I am the Chair of the conference committee for the ASHRAE 2012 Energy
Modeling Conference that will be held in Atlanta on October 1, 2 & 3 this
year (see link below). The conference is focused on bringing
practitioners, software developers, researchers and facility users together
for 3 days of in depth discussion on current modeling software capabilities
and current best practices in energy modeling. I have two questions that I
would like to pose to this group to get some feedback to help provider
richer content for the discussions planned at the conference;
1. Are the current energy modeling tools available to an energy modeling
practitioner reliable enough to allow the modeler to predict a building's
actual energy consumption with a high degree of confidence such that an
accurate energy target can be established and recommended to the building
owner for the new building or a renovation/retrofit?
2. Are the current best practices of the energy modeling community reliable
enough, and well understood by most practitioners, to allow the modeler to
predict a building's actual energy consumption with a high degree of
confidence such that an accurate energy target can be established and
recommended to the building owner for the new building or a
renovation/retrofit?
Background for the discussion:
These questions recently came up in a discussion among the conference
committee. It seems that one our colleagues from the UK indicated that in
the UK new schools have performed very poorly in comparison with their
predicted energy use.
Another comment that was made was as follows:
"Often a building's actual energy consumption is 1.5 to 2 times as much as
the results of an energy model that was used to make decisions during
design about the building's energy using systems. Is it the the energy
modeling tools or is it the processes used by energy modelers to describe
the systems and how they operate in the software? Should energy models be
used to "predict" a building's future energy performance or just be used to
inform better decisions during design?
I have reviewed a good many models and, almost without fail, I never see a
modeler start by writing a sequence of operation and I also never see the
sequence of operation used by the modeler make its way into a set of
construction documents.
Also, when I am the modeler and I am "calibrating" a model to an existing
building's actual energy consumption it is a very iterative process. I
know what things to manipulate in the model to effect demand and what
things to manipulate to effect consumption. I just keep going back and
forth until I have a model that you can almost lay its output on top of the
building's utility bill history. I also have a good understanding of how
the building is actually being operated and maintained - which I hope helps
make the model more accurate, but, does that process really give me a
better model to make decisions from?"
I invite everyone to please, tell us what you think.
Sincerely,
Link to conference webpage:
http://www.ashrae.org/membership--conferences/conferences/ashrae-conferences/emc2012
--
M. Dennis Knight, P.E.
Dear Dennis and BldgSim Community,
1. I think the tools are absolutely up to the task. My own practice uses EnergyPlus exclusively, but I know that most of the other tools are based in solid thermodynamic and physical principles - so they start on a solid foundation.
2. Current best practices is another story altogether!
a. Keeping in mind that I know only the "Best practices" for my own firm ...
b. Energy modelers of new construction are normally given scant information.
* Partly this is due to the owner not knowing exactly how the new facility will be used.
* Partly this is due to the Owner and Designer not caring about, not appreciating the importance of, or just not needing to gather detailed information about the operation of a building that hasn't been built.
* Partly it's because "as built" is never "as designed". (Think of fan and pump pressure estimates differing from actual, weather variances, occupancy schedule changes, etc.)
* The energy modeling community has, it seems, a lot of bright young men and women who are "learning the ropes". The fact that they are becoming involved is very exciting! Their education must be broadened, however, in order for them to become effective at modeling existing buildings. It's no longer just theory; there is a lot of practical, "hands on" activity that is needed.
3. Inferred above is the host of variables that differ in an actual building's operation from what may have been assumed. The older the building, the more variations there are!
4. Most building operators, if they exist within the building as a full time position, are distracted with many other details and spent precious little time optimizing energy performance. If there is no full time building operator .....
5. This is a field ripe with opportunity! The first and hardest task, I think, is to get building owners convinced that the ROI for optimal building performance is better than any of their other opportunities for investing. The next task is to streamline the process of calibration and identification of opportunities so that they is faster and more economical.
6. ... and I'd love to expand this discussion during the Q&A period at my presentation on this topic during the ASHRAE Energy Modeling Conference!
James V Dirkes II, PE, BEMP, LEED AP
Jim said most of what I was going to say. Additionally,
* An energy modeler's task for a new construction project is rarely to accurately predict energy use, but to demonstrate compliance with, and improvement over, code performance.
