At a recent heat pump summit, trainer Alex Meaney was asked about AI-powered sizing tools. His response cut through the hype: "All I need is an address? Bull$**t."
It's the kind of unfiltered take you'd expect from a panel featuring Meaney, Nate Adams (the "House Whisperer"), and Amply co-founder Eric Fitz — three practitioners who've spent years wrestling with load calculations from different angles. Ross Trethewey of TE2 Engineering moderated the discussion.
Their shared conclusion: every sizing method is a model with blind spots. Contractors who want heat pumps that actually fit need to triangulate.
Here's the uncomfortable truth: most HVAC replacements happen without a load calculation. Contractors skip the step entirely. They match what's there, maybe bump it up a size "for safety," and move on.
It's not hard to understand why. Around 80 to 90 percent of replacements happen when equipment fails. The house is cold. Nobody wants to hear "give me a week to run the numbers." And traditional Manual J — measuring every window, setting up a blower door, running room-by-room calcs — can feel like overkill for a same-day sales call.
So the real question isn't which method is most accurate in a vacuum. It's whether there's something faster and accurate enough to actually get used in the field. That's the debate this panel set out to have.
The panel kept returning to three core approaches:
Manual J turns construction details and design temperatures into steady-state loads. Done carefully, it gives a physics-based starting point. Done carelessly — padding assumptions for "safety" — it produces inflated numbers that don't reflect how the house actually behaves.
Past energy use treats the house as its own test chamber. Nate has pulled gas meters since 2014, matching bills to heating degree days. Done well, he lands within half a ton — close enough to choose equipment. But the method is sensitive to equipment efficiency assumptions, billing period mismatches, and occupant behavior nobody told you about.
Runtime data is theoretically the cleanest signal. If you know how often equipment runs and what it delivers at different outdoor temperatures, you know the load. The problem is scale — most installed equipment doesn't log that level of detail yet. And if it does, pulling the data out and running an analysis on it is clunky and cumbersome.
Each method has time costs, data needs, and failure modes. The panel's point isn't that one wins. All models are wrong. And with all models, it's garbage in, garbage out. You need to understand how each can mislead you before you trust it.
One of the sharpest exchanges came over whether you need a blower door test to run a good Manual J.
Nate's position: air leakage is such a huge chunk of heat load that without measuring it, you're guessing. Alex pushed back — not against blower doors, but against the idea that you can't make good estimates without one.
His approach leans on field clues. Look in the attic: dust streaks in insulation matching interior wall lines indicate missing air sealing at top plates. Listen when the wind blows: high-pitched sounds point to small tight leaks; low rattles suggest larger openings. Ask about comfort: do occupants feel drafts? Does a big furnace run constantly when it gets cold?
"Once you reach that threshold of 'I might be beyond loose,'" Alex said, "now you need a blower door. But I don't think you need to pull that trigger until you've got indicators."
The practical takeaway: visual inspection isn't as good as measurement, but a thinking contractor shouldn't be guessing between the tightest and leakiest possible assumptions.
Alex landed the sharpest critique of black-box AI tools. The issue isn't average accuracy — it's how they fail.
"You could have 90 percent accuracy with homes bunched together tight," he explained. "But you could also have enormous outliers. When your outliers are catastrophic, you don't get to throw them away."
Data scientists routinely discard outliers. Contractors can't. Outliers mean mold, ripped-out systems, and angry clients.
His test for any new tool: look at what goes in. If it only asks for an address, year built, and square footage, everything else is correlation, not causation. The less information a tool requests, the more skeptical you should be.
The panel landed in the same place. No single method is reliable enough on its own. But when you stack several partial views that fail for different reasons, confidence rises fast.
Alex framed it simply: "Two tools that are each 70 percent accurate, as long as they're not inaccurate for the same reason, get you to about 95 percent."
Eric pushed the point further: don't just triangulate at the design stage. Build feedback loops. Install runtime loggers. Revisit jobs on design days. See how systems actually behave. Over time, you learn where your process runs hot or cold and can adjust.
"Pick a process," Eric said. "Do it over and over. Go back to the home where you've installed equipment. Don't wait for callbacks — get data and refine."
Every load calc is a model. Treat outputs as working estimates, not facts.
Triangulate. Combine Manual J, past energy, and field observations. Watch where they agree.
Respect air leakage. Inspect attics, listen for wind noise, ask about comfort. When red flags pile up, get a blower door.
Question black-box tools. The less data they ask for, the more you should worry.
Build feedback loops. Log runtimes, check design-day performance, and refine your assumptions over time.
[00:00] - Episode Teaser
[02:17] - Intro by Ross Trethewey & guest intros
[03:22] - Nate on sizing via utility bills: "I can do it in 2 minutes"
[05:10] - Bridging the Adoption Gap
[09:16] - Alex defends Manual J: “It’s not that hard if you know your stuff”
[12:14] - How to assess infiltration without a blower door
[25:21] - Eric breaks down regression modeling using heating degree days
[34:36] - The Role of Human Behavior & Assumptions
[40:05] - The danger of outliers & the limits of AI-driven sizing tools
[52:51] - Triangulation: why using multiple methods is key
[59:06] - Final thoughts: pick a method, stick with it, and build your feedback loop
[00:00:00] Guest: Alex Meaney: One of the ways you can identify what you're dealing with in terms of this new technology. If it's a magic box, right? If it's a black box, stuff goes in and then stuff comes out and we don't know what happens in the middle. Look very carefully. What goes in. All I need is an address. Right. Well, we have historical construction year, and we have what? You're built in the square footage. Probably still, but. Okay. You know, like, how much information are they actually asking for? What are they allowing to consider? Because everything else is going to be correlations, not actual causation.
[00:00:48] Host: Ed Smith: Hey everyone. We've got a great episode for you today. It's recording from the US Heat Pump Summit, which was in Worcester in November. This is a panel with roster three moderating Alex Meaney, Nate, the Horse Whisperer, and our very own Eric Fitz from Amply. The topic is different ways to do load calculations. We cover a bunch manual j using utility data and using runtime data from equipment. It's a great back and forth between three really smart guys, who all have slightly different perspectives on the pros and cons of each. The net of is basically triangulate as much as you can, but it's a super fun listen. One comment though, on the editing we've done here, we've basically taken out when the audience asks questions, it's impossible for you to hear what the audience is asking anyway. And mainly Ross. But then the other panelists do an excellent job of repeating the question. So we just cut out when the audience asks a question, and then you can hear Ross or the panelists repeat the question. Just makes for a cleaner listen for you all. Last, a quick plug for our new article, The Heat Pump Business Model Matrix How to avoid Competing on Price and the race to the bottom and How to Win on trust. We've gotten great feedback on the article, and we're including a link to the trust checklist. It basically takes that article and distills it down to a series of items that you can look at and check the box on or not, to figure out, are you doing the things that differentiate you from other contractors? So you win on trust and can have an enduring competitive advantage and healthy margins? Both of those links are in the show notes. Enjoy it. Hope to get a lot out of it. We'd love to hear what you think. With that, let's get on to the show.
[00:02:15] Moderator: Ross Trethewey: Everybody, for coming today. My name is Ross. I'll be the moderator today. We have a great panelist right here of a group of guys we're going to go through. First some housekeeping items. If you are BPI certified, please scan the QR code to get credit for today's training. And now well, that's it for housekeeping. Let's jump into it. So three awesome guys up here. If you don't know them I'm going to go through them one by one. Alex Meaney runs mean HVAC. He's a trainer kind of world renowned in this circles. He does an unbelievable training as far as understanding taking technical subjects and making them very, very simple and funny and funny and funny and funny. Most importantly, yes, we're throwing you no pressure though. No pressure. We got Nate Adams, Nate, house whisperer, Obviously one of the guys that measures everything and will give you a untethered version of what he thinks and sees. So he is unbelievable advocate for heat pumps and for the industry. Glad to have him as well. We have Eric Fitz from Amply energy mechanical engineer. Number one, I love mechanical engineering. Pound it. He co-founded with Ed the A energy software. So it's lidar detection for manual just by scanning with your iPad in the existing house. It's a pretty amazing stuff there. So happy. Happy to have him as well. So we got some slide decks for you Nate. You want to kick it off? Yep. This ain't working. Good. Yeah. You're good.
