Question-and-Answer Resource for the Building Energy Modeling Community
Get started with the Help page
Ask Your Question

What questions do you ask when doing sanity checks on a model?

asked 2014-10-27 11:09:11 -0600

rdzeldenrust's avatar

updated 2014-10-27 14:20:42 -0600

After geometry construction and data entry you run your model for the first time, and yay, it works, you have results! But how do you know what's going on, how do you know you can trust your numbers?

What questions do you ask yourself to make sure your model is correctly representing your building? What checks do you build in to your workflow to ensure full understanding?

How does it differ between QA on your own model vs. a model someone else made? (For what's it worth, I'm using IES at the moment - but I think the question applies to all software packages).

edit retag flag offensive close merge delete


Thanks @rdzeldenrust for asking this question.

I am facing the same dilemma with my model in OpenStudio. I ran a PAT(Parametric Analysis Tool) analysis too trying different building orientations but the results for each alternative differed negligibly. There is definitely something wrong with my baseline model. What do you guys think? What should I go back and check (in OpenStudio specifically)?

dhruv's avatar dhruv  ( 2015-04-21 01:29:05 -0600 )edit

4 Answers

Sort by ยป oldest newest most voted

answered 2014-10-27 12:22:15 -0600

I love this question and am looking forward to others answers, here are a few ways we check our models:

  1. We take the overall EUI of the building and compare it to CBECS data and other information we have on whether this is a realistic number. It is either miles out or in the right area.

  2. We take the EUI of individual outputs, say, heating, cooling, fans etc... and we look and compare those EUIs to other projects we have done and use judgement as to whether this is reasonable for the climate and what we know about the building.

  3. We look at the unmet load hours and which zones aren't meeting their loads.

  4. We take a look at how the energy flows through the system, so in EnergyPlus we can do this by checking the nodes at each point in the AHU and in the zones and in the plants. From this you can check to see that all your components are working correctly. There is more on this on the question MattSteen has mentioned above.

From there it is a question of finding the issues and checking your inputs from the issues that have come out rerunning and reanalysing.

edit flag offensive delete link more


Thanks! Which CBECS data do you use? I can only find 2003 data but it seems rather outdated.

rdzeldenrust's avatar rdzeldenrust  ( 2014-10-28 09:50:52 -0600 )edit

Hi, I tried answering this yesterday and unfortunately I got a warning saying something like content banned, yikes! I will repeat but without links. We use CBECS 2003 that is the latest version although i think a new set of data is being collected. there is also a program by LBNL called Energy IQ which can help with benchmarking. Finally, a lot of buildings in Europe are required to have energy labels now. If you are modelling a high performing building you can use their EUIs for the labelling to compare against. I hope this helps! - Annie

Annie Marston's avatar Annie Marston  ( 2014-10-29 09:45:09 -0600 )edit

answered 2014-10-27 11:11:19 -0600

Oh this is too good!

... How many UnmetHours are there?

It had to be said.

edit flag offensive delete link more


See the post below for techniques to reduce unmet hours...

MatthewSteen's avatar MatthewSteen  ( 2014-10-27 11:29:12 -0600 )edit

answered 2014-10-28 06:36:27 -0600

TomB's avatar

Thanks for asking a really good question, this motivated me to organise my thoughts on things to check.

Generally, I find IES to be pretty poor when it comes to warnings, errors and reporting. I go through the following checks numerous times through an energy model process. Compared to EnergyPlus, where standard reporting of wanrings, errors and model integrity are produced automatically, IES leaves much to be desired (in my experience)

That said, the best checks to do at the early stage are those that you have to do at the end, i.e * for LEED, check everything that goes into the building summary report * sensitivity tests are a good way to spot things that aren't set up right. The pity of these, in my experience is that they are done at the end of the energy modelling process, when they are most useful more early on.

In IES, here are some checks I do

Basic Checks

  • Site orientation
  • Application of correct gain and variation profiles ( no people in ceiling voids !)
  • Variation of loads (diversity) within air handling unit group (this is a NABERs thing)

Gains and Temperature Checks

  • select all conditioned zones, range check temperature from 8am to 6pm mon-fri - are hours out of range reasonable ?
  • do likewise on all zones that should be unconditioned - to check that these are not mistakenly conditioned and using energy..
  • check magnitude and variation of internal gains
  • check solar gains are reasonable - i.e east/west facing zones are getting sun at right times
  • run the 'Cooling Report' check the cooling loads are reasonable <150 Watts/sq.m for a perimeter and ~60 Watts/sq for internal (or as appropriate to your project)

ApacheHVAC Checks

  • airflow rates across the supply air fan - do these respect minimum outdoor air flow rate ?
  • airflow rates across the supply air fan - is there an increase of outside air for the economy cycle - check against outdoor temp and humidity ( i copy results to excel)
  • air temps across cooling and heating coils - check simultaneous heating and cooling is not occuring
  • volume flow rates and supply air temperatures to VAV boxes - variation and magnitude - should find some that are maxxing out on a hot day
  • select a hot day, chart the cooling loads in all zones, find a zone that appears to be at peak cooling capacity, check air flow rate is at maximum and supply air temp is at minimum.
edit flag offensive delete link more

answered 2014-10-27 13:58:28 -0600

A few things to get you started:

1) EUI, kWh/sf, W/sf, therms/sf, $/sf -- all of these should be benchmarked as if the building was currently in operation, do the values make sense?

--- also the consumption values may change seasonally, do the patterns make sense? --- cfm/ft2 match expected values for auto-sized systems.

2) Does major HVAC equipment operate throughout it's capacity? (i.e. reconcile designed capacities with modeled loads)

--- check auto-sized capacities against ASHRAE rules of thumb if not using designed capacities. --- if you have designed values or actual values, verify against those capacities for major HVAC equipment

3) If the modeling package includes a rendering feature, does it match the architect's drawings for new buildings or photos of the actual site for existing buildings.

4) See previous thread on this site about calibration methods for existing buildings...this is an area with a wealth of resources.

5) For new buildings, do the results track with the project's goals? If not that will need to be resolved either by changes to the model or revising the goals.

6) Energy end-use breakdown should be compared to other projects, variances explained.

7) Calculate average annual efficiencies for primary equipment, are they reasonable?

8) Unmet hours, and/or zone temperature reports used to show that conditions are met appropriately. Verify that air temperatures and water temperatures are reasonable.

9) Verify that correct weather file was used, any elevation adjustments are made if needed.

10) Verify modeled building area against the design or actual area. Don't count the plenums or dummy zones.

11) Assess any key assumptions that were made. Perform sensitivity analysis to determine what effect might result if the assumption was too high/low/etc.

Lots of other factors that might be specific to your model. Too many to list!

As far as someone else's model, I'd ask them some of these same questions.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer


Question Tools



Asked: 2014-10-27 11:09:11 -0600

Seen: 1,173 times

Last updated: Oct 28 '14