Modeling Is Not a Four-Letter Word
Energy modeling has gotten a bad reputation in the home performance world. One conference I’ve attended has gone so far as to say that it’s “outside the sandbox” of topics presenters can cover. They want to see data, not modeled results. And they have good reason for that.
Energy modeling has gotten a bad reputation in the home performance world. One conference I’ve attended has gone so far as to say that it’s “outside the sandbox” of topics presenters can cover. They want to see data, not modeled results. And they have good reason for that.
The 2 kinds of physicists
Coming from the world of physics, I have a different perspective. (OK, I had a different perspective even before the book Asimov on Physics opened my eyes to the beauty of the universe when I was 17.)
In physics, modeling is essential. In fact, if you go to any physics department, you’ll find one of the two kinds of people: those who can extrapolate from incomplete data.
But there’s another two kinds as well: the theorists and the experimentalists. They need each other.
Without experimentalists, theorists would go completely off the deep end. (Some say they have with string theory.) Think of Aristotle here, with his peripatetic scholars relying only on logic to find explanations for physical phenomena and never bothering to test their ideas. It took nearly 2000 years for Galileo to usher in the era of modern science by rolling balls down inclined planes.
The harmony of modeling and data
Albert Einstein was the consummate theorist. He developed the special theory of relativity by imagining what it would be like to move along with a beam of light at the same speed the light was traveling. That was his idea of doing an experiment, and it even has a name: Gedankenexperiment, which means thought experiment in German.
Einstein’s general theory of relativity was his expansion of special relativity. He published it in 1915, and that’s when the idea of the curvature of space-time was born. An important fact about this theoretical paper was that he based it on the available data. For example, it explained an anomaly in the orbit of Mercury.
It also was testable. Four years later, Eddington confirmed the curvature predictions when he found a deflection of starlight around the Sun during an eclipse.
On the other hand, he considered the “biggest blunder” of his life to be the introduction of an unnecessary “cosmological constant” simply to conform to the prevailing idea of a static universe. Had he believed his equations instead, he could have hypothesized the expanding universe 14 years before Hubble discovered it.
That’s how science works. Ideas get thrown out there. They get tested by experiments. One negative result can disprove a hypothesis. No amount of data can ever fully prove it, but the more data you have, the more confident you can be in the validity of the idea.
The problem with energy modeling
Now, physics is a science. Home energy retrofits rely on science but are not science themselves. The late Phil Jeffers, an occasional commenter here, used to complain about turning home energy audits and retrofits into science projects. He had a good point.
It’s easy to go too far with modeling, and Michael Blasnik has exposed the flaws with energy modeling. He’s looked at program results in Minneapolis, Oregon, California, and other places and found that most modeling overpredicts the savings, sometimes unrealistically so. No matter how good a home performance contractor is, for example, they’re never going to cut someone’s energy bills by 125%. (You can download the pdf file of his 2013 Building Science Summer Camp presentation, Lies, Damned Lies, and Modeling.)
Likewise, John Proctor recently said, “We don’t need an energy model to tell us that an uninsulated house needs to be insulated and a leaky house needs to be sealed. Just fix it!”
Is modeling useful?
Joe Lstiburek has also been critical of modeling over the past few years, especially hygrothermal modeling with tools like WUFI. His company, Building Science Corporation, does WUFI analyses, and he’s open about when it should be done and when it shouldn’t. “I’m hoping two-thirds of the modeling that’s being done now won’t need to be done,” said Lstiburek at the 2013 BSC Experts Session, “and the modeling that’s needed is done correctly.”
Lstiburek was talking about hygrothermal modeling, mainly for new construction projects with assemblies that don’t have much of a track record. Think R-40 truss walls and R-60 insulated rooflines. Doing some modeling ahead of time can help avoid costly mistakes.
Jeffers, Blasnik, and Proctor were talking mainly about existing homes. The problems are usually obvious, as in the photo above, and we’ve got several decades of experience in weatherization and home performance contracting to help guide us in fixing them.
The problem comes in with programs that require modeling so the program sponsor can justify the expenditures. We could spend a long time discussing this issue and how to fix it, but this article has already gone on far longer than I had intended and is threatening to suck up the rest of my day the way a black hole sucks up everything, including light, that gets too close. (By the way, black holes are another cool thing that came out of Einstein’s general relativity!)
So let me conclude by going back to the title and saying that modeling is not a four-letter word. We need modeling. And we need real data from monitoring projects. We also need to keep it all in perspective and keep the focus on the results.
