HERS ratings & kBtu/sf/yr
Is there a direct relationship between HERS ratings and actual energy use, such as kBtus/sf/yr? Energy Star is gaining traction and making incrememtal improvements to their standards. Stretch Codes in Massachussetts pretty much make Energy Star the default route to comply. So the HERS rating is gaining traction for consumers. Its based on a scale with “0” being a Zero Net Energy Home, and 100 being a code-based home from 2003 or some such year. But I wonder how the HERS rating is related to energy use per square foot per year. Those are the standards that set by the 2030 Challenge: In 2010 buildings aim to use 60% less energy than the average for the building type in 2001. In 2030, the goal is 100% less energy use. For homes in the northeast the 2010 target is 18.3 kBtus/sf/yr. Is there an equation to relate that to a HERS rating?
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
There is little correlation between a HERS rating and actual energy consumption, and no simple formula is possible.
First, because HERS is based on assumptions about occupancy rates and "typical" building usage. Second, and more importantly, actual energy consumption is at least as much a function of lifestyle as it is the container of that lifestyle.
An occupant of the most efficient home can choose to crank up the thermostat and leave windows open for fresh air, or have lots of kids and dogs which go in and out all day long, or do a lot of cooking and baking, or leave lights on all the time (including outdoor security lights), or over-ride whatever clever control technologies are designed into the home.
On the other hand, an occupant of a less efficient home can choose to live frugally, wearing sweaters indoors and turning off all lights and appliance ghost loads when not in use.
For this reason, I have long advocated against using actual energy consumption to compare homes, since that really compares lifestyles. The only way to objectively compare the efficiency of a building is the way HERS does it - make some common assumptions about usage and create a metric for objective comparison which, like the EPA mileage stickers on new cars, won't necessarily reflect your actual consumption but allows fair comparison and ratings.
I think you can make a broadly valid correlation between HERS and Btu/sf/yr for a particular size of house in a particular HDD/CDD range. For instance, given a house of about 2500 sq ft in metropolitan Boston (5600/900), a HERS of 100 means a house that uses roughly 54K Btu/sf/yr--given a homeowner who uses the home the way REM/Rate thinks they will.
HERS is what's called an "asset rating"--meaning it reflects characteristics of the home independent of who's using it and how. Btu/sf/yr is an "operational rating"--meaning it shows actual energy usage. I have a hard time saying that one type of rating is more useful or more valid than the other; I find I use both a lot, depending on what sort of information I need--and how much money is in the planning budget: HERS ratings cost money to calculate; Btu/sf/yr "ratings" are virtually free to calculate, but still provide a lot of information.
There is no correlation whatsoever between HERS Index and even the HERS projected energy cost. let alone actual occupant energy costs. The Index is merely a comparison between the rated house and the identical house in size and configuration built to IECC standards. So a HERS 60 house in a particular climate will use 60% of the energy of the identical house built to energy code minimum standards. But a small HERS 60 house will use a lot less energy than a large HERS 60 house in the same area.
The energy cost predictions that are part of the HERS analysis is based on "typical" usage patterns for heat/AC, lights, appliances and hot water as well as on local utility costs at the time of rating. But studies have shown that occupant behavior can shift the energy consumption cost by a factor of 2, either up or down.
Lawrence Berkeley Lab did a study of the accuracy of HERS ratings and cost estimates in 1997:
http://eetd.lbl.gov/news-archives/pubs/40394.pdf
In 2007, the baseline for evaluating progress toward meeting the 2030 Challenge targets was established as the 2003 Commercial Building Energy Consumption Survey (CBECS) for commercial buildings and the Residential Energy Consumption Survey (RECS) for residential buildings. Although new building energy standards and rating systems that meet the 2030 Challenge as measured against these baselines are currently in development, they are not yet available. As a result, there is an immediate need and high demand for an interim system that enables cities, counties and states to meet the 2030 Challenge targets using existing building energy codes and standards as baselines.
Architecture 2030 has developed an interim system based on ‘code equivalents’, which are the additional reductions needed beyond the requirements of a particular code, standard or rating system to meet or exceed the initial 50% target of the 2030 Challenge. The paper also provides suggestions for ordinances that can be used to aid governments in amending their existing building code to incorporate these code equivalents.
Table A of the white paper show the most commonly used energy codes and standards and rating systems, including the HERS Index.