[00:02:32] Nate Adams: All right. So yeah this is my first slide. Here we go. But moving on. So I suggested this idea because obviously like I, I model stuff. I model stuff for a long time I've used a whole bunch of different tools. And something that has surprised me through the years is that past energy use can be remarkably accurate. I get to check at least some client homes, and I found that oftentimes I'm within 1500 to 6000 BTUs of reality by peeling apart the energy bills and putting things out. And I mean, the three of us, like, have discussed this ad nauseam and had our fights and whatnot, but we all love each other. And kind of my goal is to be within half a ton, because any closer than half a ton on sizing honestly doesn't really matter, because you can't buy in smaller increments than that. So but what I wanted to start with here, the reason I put this slide up here. So a show of hands who would like to see more heat pumps sold on the market? If he's good, he said at the conference. Yeah. Yeah, exactly. So there's a kind of a layup right there. But so with that in mind, who has seen that something like 80 or 90% of replacements tend to be emergencies. So we need to solve that don't we. Somehow. So this was something that I've been kind of chewing on for a while. And a lot of what, what I struggle with.
[00:03:52] Nate Adams: So like I pulled my first client gas meter in 2014, so I've done this for a year or two. At this point, it's like I just this stuff doesn't even scare me. Like, I shouldn't probably admit this, but one of my houses I put a heat pump on and I didn't run a load at all. The other one I actually told my friend who was running a load what I thought the load should be, and he made it that, um, so like I'm using the tools wrong. But then I got to watch what they they were breaking off. What? What? Sorry. But anyway, what I've been watching is we have generally been focused too much. So this is adoption curve. We're focused right now almost entirely on the early adopters. So these are the shiny people. Like the very early ones will try anything just because it's new. That's the first couple percent. And then you get the ones that want to have something before their friends. And so that's your first 16% of the market. But then you have to jump to the mainstream. And that chasm there is really hard to jump because they're very different markets. They want very different things. So can we help heat pumps jump from the early birds, which is mainly where we are right now to the mainstream. And this is why I've been thinking about the past energy use either as a layer on top of like manual J as a verification or it surprised me.
[00:05:12] Nate Adams: I just kept chewing on this. I'm like, I think we can probably run this. I was tempted, if anybody here has an energy bill, I'll yell out my cell. You can send me an energy bill. And while these guys are talking, I'll figure out what the load is like 2 or 3 minutes. I can usually figure out most of this, particularly if it's gas. So anyway, this is what I'm thinking about is we need to figure out how do we move to the mainstream with everything that we do. Does that make sense? So it's a different way to start a fight. Now there are two different papers have come out recently. So the building Decarb Alliance in Canada just did this. Now they were trying to get data past energy use. But these are two different modeling tools that they use for different provinces. And you can see they came up with fairly different numbers. So this is not manual. J to be clear this is F 280 another program. And that's a good paper to read. But what's going to be really interesting in the hybrid. You can call this Friday before we all came here. They are going to be collecting data this winter. And they're going to make past energy use and runtime acceptable sizing methods for Canadian programs next year. That's the goal. So if that's the case it can be pretty interesting and we get to copy them. So FYI f.
[00:06:21] Alex Meaney: 280 and manual j might as well be interchangeable when you're talking heating loads.
[00:06:25] Nate Adams: Are they that close.
[00:06:26] Alex Meaney: Yeah it's all just cltd based Ashrae okay. At worst case scenario. So yeah, there's very little manipulation there. You're very on par with the two of them.
[00:06:35] Nate Adams: Okay. Yeah I know they're similar, but I didn't know they were that similar.
[00:06:38] Alex Meaney: Oh yeah. There.
[00:06:38] Nate Adams: Yeah I think it's worth going back one point though, to say that most of the HPC replacements in the marketplace are happening when the unit fails. And so what Nate is saying is that there isn't time to say. Hold on. You have no heat. Your furnace has failed. Let me go. Run a manual, J. And let me figure out what the size of this new heat pump that we want to install is. That's not happening. And so we get the mainstream. That's one of the problems though. It's not necessarily that Manual J is bad or good. I think part of it is that the replacements are happening when units fail. So I think part of the planning here is that we want to get out across the net here. Exactly. It's talking about planning, right? Like when the unit fails, what am I replacing it with so that when it does fail you have a plan in place, right? Yeah. I'm hoping that through consumer education, maybe we can get down to 50 over 50 over time where only half of them are emergencies. I doubt we'll probably get beyond that because it's just human nature.
[00:07:30] Nate Adams: We don't fix stuff until it breaks, especially Americans, just kind of how we are. So this is one paper that's worth checking out. The other one came from their mentor at the top here. This is the the paper. And so take a look. They've got manual J coldest day watching runtime and CPR VA is based on energy use and tried to include enough here that you can kind of get an idea. So I'm doing what you shouldn't do and creating what's called a slide. It's like a document. But, uh, if the slides go out later you can check this out. So this was interesting, but note how much higher manual J is now if you do it right. So Eric, you let me use a template which actually does have a chance at the emergencies to work because it's quick. And it softened my stance on manual J. So it still like you need to know what your air leakage is because that's such a huge chunk of heat load. But the formulas aren't off as far as I thought they were. So there you go, Alex. It's probably nice for me to oh.
[00:08:27] Alex Meaney: I missed the.
[00:08:28] Nate Adams: The formulas are not off as far as I thought they were. Oh okay.
[00:08:31] Alex Meaney: So I will jump in just right now and sit back. I will push hard against the idea that there isn't time to run a manual J like it is not that difficult. It does not take that much time. Yeah, Like especially for block load. Right. If you're going to design a whole system from the ground up maybe. But if you're inheriting some information, you can do some tests, visually inspect some things and have a load done pretty quickly. And you can integrate it in the sales process. And I will shut up for a little while. And but that's going to be the my main point.
[00:09:07] Nate Adams: Yeah. Yeah. So I don't disagree with you. I just don't know how many contractors are actually going to do it. Yeah. Which is a problem with what I'm discussing too, to be frank. Yeah, yeah. Before the technology, like Apple and some other ones, you'd have to go measure every window, measure every door, measure every room, every ceiling height for the entire house. And then, oh, set up a blower door to get an air leakage target. And so if you wanted to get an accurate manual J, you had to run a lot of these items. So the question is how much accuracy are we willing to give up to get a manual? J. That's close enough. Yeah. Right. Yeah.
[00:09:42] Alex Meaney: Yeah. And I think I might take eyeballing window sizes Is over and and running the actual numbers over, you know, some pretty noisy data.
[00:09:51] Nate Adams: Three by five.
[00:09:52] Alex Meaney: Okay. What do you mean?
[00:09:53] Nate Adams: Like, the energy is noisy?
[00:09:54] Alex Meaney: Yeah, yeah, I like blending them together. That's my. I like it's a cop out and a half for this conversation, but I think there needs to be a bedrock of an actual calculation. Do we want to just.
[00:10:06] Nate Adams: Yeah. Do you want to do some questions? Yeah. We can do some questions. Question right here. Yeah.
[00:10:09] Audience: That chart is hurting. My eyes.
[00:10:11] Alex Meaney: Yeah.
[00:10:12] Audience: Yeah I got somebody you didn't I don't mind.
[00:10:17] Nate Adams: So can you say that again? I couldn't quite hear it. I said that.
[00:10:20] Audience: Chart hurts my eyes. But I don't believe that that person, Emmanuel J. Yeah. I just can't believe.
