Ian : 0:07

Hello and welcome to the Energy Beat podcast. I'm your host, Ian Perterer. There's no shortage of discourse about how AI is transforming the world at large, from AI-generated images and memes to how AI is impacting higher education, from the environmental effects of large language models to what AI means for the future of work. But what role does AI play in the energy and utility space, especially when it comes to facilities and utility energy management? That's no easy question, given the sheer diversity of facilities that utilities and facilities managers serve. We're talking cold storage, supply chain facilities, hospitals, schools, warehouses, distribution enterprises and more, each with their own needs and appetites. Today, I'm joined by Kevin Scarborough, director of Energy Services USA at Siemens and a member of the Energy Management Association, and Bob Knoedler, vice President and Executive Consultant at Hanson Professional Services and a member of the AABC Commissioning Group. Kevin and Bob are here to discuss how AI is impacting their roles, opportunities that AI opens for the industry, and more so with that. Welcome to the show, kevin and Bob.

Kevin: 1:19

Thanks for having us, Ian.

Ian : 1:22

So I really appreciate you all taking time to talk with us today. Can you quickly introduce each of yourselves and tell us a little bit about your background and your current roles, and maybe even a little bit about what led you to become an interested party in the evolution of AI? And I'll start with you, Bob. How about that?

Bob: 1:44

Well again. Thank you, ian. As you noted, I'm a vice president and executive consultant with Hanson Special Services. We're a full-service consulting firm that provides engineering planning and related services for our clients. With dual degrees in both electrical and mechanical and about 40 years of consulting experience, I've remained involved in various facility and infrastructure services for a variety of clients, which includes a number of utilities, so I've been actively involved and also have been past president of the Energy Management Association.

Kevin: 2:23

Awesome. Well, hey, I'll jump in. My name's Kevin Scarborough. I'm currently the Director of Energy Services for the US at Siemens Smart Infrastructure Buildings. So Smart Infrastructure Buildings is a division at Siemens that's devoted to creating, or helping our customers create, smart buildings that are competitive, resilient, sustainable. We touch all building types and are really leaders in driving the digital transformation in buildings, and personally I lead a team of business development leaders across the US who help our customers save money, improve their operations and drive sustainability. I've been in the industry about oh Lord, what is it? 18 years now, I'd say. I've got a bachelor's and master's in mechanical engineering from Georgia Tech go Jackets, and I'm really happy to be here. I was drawn into this industry really because of my passion around sustainability and just solving customers' problems. It's a really exciting industry to be in and, of course, the past couple of years with artificial intelligence, it's going to be a key driver in this business. So I'm really happy to be here.

Ian : 3:29

Well, that's fantastic. So, as we said up top, ai is a bit of a buzzword right now and can be discussed in a variety of contexts. For the purposes of our conversation today, and to help sort of lay a foundation for those that maybe haven't been paying attention to AI, what does AI mean to each of you in the context of utilities and energy management, and how are you and your peers currently incorporating AI into your roles?

Bob: 4:01

Hanson, Our services include design, commissioning, construction oversight for really just a wide variety of clients, both public and private, and these include government utilities and other private corporations. We often provide various energy services, including infrastructure master planning, developing energy roadmaps, performing energy audits and even doing renewable energy feasibility studies. Many of our corporate engineering disciplines and services are engaged in AI, all the way from design and planning all the way through construction. As you know, ai is actively engaged all throughout the facilities, and Kevin will talk a little bit more from the building management standpoint.

Kevin: 4:45

Sure, thanks, bob. So you know, at Siemens, we recognize that there really is a global energy transition happening. As we all know, companies are shifting to renewable energy, prioritizing energy efficiency, but also electrification of their assets, and we know that's straining the existing energy and building infrastructure. We know that's straining the existing energy and building infrastructure. It's requiring rapid transformation to maintain reliability and sustainability of these complex systems, and so I really view this and Siemens views this as a major opportunity for infrastructure to become more intelligent with AI, create that sort of future-proof technology that we aspire to. You know, we know the building sector represents about 30% of global energy demand, and global floor area is expected to double by 2060. So that's a lot of square footage that's about to be built. That's going to be putting some demand on our resources, and so electricity demand is expected to triple by 2050. So, again, huge opportunity for us to address that with optimization and AI.