Here’s another perspective: What good would physics be without modeling?
Related Articles
Should We Change the HERS Reference Home’s Energy Code?
5 Easy Steps to Understanding WUFI – Memento Style
How Accurate Is REM/Rate as an Energy Modeling Tool?
Photo of spacetime curvature from Wikimedia Commons, used under a GNU Free Documentation License.
NOTE: Comments are moderated. Your comment will not appear below until approved.
This Post Has 40 Comments
Comments are closed.
The main problem with most
The main problem with most models is they are never correlated to reality. If we can get the industry to establish feedback loops using actual post utility data to true up predictions, and apply that to future work, models would move from 4 letter words (lies!) toward being very useful. Perhaps we should clean up the sandbox to say that correlated models are within the topic discussion box? 🙂
@Mike – I wholeheartedly
@Mike – I wholeheartedly agree. Modeling and retrofits work well together, BUT they have to be trued up AND go back after the fact to watch usage to see if it came in as predicted. That’s how you figure out what worked and what didn’t. <br /> <br />We make predictions of energy savings, then watch to see if they happen. We don’t win all the time, but we learn every time! <br /> <br />It’d be good to open up that sandbox a bit… =) I ought to have more results for next year.
modeling is an eight letter
modeling is an eight letter word and coincidentally so is BS! However, another great article doctor.
Another excellent, thought
Another excellent, thought-provoking article, Allison! <br /> <br />As an old, retired guy however, I also appreciate the entertainment value of your scribblings. Provoking thought, on the other hand, makes my head hurt :). <br /> <br />Back in the days of the Carter Administration, the Energy Services Department at PEPCO in DC, provided simple walk-through audits and "what if" evaluations of a variety of common energy reduction options for all classes of customers. The "simulations" were run on a PC and the first part of the process involved inputting measured building (or home, in another program) characteristics and tweaking the base model till kWh, kW and gas use fell roughly into conformance with available records, commonly the last 12 months. <br /> <br />We had no illusions that the numbers we were generating bore anything more than a "manufactured" resemblance to the buildings we were simulating and appreciated that the real world is simply too messy to be captured by strings of ones and zeros from the simple programs we used. <br /> <br />What we felt to be meaningful from our efforts was the difference between the base model and the options we presented in our reports (lighting retrofit, temperature changes, operating schedules, etc.). On rare occasions we had an opportunity to evaluate the results of our advice and, while the numbers were generally positive, I recall the crestfallen look of a vendor who had changed a building to VAV, with no dramatic change in utility cost. Again, this was based on a simple, uncorrected (we didn’t know how) evaluation of year to year numbers. (BTW, our weather data was primitive and consisted of aggregate data for the past 10 years). <br /> <br />Hmm, there’s a point here somewhere. <br /> <br />While extant processing power and software is so far ahead of what I used, I still contend, and suspect everyone would agree, the results of contemporary simulations are humanity’s feeble attempt to put numbers on what is an exceedingly messy world, one that adamantly refuses to hold still long enough to be accurately and consistently quantified. <br /> <br />I’ll conclude by agreeing with Dr. Bailes’ final observation, (which for reader’s convenience I attempted to copy/paste here, but hey, just scroll up :). <br /> <br />All the best. <br /> <br />Allison, can you italicize "difference" above for me? <br /> <br /> <br />
It seems to me the science
It seems to me the science isn’t really there to model what happens when you violate a rule. Therefore before-and-after modeling is severely impaired. <br /> <br />Industry experts can tell you what to do, but it’s much harder to discuss with them what will be the result when you do otherwise. For some areas where the building code has changed, there should be respect that *some* buildings will not suffer following the old rules. These levels of detail are not well understood as I see it. <br /> <br />I admire modeling, just wish it were better at existing buildings.