* Modelers generally do not fine tune plug loads, occupant numbers and schedules etc. because these are kept identical for determining improved performance over code.
* Sequence of operations. The modeler is usually not the engineer of record, and should therefore not be dictating the sequence of operations of HVAC/plant equipment. I may provide the mechanical engineer with suggestions on controls strategies and setpoints, and sometimes they are receptive. However, my focus as energy modeler is energy, and not comfort, system complexity or reliability. Regardless, the exact sequence of operations, even if meticulously described in the design documents, may not be implemented, or may be changed many times during the first year or two of building operation.
* No/poor commissioning of buildings. Buildings designed to be very energy efficient often rely on complex controls for HVAC and lighting systems. Many buildings are not commissioned, and even in the ones that are, commissioning is often little more than verifying that the equipment and controls were installed as designed. The commissioning agent rarely has the time/budget/scope to determine that all control strategies are operating as designed. Also, the commissioning agent cannot change the weather conditions during which the building is commissioned, making it next to impossible to check CHW controls during winter for example.
* New buildings often go through many changes in operating conditions during their first year or two. New buildings are often in use after hours due to people moving into their new offices, or because they are nice facilities and the demand to utilize them is high.
* One of the best ways to predict energy performance is data mining of existing building performance. Hopefully, CBECS and other building performance databases will be a big area of focus for our field.
[Senior Energy Engineer 28Jun2012]
Bill, Jim, Dennis, et al
This is a tough nut. In the ESPC or ESCO world, where performance must be depended upon to achieve financial returns, there are understood and controlling processes in what equipment is bought, how it is installed, how it is maintained, how it is operated and expected duty schedule. Contracts are written to reduce uncertainty to the point that in some cases any deviance, in seemly minor matters, throws the performance expectations up for renegotiations.
While weather changes can be adjusted for after the fact (and thus readily understood as a reason for some differences between real and modeled performance), outside of the ESPC/ESCO arena, a design engineer/modeler can make their best fine tuned efforts on construction intent equipment and operations, to be undone by changes in installation, commissioning, operational practices and occupancy that are not under their control, and are frequently difficult to document after the fact.
In the design arena, modeling is a means to assess options on a level plane, and should not be offered as a predictor of actual cost unless that risk is fully evaluated and compensated for (both in labor/risk fees and in the processes to assure changes are documented). There is a real need to fully vet an industry standard approach to identifying the risks (for fair negotiation), and establishing standards in best practices.
Yes, our modeling does inform sequences of operations, as strategies can be involved, and are a combined designer and modeler effort. They do indeed make their way into construction intent documents ... but implementation and ongoing maintenance are another matter.
A calibrated model is an excellent tool for more realistically evaluating operational and systems changes. But here too, the results should be considered as a way to evaluate options on a level playing field, as control over so many variables is out of the modelers scope.
I would encourage the development of collective industry guidance towards understanding these risks and offering clarification on expectations for client informational purposes.
David Ellis
The key is to not use energy models quantitatively. Energy models should generally be used qualitatively to compare to either a known benchmark (utility bills for existing buildings etc.) or a baseline (regional energy codes for new buildings, etc.). For comparisons to utility bills we often model control sequences and other known occupant/owner/operator behavior.
Haider Khan.
After enjoying all of the posts on this topic, it occurs to me that existing energy modeling software is challenged to model certain aspects of building operation that are ?broken?. An example might be a chilled or hot water valve which never closes fully.
James V Dirkes II, PE, BEMP, LEED AP
How about modelling an uninsulated condenser cooling water line that
runs through the building outside the insulation on the South side of
the building in California?
Robert Wichert P.Eng. LEED AP BD&C
Hi all. I'd like to echo Jim's comment, and express excitement about what RMI is doing. Related to the "reality factor," I've seen a tremendous gap between what LEED/GBCI wants and what will actually take place in the building, and the lack of sanity checks in modeling software. Though it is the responsibility of the modeler to perform sanity checks, and quality control, just spending the extra couple of hours each project can make the world of difference. On that end I've created a slew of my own sanity-check post-processing tools for eQuest and feel like our deliverables have improved significantly as a result.