Kevin: 5:47

And so, at Siemens, we're currently leveraging and researching AI within what we call our Building X ecosystem to help improve delivery of reporting and outcomes to customers, along with energy efficiency. A quick example of this would be using generative AI to help identify the history of work orders for an asset to help prioritize maintenance and be able to respond more quickly. There's other applications beyond that for generative AI. We also use machine learning to optimize air handling units, as well as computer room air handlers and air conditioners, room air handlers and air conditioners. Within the past few weeks, you may have seen a press release where we recently entered into an agreement with a company to help us apply AI within central chilled water systems, and so it's a really exciting time to be part of this industry within AI, and there's so much to learn and just huge applications for our industry and, frankly, for our roles as individuals in this industry.

Ian : 6:44

Well, that's great and it's definitely at the forefront of everyone's attention right now, and it's so interesting to look back. Even three years ago it was still a fringe conversation. You go back five, six years, I mean. If you were really in the trenches, you were having conversations, maybe about blockchain and transactive energy. It's kind of interesting to see how it's moved and evolved since then, just in the short five, six, seven years. But for people who might be new to this, kevin, could you quickly explain for our listeners the difference between machine learning and AI, because I hear them both mentioned in the same conversations oftentimes, but I don't know that everyone necessarily has a key understanding of what the difference between the two is.

Kevin: 7:30

Yeah, that's a great question, Ian. So AI artificial intelligence right, you can think of that as sort of an umbrella term for all of the different things that we're going to talk about today. And so machine learning ML sometimes you'll see it referred to as that is really a subset of artificial intelligence or AI, where we're basically using data, we're feeding data into an engine that is then learning from that data and making decisions. That's the key thing. There is the making decisions. We're asking the computer to think.

Kevin: 8:03

There are other parts of AI, and one example of that that I've already mentioned is generative AI. You can think of that as like chat, gpt, for example, making a decision. What it's doing is creating something. It's creating an image or text or something based on data that is being fed to it. So there are a couple of machine learning aspects to generative AI, but those are the two different areas in which kind of fall under that AI umbrella. There are other things like deep learning as well, but probably won't go into that too much on this recording, but hopefully that answered your question.

Ian : 8:50

Absolutely. Thank you for that sort of grounding, educational moment so that as we go further into the conversation, everyone's playing from the same sheet of music. So you're both AI optimists. But as we've talked about all the headlines that you do read about AI and data centers and the like, what are the challenges that you see being at the forefront of integrating AI into the energy and utility industries?

Kevin: 9:17

Yeah, happy to jump in here, ian. I'd like to talk about three challenges. Right, I always like to do things in threes. It seems the biggest challenges I see are around data security, data governance and over-reliance. You know, speaking about security, it really feels like the Wild West right now in the industry, and, as we know, the West did remain wild until there was some oversight by law enforcement. Being a little silly there, that's an elementary way of looking at the current industry, but it really is largely unregulated, and so this is both an area of opportunity but also risk. Opportunity could arise by industry associations like ASHRAE coming together and create best practices for using AI in the industry. There's also risk here, including bad actors that can use AI in a malignant way, such as hacking.

Kevin: 10:08

The industry's got to evolve quickly to address these challenges. Data and decisions must be auditable, and there must be role-based access control to the algorithms to prevent bad actors from impacting the brains of the AI engine. I mentioned data governance, and that's another big issue. For the industry to be efficient and effective in driving results, AI must have a uniform output, especially with generative AI. As I just previously discussed, this would include standard reporting mechanisms, standard communications if the AI is communicating with a building automation system, for example, and other things, and you know for those of you who are listening, who work at large companies you know we can end up in what they call silos. This can happen with data as well. There must be standard categorizations for data to ensure the ease of communication.

