Adds Fuel Plot Graph

[trevorb1]

DRAFT PR STILL

Closes #230

Changes proposed in this Pull Request

In this PR I have added a graph to plot fuel costs. Since the marginal costs include a flat VOM cost, I have just taken 2030 VOM values to remove. So while this is technically an estimate of fuel costs, it will be a close approximation.

Checklist

• [ ] I tested my contribution locally and it seems to work fine.
• [ ] Code and workflow changes are sufficiently documented.
• [ ] Changed dependencies are added to envs/environment.yaml.
• [ ] Changes in configuration options are added in all of config.default.yaml.
• [ ] Changes in configuration options are also documented in doc/configtables/*.csv.

[trevorb1]

There is something weird going on with fuel costs I believe.

1. OCGT and CCGT prices are not coming out to be the same. Since fuel costs will not have efficiencies or anything applied, they should come out to the same.

2. The fuel costs in general are not lining up with what comes out of the EIA module. For example, in Arizona

Gas prices should be 13.706 $/MWh

Coal prices should be ~0.70 $/MWh (idk, this one is really off - also see issue #250)

[ktehranchi]

I was looking at similar issues this last week regarding mis-classification of plants between oil coal and NG, but didn't pick up on this issue.

1. It looks like the efficiencies are still included in the table you posted since CCGT is lower than OCGT? Maybe the function for pulling Fuel Costs is off? In the PR I am going to submit, I add a new fields for generators that captures the Avg Fuel Cost and the FOM cost- just so we can plot this and have the data readily accessible.
2. Coal Marginal Costs should be in the ~15 $/MWH range... (reference Azevedo Simple Dispatch paper). And that's what I am getting for final marginal costs on my end.

I'll try to get my PR in before we meet Wednesday

[trevorb1]

hmm... probably an issue with my get fuel costs function, then. Let me review this. Thanks!

[trevorb1]

Okay, Im pretty sure the get_fuel_costs function is working correctly, now. I accidentally grabbed the Eur fixed VOM costs, rather than the ATB ones; this has been corrected. Still doesnt address the issue that CCGT and OCGT are different (although closer now), or that it should be $13.71 / MWh, I believe

[ktehranchi]

hmm and confirming you used these FOMS ?

"CCGT": {  # natural gas
        "display_name": "NG Combined Cycle (F-Frame)",
        "technology": "NaturalGas_FE",
        "crp": 30,
    },
    "OCGT": {  # natural gas
        "display_name": "NG Combustion Turbine (F-Frame)",
        "technology": "NaturalGas_FE",
        "crp": 30,
    },"

Could the difference have to do with clustering? like maybe the capacity weighted averages for MC during clustering isn't applied for eff? Seems unlikely.

[trevorb1]

hmm and confirming you used these FOMS ?

Yeah, I pulled them from the resources/costs_2030.csv file which should use that dictionary to extract the ATB data? The VOM has an ATB key value associated with it

Could the difference have to do with clustering? like maybe the capacity weighted averages for MC during clustering isn't applied for eff? Seems unlikely.

Maybe! Its also a little odd that new and old coal have different fuel costs. It is at the 3 decimal point, so maybe its just a numerical error, but then its weird how gas doesnt suffer from that.

[trevorb1]

Oh, I wonder if this has to do with the efficiencies.

If I look at combustion turbine in the ATB it shows a heat rate of 9.717 MMBtu/MWh. This value is translated into the costs_2030.csv file okay.

A heatrate of 9.717 MMBtu/MWh should be an efficiency of 35% (3412 / 9717 = 0.3511). But here it is showing the efficiency of ~40% which is applied from when the generators are attached. I wonder if somewhere along the lines we end up using the efficiency in the costs_2030.csv file to do a calc instead of the individual generator (or clustered generator) efficiency?

I took a quick look and didnt see anything, but may have missed it.

[trevorb1]

Okay, I believe this may relate to the legacy code to get fuel costs, described here and in issue #235. I will switch to fixing issue 235 and come back to this.

https://github.com/PyPSA/pypsa-usa/blob/eb6e96f9bb21de896e8d8c8a57c2471a44e25a12/workflow/scripts/add_electricity.py#L1390-L1406

[ktehranchi]

Our efficiencies are actually a mix of ADS, EIA historical for existing plants, and ATB for new build plants. So we would expect those to be different. You can rerun this with the new fields for avg heat_rate and fuel_costs

[trevorb1]

Another issue, as pointed out by @ktehranchi, is that coal costs are not changing in the final network. Although, when they are applied in add electricity there is variance.

[trevorb1]

As per our (@ktehranchi and Franklyn) discussion today, I will go ahead and merge this. The minor differences in the gas prices between OCGT and CCGT are likely due to clustering. Moreover, while there is an artifact that coal prices are being read in correctly at the individual generator level, but by time the network is solved, the price is flat.

Both these issues probably deserve more exploration, but will move to a different issue ticket.

The main additions from this PR are:

• Addition of get_fuel_price function
• Plotting of fuel prices. A single figure is plotted where the first frame is system level price per fuel, then subsequent plots are individual fuels per bus.