John Proctor recently said,
John Proctor recently said, "We don’t need an energy model to tell us that an uninsulated house needs to be insulated and a leaky house needs to be sealed. Just fix it!" <br /> <br />I suspect he has no clue what it takes to sell a Home Performance project. The idea of simply going into a home and telling people "I know what you need, just write me a check for $20,000" seems to indicate a serious disconnect from the reality of the marketplace. <br /> <br />People simply don’t just spend money because you tell them to. They want some due diligence, they want proof. They want to understand what will things will cost. They want some clarity about what the benefits will be. They really want to know NET COST. <br /> <br />"A better home" is NOT going to pry 5 figures from people’s wallets for long. And the approach of simply telling people what to do without any skin in the game of the result is NOT a path to greater quality, just the opposite in fact. <br /> <br />Modeling is due dilligence. It uncovers opportunity, and allows cost benefit analysis of various measures, and comparison of various approaches against each other. It allows understanding of total cost and net cost. It allows tailoring improvement projects to meet consumer needs and budgets, and it allows leveraging EE savings into the project budget. Larger projects improve likelihood of satisfying outcomes. <br /> <br />John Proctor should stick to studying duct work. <br /> <br />" Michael Blasnik has exposed the flaws with energy modeling. He’s looked at program results in Minneapolis, Oregon, California, and other places and found that most modeling overpredicts the savings, sometimes unrealistically so. No matter how good a home performance contractor is, for example, they’re never going to cut someone’s energy bills by 125%. " <br /> <br />Absolutely correct. But "Modeling is not accurate, therefore the software is the problem" is shockingly simplistic thinking. Jumping to the conclusion that the problem is modeling rather than incentive and accountability structure surrounding process is jumping to conclusions without investing time in critical thinking. <br /> <br />If there is no reward for accurate modeling, and considerable disincentive for it, in what world do you think you would ever get accurate models? If there was extrinsic reward for accurate modeling don’t you think you might get accurate modeling?
The main problem with most
The main problem with most models is they are never correlated to reality. If we can get the industry to establish feedback loops using actual post utility data to true up predictions, and apply that to future work, models would move from 4 letter words (lies!) toward being very useful. Perhaps we should clean up the sandbox to say that correlated models are within the topic discussion box? 🙂
Unfortunately no one wants to
Unfortunately no one wants to play in a cleaned up sandbox. Post savings typically don’t align well with modeled results, such that calibrating the model throws all cost-effectiveness (in the traditional sense) out the window unless NEBs are incorporated into the equation. Some of the lost savings can be assigned to occupant behavior, or the "comfort take-back" effect. In such a case, were the savings achieved? If not, then were the upgrades/retrofit measures not "worth it"?
@Mike – I wholeheartedly
@Mike – I wholeheartedly agree. Modeling and retrofits work well together, BUT they have to be trued up AND go back after the fact to watch usage to see if it came in as predicted. That’s how you figure out what worked and what didn’t.
We make predictions of energy savings, then watch to see if they happen. We don’t win all the time, but we learn every time!
It’d be good to open up that sandbox a bit… =) I ought to have more results for next year.
modeling is an eight letter
modeling is an eight letter word and coincidentally so is BS! However, another great article doctor.
Well, according to your 3rd
Well, according to your 3rd paragraph I’m a physicist of the 1st order. Or maybe I misinterpreted that. <br /> <br />A very well presented overview of the pro and con factions of the modeling discussion. The problem usually isn’t the modeling, it’s the presentation. I’m always careful to include many and varied "YMMV" disclaimers when I present modeling results. As soon as you put numbers on a page people take them as facts. Unfortunately the way modeling results are presented (often in various efficiency programs) it would seem to be a guarantee. I modify the output results to be a range below and above the program results and then still add disclaimers. People want to believe numbers. <br /> <br />If possible I leave the promises to the installer. I’m still looking for software that will let me control other peoples quality. 😉
Another excellent, thought
Another excellent, thought-provoking article, Allison!
As an old, retired guy however, I also appreciate the entertainment value of your scribblings. Provoking thought, on the other hand, makes my head hurt :).
Back in the days of the Carter Administration, the Energy Services Department at PEPCO in DC, provided simple walk-through audits and “what if” evaluations of a variety of common energy reduction options for all classes of customers. The “simulations” were run on a PC and the first part of the process involved inputting measured building (or home, in another program) characteristics and tweaking the base model till kWh, kW and gas use fell roughly into conformance with available records, commonly the last 12 months.
We had no illusions that the numbers we were generating bore anything more than a “manufactured” resemblance to the buildings we were simulating and appreciated that the real world is simply too messy to be captured by strings of ones and zeros from the simple programs we used.
What we felt to be meaningful from our efforts was the difference between the base model and the options we presented in our reports (lighting retrofit, temperature changes, operating schedules, etc.). On rare occasions we had an opportunity to evaluate the results of our advice and, while the numbers were generally positive, I recall the crestfallen look of a vendor who had changed a building to VAV, with no dramatic change in utility cost. Again, this was based on a simple, uncorrected (we didn’t know how) evaluation of year to year numbers. (BTW, our weather data was primitive and consisted of aggregate data for the past 10 years).