Kevin: 10:57

And lastly and I'm going to touch on this a little bit later in this recording I think there's a big risk around over-reliance. If we over-rely on AI, we lose the spark of creativity. In my opinion, everything will be based on what's fed to the AI engine. Especially with generative AI, this can lead to bias and complacency. Quite frankly, the way to lead with AI is to use it to check your existing work or to optimize mundane tasks. Companies that use AI as a creativity tool to create new ideas without the input of a brilliant human mind are really going to do that at their own peril, in my opinion.

Ian : 11:35

What about you, Bob? Do you agree? Do you have a separate list?

Bob: 11:38

I do agree and you know, kevin's already brought up the fact that there's this push governmental push, you know towards electrification and also mentioned about the tremendous growth that we're seeing in demand and consumption associated specifically from this push towards electrification, whether it's from electric vehicles or mostly from data centers, the tremendous growth in data particularly related to AI, and it's been stated that there's really three R's of artificial intelligence which is relevant, reliable and responsible.

Bob: 12:14

Artificial intelligence, which is relevant, reliable and responsible, and these are really fundamental to a utility's successful implementation of AI. One of the things that we see is electric utility companies are experiencing really an evolving business model. Obviously, they're shifting from investing in larger centralized generating plants and grids to managing an increasing number of community-based and building-based generating facilities, many involving variable generation depending on renewable resources like solar or wind. Utilities are being challenged to work with software companies or to become a software company, at least at some percentage, in order to handle this increased amount of data that's being received and to accurately forecast demand and consumption for their customers. And I agree completely with Kevin, this is really going to require, particularly at the utility level, some kind of a global acceptance, governance of the software so that we can really ensure the interoperability and the transparency as these utility companies have to work together, going forward.

Ian : 13:23

That's so critical. I absolutely agree, and you touched on this idea of the utilities becoming basically energy as a service companies, and it could very well potentially become data as a service companies, as even a part of that. But when it comes to data and, kevin, maybe you can help us a little bit here there's such a huge amount of data being generated and collected today, but sometimes that data that we're collecting is not always useful, or sometimes organizations don't have the capacity to analyze and make use of insights from the data they're collecting. So could you talk a little bit about what data is most valuable for current AI initiatives and also what data might be missing? What do you wish you had access to? That would better enable you to leverage AI?

Kevin: 14:15

Yeah, sure. So, ian, one of the things I think it kind of goes without saying it's a little bit simple of an answer here is, to me, and for any sort of AI application, the most important data is accurate, applicable historical data. Right, that can include images, which is kind of a new thing that might be a little interesting. Right, you take images of a central plant, for example, or something like that that you could feed into an AI engine to help you scour the internet, for example, for design information for a device or a motor or something like that. But images reports, other things, something that's applicable to the use for which you're employing AI. So, in order to predict the future, we need to know how things reacted or operated in the past when there was some stimulus to that system. Right, an AI engine will only be as smart as the data that's fed into it. So in this case, if we're talking about generative AI, having work order reports and a full history of maintenance on a specific system could allow a specialist, somebody that's an expert in optimizing a system. It might give them the ability to move more quickly to answer a problem that a customer may have. If we're talking about machine learning, optimization. Having a building automation system with the right sensors installed, with the correct formatting of trend data and measurements that's going to be absolutely critical. We've got to have accuracy, as Bob just talked about, the really reliable amount of data, and so what data would I love to have and this is something that's getting better Our third-party APIs, apis being an application programming interface.

Kevin: 15:56

Right, that would be something you know. For example, like we often need weather data to make decisions for a central plan or for an air handler using machine learning, we might want to do that also for reporting purposes, to calculate what potential savings would be if we did something With AI, we could potentially use that data near real-time to make decisions in those systems. This is really one example, and the industry is moving in the direction of having APIs available. Thank goodness, even smaller third parties are scouring the Internet for data to help create APIs that we could then integrate into AI.

Ian : 16:39

And when you're thinking about the historical, is there kind of a minimum threshold of historical data? You would feel like you would need to feel comfortable with what you were asking AI to do. If a utility, for example, comes to you and says I've got the last three years, is that enough or do you really need more?

Kevin: 16:53

It depends on what actually happened in the past three years. Is that enough or do you really need more? It depends on what actually happened in the past three years. My background is an energy engineer, and so what we would typically do is we'd want two to three years worth of utility data. But if you told me three years ago, with the COVID pandemic, I want three years of utility data, I'll tell you 2020 and 2021 were not really relevant to forecasting what the future is going to look like, right, because buildings were unoccupied at that time.

