Dear Ka-Chun,

Could you please explain more about this application?

Not sure if this is what you had in mind, but whole building energy modeling is often painfully slow. Most of the software was written by mechanical engineers, which use good heat transfer equations, but poor programming skills.

Our office has done some limited work in this area, specifically with TRNSYS and EnergyPlus to streamline the code. In some cases we have reduced model run times from 30 minutes to 2 minutes just by streamlining algorithms. But we only look at limited applications on an as needed basis as we don't have the resources to rewrite millions of lines of code.

There is limited use of cloud computing and parallel processing, which does help greatly, but can become costly.

To manage time we often use technically inferior software like eQUEST and TRANE TRACE because they fit into our workflows and budgets.

Often calculations are developed as prototypes and carry significant additional code that "might" be needed, but is redundant or irrelevant at the end. Unfortunately few coders go back and remove it. Also less than efficient solvers are used in many software.

There are tens of thousands of energy modelers out there using software that was generally created using 1990's coding techniques.

Overall, energy modeling is too slow for a fast paced construction industry. Any help would be appreciated by our entire industry.

Dear Dr. Betz,

Thank you for your helpful information. Actually, artificial intelligence has an Interconnected with other sciences. I have never worked with eQUEST or TRANE TRACE, but , as a coder, I hope to work in this field. Could you share a dataset with me?

What do you want from that dataset? And also, talk about the duration of run time that you expected.

Do you know any problem in your field that needs very high speed e.g. use a hardware for prediction?

Thanks

Vahid,

Sorry about the delayed response.

I don't think we're at the point of high speed hardware prediction yet. The problems in the energy modeling software are more basic than that in my opinion.

Regarding data sets. We typically generate 10-20 variations of an energy model. The buildings I work on are typically 30,000-100,000 m2 in size and can have 3,000 to 4,000 spaces, each with 6 or more surfaces. Most buildings models are probably close to the 2,000 to 10,000 m2 size. The surfaces have heat transfer calculations conducted on them for every hour of the year (8,760 hours per year) based on desired indoor conditions and outdoor conditions (based on a weather file). This gives you a sense of the volume of data that is generated and processed for large energy models.

There are thousands of possible inputs to enter into a model so rather than me sending you a limited data set, you'd probably be better off downloading eQUEST or EnergyPlus and seeing it for yourself.

The run time for a large model is typically 2-4 minutes with a good desktop computer in eQUEST and can be as high as 30 minutes for EnergyPlus. During development of the models, a lot of error checking needs to be done so each model will be run maybe 5 or 10 times to get it right depending on the feature of the model being investigated. Pure run time to get a model done will be 5-10 hours of purely computational time before it is ready to be presented to the design team for eQUEST.

I'll generate maybe 5-10 unique building models per year so approximately 2-3 work weeks of my year is spent just running models. There are others on my team that are full time modelers who will have 3 times as many run hours.

Furthermore, the slow speed is a significant source of frustration as it may only take a few minutes to change a model, and then you have to wait a few minutes to see if the change worked.

Hope that helps.

Fred

Dear Dr. Betz,

Actually, I search about eQUEST and I found several versions of this program. I have never worked with this program. Could you explain more about eQUEST,please?

About the version and how can I work with that.

Sincerely,

Vahid

I would use eQUEST 3.64b as that is fairly stable. eQUEST 3.65 was just released a few months ago and appears to be somewhat unstable although there was a new build issued a month ago that appears to be better. Much of the eQUEST community still uses 3.64b.

There is a lot of content online on how to use eQUEST and there is an eQUEST email list-serv that you could sign up for.

eQUEST is a building energy modeling software in two parts, A) Wizard, B) Detailed.

The Wizard allows you to create the geometry and some basic parameters such as people, lights, equipment, HVAC and Plant. The Detailed mode allows you to finely edit these parameters, but not the building shape. Note, once you go to Detailed mode, you can't go back to the Wizard without losing all the edits you made in the detailed mode.

There are a lot of default values in the software so I would stick with those for a while until you get familiar with the software and how it functions.

For completeness sake, you might want to take a look at EnergyPlus as well. It's a more complicated software and harder to learn, but that's where the modeling industry is headed albeit slowly.

Fred

Dear Dr. Betz,

Sorry about the delayed response.

Thank you for your helpful information about eQUEST. Actually, if I want to work in this field, I need a dataset. Could you give me your dataset, please?

Sincerely,

Vahid