VO: You are listening to Cool Air Hot Takes.
Charlie Jelen: Listener, welcome to Cool Air Hot Takes a podcast about anything and everything from the energy sector and the built environment. We are your hosts. I am Charlie, and this is Dan.
Dan Gentry: Hey there. If you've made it to episode four, good for you. You probably know who we are by now, but if you’re new, welcome and we're glad to have you here.
We've been around the HVAC biz for a long time, anywhere between designing district plants to troubleshooting
Charlie Jelen: stuck contractors. We’re here and we’ve also been friends since seventh grade, and we both got into the HVAC and construction business. But unbelievably, it turns out that our long suffering wives do not actually share the same love of this business
And heavy eye rolls ensued every time we got together. They had zero interest in, in what we had to talk about, so the solution was obvious. Start a podcast. So here we are. I still think
Dan Gentry: that is, that is still unbelievable that they rolled our eyes when we talk about work. I mean, come on. Cool. Air hot takes.
Yeah. That's why we’re here.
Charlie Jelen: Alright, Dano, I feel like we need to circle back on this. You went to the A HRI show. I did not go this year. And for the listener out there, the A HR show is. The biggest show in HVAC every year. This year it was down in Florida. Dan, you went, I did. Made your takeaways. What did you like, what did you not like?
Give us a recap.
Dan Gentry: Yeah, it was, it was very cool. Lots of neat stuff. A lot of heating as I think one would assume these days. A lot of air source heat pumps. A lot of modular stuff, which I thought was kind of interesting. Seeing a lot of like
Charlie Jelen: modular, like thermo fit, like multi-stack, like modular? Modular,
Dan Gentry: yes.
Yep. Assemble ’em together sort of a thing. Lots of say existing companies developing modular products and new companies coming out with modular products. Some, uh, I think we talked about it on the first episode, but some CO2 compressors and there were a couple units with CO2 AI has a presence. Um, I went and talked to our friends at Brainbox.
It’s a good lead in
Charlie Jelen: for our guest today. It’s happening. Yeah. It is, but before we get to the guest, we gotta talk about a hot take. Daniel, did you come with a hot take today? I did.
Dan Gentry: Thanks.
Charlie Jelen: What?
Dan Gentry: Tanks?
Charlie Jelen: Tanks? What's the hot tank?
Dan Gentry: All right. Tanks are great. So you can do a lot. Tanks are great with buffer tanks, so think about air source heat pump applications.
Um, the placement of a buffer tank could be important. Air source heat pumps go into defrost. When they go into defrost, they make cold water. If we place our buffer tank on the supply side after the heat pump, when that heat pump goes into defrost, we have that nice. Big volume of warm water for when our heat pump makes cold water.
Yeah. So we can mitigate the discomfort effects to the space. So that’s again, loop volume. So stability in systems. So think like chillers, heat pumps that use scroll compressors that are banging on and off. Have a nice big loop volume. By adding a tank. Mm-hmm. You can stabilize your system. Mm-hmm. You can do thermal storage, you could do chilled water storage and like say millions of gallons of built up tanks.
You could do ice storage and say modular bricks or big built up systems. You could use thermal tanks to say store energy when you know, maybe your chiller, your heat recovery chiller can't turn down well. You can just run that bad boy. Charge those tanks, shut it off, avoid say potential surge or turn down issues with a particular piece of equipment.
So they're really versatile and you can do a lot with tanks and it's a very simple piece of equipment. Doesn, did you just
Charlie Jelen: get done reading the Ashray
Dan Gentry: manual
Charlie Jelen: on,
Dan Gentry: on Jill
Charlie Jelen: Water design?
Dan Gentry: I've been into a lot of tanks. I get asked about loop volume and tank sizing and defrost with air to water, heat pumps.
Almost every day. So I think I’m just like, I’m like really? In the tank world. Big tank guy. Okay. So yeah, I'm hot on tanks. Okay, thanks. Yeah. What about you? You gonna give us a hot take?
Charlie Jelen: Alright, so, so mine is adjacent to our space. Not exactly. My hot take is we are replacing hot water heaters too fast, and it comes down to probably the easiest maintenance procedure possible for your home appliances.