Hmm, there’s a point here somewhere.
While extant processing power and software is so far ahead of what I used, I still contend, and suspect everyone would agree, the results of contemporary simulations are humanity’s feeble attempt to put numbers on what is an exceedingly messy world, one that adamantly refuses to hold still long enough to be accurately and consistently quantified.
I’ll conclude by agreeing with Dr. Bailes’ final observation, (which for reader’s convenience I attempted to copy/paste here, but hey, just scroll up :).
All the best.
Allison, can you italicize “difference” above for me?
It seems to me the science
It seems to me the science isn’t really there to model what happens when you violate a rule. Therefore before-and-after modeling is severely impaired.
Industry experts can tell you what to do, but it’s much harder to discuss with them what will be the result when you do otherwise. For some areas where the building code has changed, there should be respect that *some* buildings will not suffer following the old rules. These levels of detail are not well understood as I see it.
I admire modeling, just wish it were better at existing buildings.
John Proctor recently said,
John Proctor recently said, “We don’t need an energy model to tell us that an uninsulated house needs to be insulated and a leaky house needs to be sealed. Just fix it!”
I suspect he has no clue what it takes to sell a Home Performance project. The idea of simply going into a home and telling people “I know what you need, just write me a check for $20,000” seems to indicate a serious disconnect from the reality of the marketplace.
People simply don’t just spend money because you tell them to. They want some due diligence, they want proof. They want to understand what will things will cost. They want some clarity about what the benefits will be. They really want to know NET COST.
“A better home” is NOT going to pry 5 figures from people’s wallets for long. And the approach of simply telling people what to do without any skin in the game of the result is NOT a path to greater quality, just the opposite in fact.
Modeling is due dilligence. It uncovers opportunity, and allows cost benefit analysis of various measures, and comparison of various approaches against each other. It allows understanding of total cost and net cost. It allows tailoring improvement projects to meet consumer needs and budgets, and it allows leveraging EE savings into the project budget. Larger projects improve likelihood of satisfying outcomes.
John Proctor should stick to studying duct work.
” Michael Blasnik has exposed the flaws with energy modeling. He’s looked at program results in Minneapolis, Oregon, California, and other places and found that most modeling overpredicts the savings, sometimes unrealistically so. No matter how good a home performance contractor is, for example, they’re never going to cut someone’s energy bills by 125%. ”
Absolutely correct. But “Modeling is not accurate, therefore the software is the problem” is shockingly simplistic thinking. Jumping to the conclusion that the problem is modeling rather than incentive and accountability structure surrounding process is jumping to conclusions without investing time in critical thinking.
If there is no reward for accurate modeling, and considerable disincentive for it, in what world do you think you would ever get accurate models? If there was extrinsic reward for accurate modeling don’t you think you might get accurate modeling?
Unfortunately no one wants to
Unfortunately no one wants to play in a cleaned up sandbox. Post savings typically don’t align well with modeled results, such that calibrating the model throws all cost-effectiveness (in the traditional sense) out the window unless NEBs are incorporated into the equation. Some of the lost savings can be assigned to occupant behavior, or the “comfort take-back” effect. In such a case, were the savings achieved? If not, then were the upgrades/retrofit measures not “worth it”?
Hey Allison, <br
Hey Allison, <br /> <br />Always a pleasure to hang out with you. I think modeling has a place, but mostly is not invited into the sandbox because it represents a distraction when unvalidated models are used to justify anything. I think that models where variables are minimal and the most important variables are accurately represented would be more likely to correlate with reality. Many consider air leakage to be an important variable, and the best way to tighten the gap between reality and model is to control air leakage to the point where it is no longer a dominate factor influencing the model, which is what the passive house folks do. So what do you think is a good input for air leakage when you don’t control it to that point, and how do you input in the model that air leakage to and from high delta T locations is not the same as controlling infiltration or ex filtration from lower delta T (or delta P) locations. I would suggest that if modeling is to be used in residential retrofits it has to prove its value. Homes that used x energy model saved x more dollar than homes that did not use x energy model. Has anyone sat across from a home owner and said I would like to spend 20,000 to fix some known defects in your home, but I want to first spend 1000 of your money and another couple thousand dollars of collective money (all that incentive money has to come from somewhere right?) to justify what I am about to do. I agree with Proctor if you have a leaky attic the money that you would have spent on modeling weather or not it’s a good idea to fix it would be better spent on air sealing the attic. We would expect a return on comfort,energy, and building durability from the the 1000 spent on air sealing, we would expect 1000 less dollars to spend on any energy improvements if we spent limited budget on modeling. The number I used is not important just a reference. I think models have their due in figuring out complex problems that testing is unlikely to occur in, understanding black holes might be a good example. Why is it that something that is all ready built, with known occupants and a record of habit is considered difficult to model and somehow it is considered easier to model a house that has never been built with occupants that have never lived in it is classified as an easier proposition? I think that modeling would be made more accurate if there were stakes involved. If modeling was sitting at the poker table and had to decide to anti up for all of it’s wagers how many times would modeling say i’m all in, I have faith in my bet and I am willing to put my money where my mouth is? By the way thanks for building a fun sandbox. <br /> <br />Best Gavin
Modeling is a synonym for
Modeling is a synonym for statistic and is used mostly in a rhetorical tautology argument. Modeling has no logical place in physics.