[00:10:26] Moderator: Ross Trethewey: He's saying the chart hurts his eyes. And the person who did the manual, J that basically got through its graph came from wasn't doing it right or an accurate layer. Yeah, it's a challenge. Yeah, yeah. The hardest part is ultimately getting people to be aggressive enough on it. And it's very uncomfortable when you start doing it. Yeah.
[00:10:43] Audience: So so I would definitely for you to agree to pay that bond, especially if you use the software suite that you just recently about that is there in that situation would have gone in and in 20 minutes and 30 minutes you're on the floor or at the end of the day, it's really all about education. That customer doesn't matter if you take the time to spend an hour. It's just that. But even though we're in a situation where you keep asking me how much you're going to spend, why don't you? You're about to make 20 or $30,000.
[00:11:17] Alex Meaney: So for those in the back, he's saying he's routinely using amply and can get it done in about 20 minutes, which does not seem unreasonable.
[00:11:26] Nate Adams: That's right, that's right. Yeah, that's a game changer for software for sure. Question here.
[00:11:30] Audience: What's your suggestion when you don't have a blower door as far as leakage rate.
[00:11:33] Moderator: Ross Trethewey: So I'll repeat the question. Yeah. So the question is what do you do when you don't have a blower door score? I mean you don't have an air leakage number for the house. Do you want to go. All right.
[00:11:44] Alex Meaney: So that's a you didn't ask a little question there. Um, but there are multiple things which you kind of do is sort of build the case. First you have to have your starting point of assumption. And if the house wasn't built, you know, post 2000, you're starting assumptions. Probably the semi loose category, which is roughly seven air changes per hour at 50 pascals in the manual J sort of point of view. I recommend people go and start looking in the attic because while air may be invisible, dust isn't right. And so inspecting the insulation for signs of yeah for for signs of you get your every ceiling penetration where there's a light you see the brown spot like okay that's kind of par for the course where you start seeing the striping in the insulation that matches the pattern of interior walls. That's a total lack of air ceiling at the top plates. And that's now and it's very common. And now your entire house is breathing. And now we're going to bump that down to probably the loose range and also have a conversation with your homeowner. The I once asked Hank Grotowski why manual J stops loose, because that is definitely not the leakiest house you can have, but not by a long shot. And so I said, why did you stop it here? And I thought I had him right. It's like, I've been doing this a while.
[00:13:05] Alex Meaney: I got cocky and I'm like, I think I got him on this one and he shut me right down. He's like, if you're any leakier than that, I don't care how well you replace the system. Homeowners not going to be comfortable. Damn it. That was a good answer. And you can spin that on its head and talk to the customer about their lived experience in the home. Are there places in the house that are particularly uncomfortable and how localized are they? Right. If if you have a spot in your house that's very uncomfortable and it's not sort of a, you know, generalized thing, you may be able to identify some infiltration issues. Using that, I like to recommend asking about what the house sounds like when the wind blows, right? Like high pitched sounds are generally coming through small, tight holes. Low rattling sounds tend to be indications of of a big leakage. And the other one is simply do you notice any drafting is there? You know, do you ever feel like the air is coming right from the outside? Do you ever feel like a little blast of humidity anywhere or a blast of cold anywhere? And again, you just sort of keep knocking it down the spectrum. And I've I've been staunchly anti. You need to do a blower door if you're going to do a manual J.
[00:14:18] Alex Meaney: But I've also been staunchly. Once you reach that threshold of oh I might be flirting with Beyond Loose right. I might be going off the scale. Well now if you want to keep your customer happy, right. Because you can look at this from the contractor's perspective. You're about to inherit all of the comfort issues that they had. And I like to put myself in the shoes of the homeowner. But if I'm going to try to convince contractors, I often will kind of make the homeowners a little bit of the enemy, right? Because the problems that they had before you got there, all of a sudden they showed up, right? It wasn't doing this before, and now you're on the hook for it. And so I sort of use that to encourage them of like, you're not going to be able to fix this without addressing the envelope somehow. And that's going to need to start with a blower door test. But I don't think you need to pull that trigger until you've got some indicators that you might be beyond that threshold. Oh, and one last one. If it's a split level, it's leaky as hell, right? Just do a blower door test. Screw it. It's. You can make a split level tight, but like all of the record breaking leaky houses I've ever seen, where a split level. You need.
[00:15:27] Nate Adams: Some more. Two and a half story houses, my friend.
[00:15:30] Alex Meaney: They're a pain in the ass. Yeah, the two, two and a half. Yeah. Anytime you have a half story. Lots of knee wall. Knee walls are never insulated properly. It's a whole thing. Come take my class. I'll tell you how to do calculations for it.
[00:15:41] Moderator: Ross Trethewey: Eric, do you have any thoughts on that?
[00:15:44] Eric Fitz: Yeah, I just I think what Alex was saying about the things you should watch for in the house, that kind of like throw up your antenna or like your radar is like going from, like, green, yellow to, like red lights, where, like, you don't want to take ownership of a situation where you're like, I think this house is wicked loose. I'm not really sure. But like, I see all these other warning signs that it's going to like, the loads are really high. The homeowner is really uncomfortable. You don't want to say, oh yeah, I can just solve this by throwing in a piece of equipment. That's where you need to talk about weatherization. You need to understand the other aspects of the home. And like be clear with the homeowner. Like there's a bunch of stuff that I'm only doing HVAC I might not be able to solve and or get out the blower door test and like, really get some data. Don't just completely guess. Yeah.
[00:16:28] Alex Meaney: Can I ask a quick question of the audience? How many of you are not from what you would consider a cold climate? Anybody here? Okay. A decent a decent amount. All right. Because the thing you generally want to focus on when you're focusing on infiltration is heating season. The delta T is much higher, which drives a much higher pressure differential, which means there's just more infiltration. And on top of that, even in hot climates, your delta T is usually double in heating mode than it is in cooling, even when you're in a in a warm climate. And so just the general comfort profile of the home if you're having humidity issues and cold issues, but the temperature never see like you know, the thermostat always seems to be satisfied. Ding ding ding ding ding. Leaky house right? That is another key indicator. But yet just discussing with the homeowner. And not for nothing. Looking at their existing system. And you know, if you're in an old leaky house, you probably don't have awesome ductwork. It's probably undersized. It's probably very loud. You can very likely have a conversation of how often do you hear the furnace kicking on and off, right? Is this always seems to be running all the time? Like that's not a very precise metric. But when the answer is yes, it's running all the time. Anytime it gets cold, it never seems to shut off. And they've got a 60,000 BTU furnace and you run a load and it says 40,000. Where's that discrepancy? There are a couple of possible places, but infiltration is by far the likeliest.
[00:17:54] Nate Adams: Agreed.
[00:17:54] Audience: I just wanted to say I'm glad you addressed the top plate because we had a house. 23 arbitrators knocked it down to eight by film boarded yet? Yeah, we had to go to 1000ft² per ton before they were comfortable. We were still at eight air changes, but that doesn't. We got in a fight with the installer to get them to cut one of the five ton systems out. You guys. Yeah, it's.
[00:18:20] Nate Adams: Hard to hear. Yeah. Yeah. So the question is basically the original. Well, the comment was the original house that they were working on was 23. Ach, 50. Right. So 23 air changes at 50 Pascal, you know, blower door test. They weatherized it and got it down to eight.
[00:18:32] Audience: All we did was.
[00:18:33] Nate Adams: The all they did was insulate the air at the top of the building, basically the attic space. Right. And so just by doing that went from 23 to 8 to basically make those homeowners comfortable, they had to put in 1000ft² per ton AC. Where was it?
[00:18:46] Audience: Kim from Dallas.
[00:18:47] Nate Adams: And that's Dallas. Okay.
[00:18:49] Audience: I wouldn't even put a sweep on the door. You could see the cockroaches coming.
[00:18:52] Alex Meaney: Yeah.