Listen up, everybody. Well, so I'm gonna ask you the reason that this is my hot take and the reason that I am concerned about this is because a couple of months ago we were at a fine establishment called diesels, and I asked you how often do you drain your hot water heater? And your answer was. Um, why would I drain my hot water heater?
Do you drain
Dan Gentry: your hot water heater?
Charlie Jelen: Yes. Religiously every year. And so since we had that conversation, have you drained your hot water heater?
Dan Gentry: Uh, I have not drained my hot water heater. All right, so you're not I've thought about it. I, so when you brought that conversation up was the first time I was even aware that draining hot water heaters was a thing.
Mm-hmm. So I've, I've thought about it and I've. I see the drain on it. I just haven't done it. I'm contributing to the problem. I'm not part of the solution. Yeah.
Charlie Jelen: Well, so real life story here. My aunt, my wife's aunt, my aunt in-law, she replaced a hot water heater after seven years. Oh, that's not too long.
And they opened it up and it was just like sludge that came out of it. So all you need to do. Once a year, put a piece of paper, tape it on the side of the hot water heater, write down the date, close off the entry valve. Open up like a, a faucet somewhere in your, on the second floor and, and just drain out your hot water heater.
You'll see all of this sludge, all of this crud come out and you'll get an extra eight years outta this thing. This thing should last for 12 to 15 years. Thank you. And people are replacing 'em every seven.
Dan Gentry: You wanna know how bad I am at this then? So I did actually recently have. A element replaced in my hot water heater and we just did it on the fly, so just pull it out.
And a bunch of sludge Yeah. And stuff did come out and I was like, huh, you know, probably should probably drain these bad boys once in a while. There you go. Maybe I'll do that this weekend. Probably not, but I'm, I'm
Charlie Jelen: probably not. Alright, here's what we got lined up. HVAA. Headlines, of course, we'll get you caught up on what we thought was interesting from the week.
Our interview today is fantastic. His name is Jean Simone Ven. Thank you very much Elena. In parentheses, she wrote pronounced the French Way 'cause he is French Canadian. He's gonna talk to us about. AI in the HVAC space. He is the co-founder, chief Technology Officer at Brainbox. It was such a good interview that we're actually gonna extend it out and, and we're gonna have a longer interview and we're gonna go straight to the stat
Dan Gentry: of the day.
Charlie Jelen: That's right. All right, here's your headlines,
VO: HBAC. Headlines your news today.
Charlie Jelen: Headline. Number one, Boston expands net zero emission requirements to new buildings and large additions. This one is from Utility Drive. We wanted to bring you this one because headlines like this and new policy like this that rolls out, especially from these large cities like Boston, New York, Seattle, la, San Francisco, they are either part of a larger trend or they're, they're starting a new trend, and so we wanna bring this to you.
Especially now with where the current administration is pulling back from various climate initiatives, the federal government's not gonna go after it states or cities can, and that's what we're starting to see is some of these pockets around the country that want to go after electrification. They're starting to to do things like this.
And so what this headline is is the Boston Zoning Commission approved an amendment that requires most, not all, but most large buildings, to have net zero carbon emissions. From the day they open. Wow. It's aggressive, which is tough. All right, so if you're not familiar with net zero, net zero emission means you need to offset all of the emissions that you create or create no emissions.
Dan Gentry: When does it start?
Charlie Jelen: It goes into effect for new projects that are filed after July 1st, 2025. So like now? Yeah. Yeah. It's like going into effect. Now, the easiest way to think of what a building needs to do is think of electrifying a building. So the only energy that the building uses is electricity.
Mm-hmm. And then you add enough onsite generation through solar panels or wind to offset all of the energy that you use. So all of the energy is from renewables, zero emissions. That's where the concept of net zero comes from. You net it out to zero.
Dan Gentry: This is interesting. Are they gonna be able to, can you buy carbon offsets or is this truly you have to
Charlie Jelen: do it?
Yeah, there's gonna have to be provisions in there for how they procure that. Because you know, some of those buildings downtown, you're not gonna be able to put enough onsite generation for it.