Well, according to your 3rd
Well, according to your 3rd paragraph I’m a physicist of the 1st order. Or maybe I misinterpreted that.
A very well presented overview of the pro and con factions of the modeling discussion. The problem usually isn’t the modeling, it’s the presentation. I’m always careful to include many and varied “YMMV” disclaimers when I present modeling results. As soon as you put numbers on a page people take them as facts. Unfortunately the way modeling results are presented (often in various efficiency programs) it would seem to be a guarantee. I modify the output results to be a range below and above the program results and then still add disclaimers. People want to believe numbers.
If possible I leave the promises to the installer. I’m still looking for software that will let me control other peoples quality. 😉
Seems some here would attempt
Seems some here would attempt to build a home without architectural plans! The argument "just get started fixing, don’t bother putting together a plan" wouldn’t work when building homes? <br /> <br />Why does not having a clearly thought out plan when spending others peoples money seem so OK to people in this "profession"? <br /> <br />"What do we need load calcs for, all that pointless measuring and modeling… We’ll just best guess, put in 2 stage equipment and save us both headaches." <br /> <br />"Measure duct leakage? What a stupid thing to do. Just seal it, think of the savings!!" <br /> <br />"Measure the 2 x 4 before cutting it? Why bother!" <br /> <br /> <br />No wonder consumers have less trust in us than used car salespeople. In the rare occasion integrity exists, its not accompanied by any critical thinking skills whatsoever. <br /> <br />We have more and more people hiring us just for building science education, modeling, and work-scope. Often after having some rip off artist "try to sell them insulation and air sealing." Sometimes we see these people AFTER they’ve been ripped off, and we have to UNDO and REDO what was done. <br /> <br />Without the model where is the commitment or accountability for results? <br /> <br />I’ve never seen so many people so committed to avoiding accountability as I’ve seen in this industry. It’s really disgusting.
Hey Allison,
Hey Allison,
Always a pleasure to hang out with you. I think modeling has a place, but mostly is not invited into the sandbox because it represents a distraction when unvalidated models are used to justify anything. I think that models where variables are minimal and the most important variables are accurately represented would be more likely to correlate with reality. Many consider air leakage to be an important variable, and the best way to tighten the gap between reality and model is to control air leakage to the point where it is no longer a dominate factor influencing the model, which is what the passive house folks do. So what do you think is a good input for air leakage when you don’t control it to that point, and how do you input in the model that air leakage to and from high delta T locations is not the same as controlling infiltration or ex filtration from lower delta T (or delta P) locations. I would suggest that if modeling is to be used in residential retrofits it has to prove its value. Homes that used x energy model saved x more dollar than homes that did not use x energy model. Has anyone sat across from a home owner and said I would like to spend 20,000 to fix some known defects in your home, but I want to first spend 1000 of your money and another couple thousand dollars of collective money (all that incentive money has to come from somewhere right?) to justify what I am about to do. I agree with Proctor if you have a leaky attic the money that you would have spent on modeling weather or not it’s a good idea to fix it would be better spent on air sealing the attic. We would expect a return on comfort,energy, and building durability from the the 1000 spent on air sealing, we would expect 1000 less dollars to spend on any energy improvements if we spent limited budget on modeling. The number I used is not important just a reference. I think models have their due in figuring out complex problems that testing is unlikely to occur in, understanding black holes might be a good example. Why is it that something that is all ready built, with known occupants and a record of habit is considered difficult to model and somehow it is considered easier to model a house that has never been built with occupants that have never lived in it is classified as an easier proposition? I think that modeling would be made more accurate if there were stakes involved. If modeling was sitting at the poker table and had to decide to anti up for all of it’s wagers how many times would modeling say i’m all in, I have faith in my bet and I am willing to put my money where my mouth is? By the way thanks for building a fun sandbox.
