more detailed docs on electricity sector

[rjplevin]

I have been trying to understand the electricity sector to implement an RPS. (Which I've now implemented with much help from Pralit. Simplified approach coming soon in pygcam, but the following is still confusing.)

I see that several electricity generation technologies feed into a sector called electricity_net_ownuse, which add a loss factor (input/output coefficient) representing electricity consumption by the power producer, which apparently includes both electricity consumed in the production of exported power, as well as electricity generated for own use that (largely) isn't exported to the grid.

If "own-use" (as implemented in GCAM) represents mainly electricity produced and consumed within a facility (e.g., a refinery, not a grid-exporting power plant) then why is this a multiplier on exported electricity? It seems like auto-production would not be a function of total electricity exported to the grid, but of the scale of the industrial activities (e.g., refining) that tend to auto-provide electricity. I'd love to see an explanation for why this is implemented as a multiplicative loss factor.

An additional paragraph or two in the docs explaining all of this would be super helpful.

[pkyle]

Rich,

Your description of the sector structures is correct: electricity generation technologies produce “electricity”, which is an input to “electricity_net_ownuse”. Electricity_net_ownuse corresponds to the IEA’s category called “EPOWERPLT”, or “Own use in electricity, CHP, and heat plants”, which “represents the energy used in main activity producer electricity, CHP and heat plants.” The difference between the input and output of this GCAM sector in each region is equal to the electricity for “EPOWERPLT” reported in the IEA Energy Balances. The output of this GCAM sector is then an input to elect_td_bld, elect_td_ind, and elect_td_trn, each of which has the same input-output coefficient. The sum of the differences between input and output of these three electricity T&D sectors, in turn, is equal to the IEA’s category “DISTLOSS”, or “Losses”, which “includes losses in energy distribution, transmission and transport.”

So, as you note in your question, the interpretation of “electricity_net_ownuse” or “EPOWERPLT” is vague; from the text, it would seem to include the following:

• Electricity demands of power plants, for their operations. Examples would be coal crushers, boiler pumps, cooling water pumps, on-site lighting and other operations.
• Electricity demands of on-site industrial manufacturing. This type of energy demand is compiled by the U.S. EIA, under the term “direct use”. Copied from their online glossary (https://www.eia.gov/tools/glossary/index.php?id=D):

“Direct use: Use of electricity that

1. is self-generated,
2. is produced by either the same entity that consumes the power or an affiliate, and
3. is used in direct support of a service or industrial process located within the same facility or group of facilities that house the generating equipment. Direct use is exclusive of station use.”

However, what is actually included in the figure reported for each country is a bit unclear. What is clear is that in the IEA Energy Balances, “EPOWERPLT” is not disaggregated to the various generation technologies, so without supplemental bottom-up analysis, there is no way to disaggregate this electricity demand to the specific generating technologies that were responsible for the reported own-use electricity. For instance, natural gas power plants don’t need to crush coal; such a distinction is not reflected in the source data, nor in the model. Similarly, the IEA energy balances do not disaggregate the own-use electricity into the types of facilities that use it (e.g., refineries, stand-alone power plants, or different types of industrial manufacturing plants). Furthermore, there is also no way to tell what the reporting conventions are, and how those conventions may differ over time and across nations.

All of which is to say, "own-use electricity" is defined vaguely, in the IEA Energy Balances and in GCAM as a result. To improve upon the existing representation, one would need to use supplemental data sources and potentially bottom-up analysis in order to disaggregate it to the relevant parts. Anyone interested in looking further into this is welcome to contact me offline, and I can help with data sources and/or contacts.

[rjplevin]

Thanks, Page. So it seems that GCAM approximates this vague quantity by adding about 5% to all electricity produced (apart from rooftop solar, which doesn't flow through elect_net_ownuse.)

Is it correct to say that as RE and electrification (and total electricity production) increase, the approach of adding 5% would become increasingly an overestimate?

[pkyle]

Note that the 5% figure is just the USA; the electricity_net_ownuse consumption percentages range from 1.1% (Brazil) to 10.5% (Central Asia). Also note, own use is deducted from reported electricity generation, not added.

As for how the fraction of electricity generation that is used for "own-use" would evolve in a high RE scenario, I can't say for sure without knowing what exactly own-use electricity is. While it seems obvious that electricity used for coal crushing and on-site transport would decrease if coal electricity is phased out, that type of demand would also increase if biomass electricity gains market share. Water-related electricity would decrease as thermo-electric generation decreases, but CSP and geothermal are renewables, and use more of this type of electricity than their fossil counterparts (per unit electricity generated). Re: on-site manufacturing, it seems intuitive that there would be a lesser degree of co-location of industrial manufacturing and electricity production in a high-RE scenario. However, industrial rooftops might host PV, and electricity-intensive manufacturing often co-locates with geothermal and hydro. How the manufacturing-related electricity demands at co-located facilities are classified (whether ownuse or final consumption) at different types of facilities is anyone's guess as far as I know, but is relevant for correctly answering your question.

Ideally, we'd know what the "own-use electricity" demands are of each specific generation technology, and the types of facilities using this electricity, so that it could be specifically attributed in the model, and there would be no "electricity_net_ownuse" sector. As it is, we don't; moreover, given what we (don't) know, it's hard to say how own-use electricity should evolve in a high RE scenario.

[rjplevin]

It would great if this information were added to the documentation.

I wonder if it would be useful to create a page with links to aspects of the model where additional research might provide useful refinements. Some PhD student might take you up on it.