Dan Gentry: Yeah, that is wild. And I mean, I think like to your point at the beginning, we're gonna, we're gonna continue to see this, we're gonna see cities and states do this like on their own, however they.
You know, feel fit for their
Charlie Jelen: Absolutely. Their people. Yeah. New York City's got similar rules in place and for Boston. This is on top of Alberto, which is a building performance standard that is regulating the existing building portfolio. I. So this is, this is only for new buildings that we're talking about, but they already have existing buildings too.
So,
Dan Gentry: and Boston really is, you were just out there, right? I was, I was talking to the folks out there about this stuff, and a lot more than this. This is just one example, but they're doing a lot of this kind of climate initiative stuff in many different facets, and they really are, it seems like the kind of, if not, you know, one of the major leaders as far as cities go to, yeah.
Implement and drive this stuff.
Charlie Jelen: What were the solutions that were. Percolating to the top.
Dan Gentry: Well, I guess one thing that I thought that was, I'm gonna say that was just really interesting and cool, is they were looking at a district energy plant that was out for a district energy conference, and they're putting in a 10,000 ton heat pump that is sourcing water out of the Charles River all year long and then making like.
300 some degree hot water, which is pretty cool, I think.
Charlie Jelen: Yeah. Once you get above two 12 or a little bit above that,
Dan Gentry: it's a steam converter can turn it into steam. Yeah. I don't know how steam converters work, so if anybody email the show, tell us how they work, I would be very curious. But you know, it's, to me it's just that little box and hot water to it and steam comes out.
Um, but it's, it's very interesting stuff and just, I would say a lot of. These exact things just tightening up, you know, carbon emissions for the built environment. Yeah. It's going all over, you know, from, you know, to electric appliances, you know, everything. There you go. Yeah. Uh, up next we have special guests, Jean cmo, van co-founder, and CTO at Brainbox ai.
He will be joining to talk to us about leveraging AI in the HVAC space.
Charlie Jelen: Dano. Lisa, can you hear it? Hear, what if your building's not speaking to you? It's time to connect with Trane's Tracer. SE Plus Tool Tracer has the power to elevate performance for buildings of any age, size, or brand of equipment.
Dan Gentry: Ah, yes. Nothing like a building reaching its full potential. Where can our listeners find out more?
Contact
Charlie Jelen: your local Trane account manager today, or visit trane.com.
Alright, listener. I am very excited about this next guest since we launched the podcast. This is one that I had on my list, so I'm very excited to welcome Jean Simone, welcome to the show. Hey, Charlie. Thank you. It's a pleasure to be here. All right. Now before we get into the technical aspects of the work that you do, I wanna share a few nuggets of information.
Alright? So we happen to have a common love of two things. Ooh. Mm-hmm. Okay. The first Alpine sports, really? No, really? The second. A fine lager beer called 1664. How? How can it be like,
Jean-Simon: I think you're the first human being that I encounter, which have these two, two together. I love it.
Charlie Jelen: So Jean, where do you live?
Where are you skiing at? Give us a little bit of that.
Jean-Simon: Yeah, absolutely. I'm based in, uh, Montreal, uh, Canada. My home ski town is, uh, bla so, um, bla I live there. I work there in term of, uh, weekend, uh, job, if I could say like that. So I'm an official in a ski race and that's my kind of second life.
Charlie Jelen: All right.
That's awesome. That's fun. Alright, well the name of the show is Cool Air Hot Takes and when our guests come on, we ask them for their hot take. And this could be anything in the world that they live in. It could be in their personal life. So Jean, do you have a hot take?
Jean-Simon: Absolutely. And I might create some shock here.
Charlie Jelen: Okay.
Jean-Simon: So the HVAC business will become a data business within the next five years.
Charlie Jelen: Five years. Wow. Five years.
Jean-Simon: Five years. Mark my word.
Charlie Jelen: So does that mean that products, they get to the point where they're commoditized, like the controls, the AI architecture doesn't really matter what they're controlling, they're just gonna be so optimized.
It's all just commoditized.
Jean-Simon: It's all about data, right? So you basically, once you extract that data and you start to create value and layer and layer of data with that. Yeah. So whoever. Control the data will control not only the equipment behavior, but will also create that tons of value, which gonna be what people are buying.