Best Gavin
Modeling is a synonym for
Modeling is a synonym for statistic and is used mostly in a rhetorical tautology argument. Modeling has no logical place in physics.
Seems some here would attempt
Seems some here would attempt to build a home without architectural plans! The argument “just get started fixing, don’t bother putting together a plan” wouldn’t work when building homes?
Why does not having a clearly thought out plan when spending others peoples money seem so OK to people in this “profession”?
“What do we need load calcs for, all that pointless measuring and modeling… We’ll just best guess, put in 2 stage equipment and save us both headaches.”
“Measure duct leakage? What a stupid thing to do. Just seal it, think of the savings!!”
“Measure the 2 x 4 before cutting it? Why bother!”
No wonder consumers have less trust in us than used car salespeople. In the rare occasion integrity exists, its not accompanied by any critical thinking skills whatsoever.
We have more and more people hiring us just for building science education, modeling, and work-scope. Often after having some rip off artist “try to sell them insulation and air sealing.” Sometimes we see these people AFTER they’ve been ripped off, and we have to UNDO and REDO what was done.
Without the model where is the commitment or accountability for results?
I’ve never seen so many people so committed to avoiding accountability as I’ve seen in this industry. It’s really disgusting.
Ted, your rants about not
Ted, your rants about not measuring have nothing to do with the reasons that modeling so often turns out for the worst in the HVAC industry. Of course we have to measuere, but modeling uses too many hypothetical measurements to not be subjected to being corrupted. <br /> <br />Maybe modeling isn’t a four letter word, but it is certainly often described by four letter words. Take for instance load calcs, which when done honestly are a lot less modeling and more physics oriented calculation tools. So, a contractor in New Jersey, let’s just pick NJ becauase this is where I ran across this incident in real life. So, this contractor does a load calc (mandatory in NJ) based on known parameters of data in his area. The load calc comes up with 28,530 Btu heat load and 59,980 Btu heat loss. This heat load/loss difference is about the average ratio for this area. So, the contractor looks for an AHRI matching system to quote for installation, but the HO is requiring the system qualify for certain rebates available in NJ. The contractor cannot find a 2.5 ton cooling system that will meet the rebate qualifications with any 70k Btu furnace. In order to meet the requirements for the rebate, the contractor must go to a 90k Btu furnace. The contractor tries to explain to the HO that it is better to have a less effiency "rated" system installed with the proper sized furnace, but the HO wants that rebate and wants what is perceived to be the more efficient system because government regulations deem it to be more efficient. <br /> <br />To keep from losing the job, the contractor changes the parameters of heat loss on the load calc so that the load calc justifies installing the rebate qualifying system. The contractor gets the job, installs it to perfection, and there are problems with the heating. This is when I am called in. <br /> <br />Turns out the duct system was just barely sized to provide 1,000 cfm of air at the rated SP for cooling, as long as a standard, low efficiency fiberglass filter is used. Since a 90k condensing furnace requires approximately 1350 CFM of air to maintain a desired temperature rise, there is going to be overtemperature issues. <br /> <br />SO; in the case of load calcs, as a properly calibrated calculation tool, they are great, but, as a modeling device, they suck (four letter word).
Allison, <br /
Allison, <br />Great post, again! Thanks. <br /> <br />I thought confidence was always the problem, not modeling. (Or, was that causality?) <br /> <br />Foster
If we truly dissect the
If we truly dissect the causality of issues as they pertain to modeling, I suppose it really is a lack of confidence that ends up perverting the model just to make a sale. <br /> <br />As a contractor who won most bids and worked way too much to earn a living, I learned it is much better to lose jobs that are going to cost you in the long run, because the jobs you win on your terms are going to provide a decent living without having to work as hard.
Ted, your rants about not
Ted, your rants about not measuring have nothing to do with the reasons that modeling so often turns out for the worst in the HVAC industry. Of course we have to measuere, but modeling uses too many hypothetical measurements to not be subjected to being corrupted.