The white paper can be downloaded at http://www.architecture2030.org/pdfs/2030Challenge_Codes_WP.pdf
Thanks for the comments and links. The 2030 link is from 2008 and basically says that in lieu of better standards, a RESNET HERS value of 65 should achieve the (then) target of 50% energy reduction. For 2010, I assume the new 60% reduction target would be a HERS rating of about 50?
.
The HERS accuracy study link is from 1997 is a pretty thick read with some interesting points. It showed that the ratings were not very accurate in terms of energy use mostly because of variations such as: the size of home, number of occupants, temperature settings, cooking & laundry, appliances & electronics, time away from home, fuel type (including off-grid sources) and the 'take-back' phenomenon - homes with better scores tend to have higher usage. It also notes that accuracy is not a stated feature of HERS scores, but that does not seem negate the 'success' of the program. Yet the ratings might influence financing and purchase decisions, with the potential that consumers could be misled by equating HERS score with actual energy costs. It suggests that a Prescriptive system based on real energy-use data could be a more effective tool than the HERS Simulation-based ratings for pay-back decision-making on energy upgrade features.
More recent and climate zone specific studies would be good to have. Energy Star is also tweaking their requirements, and 2010 will bring in at least some recognition of house size into the ratings.
So then, I agree that both the HERS rating and Actual Energy Use are useful numbers, but they do not have a direct relationship. The HERS rating gives us a relative score, accurate enough in at least that respect, and points the way towards greater energy efficiency. But for individual consumers, the number that really hits home is the Actual Energy Use. And as individuals we'll find our own comfort level with consumption.
Except that "actual energy use" is far more a measure of the lifestyle of occupants than the quality of the house, or the future value of a present improvement. Objective and comparative measures have to exclude occupant idiosyncrasies and be based either on thermal properties alone or on some reasonable assumption of "normal" occupancy usage.
I have to disagree. Actual energy use -- especially when it comes to heating loads in cold climates or cooling loads in hot climates -- is mostly a function of the efficiency of the home and equipment, not behavior. Certainly the impacts can be large in some homes, but not in most homes.
For example, I looked at the actual electric use of about 2500 new homes in Phoenix and compared the cooling loads for the same homes in 2000 and 2004. There were 1384 homes with the same occupants and 1289 homes with differing occupants. The median change in cooling load between these years was 14% for the stayers and 21% for the movers -- not exactly an overwhelming impact.
I have also found that there is typically a fairly strong correlation between actual space conditioning loads and building characteristics in a wide range of climates and housing types. If there weren't, we'd all be wasting our time trying to improve insulation levels, air tightness and equipment efficiency in homes since occupancy is the only thing that matters...
Michael,
No one said "occupancy is the only thing that matters", and caricaturing the counter argument adds no credibility to yours.
Reporting only the median values in a very large statistical sample tells us next to nothing. My father, who taught statistics at the graduate level, used this simple example: If you lay across your kitchen table with your feet in the icebox and your head in the oven, on average you'd be perfectly comfortable.
Average values are meaningless without context. What is the shape of the curve? Where is it weighted? What is the standard deviation?
My point is that one building cannot be compared to another in terms of energy consumption if the occupancy patterns are dramatically different, and they can be by a significant factor. Other surveys have demonstrated this. But those same two houses can be objectively and accurately compared using identical occupancy assumptions, blower-door testing and fine-tuned computer analysis.
That remains the only valid method of comparison.
If this is what you meant to say with your other comments, then we are in agreement. But what you actually said is that actual energy use is "far more a measure of lifestyle of the occupants than the quality of the house". This latter statement is what I objected to.
Your knowledge of statistics is lacking -- the median of a large sample can be quite informative -- it all depends on your question. I cited the median absolute percent change because it is a fairly clear and very robust (statistically speaking) way to characterize the typical change in a value -- which is quite applicable here. In this context it means that half the homes with no occupancy change had usage changes of less than 14% and half had larger changes while for homes with different occupants half had changes of less than 21% and half had larger changes. There are certainly many statistics that can be used to describe and compare distributions But my point was about the typical magnitude of occupancy effects -- median is quite suitable for this task I wasn't arguing that extreme occupancy effects don't exist -- just that the typical effects are not as large as you claim.
I have some data to support my claim. Can you share with us some data supporting your claim that actual energy use is far more related to lifestyle than building quality? That doesn't mean citing the extremes, but the typical values. Can you explain why the median difference in cooling loads wasn't much bigger when occupancy changed? Can you explain to me how I found a correlation between floor area and cooling loads of 0.67 in a sample of more than 10,000 homes in Houston? Can you explain to me how cross-sectional regression modeling of actual energy use in large samples of homes can have regression R-squared values of 0.5 or larger if building quality is far less important than lifestyle? Wouldn't you expect a maximum R-squared of maybe 0.3 or less (depending on how you define "far more")? Please share some data...