Dan Gentry: I love it. That's amazing. I mean, we're talking about like a hundred plus year old industry and massive changes within years.
Jean-Simon: Yeah. Think of uh, cooling as a service. Yeah. Eating as a service. Mm-hmm. What are you buying? Are you buying a piece of equipment or are you buying a controller or you're buying just the fact that you wanna maintain the temperature in the room you're in.
So that's kind of what's happening very rapidly, right? Actually, when you look at the airline industry, it's already happening. When you look at the engine manufacturer, I'm not gonna give any name here, but the airline, like they don't even own the engine on the plane anymore. They're leasing it, but actually they're not leasing it.
They're paying by mile fly on the engine.
Charlie Jelen: Really. I
Jean-Simon: didn't know that. So on the CapEx, it's great for them, right? They don't have to, uh, buy these things. They don't have to maintain it. They don't have to have a mechanical crew to maintain these things, which are very complex. They just pay by mile that they're flying the plane and everybody's taken care of.
Charlie Jelen: Yeah, that's that. That's interesting. On, on the control side of our business, we've seen more of that as a service model on the actual output, on the cooling, on the heating side. The actual BTUs A little bit of it, but we haven't seen a big push into that yet. No, not yet. Not yet. But you're thinking next five years, huh?
Jean-Simon: But to offer it, you need to control the data. Otherwise, like, you're not gonna be able to offer that service, so, right. You really need to, to control the output and make sure that you're delivering it. At the same level all the time, and only then you could start to offer that kind of, uh, cooling as a service.
Right. I love
Charlie Jelen: it. Great Hot take. Yeah. Very good. Hot take. Okay. So brain back. So let's get here, let's start at the beginning. How do you get to, you know, a young lad in Canada to the spot where you're founding an AI company for building control? How, what, what, what was the genesis of this?
Jean-Simon: Oh boy. So, you know, I go back to the eighties, right?
So I always played with, uh, technology, a lot of frustration about you wanna do things, but the technology is either not ready, not existing, really to do what you would like to do, or if it does exist, it costs way too much. And then you had the pleasure and actually the frightening experience to sit in the autonomous cart about nine years ago.
And it's really like. It's working right? It's like if they're capable to do that, we should be able to do autonomous hvac. So that was the kind of the reason why. Okay, let's try to do that. Let's do an autonomous HVAC system. Mm-hmm. That will basically do the continuous commissioning for a few penny on a dollar if you compare to a human doing it.
Charlie Jelen: So then let's go from there. What is Brainbox ai?
Jean-Simon: We extract all the data we can from the existing controller in the building. We clean that data in term of putting a, a standardized nomenclature. We tag it a lot of tags, we enrich it, and then we take external data, uh, think of the weather. So we're taking, uh, that detailed weather that.
Airline type of pilot weather, because I wanna know the thickness of the cloud at any given time in that location. So I could basically derive the, the solar radiation. And then we basically train a neural network. So these neural network, what are they very good at? So we're talking about the deep learning.
Mm-hmm. So these neural network, where they're very good at is the, if you've trained them with enough. Data, they will give you an accurate prediction. So what is the prediction we wanna know? So we basically train these neural networks. So they give us a prediction of how the temperature and the humidity will fluctuate in each of the zone in that building.
And they give you that prediction accuracy at the 99%. So I actually know in the room you're sitting in, if we train a Neur network, how the temperature and humidity will fluctuate over the next six hours. Very precisely. So that's basically the first component of Brainbox, that knowing the future. So it's a bit like the movie Back in the Future, right?
We were the capability to see the future. It's a gray movie, by the way, gray movie. Yeah. So we go in the future, we see like, oh my God, in three hours it's gonna be very hot in the room you're in right now. And the system will react, right? The system will say, mm-hmm Oh my. Start a cooling. Start a cooling, and then in the temperature will go back to whatever is a set point you like or you set on the thermostat.
So knowing that we basically run different algorithm, which are typical operational research algorithm, but they were basically. Coded into, if you want, the ASH 3 36 guideline. Mm-hmm. So I know how to operate a pump, I know how to operate a damper. I know how to operate a stage one, stage two cooling, and basically we run all of the different scenarios.
