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

What are the reasons for abnormally long simulation time?

asked 2019-07-23 04:17:03 -0600

Rocky Lin's avatar

updated 2019-07-23 17:22:55 -0600

Dear all,

I'm modelling a project with two buildings totaling 378000 m2 of GFA. The total zones are 1288 and total building surfaces are 3496.

After exporting the geometry from Design Builder, I did a trial run with IdealLoadAirSystem, which took around one hour.

After further adding HVAC systems (mainly VAV and PAU+FCU), the simulation time could take up to 22hrs, which is abnormally long. There are exhaustive severe error messages for zones like below:

CheckWarmupConvergence: Loads Initialization, Zone="XXXXX" did not converge after 25 warmup days.

Can anyone kindly suggest the possible reasons for the errors and long simulation time?



edit retag flag offensive close merge delete

3 Answers

Sort by ยป oldest newest most voted

answered 2019-07-23 14:51:28 -0600

You mention you have two buildings. Split it into two different models.

378000 m2 is enormous, and if built they would be the 5th and 6th largest buildings in the world if the two buildings are of equal size. Energyplus just isn't going to be fast if it has to run convergence on HVAC systems across 1288 zones and 3496 surfaces. Ideal loads is faster because it doesn't have to resolve the operation of the plant equipment; it meets the zone loads exactly.

edit flag offensive delete link more

answered 2019-07-23 12:28:05 -0600

updated 2019-07-23 12:29:59 -0600

The number of zones that you have may be the culprit. You can try setting you numberOfTimestepsPerHour to 1 to try and reduce the simulation time.

Also, as for the warmupconvergence error, it can be either you have really complex constructions or you have errors in your geometry. Each zone normally has at least 6 surfaces, but currently, you have an average of 2.7 surfaces per zone. I would argue that the export from DesignBuilder failed.

edit flag offensive delete link more


numberOfTimestepsPerHour will reduce simulation time, but also reduce simulation accuracy. Generally should go beyond 20 min timesteps.

mdahlhausen's avatar mdahlhausen  ( 2019-07-23 14:45:17 -0600 )edit

Thanks so much for both of your advice. However for this case the two buildings are served by the same central cooling plant, making it difficult to separate into two. And the 22hr runtime is already the trial result of timestep=1... Is there any way to further investigate which zones or other possible factors that are lagging the simulation? On the other hand, is there any further way to reduce the runtime during diagnostic runs?

Rocky Lin's avatar Rocky Lin  ( 2019-07-29 22:31:08 -0600 )edit

answered 2019-12-02 04:58:55 -0600

jadinandrews's avatar

You definitely have a very large amount of zones. What I have done, which is also the recommendation in the manual, is to reduce the number of zones. If an office block for example has 22 cubicles, rather have just one zone for the whole office block, than try to simulate each individual cubicle. In some cases it is best to have a single zone per floor in a multi storey building and to approximate loads and occupancies across that.

Another thing you could try is to not simulate an entire year in one go, but to set up multiple simulations for different months. You could potentially speed up simulations by a factor of 12 if you simulate one month at a time.

I am honestly surprised that your simulation even completes at all with that many severe errors.

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

Training Workshops


Question Tools

1 follower


Asked: 2019-07-23 04:17:03 -0600

Seen: 557 times

Last updated: Jul 23 '19