What some of the data our clients are sharing has begun to express the overwhelming need for these checks in all modeling, which includes analyses I've reviewed on behalf of our clients. In recent projects I've gone so far as to assume building-user error, reflected in schedules and temperatures, which has been spot on when comparing to metered data. This is somewhat taboo, because not many people are likely to voice these foreseeable shortcomings when delivering EUI's during design that look inflated, for a new "high performance" building, but as sanity checks and associated tools develop it may become easier.
What we can do, as an industry, is get better at setting expectations parallel to educating the building community about the usefulness/value of a good model.
Great thread folks.
Matthew Higgins, CEM, HBDP, LEED-AP (BD+C)
Two things:
1. Some modeling software has lots of controls built into default HVAC
system types. These controls are often better than what is planned or
installed for the building. For example: economizers, demand controlled
ventilation or outdoor air reset might be assumed. Similarly, on many
projects the HVAC controls drawings or sequence of operation is either
missing or not useful at all. Without good information, it's tempting as a
modeler to assume efficient control of equipment. However, on the actual
building, the opposite is likely to happen.
2. Once I account for usage patterns, if a building I modeled during
design performs poorly in comparison to my model, I generally view at as
the building's "fault", and not the fault of my model. The model can be a
great diagnostic tool for discovering where/how the building is not
operating as was modeled. With a cooperative engineer and facilities
manager, we can usually make adjustments in building controls to bring the
building more closely in line with the model. If the building was not
insulated properly, that is another matter, and post-occupancy adjustments
are difficult/impossible to make. However, I would argue that the solution
is not to model poor construction and execution, but try to find a way to
use the model's projected energy data to keep the energy target in sight
through construction.
--
Karen
Hello,
There are lots of interesting threads on this conversation. However, I think that there are also some misconceptions.
First, model as you will for new construction, at best, in my opinion, you are performing a rating of the building's assets compared against a standard, much like the mileage ratings on a car that you purchase. If you are using the IECC, 90.1, 189 or the IGCC, you have to use specific rules to guide the input of internal heat gains, etc. Such rules have been carefully crafted to avoid gaming and to attempt to deliver the same result regardless of who performs the simulation.
Therefore, it is my opinion that such a simulated asset rating of a building should not be construed to match the actual energy use of the building (period). If does happen to match the actual energy use, then a miracle has just taken place because the probability of this not happening is large.
Second, if you do attempt to match the energy use of a real building, there is advice about how to go about this in the literature, including the results of many ASHRAE projects, for example: RP827 for in situ equipment, RP 1404 (ongoing) for how to use short term/long term measurements to match a real building, RP1051 for advice about overall calibrations, RP1052 for advice on AHU modeling, RP1004 for advice on thermal storage modeling, RP865 for advice on secondary system modeling, and soon RP1468 for advice about modeling and BIM to thermal. So, there's hope that one can actually match a real building, if you have measured data and if you have real systems that match your model's systems, and if you have real weather data and if you have a large budget and tons of patience, etc., etc. Otherwise, getting a "match" is really an exercise of "matching lumps" against "measured lumps".
Third, don't ever forget that simulations of building envelopes are just great big R-C networks, and when you attach a system simulation to this you are perhaps adding a bit of psychrometrics, and then a bunch of curve fits since most widely used programs do not have first principle models of their chillers, boilers, etc., just curve fits.
Fourth, whole-building simulations do a really bad job of simulating faults or broken systems. For
example, when a building is badly zoned (i.e., the thermostat and zoning was wrong from the building and parts of the building are always too cold or too hot, etc.). In such cases, you can't simulate it very well. In other cases, where valves are broken and/or the building has weird controls -- best of luck.
Fifth, simulations of building envelopes is still REALLY CRUDE. For example, simulation programs have no wall thickness. Think about it. The wall is just a thin 2D construct with the thickness only accounted for in the material definition. Hence, to get it right, a simulator has to ?setback? the windows, since most windows are located midway in the wall and have a small 2 to 3 inch shade from the overhang of the lintel. Or consider that simulations do not consider the corners of buildings, just the area of the wall. So, you?ve got lots of corners in your building you might need to compensate.