So what are all of the control strategy that we could possibly. Run on these equipment and what is the outcome in terms of the future? And then we've tried to do what is the optimal control strategy to optimize all my objectives. So you're really into a multi objective optimization like tank of a Perato front optimization pattern where you basically say, I want to reduce the cost, I wanna reduce the emissions.
So what is the source of energy being? I want to reduce the cycling because too much cycling will damage the equipment or will shorten the lifespan of that equipment. I wanna optimize the comfort. So considering all these objective, what is the optimal control strategy that I should apply right now for that equipment or for the entire equation of the equipment in that building?
And we move five minutes in the future and we restart everything that I just described. Mm-hmm. So it's quite interesting, right? Because it's crunching a lot of prediction and then control strategy and then it decide what it should do. So it's kind of flipping from a reactive control. Schema, which is what we have in all of our billing on the planet into a preemptive control schema.
So of course, if you're in Texas, July super humid, 99, uh, ferry Faite. Yeah. There's only one strategy. Hmm. All out, run to the floor and hope for the best. Yeah. So no value, but most of the time during the year, you're not in full throttle strategy. And then that kind of preemptive, that's where you see the value.
That's where you bring money and you save energy, you save emission and you reduce cycling.
Charlie Jelen: And I'm assuming it's kind of twofold in terms of where you see the savings. It would be, to your point, it's those off shoulder where you're not running everything full tilt. But then also. The age of equipment or or age of systems, like the further you get from that day one commissioning that you were talking about, the more opportunity you likely have for savings.
Absolutely. Throw in COVID, the amount of outside air that you need, the amount of people that are in the building, and you're ripe for savings.
Jean-Simon: It's a change of pattern, right? Change of behavior. That's where you start to basically see the automatic adaption. You could still do it with human. You need to go get and revisit your control sequence.
Okay? There's a new reality in that building, so I'm gonna redo my control sequence. But the problem is like very few people are really doing that. Mm-hmm.
Charlie Jelen: Alright, so let's talk about actual applications. Where are we deploying Brainbox AI today? What types of systems are we seeing it on the day?
Jean-Simon: Well, we, uh, for a business, uh, decision, we said let's just focus on commercial building.
So we're pretty much covering all of the commercial building space. So it goes from a dollar tree, a small retail facility should be like a rooftop unit. Single zone rooftop. Yeah. Maybe two, three rooftop. We're doing some of the story with only one rooftop. So as soon as you have a system which is as some kind of control, and it could be only a thermostat, we could basically start to do something all the way to an airport, right?
So we're doing very large airport, which are very, very complex. So we cover the air side. The water side and of course the two that are working together. And that's usually what is quite interesting. Most of the water side, when they operating chiller plant mm-hmm. They're producing a quantity of chilled water.
Right? Yeah. And very often that quantity is basically circulating through the entire facility and is ready to be used. But what is the exact quantity of ton that you need per hour? Really? So don't, instead of producing a standby capacity, which is ready to be used. Produce exactly. Once you have the prediction, you've produced exactly what you need.
Still keep a bit margin, but don't approach it as a static kind of, I'm producing that quantity and, um, it's you ready to be used in case produce exactly what you need.
Charlie Jelen: When you're doing a large chilled water plant like that, do you need to know the thermostat set point or the sensor set point in every room?
Dan Gentry: That's what I was getting at the, the thermostat part. Absolutely. Would these systems allow for fully adjustable thermostats, like I'm imagining, you know, humans going and adjusting things. Does that mess the stuff up or is it.
Jean-Simon: That's, uh, that's one of my favorite topic after the, I call it the thermostat war, right?
So we see that all the time, right? So let's say you have an open space and somebody goes to the thermostat and they say like, well, I want 75. Um, half an hour later, somebody else go to the same thing. Say, no, no, no, I want 70. What is this thing? So the target keep changing, right? So the AI is starting to chase that rabbit, which is moving like this because the human is not agreeing on what is the desired temperature in that space.
There's not much we could do. The AI will just try to follow whatever they. Put the target, but, but the eye is not deciding the target, so the human is deciding the target and then you just try to serve that target. But generally speaking, it's pretty stable. But yeah, absolutely. The set point, the desire set point by human is a key component.