[00:18:53] Nate Adams: Yeah, yeah. Crazy. Well, I mean, 23 to 8, that's a huge drop in air leakage.
[00:18:58] Alex Meaney: And side note, if you are from an area that doesn't have, like the massive program, it's going to weatherize your house for you, basically sealing the lid. You got it down to eight air changes per hour at 50 pascals. It was probably outperforming that. Doe did a study in right after the building crash in zero eight, where they bought up a bunch of houses in Vegas, and they found that a hole in the side of the house created one third as much air changes or infiltration than a hole in the top of the building based on decay testing. And that's because both in summer and in winter, you're because of mild stratification. Okay. Two storey houses with 80 degree second floors. That is not stratification. That's bad building and hot attic infiltration, FYI. But due to mild stratification, your warmest area is going to be at the top. And that's right up against the coldest air. And so that tends to drive the stack effect. As you know, everybody's probably heard the term the air leaves through the top of the building at least hot pictures. The air leaves to the top of the building and tends to get drawn in through the sides unless the wind is blowing. And then in the summertime, that reverses because you've you've enclosed high temperature air. And as anybody tangentially related to the HVAC industry should know, hot means high pressure air moves from high pressure to low pressure, so the air will penetrate down through. But the driving force tends to be the top of that building. And so that makes sense on a by the numbers thing. But our numbers are not always very reflective of real world performance. And it turns out that the improvements you make to the lid of the building are probably three times as effective as the ones you're putting around the perimeter in terms of the lived experience.
[00:20:42] Nate Adams: Yeah, if you do those.
[00:20:43] Alex Meaney: Yeah. Lowest point and highest point. Yeah. Lowest point also helps. Highest point is still bigger. But like yeah the lowest highest point. So that's true if you have vaulted just pulling.
[00:20:53] Nate Adams: It back to what I was originally talking about. Oh yeah. You're up next Alex. So again, what I'm trying to think about is the larger market, because everybody here, we are nerds, but we're only going to do 1 or 2% of work in general. So how do we get better work out to significantly more people. So like I want to build it like I need to learn vibe coding here. But I think with just like it's easiest with gas by far, but with annual gas use, I think I can get within a ton just with one number in most of the country. So that's the sort of thing that I'm trying to think about here. And that's I also model too. So I don't want to say that like I'm devaluing this stuff because there's a lot of pieces of the energy use that don't work well. Friend of mine in San Francisco. He's like, some people just kind of suffer or they'll just run a space heater. And so the gas use there is not valid like that. That isn't a good picture. Well, I was just texting with Jim Bergman yesterday who has a hybrid thanks to me, and he sends me his energy use. And like, this is really hard to tear apart now. He's like, oh, I changed this in February. I'm like, dang it, I can't undo this easily. Like, this is such a hard math problem now. So there's there's instances where it's not good new builds got a model you don't have past energy use on a model. Yeah.
[00:22:07] Alex Meaney: So there's I worry a lot about like long hot showers. Right. Like like literally the way we live in the house. Like, it's there is a lot of noise in that signal. And it is not a it's hard to dive deeper into that data without just getting anecdotal information.
[00:22:25] Nate Adams: Well summer bills is enough. Yeah. So I find that.
[00:22:28] Alex Meaney: Yeah. So use the summer build.
[00:22:29] Nate Adams: 15 terms per month for hot water depending on what they are. And then you multiply that times 12 and you subtract it from the annual usage and you're within 20% on those.
[00:22:38] Alex Meaney: Anybody take longer hotter showers in the winter?
[00:22:44] Nate Adams: How much will it affect it though? What's that?
[00:22:46] Alex Meaney: I don't know. Right. Like it's.
[00:22:48] Nate Adams: You gotta run some more loads this way, man.
[00:22:50] Alex Meaney: I do, I yeah, yeah.
[00:22:53] Nate Adams: Are you.
[00:22:54] Alex Meaney: I've already. I've. Yes. So and actually let's go back to I don't trust these these manual J numbers. So we actually kind of know that especially in terms of energy usage, to expect that those numbers should be off by about 30%, like when we do energy modeling like simplified energy modeling and have since like the 70s, we've just added a 0.7 multiplier to our heating numbers because we just couldn't reconcile the end data. And it's because you don't take sunlight into account. You don't take internal loads into account. You don't take thermal mass into account. That's the reason why the F2 80 and the manual are so close together is this isn't a methodology, it's just math. It's just physics, right? It's like if somebody tried to like, you know, worst Bragg in the history of Bragg's. I could do a I mean, I could do a heating load off the top of my head with a pencil and paper. It's anybody could it's like three formulas. And so we've known about these limitations for a very long time. And I do believe in when you can reconciling the information. But this idea that it's going to be faster and better than just doing the damn manual, J doesn't feel right. Because in order for me to trust those numbers I got to do, I've got to do more.
[00:24:13] Nate Adams: I see so many bad manual J's, though. My assumption is I can cut it in half to start like that's my assumption. Well, and then I go deeper when I can.
[00:24:21] Alex Meaney: You may also largely be dealing with people who knew what they wanted and reverse engineered.
[00:24:28] Eric Fitz: I'm going to call an audible. Should I do my slides next? Like, we can dive into exactly what you guys are talking.
[00:24:33] Nate Adams: Let's do it. Let's do it. There you go. Yeah. Oh, yeah. You need the clicker.
[00:24:36] Eric Fitz: Great. Thank you. So we started talking about these different trade offs between manual J historical fuels, even mentioned runtime data. And the thing I want to the most important thing I want to point out is that these are all models. And with any model, if you put garbage in you're going to get garbage out. And so like a lot of you know what we've just been talking about, you know, looking at those slides that Nate was showing like what David was saying in the audience here, if you are padding your load calculations with assumptions about, oh, you know, I'm going to assume a single pane clear glass, I'm going to assume I've got the worst air leakage. You're going to have some pretty crazy load calculations, because if that doesn't actually match the reality in the home, you're going to get some bad results. The same thing is true with other methods. They are models. They take inputs and assumptions and data and there's different amounts of uncertainty. There's different amounts of risk. There's different amounts of sensitivity to those different assumptions. And across these different methods, the one real common aspect is that they all incorporate something to do with the weather, which is makes a lot of sense because ultimately, fundamentally, things that are driving loads, it's the outdoor temperature of the home. So we've got to understand whether there are differences. Right. So like in manual J, we're a lot more focused on the construction methods and the materials. We don't incorporate details about occupant behavior. We're not understanding that historical fuel use.
[00:26:09] Eric Fitz: But obviously the past energy method is really going to focus on those details because that's the primary data that it's using to do the modeling. Similarly, for runtime data, yes, you can get some information maybe from a if you've got a thermostat that can record how often the equipment is running, but you've got to really pay attention to the manufacturer equipment, performance data and do a lot of careful matching to make sure that that works out well. We've been spending a lot of time talking about this sort of historical fuels based method, so I thought it would be helpful just to do a little level set of like, what are the main variables and what is the most basic way to understand how these models work? They're pretty straightforward, at least in the spreadsheet methods. They take the two most important inputs that I that I mentioned. You've got weather data. And most commonly that's in the form of heating degree data. So if you're not familiar a single heating degree day number, you can calculate it using sort of a baseline temperature is what it's normally referred to or a balance point. So the standard is 65 is what most data North America. And you compare that to let's say the average outdoor temperature your project location for a single day. So if the average temperature was 45 degrees outside and your baseline temperature was 65, your heating degree days is 20 degrees. And so you can do that across hours, days, months, years. And you can add up all these numbers.
