Welcome! Here are the website rules, as well as some tips for using this forum.

# Possible solution for all those DIYers struggling with heat loss calculations

Options
Member Posts: 18

Hi All,

After a lot of frustration (and errors) trying to do my own heat loss calculations I recognized that there is likely a very handy tool out there to run these complex calculations in an extremely customizable way. And the best part is its free!

You may know it as ChatGPT. You can give the model as much or as little info as you have and it will tell you what else you need, or make clear what it is assuming on your behalf and lay it all out for review. You can iterate by changing the assumptions it is making on any variable.

Also helpful that you can provide it other context like the year range and municipality of construction, tree cover, etc and it will instantly adapt its calculations if its relevant.

I'd be interested to know if any of the veterans on here have given this a try and/or if there are any thoughts/reactions. AI models have been know to make stuff up, but this seems like a neat solution for a lot of DIYers.

Bryce

• Member Posts: 781
edited June 27
Options

As generative AI's are at best "unconcerned" with the truthfulness of statements they generate, I'd take a hard pass on that one.

Just suck it up and do the math.

Trying to squeeze the best out of a Weil-McLain JB-5 running a 1912 1 pipe system.
• Member Posts: 18
edited June 27
Options

Totally understand your sentiment, yet having done the math first, and then thinking trying the AI model, the math is highly accurate. My take is that its not a "statement" style ask/answer that gets AI into trouble, but an application of know mathematical formulas.

Happy to hear if someone has seen it go awry, but it was bang on for me having already "sucked it up".

Only question I had was that it multiplied all its calculations by 1.2 for "safety". 20% seems like a lot.

• Member Posts: 23,635
Options

2% is actually pretty conservative for a safety factor — since it is unlikely that any of the input information is good to better than that…

Br. Jamie, osb
Building superintendent/caretaker, 7200 sq. ft. historic house museum with dependencies in New England
• Member Posts: 298
Options

Just because ChatGPT is accurate in one case doesn't mean it will be accurate in all.

• Member Posts: 18
Options

Based on the way it is running the calculations and describing each step in detail its a case of garbage in - garbage out. Definitely worth double checking the U-values if you're asking it to make assumptions - had to correct its numbers for a slab-on-grade to get it extremely close to my own numbers. Also make sure you give it the correct delta T for your area and preference. Otherwise, its very transparent.

• Member Posts: 7,605
edited June 27
Options

Aside from the AI component, Any heat loss calc without a blower door test is a serious SWAG! Just take a look at how big the air infiltration number is compared to the whole.

"If you can't explain it simply, you don't understand it well enough"
Albert Einstein
• Member Posts: 18
Options

agreed. Its part of the calculation and the AI asks for it and advises against using the results without adding that value in. Again, garbage in - garbage out.

• Member Posts: 1,415
edited June 28
Options

Isn't this true for flesh and blood humans too? 🤔

• Member Posts: 781
edited June 28
Options

Fair enough, but at least humans generally try to generate true statements.

Trying to squeeze the best out of a Weil-McLain JB-5 running a 1912 1 pipe system.
• Member Posts: 301
Options

The work in a Manual J isn't doing the math, it's gathering the underlying information — measuring the building. The output is only going to be as good as the input.

• Member Posts: 22,472
Options

In older homes, 2 big assumptions are infiltration, and insulation quality.

So for a tighter, more accurate number a blower door test and infrared scan will help.

Bob "hot rod" Rohr
trainer for Caleffi NA
Living the hydronic dream
• Member Posts: 2,083
Options

Slant fin has a very nice heat loss calculation app. I'm pretty sure it's free and easy to use.

Has anyone used this or other apps?

It's been a while, but I had great success the old way with "the sheet" and pencil.

• Member Posts: 23,635
Options

I had used SlantFin's program in the past, and it was both easy to use and gave reliable results.

However…

It seems to have disappeared, and I suspect I know why: it — like all flavours of Manual J calculations — was very much affected by the quality of the input information, and I have a feeling that there was concern about untrained individuals using it, putting in over-optimistic information, getting results which were, as a result, in error, and then installing equipment which was incorrect. And then going after SlantFin. And corporate decided, quite reasonably, that that was a headache they didn't need.