And then the behavior of that zone. This is what the prediction is being trained on in term of neural network, uh, giving you exactly how the temperature will fluctuate. Mm-hmm. And it's very different from one building to the next. Right. So each building is really a snowflake when we say that, like it's really true.
Even when you look at the retail, we're doing this, uh, chain of, uh, a mattress furniture store in Canada called, uh, sleep country. They're pretty much all the same in term of configuration, but it's still all unique. And it could be the type of window, the wind that. Door, which is not closing properly, could be the sun, which is the orientation of the store.
There's always some specific for each. So you really need to have a neural network learning each of the store independently.
Charlie Jelen: So talk to us about results, right? What are the results that you're seeing and and maybe a case study is a good one to do here. 'cause I'm sure it's totally dependent on where you start and where you're at.
But yeah, what's the result or what have you seen from deploying this, either at scale or at an individual site?
Jean-Simon: So I'll give you the case of Dollar Tree. I've been mentioning that name a few times already, but, uh, let's keep at it.
Charlie Jelen: We're looking for all the sponsors we can get. So yeah, keep throwing out brand names.
Just keep going.
Jean-Simon: So, yeah, I mean, dollar Tree, uh, we're now modulating I think around like, uh, 2000 store in the US across uh, pretty much all the states. And really when you look at the number, it's quite interesting now, like, 'cause we've been at it now for about a year and, and now we start to have the measure and verification company so.
You're looking like, uh, eight gigawatt hour save, uh, 5,000 tons of emission save, and a $1 million in energy cost save. Um, so this is what AI is doing, right?
Charlie Jelen: On a per store average. What is that roughly?
Jean-Simon: It vary in between 15 and 25% of the energy saving. Did any stores get worse? No. No? Okay. No. What's happening is you might have a case where we cannot save more than 7% and it is what it is.
So, but when you look at the average, it's quite interesting.
Dan Gentry: 15 to 25% just with some smirk. That's, that's a huge amount. Yeah.
Jean-Simon: There's a lot of low winging fruit. Right, because there's, you would put an engineer or a control technician on that store and say, try to get me some value. That person will find value.
Mm-hmm. And will save you money. The problem is that person is also costing you pretty high rate, right? Yeah. So then you put the two together and the equation is not that great, but the AI is running like maybe a 5 cents per hour, right? So no problem. Yeah, let's get the ai, do the work, and then you get all of the saving for you.
Except that 5 cents, right?
Charlie Jelen: So for like Dollar Tree, and I don't know if you can share this, but like what were the algorithms or what were the processes that were saving the most?
Jean-Simon: There's usually one of the key algo is, is what we call the Kronos. So Kronos is a, an algorithm which basically optimize the start time and the.
Close time of the schedule. So the transition from occupied to unoccupied and to unoccupied to occupied. And you'd be surprised how much we could glide at the end of the day. So yeah, the store, I don't know, close at 7:00 PM the regular system will shut down everything at 7:00 PM 'cause we go in unoccupied period.
Kronos will start to basically glide you probably at six. Mm-hmm. Because it knows that I got my terminal mass. I'm looking, my prediction, I'm good. I could start to slow down and close it and nobody will notice because I'm gonna still be on the set point by seven. So that's kind of the saving. But then we have all kind of other algorithm which will play with the damper and the cooling one and cooling two staging.
Yeah. Same thing on the eating side. And it's using that prediction, right? So we know in advance what you're gonna need in term of, uh, a reaction. So we try to basically canceling out in a preemptive, uh, strategy. So we have several algorithm, but. Each store might have a different mix of algorithm based on what's available and what's available in term of points.
Yeah. For us that's a key criteria. Like say, can we access these point, can we read them? Can we write back to the right point? We need to write back? 'cause that will basically limit what you could deploy in term of algo in a given building.
Charlie Jelen: So you're hitting on this a little bit with operators, but who should be the most excited about Brainbox and should anybody be nervous for their job?
I.