[00:27:36] Eric Fitz: And you get heating degree days for a whole year, for a whole month, whatever period of time you want. So you get that weather data and then you get your historical fuels data. And ideally that is in at least a monthly interval. Even better, if you've got time of use rates or 15 minute interval data, like a much more granular amount of data. But at least you've got monthly. And it really needs to be for an entire year. You can run a bunch of issues if you do, like a single month. It can be challenging, but you take that weather data and you compare it to the utility bill data, and you can plot a simple line, basically a ratio between those two key inputs, and then you can kind of look at the how that data is sort of laid out using a regression method. So you can make a basically draw a line, a best fit line. And if you take the slope of that line and multiply that by your equipment efficiency, your design delta t, you've got a heat load. So it seems super straightforward but there's a bunch of gotchas. So for heating degree data alone it has some of the same challenges as the weather stations that we're using for manual J. So we have a lot of different weather stations around the country that have been collecting very accurate data for long period of time. But there's, you know, roughly 1500 in the US that does not cover every single point in the entire country. Right?
[00:29:06] Nate Adams: Microclimate.
[00:29:06] Eric Fitz: Yeah, exactly. Microclimates, all that kind of stuff. So there's a bunch of different tools out there that are available to get heating degree data, and they're doing different averaging and other methodologies to kind of figure out what are the heating degree days for this specific.
[00:29:19] Nate Adams: Yeah, probably the big one. Because if you use a 15 year you're going to get a very. If you use a 15 year average right now, you're going to get a very different difference number than 30. Yeah. Like where I live is supposed to technically be climate zone five. We moved and I looked afterwards. And there's this little point that comes down in West Virginia where we are still in five. I'm like, dang it. I wanted to get warmer, but it is climate Zone four where we live 100%. So that has changed totally.
[00:29:45] Eric Fitz: So you have to you got to really understand your source of heat degree data is what I'm trying to say. And this topic of this baseline temperature 65 degrees, that is a standard because we had to pick something. But in reality, if you're doing this correctly, you're actually iterating over the model to figure out what is the actual homes balance point. There's a methodology to do that. I don't know of very many systems or tools out there that are, but it is possible to actually figure out for this particular home given the internal loads, given the solar gain. This home doesn't actually need any heat all the way down until we hit 57 degrees.
[00:30:21] Alex Meaney: That's with momentary data. That's what finally parse data into, like, a monthly energy bill. Sure.
[00:30:26] Nate Adams: Yeah.
[00:30:27] Eric Fitz: Question.
[00:30:28] Audience: One of the.Facts that I found as well at the UK. So different climate but same challenges. Yeah. Thinking houses. Well the kind of fudge factors, if you like, or this fantastic theory doesn't take into account is additional sources of heat within the roles. Yeah. Yeah.
[00:30:50] Alex Meaney: That's that's what we mean by internal gains.
[00:30:52] Audience: Yep yep.
[00:30:54] Audience: Electronics within homes increasingly.
[00:30:57] Moderator: Ross Trethewey: Yeah.
[00:30:57] Moderator: Ross Trethewey: So the do you want to repeat the comments of people in the back are basically they can't hear. So in the UK you know there's you know different ways of looking at it and different fudge factors. But the main thing is internal gains is the comment. Right. People give off heat, appliances give off heat, and some of those are not factored into the to the graduation. Obviously in balance school. Sorry.
[00:31:19] Nate Adams: What's the lowest that you have seen? Like I've seen some houses that don't need heat till 40 or 45. It's got to be pretty tight. Yeah, but I've seen them that low. Like passive House is definitely in that ballpark. But, I mean, I've got some of my client homes down, like in the 50s. No problem. Sure. Where they slide really, really slowly. So one of the things I wonder and like, I don't know how to collect the data, I really don't. Could this line like I think for leaky homes it's not quite straight line, but it could also be a different slope. Like I just I don't even know how to collect the data for it. I'm thinking it's like a one and a half power thing when it gets really cold. Well.
[00:31:52] Eric Fitz: All of these models are they're they're assuming steady state. So if, if the wind is blowing. Yeah. Your air leakage is changing in real time as the pressure differential changes. But generally speaking, you know, our understanding of why these models work at all is like we see this linear this line. Right. That's a home. There's a direct proportional relationship to the outdoor air temperature and the energy that's used in the home. Like, the energy flows are happening in a linear fashion. And that's generally that's how homes operate in dynamic conditions. Yes, but this is trying to model steady state over a long period of time.
[00:32:30] Nate Adams: Which doesn't happen.
[00:32:31] Alex Meaney: One of the reasons why we have that 30% discrepancy between manual J and anything energy modeling is I still have to cool the heat. The house at midnight. Right. And so when I do a monthly regression, if I have a house that had a lot of windows front and back and it was facing east west, it would have a much shallower slope, right? I would have the line would be down here if I flip that around and make it face north south, the line is going to be up here. The load on the on that building doesn't change in heating once the sun goes down. Right. Like once once. I need to maintain heat levels in the house on a on a cloudy week or something. The fact that those that they're being averaged out over the month, that's good. But it's not great, right? Because I still need to heat things when the sun goes down. Right. So like that major factor introduces a lot of noise into the signal. This is in my opinion, this is a very good companion. And it's a it's a great way to do large scale studies, like a great way to do large scale studies. But if I want to replace somebody's system, I have all the data I need, like standing in their house.
[00:33:44] Audience: You still need to heat the house after a week. Cloudy weather.
[00:33:48] Alex Meaney: Yeah, a week of cloudy weather, he's saying. Yeah.
[00:33:51] Audience: Yeah.
[00:33:52] Audience: The other factor I think, is just human behavior. Some houses set their house to auto their heating it when it's 68 outside or whatever. We're doing. Other houses, all New Englanders throw the windows open and close the windows, just behave their own. So what's the balance point. For similar houses. Move?
[00:34:11] Eric Fitz: Yes. Economy is around human behavior. People do weird stuff. The open windows, you have setbacks. There's all these different factors that you need to pay attention to. Because if you don't know about these kinds of behaviors, it really changes your data. If you've got someone who, you know, they're they're like, we say in Maine, they're snowbirds. They go down to Florida for for the winter, and they're leaving their house at 55 degrees, set point for three months of the year, a big chunk of the heating season. That's obviously going to dramatically change the dynamics of the home and in the data. And if you're not aware of that, like you can get some really crazy numbers. The other thing I just wanted to point out, so there's there's a bunch of different things you need to pay attention to these, these other fuels that are being used. Do you have domestic hot water like an indirect water heater off of the the boiler that's also consuming your primary fuel like natural gas. But equipment efficiency, equipment efficiency is probably the most sensitive variable in this kind of modeling. Every 1% difference in your assumption for equipment efficiency will change the load calculation by about 2%. So it's like A2X factor. So and this is really important. Like if you if you've got an old, you know 80 Afue furnace, it's pretty much you know you can use a combustion analyzer. You can you can measure the efficiency. But generally an older furnace, they're going to be operating pretty close to that. But if the homeowner has a condensing furnace or a condensing boiler, you're like, oh, this is like 95 Afue. Well, depending on how they set up that boiler or that furnace, it might not ever really be condensing. So it might say on the nameplate that it's got this rated efficiency, but you actually need to get some good data about how it's really performing over time. How often is it actually getting into condensing mode? What is the real efficiency? Or you've got some big issues again, with using this approach for for modeling.
[00:36:02] Alex Meaney: And now instead of teaching them how to do a load calc, I teach them how to use a combustion analyzer properly.
[00:36:09] Nate Adams: For what it's worth, though. Combustion.
[00:36:13] Alex Meaney: Yes. Yeah. This is actual efficiency. Yeah.
[00:36:15] Nate Adams: For what it's worth, though, the energy use does end up showing most of those things. Not always, but I had one client who I asked, like, so what do you keep it in winter? It was like 55 degrees. Like, who is heating their house to 55? So all of a sudden, the 800 therms he was using, I'm like, oh, it's a low load house. I'm like, no, it's leaky. And I put the blower door up. Now it's I had the benefit of all the other data. Yeah, but it was a really leaky house. And then I came back like it ended up six months later when I did the audit and I turned the air conditioner down, it was 80 degrees out. I turned it down four degrees. It never moved. I'm like, all right, we got all kinds of other issues here. So yeah, it's but like it did show in the energy use, but I had to ask him what his set point was. But in general, we're talking edge cases where most people live in their houses all the time. Sure. So like 70, 80, 90%. We can talk here, but there's always the exceptions. Yeah, yeah.