Kevin: 17:21

So, ian, that's a great question and sort of the unsatisfactory answer is it really depends on the system and the building and what you're trying to do for it. So I'll give you an alternate example here. If you've got an air handler in a conference room, for example, and I need a data for that, and it really doesn't change that often maybe there's a meeting once a day in that conference room I could probably get by with a week or two's worth of data. If the temperature outside was representative of a greater quantity of hours at a certain temperature, we could get away with that. It really depends on the application, but it's definitely important to think about the past and how global events might have shaped the actual quality of the data that we receive.

Ian : 18:08

Yeah, and when I think about that, I also think about the evolving climate, and then you start to look at historical data. How easy is it to use the past data to modulate or figure out what's going to happen in the next three to four years? I think that's probably something a lot of resilience people are struggling with right now as we talk about the evolving landscape around energy and data. Much of that is due to the increased adoption of distributed energy resources, or DERs, and you know, along with DERs, there's the other side of the knife that there are many more assets to manage. So what kind of challenges does that pose to utilities, and how might utilities use AI and data to better manage those distributed assets?

Bob: 18:53

Well, that's also a very good question, ian, and, as we talked about a few minutes ago, this is part of that evolving business model that utilities are facing right now, particularly electrical utilities, going to come from managing this increasing number of assets or, as you said, distributed energy resources, which could include microgrids, solar PV and wind farms, private storage panels, energy storage units, often even grid-scale batteries, which we're starting to see come into play. Some of the statistics that we have out there now Europe is really projected to have about 36 million assets later this year, with a projected increase to 89 million assets by the year 2030. The median power plant size in Europe was about 800 megawatts in the year 2012. And then that decreased to about 562 megawatts in 2020, and is now projected to decrease to just 32 megawatts by 2050. And although the US is maybe slightly behind Europe, the US is really reflecting a similar increase in the number of assets, with a decrease in the asset size size. So, along with the other data provided by various components of the grid and smart meters, including this historical demand and consumption data, there's fuel costs, maintenance scheduling, real-time weather data.

Bob: 20:23

This really represents an enormous challenge in forecasting and in control, which really can only be managed by some advanced automation like AI. Utilities are currently even developing and using digital twins. They're virtual and often real-time to represent their physical assets so that they can run various scenarios. Really comprehensive utility energy management systems are going to need to track thousands of energy sources and uses. These AI systems will have established the basis for nimble capacity management, dynamic pricing models that can adjust energy prices and then steer the production, as well as usage patterns based on the market conditions, consumer demand, weather patterns and any other inputs that are out there.

Bob: 21:16

These systems are going to enable far more accurate supply and demand forecasting for the various utilities and much more precise energy storage and load balancing than is possible through human operators today. These AI systems will also be critical in meeting what are really ambitious carbon reduction targets. You know a lot of the high-tech companies like Google that had forecast when they would be able to become carbon-free have backed off of that now because of the tremendous growth and demand of a lot of their data centers. Tremendous growth and demand of a lot of their data centers. So there's going to be a real push now to see what can we do to try to reduce the greenhouse gas intensity and impact that we see in a lot of these fossil fuels.

Ian : 22:08

It's kind of interesting with the increase in DER and AI, adoption has the ability to make us greener through more efficient energy use. But also you referenced the energy intensity that's behind that, which also feeds back into the load management equation, sort of balancing act that you have going on. You know, how is your organization approaching load and demand management today, maybe compared to what it was doing five years ago?

Bob: 22:38

Well, one area of our services Hampton Services involves, as I noted, working with clients to help evaluate their current energy needs you know, really their baseline demand and consumption and then to try to assist them by developing, like I mentioned, energy roadmaps or master plans to consider their future needs.