Jean-Simon: I'm gonna give you the example of Aria and then I'm gonna answer your question. Okay. So Aria is our virtual, uh, mechanical engineer and we released, uh, we made the announcements last year and it's now in beta. And really, aria is this kind of assistant, which is looking at how the equipment are performing using, uh, the generative AI technology as an agent.
And it basically. Do the root cause analysis of behavior, which are troubling. So instead of basically asking the human to spend one or two hours in front of the BMS and try to understand what is going on, the cooling stage is not starting. It should be starting. Why is it not starting? Is it the malfunction?
Is it like a refrigerant problem? Is it like fan? What's going on? So Aria will be doing that. Root cause troubleshooting, analysis, and basically come down with the insights, which are, I looked into it. I look into the trend data. I actually read the document of the manufacturer of the equipment and look at the different possibility, and it's pretty clear.
That in this situation we have a leak of the refrigerant. So yes, we should send a truck roll and be prepared to basically attack the refrigerant question when you get in on site instead of just trying to find out. So suddenly like that one to two hour of, uh, troubleshooting and finding the insight to understand what's going on is taking a few minutes.
So it's a productivity gain.
Dan Gentry: Yeah. That's awesome. So technicians can just, yeah. They know what they want to go look at. They can do it quicker and then they can get to more sites. Yeah. Within a day.
Jean-Simon: Exactly. 'cause when you look at a typical day, right? Let's say a building manager will say, it's too cold in the building.
Send somebody. So truck roll, we're sending the technician, the person arrive on site, start to figure out, goes on the roof, check the equipment, check the dogs, and eventually discover that, oh, now I don't understand what's a problem. Oh, I don't have that spare part with me. I'm gonna come back tomorrow.
VO: Yeah.
Jean-Simon: So you see where suddenly you're gaining an extremely well on productivity and the fact that there's a shortage of labor in that specific field of field technician. Mm-hmm. So most of the ticket, you know, they don't get really served the way they should because they're just not enough technician on the road to serve all of these tickets in a given day anywhere else in Canada or in the United States.
So suddenly with the existing. Population of technician. With that type of tool, you could increase your productivity and have enough now men labor to do the job at a much lower cost. That's awesome. So it's benefit people, it has become a tool. Right. So I like to give the example, like, you know, now we all have our intelligent phone and we using email.
When I started working, uh, there was no internet, no mobile phone and no email. We were sending letters by fax. So do I still have a job? Yeah. Do I still use a fax? No.
Dan Gentry: The tools just change.
Jean-Simon: So am I much more productive than I was back then? Absolutely. So it's just another tool that's gonna make our life more productive.
Charlie Jelen: Last part of that question. Should anybody
Jean-Simon: be worried about this? Actually, I like to say some. I'm gonna repeat a sentence that you probably heard, uh, numerous time over the last 12 months. AI will not replace humans. Human using AI will replace you. So I think that's basically what you have to keep in mind.
So if you're not embracing new technology as they're becoming available, somebody else which is using them, might displace you.
Charlie Jelen: What are your thoughts on some of these studies or some of the reports that show the energy impact, the environmental impact of ai? In terms of the energy density, right? And so like using AI to save energy, how do you balance that, that net positive there?
Jean-Simon: I think if you're like Netflix, uh, you're in trouble, right? Because you're, when you look at the quantity of data that you're moving and you're storing it is, uh, mind boggling. Absolutely. In our case, we're not moving movie or, or voice. We're moving like, uh, temperature, pressure point percentage. So it's so.
Tinny that really there's a very, very little impact. Mm-hmm. And even when you look at the gen ai, right? All of the LLM large language model. Yes. The huge one, which are trying to be capable to answer all of the question of the universe. They are heavy, they basically have an impact. But when you're using the very small ones, which are super niche and very narrow.
They are running actually on a computer. You don't even need a GPU to do the inference for it. Mm-hmm. So, of course these very small, teeny model, they're only capable to do, let's say, building engineering. Don't ask them how to do a blueberry cheesecake 'cause they have no clue. But when they run, if you keep it at building engineering, they are running at a very small CPU and memory space.
So, so it's all about, you know, use these things in intelligently.
Charlie Jelen: All right, so if, if anybody doesn't wanna deploy AI on their building after this conversation, I don't know who that is, but for the people that wanna deploy AI on their building, what are the first steps that they need to take?