[00:37:09] Alex Meaney: I think he had it first.
[00:37:11] Audience: Yeah. I'm sure. Their ventilation comes into play in terms of education. The importance of mechanical ventilation. How proposing supply exhaust all your whole house impacts. And yeah bringing them afterwards the harmonization and repeat the question.
[00:37:31] Alex Meaney: Yeah. So he's asking about how ventilation would impact the process. And the good news is, is very simply I don't care if you're exhausting supplying balancing. I mean, I do, but for a load calculation a CFM in is in, CFM out. And so when I'm sizing a system, how many cfm's going out? Okay. Well that's how many cfm's coming in. How many CFM coming in. That's how many CFM coming out. And so if it's a quantifiable amount and I'm in control of it, then that's a very simple thing to add to a manual. J um, so I mean that is a big argument for having ventilation is putting you in control of the outside air, so it actually simplifies the process. Now as far as homeowner education. And I'm going to duck that one and call it a little off topic because that's a bit of a whole. And I have a some radical opinions on that. That will definitely cause us to go down a rabbit hole. Uh, so I'd rather avoid that.
[00:38:26] Audience: Just wanted to point out it's a little more complicated if you factor in energy recovery.
[00:38:32] Alex Meaney: Sure. I'm just gonna mention that. Yeah.
[00:38:34] Audience: Energy recovery. Ventilators. Right. Air to air exchangers. Those are going to have a certain efficiency with them, both in terms of sensible and latent efficiency. So it does matter a bit.
[00:38:42] Alex Meaney: And that's just a number. And my favorite thing about Ervs and Hrvs is the efficiency on them is the one damn data point in this industry that's not hidden from the whole damn world. If I if I want to know what it is, I could just look it up. That's right. Everything else is, like, tucked away. And I need to log in and like, it's, you know, can they publish three different numbers in three different documents. Like it's nuts. Rv and RV. Here's how efficient.
[00:39:08] Audience: Can the heating and cooling manufacturers please publish their engineering data? Good lord. Yeah, man. Yeah. Wait. It's hidden behind six paywalls, right? Yeah.
[00:39:17] Alex Meaney: Super annoying. Yeah, I guess I'm supposed to. Yeah. Pick up. Besides. Here. Yep. So I basically just anticipated anytime I'm asked to talk about next gen load calculations. Somebody asked about AI, right? Like AI is going to do this for I had ChatGPT do my load. Like, you know, I've heard it and like, oh, Lord. Well, and the thing is like, did it was it that terrible? Like, hey, so I have an operating theory. We have an like, in case you are not picking up the undercurrent here, we have a very long standing debate as to whether or not blower doors are required and like the accuracy of fuel bills. And I think one of the secrets to Nate's success with using fuel bills is Nate. What? What? Humans are incredibly good at compared to computers up until very recently, is pattern recognition. Like we're really good at it. Like on a subconscious level. Good at it. And we are just now starting to have computing power that can do the same sort of pattern recognition that we can do. And then, of course, you add the power of a computer to it and you can start drawing some insane conclusions. Right. Like when data correlates, like if you're not asking the program why? Why is it that when they're built in this year it tends to be this? If I don't know why, I don't know what to look for in terms of exceptions, where my red flags are, where this won't apply, etc.
[00:40:54] Alex Meaney: like, oh, we might, you know, we took the data from blah blah blah, Saw somebody did a study of a thousand homes in this area. We had very good data on it. We plugged it into our model. We cross-referenced it against when it was built, and the square footage and whatever other data points we can find. And now we're and now we're confident that it can actually come up with a load. You know what? You know, you've built there. You've built a tool that can come up with the load from that data set and that data set alone, because we are not explaining why this works. We are reverse fitting models to piles of data. And we're hoping that if the data pile gets big enough, it's going to start being right. And and here's the thing. It is right a lot like a lot of lot like 90% a lot probably. Here's the problem with that. The top right box here, those are all 90% accurate Okay. Like you could have them bunched together kind of tight. You could have them bunched together incredibly tight and have enormous outliers. You can have more frequent outliers with them tightly grouped in the middle like that.
[00:42:12] Alex Meaney: And here's the thing. Like within that tight range, it costs me nothing to be a little off, but to be on the furthest ends. I'm tearing shit out and replacing systems I've grown mold in somebody's house. When your outliers are catastrophic, you don't get to throw them away. And you know what data scientists do with outliers? They throw them away. They throw that it's standard operating procedure for coming up with large scale models. And when you use large scale models in the large scale, they're awesome. And when you draw conclusions from large scale models and you use them on my house, there is absolutely no guarantee that it is going to work. And if the if I was within, if I was the top line, if I was within 10%. And that's what we meant by 90% accurate. Always within 10%. Do it. Absolutely. Every time. Absolutely. But when I run the risk of being one of the outliers, I take a very dim view of blindly trusting any tool that's going to do that. Any tool that tells me I can have the size of a system with the click of a button. Bullshit. No way. Right. Houses are too varied for that to possibly work. Now we start the fuel. Data is data. Like I'm starting to feed it. I'm way more like trustful ground truth.
[00:43:37] Nate Adams: Ground truth? Yeah, yeah.
[00:43:39] Audience: Yeah, but plug in space heaters and humidifiers with handle that 90%, 10%. Come on.
[00:43:45] Moderator: Ross Trethewey: Plug in space heaters and humidifiers. Dehumidifiers will cover. That is what he said.
[00:43:50] Nate Adams: You're not wrong. But that's part of why like personally, I want to see monthly data. But what I'm kind of talking about here for a really rough thing is let's put in annual use and get plus or minus a ton, something like that. And then it's a trust but verify. It's not like you should size from this like the voice of God has spoken. Um, it's well, it's probably in this range, and you should probably have someone come ask you a couple of questions to narrow that in. So I think a lot in ranges because I mean, well, just for instance, like we're working on a house that's going to be all steampunk themed, we do these crazy Airbnbs and we spray foamed it. Um, and I ran the load with Ampli, and I came up between 18 and 36,000 BTUs based on just blower door changes from the low end of what I've seen in reality. And the high end now, you're not going to get, you know, it's a 900 square foot house and it would be like a 5000 blower door. Um, I don't even know what it is in ACH 50, but like 30, something like that. But I've seen it, so I'm like, well, let me just put the low, high in here. And so that is possible to see that on a house that size. And that's one of the challenges. Now, another one of our houses we just bought, I got past energy use. It used 520 Therms for heating, and that works out to right about a two ton load in our climate zone. And I'll watch it. I'll put an ecobee on it. So I'll show you on a design day. Uh, should.
[00:45:07] Alex Meaney: I don't I don't think that's a realistic range. I don't think you can live in that house with 5000 cfm of leakage and not know it. I don't know, like leaky. Yeah, this is what I'm saying, but.
[00:45:20] Nate Adams: I've started with multiple houses in that ballpark. It's like a 3 to 1.
[00:45:23] Alex Meaney: When you look at the swing between a really good house, which I can tell I'm in, and a really terrible house, which I can tell I'm in, that's not a like anybody who's thinking and doing a manual. J should not be susceptible to that range, right. Because they're not going to choose the tightest possible construction when it's obviously not right. The range a human. Am I saying visual inspection is as good as a blower door test? Hell no. Like not even Close.
[00:45:53] Nate Adams: I'm here. You can't.
