Tips for model development with multiple modelers

[cyang]

I have a question about how folks have developed models with multiple people who are contributing to different sectors.

Specifically, we have folks who are working on the electric supply sector and others on transportation and vehicles and others on buildings and end use demands.  Right now we have the Base Year files separated by sector but the SubRES-B-NewTechs is one sheet and it is somewhat challenging to keep this file up to date with all of the parallel changes that we are making. 

So my questions are:
1) can we split the SubRES-B-NewTechs into multiple sectors (i.e. an Electricity versions, a transportation version, etc. . . ) like we do the base year files?
or 2) can we just put all of the new technologies into the base year files so that we just have to import the appropriately updated files?

Or are there any other options and tips for making sure that we have the latest version of each sector in an easy to manage way?

thanks,

Chris

[AKanudia]

Good question Chris.

I would prefer your first option and make files like SubRES_NewTechs-<SectorName>.

Remember that VEDA still needs to know the sector when it encounters any new technology/commodity. In the case of base year (BY) templates, this information comes from the file name. In SubRes files, it comes from the sheet name. But you can have several sheets for the same sector by using an "_" after the sector name. For example, if ELC is one of your sectors, you could name sheets like ELC_Nuclear, ELC_Fossil, ELC_Renewable etc.

In principle, you could also include new techs in the BY templates without much of an issue provided you have only one template per sector. But if you decide to add more regions (super-regions, actually), then you will need to copy the same set of techs across multiple workbooks, which is clearly undesirable. The other reason I would recommend the first option is that, generally, one would edit the new technologies more frequently compared to the existing technology stock. If new techs were in BY, then all the existing techs would need to be reprocessed each time you change/add new technologies.

With multiple people working on different parts of a model on an ongoing basis, I would set up a central location for the "official" version of all templates. This could be a common drive on a network or even a dropbox folder. This could be supported by a google doc/Springpad/Evernote tracking system where sector/region analysts maintain a log of their updates. One advantage of using dropbox would be that it would keep previous versions of files automatically. Otherwise you can ZIP the entire set of templates periodically.

[cyang]

Thanks Amit.  That is helpful.  We will go with your suggestion and let you know if we have any issues.

The dropbox suggestion is also helpful.

Thanks!

Chris

[saleh]

Hi,

We have problem in running the model in disaggregated form.

First, for some technologies we do not have output when we see them in process master, at the same time if you look at the commodity master you can see that these technologies are assigned to a commodity (also they are fine in RES).

The other thing is in VEDA BE, we do not have any dummy imports. However, when we look at COMELC and COMNGA (Electricity and NG which are inputs of some technologies) they have only value in 2010 (base year), but when you look at VAR_FOUT of technologies which their input is COMELC or COMNGA they have output. Also we do not have any important import warnings/errors. I would appreciate your help.

I have attached the pictures below.

[Antti-L]

Just a few small comments and questions:

1. The  picture for the COMINLL commodity shows that the commodity is in the set .DMD. However, that set is meant to be used for demand processes only, not for commodities. The set to be used for demand commodities is DEM. Possibly this could be causing some of the anomalous behaviour.

2. The PCG of the process CLT2LF54HBLO06 appears to be COMELC, i.e. the input commodity. That is an unusual choice for demand processes, and it raises the question how you are defining the process transformation, between the main input and output. If you are using EFF for defining the the output-to-input ratio, I fear that the shadow group of the process may not in this case be correctly identified by TIMES (due to the original design of TIMES), and therefore that might be another source of your problems. (You can define the shadow group explicitly by using ACT_EFF instead of EFF.)

3. Have you looked at the file QA_Check.log (in the VEDA-FE working folder)? It might contain some helpful warnings....

[saleh]

Hi Antti,

The problem was DMD, I fixed that and it works perfectly now.

Thank you.

[cyang]

Thanks for the tips Amit.  Things are working well with our disaggregated model.

[AKanudia]

good to know!