Jean-Simon: The data, the data question.
Like, can we access the data which your control system is right now generating? 'cause if we cannot, it's uh, end of the discussion right there. So it's a number one question.
Dan Gentry: Is that data, is that like a backnet point or is it Yeah, exactly. Hardwire point Or can it be anything?
Jean-Simon: When I started Brainbox, I was a bit worried, right?
Because coming from the space, I know there's a lot of different control system out there and, and how do you plug yourself to go read that data? So I did a, a kind of an inventory, and then at that point, I think I nearly decided not to create Brainbox because I discovered it's 700. HVAC control protocol on our planet.
That's gonna be the barrier that's gonna kill this, uh, this initiative. Um, but when you look into the detail, there's about 17, 18 of these 700, which are 80% of the pie. Mm-hmm. And Backnet is one of them, right? Mm-hmm. Three Dium Niagara is another one. Mud Bus is another one. So, and then you say, okay, okay.
Okay. So I'm gonna forget all these 700. I'm just gonna focus on the 1780, make sure we could. Translate and basically have a gateway or talk that language, and then I'm gonna focus on that. So sometime we, we do hit customer, which are in that 20%. Then unfortunately we tell them, I'm very sorry, but there's not much we could do for you.
Uh, you're in that 20%. Um, maybe you should call your control company for an upgrade. Yeah, time for an upgrade.
Charlie Jelen: All right, listener like Doc Brown and back to the future. John's got all the answers. John, thank you so much for coming on. This has been awesome.
Jean-Simon: It was a pleasure.
Dan Gentry: Up next stat of the day,
a higher building IQ is a no brainer, right? You betcha. Train autonomous control. Powered by Brainbox, AI uses artificial intelligence to make AI work for your building. 24 hours a day, seven days a week, so you can focus less on the little things and more on the most critical bingo. Go to trane.com/autonomous control to find out more about trane's latest AI offering.
Charlie Jelen: Here comes Joe Day Stand. Other Dan, instead of the day. Dan, are you working on any large chilled water plants right now?
Dan Gentry: Um, yes. I would say at any given time, I would, you know what, what is a large plant? Let, let's, let's define something.
Charlie Jelen: Well, I don't, Dan, I don't, I don't know if size matters here, but I would say anything over a thousand tons, you know, ish.
Oh, yeah. You know, it's like a cute little plant. Cute little plant. I mean, okay. What's the biggest plant you're working on right now?
Dan Gentry: Over a hundred thousand tons. Um, big data center. That's pretty big. That's big. And I think for me too, I maybe from my vantage point, I deal with a lot of campuses, district energy stuff.
Yeah. So like, you know, 10,000 tons that's, you know, getting there. Okay. We get everything, you know, so I think size can matter, you know? Okay. Big plants, you know. Yeah.
Charlie Jelen: Alright. So maybe the stat of the day won't shock you, but the world's largest district cooling plant is the stat of the day. It's in Dubai and it's called the Business Bay District.
Chilled water plant. Mm-hmm. 241,000 tons of cooling. That's the capacity of this thing. Okay. That's a big plant. That's huge. That crazy. That is a big plant. That's really cool. Alright, so I put this, I wrote this out, the average American home. It's about two and a half tons of cooling. That would mean that this plant can serve a hundred thousand US homes.
That's huge. That's
Dan Gentry: a lot of cool.
Charlie Jelen: That is so cool. Those big plants of the day.
Dan Gentry: Thank you.
Charlie Jelen: You're welcome.
All right. Thanks for listening to this episode of Cool Air Hot Takes. Drop us a message if you want to get ahold of us or if you have any questions or if you have any edits that we need to know about. Cool air.hot takes@trane.com.
Dan Gentry: And as always, we want to hear from you guys. We want to hear your questions.
We would love to hear your hot takes. We want to hear any funny or maybe not so funny stories, jobs that went sideways, anything like that, fixes, hacks. Send those over to us. And as always, if you like the show, please like and subscribe. Give us a rating and a review on what. Whatever podcast platform you listen to, and until next time, stay cool and keep those takes hot, hot.