Bob: 22:57

We examine their capital plans, any projected changes in their space usage or occupancy. We also work with the utilities and we look at alternate utility rates and really any other fuel sources. We do a lot of feasibility studies for renewable energy generation. We have a lot of clients that perhaps have the available space or land and they're very anxious, both for reliability and resiliency, to at least have some measure of generation that they can control on their own. We recently assisted a major client that was adding about 2.5 million square feet of space over a five-year period in helping them develop an energy roadmap and then integrating that roadmap with their sustainability master plan and then integrating that roadmap with their sustainability master plan. So this is an area for us in the consulting field that we believe we can help and we have worked with, again, a number of both public and private clients you know, hospitals, medical groups, schools, campuses, universities in moving forward in this area.

Ian : 24:11

Campuses- universities in moving forward in this area. So, kevin, you know, with all these emerging tools and things that we're talking about, what are the opportunities for AI in this area of load management and demand flexibility? What are the opportunities for AI to actually help, be a net positive, to solve some of these problems?

Kevin: 24:33

for AI to actually help, be a net positive to solve some of these problems. Yeah, I can speak a little bit about what we're doing at Siemens and our approach with AI in this space. So we have software, and this is a big push for us is creating a digital twin, or doing a digital twin approach, as you mentioned, where we aid with grid planning and operations, which should increase the lifetime of electrical assets, and so we do this by detecting faults in these systems before they become major. We also have a service platform called ElectrificationX. You're going to hear the word X a lot from Siemens. It's part of our accelerator initiative or platform that we use. That's part of our accelerator initiative or platform that we use.

Kevin: 25:15

Electrificationx is built on cloud services and is designed to manage, optimize and automate electrification infrastructure.

Kevin: 25:20

It's integrated IoT suite that provides holistic view of substations and assets. It helps renewable energy operators, transmission system operators and others increase uptime and improve reliability, as well as cybersecurity, and so, by adding a layer to this ElectrificationX, we have something called ElectrificationX Asset Management, which uses insights from machine learning models to monitor transient behavior for disturbances before they become outages. And lastly, very similar but a separate thing that we do is something called GridScaleX, which provides easy to deploy modular software where we can integrate in any IT or OT landscape. It provides end-to-end grid management and leverages AI to analyze vast amounts of data generated by decentralized energy systems. And adding another layer to this is what we call GridScaleX DER insights, which uses AI to uncover behind-the-meter DERs, their behavior and their impact on grid equipment. It helps with forecasting, analyzing and extracting useful insights. So a lot of things out there. That's just a little taste of what Siemens is doing, but the industry is moving in this direction of leveraging machine learning and AI in order to optimize and right-size the load on the grid and the systems.

Ian : 26:44

Yeah, and so we've talked a lot about the tactics and platform software solutions and products, and all of these things, I think, rest on a macro-level foundation of competent skill in the workforce to be able to understand and make informed decisions about how to best leverage and integrate tools like this.

Ian : 27:06

At AESP, we have a conference that we have every year in the summer FlexConnect that AI is one of the central parts of this conference, and so one of the things we did is we had an AI 101 workshop last year and that was very educational for us, not just on the content level, but just getting a real sense of where the industry is in terms of its readiness to take on some of these potential solutions. So question you know that I would posit to both of you is an AI is becoming more integrated as part of our work. How are your organizations approaching upskilling and training, and where do you see the most critical need for this education in order for us to be able to leverage AI effectively for this education in order for us to be able to leverage AI effectively.

Bob: 27:54

Well, upskilling will be very important for both utility and facility staffs in various buildings, and it's got to be addressed proactively.

Bob: 28:07

You know, the first thing I see is the need for what I call data literacy, and that's really being able to understand and interpret the data that's generated by the AI systems. And then there's really going to be a need, a complementary need, to learn how to manage and oversee these AI systems, being able to collaborate and work with AI. There's also a need for cybersecurity awareness Kevin mentioned that as well as a reliance on digital systems starts to increase within the utility arena. An increase in the number of assets, which we've been talking about, really corresponds to an increase in the number of entry points to the grid for malicious actors, so that's going to be very important to be aware of in terms of security going forward. But I think that Kevin and I agree that there's no reason to fear AI, but really to embrace it and to learn how to collaborate with it for greater efficiency and optimization. No matter what area you're working in, managers need to commit to a program of continuous learning regarding both new and best practices with respect to AI.