Sixth, shaded windows in the DOE-2 family of codes use George Walton?s algorithm to assign the calculation point for the shading. This has 10 divisions by default, it splits the window in half vertically, then each half into fifths?or if you choose 20, 30?up to 40 it keeps going. For most rectangular windows this doesn?t make a difference, but for a shaded, curved window in elevation, to get the most accurate results, you end up with a bit of a quilt-looking window ? no kidding. RP1468 (forthcoming) will shed some light on items 5 and 6.
So, before we make claims about how to do things better, we need to remember just how rough we are REALLY doing them now (actually it?s can be pretty good), and then choose our words carefully when we make claims about what is being simulated ? I?m just as guilty of this as anybody.
Jeff S. Haberl, Ph.D.,P.E., FASHRAE
Thanks, Jeff for the great synopsis and list of sources for calibration
methods, and for others providing an enlightening discussion. Discussions
like this are why I subscribe to the bldg-sim.
I've felt for a while that standardized calibration procedures for
different kinds of buildings would be highly useful for the community of
modelers working on calibrating to existing buildings (like those of us
doing impact evaluation measurement and verification work for utilities and
regulators).
I think the whole idea of an uncalibrated model being able to predict
energy consumption of a real building is off base. Look at the BESTEST
results for different simulation engines looking at simplified results. I'm
not sure what a reasonable standard would be for prediction when you have a
well-specified set of building operating parameters, but my gut says its at
least +/-30%, and more like +/- 50%.
This is where modelers need to have a consistent, clear message on why
building energy models make sense for new construction. I'm not a new
construction modeler and never have been, but I see the value from new
construction models being to evaluate different design options for saving
energy compared to some baseline, and for estimating what the savings
fraction relative to a baseline. While these kinds of results aren't
completely independent of actual building occupancy and usage, they are
likely to be accurate enough to provide useful information to the design
process and create real value. We need to remember the inherent uncertainty
in the results, but also remember that the information is still useful for
making decisions about a building's design and which design paths are
likely to offer more cost-effective means of achieving energy efficiency
goals.
If we want to come up with target EUIs for newly constructed buildings,
then I think major benchmarking studies are much better ways to go, i.e.
what is the typical consumption of new buildings having X application with
X occupancy patterns in X climate, and what is the distribution. If you're
a standard deviation below the mean, you should have a big opportunity for
improvement.
Last, I think we should have some way of quantifiying the resiliency of
savings associated with different kinds of energy efficiency measures. This
comes down to how sophisticated the building operator needs to be to
maintain the savings as designed. For example, an improvement to the
envelope is pretty close to permanent. A high efficiency chiller or other
primary HVAC device should offer savings regardless of what the operator
does. An enthalpy-controlled economizing system is not very permanent, and
any kind of reset schedule for various airside or waterside heating and
cooling delivered to the space is only going to last with an adept building
operator in charge. It would be nice to standardize a way to differentiate
energy scoring of new buildings on the basis of where the savings are
coming from and have some way of discounting controls-based savings against
envelope or equipment-based savings as a means of showing the true value of
savings that are more resilient to the whims of building operators with a
different objective than energy savings.
Regards,
Justin Spencer
Justin,
I agree with you that it is important for modelers to have a consistent,
clear message on why building energy models make sense for new construction,
and want to reiterate what Jeff wrote below, that their main intent is to
rate building's assets from efficiency standpoint. In this sense,
performance rating calculated via 90.1 Appendix G is conceptually similar to
efficiency ratings of HVAC equipment (EER, COP, etc.), MPG fuel economy for
cars, or kWh consumption on appliance's Energy Guide label. I don't think
there is any disagreement regarding the usefulness of the standard EER, COP,
MPG, etc. ratings. It is also widely understood that these ratings represent
performance at standard rating conditions and thus will not match the actual
equipment efficiency achieved in a particular installation. All buildings
are different, making it impractical to use field testing to rate their
performance (for one, it would be too late to improve building efficiency if
it is already constructed), so we have to resort to energy modeling. 90.1
App G protocol is not intended for predicting the actual post-construction
building consumption. This is made very clear in Note 2 to G1.2: "Neither
the proposed building performance nor the baseline building performance are
predictions of actual energy consumption or costs for the proposed design
after construction. Actual experience will differ from these calculations
due to variations such as occupancy, building operation and maintenance,
weather, energy use not covered by this procedure, changes in energy rates
between design of the building and occupancy, and the precision of the
calculation tool."
Maria