[00:45:54] Alex Meaney: But it's not this. But it's not this range, right? It's like this range. Now, when you bring it down to the worst stuff, it grows. Which is why the worst the building is. The more I'm going to encourage people to. Let's slowly. First of all, if your house is in that range, this is the heat pump summit. Don't put an effing heat pump in that house. Okay? Done. Right? Done. If you are not going to air seal that house, that is not a heat pump house and no dual fuel, maybe dual fuel.
[00:46:30] Nate Adams: That's what I mean. Dual fuel maybe.
[00:46:32] Alex Meaney: But at that point, like.
[00:46:34] Nate Adams: Not straight electric. Just to define okay. Yeah, yeah yeah.
[00:46:37] Alex Meaney: Everything should be dual fuel I make makes an excellent point. Make everything a heat pump. Why do we even make air conditioners like it's dumb?
[00:46:45] Audience: Yeah, yeah, I'm a writer.
[00:46:47] Audience: Not really. Do you take the slope this point out? Technology? I had a question? You know, maybe what inspired me is that I'm really interested, like Bill is it is better. 90% is a huge step forward. So that bad compared to Korea right now. But all right. So part of that it relates to colors of sparkle okay. Next to this you already repeat the question.
[00:47:22] Alex Meaney: All right. So that was a two parter. It was maybe three parter. Yeah. Well I'm combining the first two uh, is given the fact that so many people do poor load calculations when they do a manual J that they're seeing, they're seeing all of these distributions of accuracy from humans. Right. And sometimes even worse, like this might represent an improvement. Bad actors are going to bad act, man. Right. Like that. Simple as that. Right. If I know what size system goes in this house. God damn it, I'm gonna. That's what I'm gonna do, right? Like, there is no AI tool that's going to make that guy do a better job. Um. Uh, you know, when the conversation turns toward that, I start beating the drum for licensing, you know, stricter, stricter licensing standards and better training because it's not the only thing that's going to going to combat that, I think. And as with this, isn't this the goal of smart home technology? I will immediately surrender. If you've got 15 minute incremental data, awesome. You have 15 minute incremental data from the runtime of your system. We 100% know what the load is on that. Like, we May 1st day live in a day where I don't need any of this. I can just look at the runtime data that is very realistic. There is still some some noise in that signal, but you're never going to get a perfect signal, right? Like so. But that is going to limit the usefulness of the methodology to people who have already collected the data.
[00:48:51] Nate Adams: That's a problem.
[00:48:52] Alex Meaney: And that's a huge problem. So unless you plan on mandating data collection from our thermostats and smart home technology, that's going to be a narrow market, right? And so yeah, that's.
[00:49:04] Moderator: Ross Trethewey: Do you want to add something?
[00:49:05] Eric Fitz: Yeah. I just wanted to say that what I was trying to point out with these different models, there's there's ways to go horribly wrong in all these different methods. And so.
[00:49:15] Nate Adams: There are.
[00:49:16] Eric Fitz: It's, you know, it's it's a tool. It's a model. If you use a screwdriver as a hammer, like, you can kind of you can kind of come on something in. But like, so like you gotta understand the trade offs with the tool you're using. And you know, there there are things that are surprisingly sensitive depending on the different tool that you're using. So I just want to say another point like, yes, utility bill data. It seems very straightforward on the surface, but there's a bunch of subtleties that, if you're not paying attention to, can bite you. So if you don't very carefully look at what are what is the metered date versus like your billing days compared to your heating degree day dates, you can get wildly crazy numbers and not realize it. So you gotta you gotta pay off.
[00:50:06] Nate Adams: Would you say.
[00:50:07] Eric Fitz: It? It depends. It depends. So if if you're in a heating dominated climate and you're not paying attention that February has 28 billing days and January has 31, and you're using like a proper day normalized weighted model, you can get like 15% probably. Yeah. I don't know the specific I don't I've not done the analysis on like exact sensitivity. But it, it it definitely has an impact.
[00:50:36] Nate Adams: As you said annual data is is clutch. You need to know your annual usage.
[00:50:41] Eric Fitz: Right, right. And so I guess.
[00:50:42] Nate Adams: My annual.
[00:50:43] Eric Fitz: My point is, is even like if you get the utility bill data like you still there's a human that has to then still make decisions, enter in some information. There's there's technology that allows a homeowner to enter in their credentials. And like you can share that data directly, but there are still challenges with sharing that data. The data formats, there's a bunch of a bunch of challenges now. There's a bunch of challenges with manual J. There's a bunch of challenges with runtime data. So what I really like about all of these different methods, and I think both of you guys have have talked about this, that there's not a perfect way to do this. Use your brain, come at it from different angles, pay attention to what the homeowner is saying and you from these different angles, you can triangulate and get to a great solution. You know, you know, as a business, we're constantly trying to improve the technology side to make this stuff easier. But man, this is a hard problem and it's going to take take some time to figure it out.
[00:51:40] Nate Adams: Yeah.
[00:51:40] Alex Meaney: And I want to emphasize a point that he just made that may have flew under the radar. The noisier the signal is, the more I'm in. I'm in a house that I believe is very leaky, and I don't know how to quantify that because it's very leaky. It could be nine interchange. It could be 20, right. Like at that point you stop knowing right there. Let's look at your fuel bills. Let's look at this. Let's look at that. It's a it's a thing I often I've been trying to try to slay a dragon here for a very long time of how to get people into building science without having to spend about $10,000 worth of toys. And one of the things I've come to the conclusion is, is use multiple test points, right? Like, yes, this tool is only about 70% accurate and this tool is only about 70% accurate. And and it turns out statistically when you use two tools that are 70% accurate, as long as they're not inaccurate for the same reason, you get to about 95% accuracy, right? And so using multiple data points is an excellent, excellent, excellent way to go. And using a sort of. And this is why I say manual J is bedrock, right? You do your manual J and you have very good reasons not to trust it. Okay. Let's crack open. Let's. But I don't think an approach that claims it's going to be easier than manual J or faster than manual J is going to hold up to scrutiny. I think it's a wonderful backstop and an addition to a pepper for the gumbo, if you will, to make that better.
[00:53:10] Nate Adams: We're going to find out in Canada next year. But to second your point. My preference is to triangulate. Yeah. So like it's like it's going back to geometry. If you have three points now you kind of know where the plane is or like you can triangulate between those. Uh, what was it, Fe you were telling me that your your dad was a fire chief, and they had fire towers, and they would estimate what angle and what distance it was, and they could pinpoint fires in the 30s. He said it would show up like somebody was having a fire in their backyard. Like, what are you doing here? Um, so, yeah, if we can pinpoint. That's good. So if you can have multiple data points, that is the ideal. But what I, what I am trying to say here is, is there a way where we can use one data point and get reasonably close with a bunch of if, if, if. But I think we can. I'm actually I'm working on a course about this because there are so many different ways. If it's an all electric house, they're hard. If you have propane and oil, you need like three years worth of bills to have any certainty on what you're using every year. So there's there's a whole lot of ifs on this, but if you have a gas heated home with a gas water heater, they're easy. Two minutes. No problem, I got it. But there's lots of other ones that are harder. Yeah, or if you have like a heated pool. That one really threw me for a loop one time.
[00:54:22] Nate Adams: Like, why are May in October freaking on the moon? Like, oh yeah, they stole the pool off just to hit this one home. I feel like the main takeaway is I feel like that I'm at least gathering from this conversation is that the three different ways of actually getting to a load calculation for a building? There's three different, three different ways of doing that we currently know of. They all have pros. They all have cons. Yes, but if we can start to hone in and triangulate between. If we can ideally run all three, use runtime data, use energy bills, and use manual J, then we can triangulate and we can actually get into a much more accurate spot. You know, and I think to the gentleman's point over in the back there is that as technology gets better, I think a lot of these manufacturers are looking to include smart technology into their devices. This has already started in Europe, where they measure leaving water temperature, entering water temperature and flow rate from an air to water heat pump as an example. So guess what? I know exactly what that unit is delivering in terms of BTUs to the house, and I can map that to outdoor temperature. And now I have a much more concrete number. So I think technology will play a role here in the future. It's just a question of how fast that gets, you know, into the marketplace, etc.. But I think I think it's a good, good, good question.
