for the analysis, along with their sources. As I was explaining, these values can be adopted depending on the region, the specific life cycle stages that apply to the project, or alternate LCA data sources.

Next, we dive into the LCA numbers for each MEP system category. This section provides detailed insight into why certain systems have a higher impact. Factors such as the material used and the size of the system play a crucial role.

Materiality, in particular, is a key focus for many of our clients and designers, which is why we've added a slicer to allow for easy manipulation of this variable. In the next tab, we have the SIFT visualization. It takes a second to load, just like the ACC view.

Here the model is colorized based on each system's contribution to the total global warming impact. For instance, the chilled water return system contributes a significant 37% to the global warming impact, which is why it's highlighted in a darker orange. In contrast, the sanitary system contributes just 2%, so it appears in green. All of the systems can be displayed and colorized simultaneously. This is very helpful for clients.

Next, in the global warming insights sections, we attempt to highlight some of the key factors that might be driving a system's impact when it comes to global warming. When comparing the chilled water return system to the sanitary system, for example, we can see that the chilled water system has a longer length, a larger average diameter, and is constructed from steel, factors that increase its impact.

On the other hand, the sanitary system is shorter, has a smaller diameter, and is made of PVC. Similar tabs exist for every MEP category, ducts, cable tray, and conduit, each offering detailed insights into their respective impacts.

So at Jacobs, the next steps for us is creating an equally efficient workflow for MEP equipment, which comprises an even higher percentage of global warming impact for data centers. Looking forward, we're researching the possibility of using Alluvial SIFT and artificial intelligence to automatically match environmental product declarations with what appears in the Revit models. This is currently a very time-consuming task, but it's one that actually aligns squarely with the strengths of artificial intelligence.

I'd like to end by returning to this slide, which highlights the significant contribution of MEP systems to overall global warming impact. On average, MEP systems account for 15% of total impact, and MEP equipment accounts for 43%. It is crucial that we continue to investigate MEP impacts, not just for data but across other building types as well.

Ignoring MEP systems means missing a huge piece of the sustainability puzzle. The future of life cycle assessment must include a strong focus on MEP systems to drive meaningful change. I'd like to encourage you to push for incorporating MEP LCAs into projects armed with the knowledge from this presentation.

Also, I want you to know that calculating MEP impact is possible to do in an effective, consistent way. And I think that we're on the precipice of having tools that accomplish this at scale. Thank you.