Kevin: 29:22

Yeah, sure. So, to answer your question about how we're approaching upskilling and training, we have about 1,500 AI experts around the globe, with 200 what we call AI champions within Siemens Smart Infrastructure. We filed 3,700 AI patent applications and have dedicated AI labs in our buildings facilities industry. We're actively training our boots on the ground staff on every AI offering that we have, and they're coming rapidly, and so we need to make sure that our staff has the best skills to ensure we're hitting the mark with our customers. Critical need for upskilling I would say the most critical need we have is being able to leverage AI appropriately. We want to make sure that people are leveraging AI appropriately, and specifically, what I'm going to speak about is around generative AI.

Kevin: 30:17

You know there's a lot of training out there, and I believe that there is a human tendency that we'd all agree on that people like to take the path of least resistance, and so this can lead to, like I mentioned before, over-reliance, and over-reliance will inevitably lead to inefficiency and productivity of our people, but also the results that we're seeing in the industry.

Kevin: 30:38

We know the saying garbage in, garbage out. Ai will only be as good as the data that is fed into it, and if we rely too heavily on it and not feed it data, we lose that creativity. We don't want to do that. We run the risk of having the wrong answer as well.

Kevin: 30:54

You need to, as Bob said, train people on discussing things like mirages, which may come from overusing AI or not using it appropriately, and what that's defined as a mirage is where it's providing an answer that looks absolutely right, but it is completely wrong, and it's because you've trained the AI engine to apply to a certain application, but now you're asking it to think outside of that application. It will say, yeah, this is the right answer based on the data that you fed me, and so we need to make sure that people are trained to identify when those things happened, what to do in that case, right, and so just the key thing to keep in mind is what worked in one scenario may not work in another, and to the point that this kind of ties into a lot of anxiety people have about AI, humans will always have a place in this industry and humans need to be leading the charge with AI and using it as a tool, not an answer.

Ian : 31:54

I think those are two very important points A, being able to understand the data that you're receiving from AI, and also just knowing when it's appropriate, because, especially with generative AI, it is not appropriate for every scenario. There are still some things in this world. I tried to use it for some accounting tasks recently. There are still some things in this world generative AI cannot do. I've got one last question for you both before we wrap up. Looking ahead, what are some current trends that industry leaders should be paying attention to in order to stay ahead of the curve when it comes to AI?

Bob: 32:31

Well, I would say that to recognize that most industries and companies, maybe even your competitors they've recognized the value and are looking for new ways to utilize AI, specifically in collecting and analyzing large amounts of data as well as customer interactions. Ai is really touching a lot of industries across the board. Kind of as an aside here, I recently read an article about a chief information officer of a company that only made lipstick a chief information officer of a company that only made lipstick, and until recently he had stated that lipstick would never be digital. But as it turns out, he had an aha moment and admitted he was wrong. Everything about the production of lipstick now in his company from consumer data to how people feel about the product, to how colors and aromas are developed and even finally to the procurement of materials and distribution, is all now digitized. So there has to be a recognition that AI is coming and it's going to have an impact on everyone.

Kevin: 33:41

Yeah, ian. So I would say keep an eye on the regulations that are coming, not only, of course, in the United States, but also globally as well. Stay on top of those, as they may impact how AI develops in the future. I would say and we see this weekly is be on the lookout for new technologies as they come idea. Some spark out there that you really see an application for your business or for your customers right. Study them and see how they could apply to your business. And I would leave it with this have an open mind. Embrace the changes that are coming. Run toward it with open arms. The changes are coming. Be cautious, don't be afraid. Embrace the future. Ai is not going anywhere. Be a leader in your organization. Learn all that you can about it and try to grow your business using it where it's applicable. It's really exciting times right now in the industry.

Ian : 34:41

Well, thank you. It's so wonderful to be able to end on a great like expiring call to action. So definitely lean into the change the AI landscape for utilities. You can check out AESP's State of AI in the Utility Industry Report, which is research conducted in partnership with Bidgely and that will be linked in our show notes. Thank you all for joining us and we'll see you next time on the Energy Beat podcast.