Replace Variable Renewable Electricity Backup with Profitability-Adjusted LCOE (PLCOE) #436
Conversation
…this commit; data changes to follow in a separate commit. Model runs and solves normally, but results have not been explored / vetted.
…this commit (code changes were in a previous commit). Model runs and solves normally, but results have not been explored / vetted.
… hard-coded values.
…l XMLs; GCAM runs and sovles cleanly; results have not yet been vetted.
… in module_water_L2233.electricity_water.
| DEFINE_DATA_WITH_PARENT( | ||
| IBackupCalculator, | ||
|
|
||
| //! TODO: leave an explanatory comment |
There was a problem hiding this comment.
Do you need to add some explanation still?
| const double aAverageGridCapacityFactor, | ||
| const int aPeriod) const | ||
| { | ||
| // This is a placeholder function so we can use the IBackupCalculator class |
There was a problem hiding this comment.
okay, so my rust is showing. Is it because it is pure virtual and needs to be defined? could it instead be removed from the ibackup_calculator, or do you think it might be useful for something else so should be left in?
There was a problem hiding this comment.
i guess we need to see how to deal with Page's comments on maintaining the old ways as an option perhaps
There was a problem hiding this comment.
I don't think we want to support multiple ways of representing intermittency; pretty much every time we have ever tried to maintain multiple ways of doing something, the “non-default” option only gets maintained for a few months. We don't want to require ourselves to include multiple representations of intermittency in our standardized testing suite.
The issue that I have raised with this proposal is that right now, in high renewable scenarios, we see lots of single-cycle combustion turbines, whether powered by hydrogen or natural gas. Almost all of this capacity is in the backup electricity sectors. Perhaps coincidentally, the total capacity and total electricity output of these combustion turbines is reasonably consistent with ReEDS; high renewable scenarios in the USA tend to see between 500 and 1,000 gigawatts of combustion turbine capacity. In my opinion that technology should not be turned off in this revised approach; there should still be some demand for the services provided by combustion turbines, and their costs should be passed forward to electricity consumers.
There was a problem hiding this comment.
Hi @pkyle, thanks for your review comments.
One thing to note is that the new PLCOE method does allow us to calculate a total integration cost per region / technology / model period by taking the difference of PLCOE and LCOE. This integration cost reflects a variety of technologies that provide system flexibility and support VRE integration (e.g., storage, transmission, and firm capacity resources), and could be used to post-calculate implied combustion turbine capacity or some mix of combustion turbines, storage, and transmission. It’s not obvious that single-cycle combustion turbines are the primary system integration option supporting VRE. Per the table below for the ReEDS 2023 Standard Scenarios Mid Case, the growth in battery capacity from 2010 through 2050 is twice that of CT capacity. Doing a robust competition between Storage, CT, and other integration options is beyond the scope of this CMP, but we have done some preliminary thinking on a hybrid approach that combines PLCOE-based competition with dynamic calculation of storage/firm capacity. This is a good opportunity for future model development.
| 2010 | 2020 | 2030 | 2040 | 2050 | |
|---|---|---|---|---|---|
| Wind Capacity (GW) | 40 | 118 | 441 | 641 | 762 |
| PV Capacity (GW) | 1 | 72 | 352 | 753 | 1095 |
| VRE Capacity (GW) | 41 | 190 | 793 | 1393 | 1857 |
| Battery Capacity (GW) | 0 | 0 | 50 | 197 | 369 |
| Gas-CT Capacity (GW) | 117 | 128 | 123 | 173 | 300 |
| H2-CT Capacity (GW) | 0 | 0 | 0 | 11 | 11 |
| CT Capacity (GW) | 117 | 128 | 123 | 183 | 311 |
| Battery-VRE Capacity Ratio | 0.00 | 0.00 | 0.06 | 0.14 | 0.20 |
| CT-VRE Capacity Ratio | 2.87 | 0.67 | 0.16 | 0.13 | 0.17 |
Table from ReEDS 2023 Standard Scenarios Mid Case
RE: supporting multiple ways of representing intermittency, @mbins talked to @pralitp and worked out an approach for maintaining the previous approach in code that should be reasonably robust. You’re right that the “non-default” option will be more susceptible to lapses in maintenance. While we feel that the PLCOE approach proposed in this CMP does a better job of representing VRE competitiveness (and is more transparently parameterized), we acknowledge that removing the ability to estimate firm backup capacity (and associated energy use / emissions) does remove a capability that’s valuable for some projects. Thus, we agree with @mwisepnnl's suggestion to maintain the ability to use the current approach for the subset of projects for which estimating firm backup capacity is important.
There was a problem hiding this comment.
Thanks for the context, and comparison with the published ReEDS run. As I noted in my initial comment, what is most important here is how this proposal impacts the system level demands for natural gas and hydrogen. Levels of deployment of hydrogen combustion turbines in ReEDS are sensitive to the scenario chosen, through the ratio of total combustion turbine capacity to renewable energy capacity is presumably more stable. And from GCAM’s perspective, the battery option would be a system cost and electricity loss that is passed on to end use consumers of electricity, similar to T&D in general.
I suppose for my own purposes, if I can continue to use the backup electricity representation that we have right now for the next year or two, while we coalesce on a strategy for how to represent short term capacity and generation, which would value combustion turbines appropriately, that does suffice. But longer term I would not want to see the demand for hydrogen combustion turbines go to 0 simply because of a different modeling convention that we adopted. Right now this is one of the major demands for hydrogen we see in GCAM, system wide in net-zero scenarios, and the deployment levels are consistent with the current literature.
| DEFINE_VARIABLE( SIMPLE | NOT_PARSABLE, "average-grid-capacity-factor", mAveGridCapacityFactor, Value ), | ||
|
|
||
| //! State value necessary to track tech output ration | ||
| //! State value necessary to track tech output ratio |
There was a problem hiding this comment.
awesome. i think that was my typo from 20 years ago!
| @@ -57,23 +58,8 @@ class IInfo; | |||
| * resource. | |||
| * \details An intermittent subsector represents the production of a good, such | |||
There was a problem hiding this comment.
i probably caused this, but i am now confused as to whether it is teh subsector or technology that is intermittent and why?
There was a problem hiding this comment.
That comment was committed in October 2008 (!), and was itself copied from an even older commit in a different repository that I don't think we have archived. In any case, what is described in the comment is not true of the structure; in the core model there are intermittent technologies, not intermittent subsectors, and the technologies have multiple (non-substitutable) inputs: the variable resource (e.g. wind) and the backup electricity.
This structure allows us to have multiple technologies which may or may not be intermittent (e.g. wind and wind_storage) competing within the same sub sector (wind).
This Core Model Proposal (CMP) updates the cost adjustment applied to variable renewable electricity (VRE) generation (wind and solar, labeled “intermittent” technologies in GCAM) based on technology value parameterizations derived from the Regional Electricity Deployment System (ReEDS) model. In addition, share-weights of wind and solar electricity are set to converge to one by 2030 (rather than 2100).