Can you point me to research that shows that these models are accurate? I guess you haven't read many studies that have tried to compare computer models to energy use. You may want to start with a recent study done in Oregon that looked at measured and modeled energy use in a couple of hundred existing homes. see here http://www.earthadvantage.com/eps.php . The large bias implies that lifestyle isn't the problem as much as bad models..
I never suggested that energy modeling can predict energy consumption. What I has said, in fact, was quite the opposite: "There is little correlation between a HERS rating and actual energy consumption, and no simple formula is possible."
My point is that, for an objective rating of the energy efficiency of houses which allows some sort of comparative ranking, occupancy variables have to be eliminated (just as they are with MPG ratings for automobiles).
Neither I, nor most designers, builders or home-owners are interested in large-sample statistical averages but in how this particular house design compares with that particular house design, or how this house compares with a code-minimum house.
But your statistics demonstrate that change of occupancy had a 50% greater impact on energy consumption changes than with unchanged occupancy. That's a dramatic difference.
Thanks Michael B,
We now have quote boxes!
Robert-
I guess I wasn't clear or maybe you just don't like the idea of empirical data.
If the energy modeling that you believe in is accurate but occupant lifestyles vary widely, then we would expect that the models can predict the average energy usage of a group of homes fairly well but that there would be a lot of variability from house to house. In addition, projections of the energy savings from various retrofits would be reasonably accurate if you averaged across groups of homes but not be reliable for individual homes. But the empirical data generally shows that, especially in older and less efficient homes, the models tend to overpredict energy usage and especially savings, often substantially. I think these discrepancies are due to a combination of poor assumptions and errors in some algorithms. Some of the problem areas include assumptions about effective R values of low R building components and modeling problems with basements, crawlspaces, ducts, air infiltration.
I think we need to use empirical data to improve our modeling and we need to do this through a combination of large scale analysis of actual energy usage and more directed research projects about specific issues.
Occupants are a convenient scapegoat for bad models and modeling. I think that the models are not good enough to provide reliable estimates of the energy use of homes in many circumstances.
By the way, your assessment that a 21% median change in usage vs. a 14% median change supports your case is interesting. I would think that if lifestyle were "far more" important than the building that we should see a much larger difference -- not just 7 percentage points. Maybe the mix up is that i define "far more" as much larger than half while you define it as somehow less than half?
As I stated several times, energy modeling cannot predict energy use - there is no correlation. That's not the purpose of ratings such as HERS. They are designed to compare the relative efficiency of a house with the same house built to IECC standards, making assumptions about occupancy patterns that don't have to correlate with actual usage but have to be consistent and reasonable.
If your data shows a 50% median increase in usage variability, then the difference in the entire range of variability of the two data sets within the 3 sigma standard deviation is likely considerably greater. Houses aren't occupied only be "median" occupants, and energy usage patterns include a significant range of lifestyles. For that reason actual energy consumption is an unreliable standard for judging the efficiency of a physical structure.
I stand by that conclusion, and you will undoubtedly stand by yours (and probably continue to make insinuations about my resistance to empirical data and my ignorance of statistical analysis).
Please stop making these clearly erroneous claims. I have done many studies that do show a correlation -- it is there -- it better be there. If there were NO correlation between modeling and real world energy use then the modeling would be a complete waste of time since it's real world energy use we want to reduce with our efficiency upgrades, not just computer print outs. Model accuracy does not mean the model will be very accurate at predicting a given home's energy use, but that it gives a good answer on average. If the model does not give a good answer on average, then it is a bad model and should not be used to help design buildings or compare one to another. Assumed occupancy patterns are just one part of the modeling exercise but they too should be reasonably accurate. If I wrote a model that assumed an 85F heating thermostat setting it would be a bad model and would not properly estimate the impacts of various strategies to reduce energy use. But the problems with the models are also about the physics of how buildings work, not just, or even mostly, occupancy .
Your attempt to reinterpret the data i presented to support your claim is really grasping at straws. The data show pretty clearly that occupancy differences don't have a huge impact on usage for most households. I am not denying that it can have a big impact on some households (out in the tails at 3 sigma), just that it's not such a dominant factor. The median household does matter.