[00:55:28] Alex Meaney: And I think you can fear those things. And I think actually real good runtime data is the top of the list, right? I think runtime data is here, right. But I think manual J is next. And I think the overall energy bill is third.
[00:55:41] Nate Adams: Time will tell.
[00:55:42] Audience: Yeah, yeah.
[00:55:42] Eric Fitz: I'll say it depends because like all these things, it's trade off in like, you know, time is money. The different tools you need to like do certain get certain data. You know, what your sales process is like. If a homeowner is actually comfortable sharing like a bunch of utility data with you or not, or even thought about it in advance, do they have it available? So like it depends on the situation and how much, you know, money you're willing to spend or time you're willing to spend relative to the uncertainty or the problem you're trying to solve. So you talk about like pattern recognition. If you're in a home that you've in the same housing development you've been in 50 times and you've, you know, was the same contractor that built it, you've done a manual J on it a bunch of times. Maybe you're not going to spend the extra effort to get pull the utility bills. That's not necessary. But if you're seeing all these other red flags, maybe I should go after rental. I should go after a blower door test. I should do this other stuff.
[00:56:35] Alex Meaney: I mean, if you had access to to runtime data. You just use that. Yeah, yeah. Unfortunately, it's not widely available for us to be using in a way that we can really trust. But if it were, I think that would be the hands down winner and I want to jump. I have one quick thing and I want to jump back on process, because I think that's where you're going to find a lot of common ground between the three of us. Number one. Number two, I think it's probably the most important, like conversation about implementing this stuff is not so much what you pick is how you implement it as a process. But I will say as new technology rolls out and we start worrying about this kind of thing, and particularly actually this slide is probably more relevant. One of the ways you can identify what you're dealing with in terms of this new technology, if it's a magic box, right, if it's a black box, stuff goes in and then stuff comes out and we don't know what happens in the middle, look very carefully at what goes in. All I need is an address. Bullshit. Right. Well, we have a historical construction year, and we have what year? Built in the square footage Probably still bullshit, but. Okay. You know, like, how much information are they actually asking for? What are they allowing to consider? Because everything else is going to be correlations, not actual causation. If I'm looking at things that go into oh we run it. Yeah.
[00:57:50] Nate Adams: Are we done at noon?
[00:57:51] Nate Adams: Yeah. Okay. One last point here. So the tool that I have used wrong, I use it backwards is actually a right tool. So it's carrier and right soft. If anybody wants to come up to me I'll show you the link. And that takes your hourly heating into account. So I game it so that I can figure out what the usage is. And I match the usage to what the load is. And I find that's pretty close, but it's based on a solid model. So it writes off this solid. I know you would agree it has its challenges, but like if it's used well it's amazing. It just is.
[00:58:22] Alex Meaney: Anything Acca certified it's going to perform just as well. Yeah, yeah, I shouldn't get the last word I've done all the time.
[00:58:30] Moderator: Ross Trethewey: All right. Quick last word. I think we're out of time.
[00:58:32] Eric Fitz: Otherwise I think one thing we missed. Just do one of these methods. Do it over and over again. Yes. Go back to the home where you've installed equipment and see how things are going. Don't just like, wait to hear a call back, like get a feedback loop. And so when you've done any one of these processes over and over again, you will start to really learn. You will understand the sensitivities and you will get better and better. It'll get faster and faster. It's a great thing. So get those feedback loops in place. Pick a process. Yeah, go for it.
[00:59:05] Nate Adams: Pick an ecobee. Like every time if you've got a basic 24 volt thing. So because they make the runtime the easiest there's other ones that do too.
[00:59:12] Alex Meaney: I know I said I don't want to do it. Dude come on, you're all nerds. You're buy a book on Bayesian reasoning if you don't know what that is. Bayes. Bayes. I would recommend Bernoulli's Fallacy if you want a recommendation for a specific book. It's all about starting with a reasonably good data point and then iterating from there and how you can be more accurate doing that. That's basically the philosophy behind it. It's this is a nerdy crowd. You'll love it. Like it'll make a whole lot of sense. Awesome.
[00:59:42] Alex Meaney: Speaking of books, building science here. So come see me if you want. Sorry. Yeah. Shameless. Yes. Very shameless. Okay, I think we're. Are we?
[00:59:53] Alex Meaney: I'll be. I'm not going anywhere.
[00:59:54] Nate Adams: Yeah, I'd like to thank Eric, Nate and Alex for their time. Obviously, their wealth of knowledge. I guess we could do a couple more questions if you guys are up for it.
[01:00:04] Alex Meaney: Yeah, I'm. I'm getting for it. Yeah.
[01:00:06] Moderator: Ross Trethewey: Okay. Great question.
[01:00:08] Audience: So a couple of things I had, first of all the the customer expectation factor. Yeah I know they live in a cold leaky hole. The crappy old furnace, you know, the super sexy heat pump system. They're going to be warm. Their energy bills are going to be long. You know where I'm going? Yes. So there's that kind of fact. The other point I just made was also in regards to use of utility bills. I always wonder I was doing this at first. So if you're going to improve fabric and obviously worse is more worse you.
[01:00:50] Moderator: Ross Trethewey: Get the envelope. Yep. Yeah. Yep. Yep. Yeah. We knew we had it for everybody else. Yeah.
[01:00:57] Audience: Just like usage rules. Yes. It's like, well, that's the worst case. We're going to be improving. So they you have starski method of glucose spike for baby to get.
[01:01:12] Alex Meaney: Yeah. General rule, the worse it is, the easier it is to make it better.
[01:01:15] Nate Adams: Yeah.
[01:01:16] Alex Meaney: Yeah, totally. So it's just the law of diminishing returns in reverse, right? Yeah. And a more reading material or things to look up. Check out Robert Bean's material on mean radiant temperature. Because when you are dealing with old, crappy leaky houses with no insulation, you are you are suffering from radiant temperature problems nearly as much, if not more, as you're suffering from low problems. So, like the science behind comfort in terms of managing homeowner expectations is is really important. And it gets trickier. Like it's not as easy as just having enough heat in your house. Okay.
[01:01:52] Nate Adams: Beans. Amazing. Read this stuff. Check this out. Yeah. One counterpoint. So we've ended up owning a bunch of houses here, and our worst insulated one was there. We we stay in them usually a week, a year, something like that, just to kind of see what needs fixed. And my wife is like, this is the most comfortable home that we own. The walls are empty. It's a 1950s ranch and there's two inches of rock floors in the attic. So the the right system that runs all the time actually does make a giant difference, which is why I pull for hybrids. Like, let's at least get it running all the time. Um, and then you may need more guts to actually heat the house when it's cold. Yeah, yeah, yeah.
[01:02:26] Eric Fitz: And I'll just say in homeowner expectations and we are at the Heat Pump Summit. But we can't solve all problems with the equipment. We got to address the envelope like you're talking about. So like sometimes you just need to tell the homeowner, like, we can't. There's no equipment that's going to solve this problem. We got to address the air leakage and the insulation before we do anything that happens.
[01:02:44] Alex Meaney: For what it's worth, I'm going to talk a little bit about that in my 330 or 3:00 whenever I'm on. All right. Come on up if you want to talk now we're done.
[01:02:51] Nate Adams: Yeah, yeah.
[01:02:55] Eric Fitz: Thanks for listening to the Heat Pump podcast. It is a production of energy, and just a reminder that the opinions voiced were those of our guests or us, depending on who was talking. If you like what you've heard and haven't subscribed, please subscribe on your favorite podcast platform. We'd love to hear from you, so feel free to reach out! You can reach us once again at hello@amply.energy. Thanks a lot.