Maybe modeling isn’t a four letter word, but it is certainly often described by four letter words. Take for instance load calcs, which when done honestly are a lot less modeling and more physics oriented calculation tools. So, a contractor in New Jersey, let’s just pick NJ becauase this is where I ran across this incident in real life. So, this contractor does a load calc (mandatory in NJ) based on known parameters of data in his area. The load calc comes up with 28,530 Btu heat load and 59,980 Btu heat loss. This heat load/loss difference is about the average ratio for this area. So, the contractor looks for an AHRI matching system to quote for installation, but the HO is requiring the system qualify for certain rebates available in NJ. The contractor cannot find a 2.5 ton cooling system that will meet the rebate qualifications with any 70k Btu furnace. In order to meet the requirements for the rebate, the contractor must go to a 90k Btu furnace. The contractor tries to explain to the HO that it is better to have a less effiency “rated” system installed with the proper sized furnace, but the HO wants that rebate and wants what is perceived to be the more efficient system because government regulations deem it to be more efficient.
To keep from losing the job, the contractor changes the parameters of heat loss on the load calc so that the load calc justifies installing the rebate qualifying system. The contractor gets the job, installs it to perfection, and there are problems with the heating. This is when I am called in.
Turns out the duct system was just barely sized to provide 1,000 cfm of air at the rated SP for cooling, as long as a standard, low efficiency fiberglass filter is used. Since a 90k condensing furnace requires approximately 1350 CFM of air to maintain a desired temperature rise, there is going to be overtemperature issues.
SO; in the case of load calcs, as a properly calibrated calculation tool, they are great, but, as a modeling device, they suck (four letter word).
Allison,
Allison,
Great post, again! Thanks.
I thought confidence was always the problem, not modeling. (Or, was that causality?)
Foster
If we truly dissect the
If we truly dissect the causality of issues as they pertain to modeling, I suppose it really is a lack of confidence that ends up perverting the model just to make a sale.
As a contractor who won most bids and worked way too much to earn a living, I learned it is much better to lose jobs that are going to cost you in the long run, because the jobs you win on your terms are going to provide a decent living without having to work as hard.
Why is it we so often see
Why is it we so often see modelling blamed for problems when perverse incentive is at fault? <br /> <br />Seems only light weight critical thinking skills would be required to understand modelling isn’t to blame when a contractor installs a grossly oversized furnace, at the encouragement & blessing of program of homeowner and program. <br /> <br />But you find contractor, homeowner, and program, all of which are fully to blame, looking for a scapegoat, and a bunch of lazy illogical sheep thinkers willing to accept that line of thinking as reasonable. <br /> <br />"Baaa-aaaa-aaaa"
LOL, Ted. You are feisty on
LOL, Ted. You are feisty on this one. But as you should know by now, especially since you and I agree more often than not, I am certainly no sheep. Unfortunately, consumers and too many contractors are. They are sheep who are willing to follow bad modeling data right into the realm of high energy bills, warrany claims and uncomfortable homes. <br /> <br />Maybe it is not the fault of modeling in and of itself that is to blame, but it is the ability to pervert the data in the model that allows modeling to become a very deceptive way of doing business. <br /> <br />With any modeling, as with statistics, anyone can justify anything they want. How is this not something that is going be referred to with the use of four letter words? <br /> <br />All AHRI system ratings are based on models in which the data imput can range from data used in Maine, Key West, Salt Lake City, Honolulu, Fairbanks AK, Brownsville TX and every point between these extremes. Unfortunately, consumers and even contractors too often believe that if a system is AHRY rated, based on modeling, it must be ok to install anywhere.
Why is it we so often see
Why is it we so often see modelling blamed for problems when perverse incentive is at fault?
Seems only light weight critical thinking skills would be required to understand modelling isn’t to blame when a contractor installs a grossly oversized furnace, at the encouragement & blessing of program of homeowner and program.
But you find contractor, homeowner, and program, all of which are fully to blame, looking for a scapegoat, and a bunch of lazy illogical sheep thinkers willing to accept that line of thinking as reasonable.
“Baaa-aaaa-aaaa”
LOL, Ted. You are feisty on
LOL, Ted. You are feisty on this one. But as you should know by now, especially since you and I agree more often than not, I am certainly no sheep. Unfortunately, consumers and too many contractors are. They are sheep who are willing to follow bad modeling data right into the realm of high energy bills, warrany claims and uncomfortable homes.
Maybe it is not the fault of modeling in and of itself that is to blame, but it is the ability to pervert the data in the model that allows modeling to become a very deceptive way of doing business.
With any modeling, as with statistics, anyone can justify anything they want. How is this not something that is going be referred to with the use of four letter words?
All AHRI system ratings are based on models in which the data imput can range from data used in Maine, Key West, Salt Lake City, Honolulu, Fairbanks AK, Brownsville TX and every point between these extremes. Unfortunately, consumers and even contractors too often believe that if a system is AHRY rated, based on modeling, it must be ok to install anywhere.