You are right that I will stand by my position and I guess you will stand behind yours. The difference is that I have lots of empirical data to back up my claims and you have nothing but unsupported claims -- including the latest easily refutable "no correlation" one...
Michael,
Because my interpretations and conclusions are different from yours does not make them "erroneous". In fact, they are logically necessary.
And what I've been saying all along is that most architects, builders and home-owners aren't interested in statistical averages - we're interested in how this one house will perform or compares to others or to the code standard house.
The difference between our interpretations and conclusions is due to the fact that you're interested in statistical norms and I and most of us are interested in the single case.
For the majority of people immersed in the real world, there's nothing in the center of the road (the median) but yellow lines and dead armadillos.
Your claims are demonstrably erroneous. It is not my opinion but a fact. You said there is no correlation between actual energy use and modeling projections. I have data that absolutely refutes that -- lots of data from multiple studies, climates, and housing types finding correlations ranging from about 0.5 to more than 0.6. (on a scale of 0 to 1 that makes them clearly not 0). You are certainly entitled to your own opinions, but not to your own facts.
In an analysis of gas usage of new homes in upstate NY, the REM/Rate predicted heating use was within 20% of the billing data in 80 out of 108 homes.
It's funny you should try to lecture me about the real world. You say that models (which, by definition, are NOT the real world) are the only accurate way to compare the efficiency of homes yet you also say that their results have no correlation to actual energy use. These beliefs are fundamentally inconsistent.
To me (and many other people I would guess) the real world is actual energy use -- not model projections. Models are only useful to the extent they represent the real world in a reasonably unbiased way. How can we tell if our models are any good? By comparing their projections to the real world. How do we deal with random variations (such as occupancy)r? We use statistics to assess accuracy in groups of buildings. This approach is a fundamental aspect of the scientific method -- but something that seems to elude you. If models have no relation to reality, then what good are they? .People in the real world do want to hear things like -- if we do these retrofits then you should save about $400/year on average Your attempted statistical joke about armadillos is a non-sequitur at best.
Michael,
I'm not "lecturing" you - I'm stating the obvious. Statistics are not "facts". Numbers don't tell us anything about the world - they are abstractions (but since Plato, philosophers and scientists have perpetuated that error). What tells a story is the interpretation of numbers or the broad collections and manipulation of numbers we call statistics.
And if you were truly the master of statistics you claim to be, you would know that numbers can be interpreted in an almost infinite variety of ways. You've chosen one interpretation, which you would like to believe is "true" while all competing interpretations are, ipso facto, false.
And, since statistics are not empirical facts, any conclusion based on them are also abstractions and not descriptive of empirical reality.
But rather than acknowledge that there are other valid perspectives, you repeat yourself ad nauseum and resort to ad hominem attacks. That's hardly the scientific method which you claim as your mantle.
We could simply agree to disagree, but you seem to get increasingly incensed at anything that contradicts what you're certain is true or anyone that does not accept your version of reality. That's hardly the scientific approach.
Let me add that the essential problem with "proving" that variations in energy use are primarily dependent on structure, and hence identifying structural "solutions" to our energy crisis and global warming catastrophe is that it mistakes a human problem of attitude, belief and behavior for a simple mechanical challenge.
This is exactly what Einstein was referring to when he said that "no problem can be solved with the same mindset which created it". Almost every anthropogenic problem we face today is being addressed - by scientists, engineers and architects - as Newtonian mechanical challenges that require only or primarily technical "solutions". However, every problem we face has been exacerbated by previous technical "solutions" which invariably have unintended consequences.
Most of the "green" homes being built today, for instance, are many times larger and far more complex than basic shelter would require. Since the primary determinant of energy consumption by a building's mechanical systems is size, and the secondary determinant is complexity of shape, simply improving envelopes and mechanicals without altering the attitudes and consequent behaviors that demand excess and waste as essential elements of unsustainable lifestyles will do nothing at all to solve our crises.
For instance, in spite of significant increases in the efficiency of domestic appliances, we are now using more electricity per capita for appliances than before. Buying hybrid cars so that we can drive more miles "guilt-free" does not solve the fossil fuels crisis. Building more efficient houses and then turning up the thermostat in winter or down in summer does not necessarily save any fuel.
The bottom line is that we will not find solutions if we are looking in the wrong place - like the man who was asked, while searching the sidewalk under a streetlight, if he had dropped something there and answered "No, I dropped my keys over there but there's more light here."