Br. Jamie, osb
Building superintendent/caretaker, 7200 sq. ft. historic house museum with dependencies in New England
• Member Posts: 1,105
edited June 28
Options

I thought it disappeared since it was supported by Slant/Fin boiler which no longer exists. Slant/Fin baseboard has been a separate division under separate ownership for several years. Still comes up with a google search, but I haven't tried to download it. For a DIY you can download a 30 day trial of loopcad, or possibly a similar program. If you are going to use it for more than 30 days, or for professional use, you may as well purchase one. Manual J software is well worth the money if you are doing more than one per month in my opinion

• Member Posts: 2,040
Options

Just tried it and the numbers were crazy. I took two rooms. Everything exactly the same, including window size.. only difference was that one room was two and a half times the size. The two rooms hnd the exact same heat loss, according to GPT

• Member Posts: 18
Options

Look at the parameters and check the assumptions its making, it must have shifted an assumed value that you didn't specify. As noted above, it sometimes randomly changes the assumptions for heat loss of specific elements. The less you let it assume, the better it is as with all the different heat loss calculators I've tried.

It shows all its work, so you should be able to find where its gone wrong.

• Member Posts: 1,408
Options

In the future sizing will be different.

Grid aware heat pumps will get shut down for short (hopefully) periods of time to prevent mass power outages.

When the heat pumps are allowed to go back to work, they will have to play catch up.

So you may want to over size.

• Member Posts: 5,792
Options

Oversizing just in case something were to occur that has no basis in reality…interesting idea. Where to stop?

NJ Steam Homeowner.
Free NJ and remote steam advice: https://heatinghelp.com/find-a-contractor/detail/new-jersey-steam-help/
See my sight glass boiler videos: https://bit.ly/3sZW1el

• Member Posts: 1,408
edited June 28
Options

Sorry Paul. HPWHs are already grid aware. We recently had a discussion here about one that didn't go back to heating after it was told it could go back to heating.

What does have no basis in reality is how the great electrification is going to happen without causing power outage related deaths.

My son works at a Nuclear Plant on the East Coast. They recently experienced their highest temperatures. Usually doesn't happen until August. They, like solar PV, cut back their output on the hottest days. They are operating under conditions where they defer any tasks they can, that have a remote chance of causing a shutdown (SCRAM) of the plant. The reason? People DIE when the power goes out.

That's Reality!

• Member Posts: 5,792
Options

There are already power outage related deaths, ask Texas. PM me when you know someone personally who has had their home heating or cooling system shut down by the utility. I'll buy you a coffee.

NJ Steam Homeowner.
Free NJ and remote steam advice: https://heatinghelp.com/find-a-contractor/detail/new-jersey-steam-help/
See my sight glass boiler videos: https://bit.ly/3sZW1el

• Member Posts: 1,408
Options

I wasn't clear above.

In the future, because of Grid Enabled, it will no longer be possible to guarantee 70F indoor temp 24x7x365.

No matter how much time and money is spent on design, and no matter the heat loss tool.

Indoor temps are going to fluctuate.

The goal is to avoid wide spread extended power outages.

In the past, contractors oversized, to guarantee 70F 24x7x365. But they could count on uninterrupted (mostly) fuel.

• Member Posts: 1,408
edited June 28
Options

My brother had a Grid Enabled shutoff on his home AC 10 years ago. ComEd gave him 40 bucks a year for the privilege to shut down his home AC for short periods of time when needed. They did exactly that about 4 times.

Thanks Paul. I'll have a McDonald's Sr Coffee Black. I've been ordering those since I turned 40. I'm prematurely gray. So I would not make a good politician. (and too often I rub people the wrong way).

🇺🇲

• Member Posts: 1,408
Options

Re Texas. Pablo Vegas expects Texas power usage to DOUBLE in 5 years.

• Member Posts: 2,040
Options

I did it like three different times. I even had a conversation with them about the differences. Something to do with the fact that the smaller room had more windows relative to the volume. And regardless, not so sure I want to rely on something where I would constantly need to double and triple check to make sure they didn't change the parameters that I gave them

• Member Posts: 18
Options

Did you look at the actual calculations it spit out for each segment of heat loss? Those are definitely worth double checking. I know when I did my own manual calculations I double and triple checked everything and still made a few errors the first time through. No matter what method you're using, the rule of thumb is garbage in - garbage out.