Ted Your comments are
Ted Your comments are offensive, I will not defend myself against stuff. Please make you point without attacking someone else. Marketing home performance is difficult and the size of the market is small (people who will part with that amount of money). I think everyone who has been in the business very long knows it. Nevertheless there is a problem with a system that loads on a lot of modeling overhead. That overhead drives up the cost and lowers the size of the market.
Ted Your comments are
Ted Your comments are offensive, I will not defend myself against stuff. Please make you point without attacking someone else. Marketing home performance is difficult and the size of the market is small (people who will part with that amount of money). I think everyone who has been in the business very long knows it. Nevertheless there is a problem with a system that loads on a lot of modeling overhead. That overhead drives up the cost and lowers the size of the market.
"the ability to pervert
"the ability to pervert the data in the model that allows modeling to become a very deceptive way of doing business. " <br /> <br />John, glad you came around on that one. <br /> <br />That there is much perverse incentive to exaggerate savings projections, and effectively NO counterbalancing accountability or incentive to tell the truth, is something I can definitely agree with you on. <br /> <br />That the butcher puts his thumb on the scale does not make the scale bad, nor mean we should throw away the scale. We must create reason to remove the thumb.
“the ability to pervert
“the ability to pervert the data in the model that allows modeling to become a very deceptive way of doing business. ”
John, glad you came around on that one.
That there is much perverse incentive to exaggerate savings projections, and effectively NO counterbalancing accountability or incentive to tell the truth, is something I can definitely agree with you on.
That the butcher puts his thumb on the scale does not make the scale bad, nor mean we should throw away the scale. We must create reason to remove the thumb.
Scales are based on precise
Scales are based on precise known measurements, models are not. Models are subjective from design to first abuse, which in many cases is the initial design. <br /> <br />I think a main issue here is what we are individually defining as models versus proper measurements based on known data. Manipulating a manual J by going outside of the parameters for the specific region is akin to putting a thumb on the scale. Basing a system efficiency on some hypothetic number during some assumed length of time.
Scales are based on precise
Scales are based on precise known measurements, models are not. Models are subjective from design to first abuse, which in many cases is the initial design.
I think a main issue here is what we are individually defining as models versus proper measurements based on known data. Manipulating a manual J by going outside of the parameters for the specific region is akin to putting a thumb on the scale. Basing a system efficiency on some hypothetic number during some assumed length of time.
Allison, <br /
Allison, <br /> <br />You’ve reminded me of that episode of "Big Bang Theory" where our mutual hero, Sheldon Cooper, exclaims: "Engineering is just the younger, slower brother of the theoretical sciences. Now, somebody help me figure out how to open this toolbox…". (I think it was the fighting robots episode…) <br /> <br />Seriously, though: I’m of the opinion that energy modeling is essentially hypothesis building; or more precisely, the creation of a particular instance of an hypothesis. And sure, you absolutely should confirm it with experimental data. <br /> <br />But I believe the problem with that are the practical challenges to setting up and managing a follow-on program of data collection and analysis. <br /> <br />As I’d mentioned to you, I’ve been collecting passive performance data from several properties for well over a year now. But these are properties I either control or otherwise have ready access to. <br /> <br />I could see where private homeowners might be less than thrilled by the idea of instrumenting their homes for a year of ongoing data collection after you’ve completed construction or retrofit work. Unless they’re really engaged in the process and you can convince them of the value of doing so upfront. <br /> <br />~ John
Allison,
Allison,
You’ve reminded me of that episode of “Big Bang Theory” where our mutual hero, Sheldon Cooper, exclaims: “Engineering is just the younger, slower brother of the theoretical sciences. Now, somebody help me figure out how to open this toolbox…”. (I think it was the fighting robots episode…)
Seriously, though: I’m of the opinion that energy modeling is essentially hypothesis building; or more precisely, the creation of a particular instance of an hypothesis. And sure, you absolutely should confirm it with experimental data.
But I believe the problem with that are the practical challenges to setting up and managing a follow-on program of data collection and analysis.
As I’d mentioned to you, I’ve been collecting passive performance data from several properties for well over a year now. But these are properties I either control or otherwise have ready access to.
I could see where private homeowners might be less than thrilled by the idea of instrumenting their homes for a year of ongoing data collection after you’ve completed construction or retrofit work. Unless they’re really engaged in the process and you can convince them of the value of doing so upfront.
~ John