Nice thing about doing it via AI is that you can just ask it to correct parameters and it reruns all the calcs within in a couple seconds.

Its not perfect, but if used carefully its a much more comprehensive tool than anything else I've found.

• Member Posts: 2,040
Options

Everything was exactly the same. Except the room size. For what it's worth, I asked how to size a replacement steam boiler and they said to do what he was calculation

• Member Posts: 18
Options

If everything was exactly the same it would have given the same values for heat loss. Its made a different assumption somewhere, you just have to suss it out. Ask it to show its work for all calculations. Compare them. You'll find the difference. I didn't run a test like you did, but for my calculations I asked it to show all calculations in detail for each room so I could sense check them and make sure it a) had all the right parameters in the formulas b) had included the data I gave it and not made any crazy assumptions and c) evaluated any assumptions I asked it to make/identified values it was including that I hadn't provided.

I've now tried it with three different AI models and its worked fine. The hardest to straighten out was the Microsoft Pilot one. Easiest was ChatGPT and Google Gemini was somewhere in the middle.

Considering I've managed to run three accurate (+/- 5%) heat loss calculations in what amounts to less than an hour, plus many many tweaks and variations/test cases for improving heat loss, I think its a great tool. Use with caution, but a great place to work these things through. A lot easier than the manual calculations and waaaaay easier to resolve any errors. Use with caution though.

• Member Posts: 2,040
Options

I did ask them to write a song about steam heating. Came out pretty nice. And when I asked for the best website for steam heating info, Heating Help was on the top of the list

• Member Posts: 18
Options

Thats amazing. You should post the song.

• Member Posts: 5,792
Options

OK no problem, but let it be known that your brother decided of his own free will to have his power regulated in exchange for cash. I was talking about without homeowner choice, although it's true I didn't state it.

NJ Steam Homeowner.
Free NJ and remote steam advice: https://heatinghelp.com/find-a-contractor/detail/new-jersey-steam-help/
See my sight glass boiler videos: https://bit.ly/3sZW1el

• Member Posts: 3,696
Options

My neighborhood was shed to keep the lights (or maybe AC?) on in another neighborhood during a recent outage. Does that count?

(Now, it's possible that it was a neighborhood that's poorer than mine, but the odds suggest otherwise.)

• Member Posts: 5,792
Options

what’s “shed”?

NJ Steam Homeowner.
Free NJ and remote steam advice: https://heatinghelp.com/find-a-contractor/detail/new-jersey-steam-help/
See my sight glass boiler videos: https://bit.ly/3sZW1el

• Member Posts: 23,635
Options

"shed" here refers to cutting power to a portion of the grid — or in some cases to individual customers or individual uses for a customer. At one time, way back when, that involved shutting off parts of the grid at substations or, much more rarely, at line switches (substations can be remotely controlled; line switches only rarely so). Now some large users which can tolerate interruptible power can be cut off by the power company individually. Increasingly, with smart meters, individual houses and users can be cut off remotely. Some users have uses — such as water heaters — with wifi control which the utility can shut off remotely. They usually pay a lower electricity rate for that use.

Increasingly I think we will see a drive for any wifi controlled device using significant power — water heaters, ranges, space heating or cooling — to be registered with the grid operators and used for load shedding. This would at least keep lights and more critical equipment operating, such as medical devices.

One interesting aspect of this is that the shutdown characteristics of the two main power sources — rotating machinery on the one hand and wind and solar on the other — are drastically different, and developing control strategies for them is still being worked on. Rotating machinery was pretty easy — on overload they slow, and for stability they were cut off (or load shed, preferably!) when the frequency dropped (typically 0.2 hertz, as I recall). Wind and solar are frequency controlled by the grid itself and the rotating machinery elsewhere, but their overload character is that of a circuit breaker — on overload, they cut off completely and instantly. If this isn't predicted and load shed to prevent it, and they are a significant fraction of the instantaneous grid power, this will throw an overload elsewhere, resulting in a cascading (and possibly total) failure. Which can get very interesting…

Br. Jamie, osb
Building superintendent/caretaker, 7200 sq. ft. historic house museum with dependencies in New England