PyPSA-EUR: Optimising dispatch only - a market model

[Hyemin Park]

Hi PyPSA community,

I am trying to run a short-term market run using pypsa-eur to only optimize the dispatch (no capacity expansion). Could you advise how I can alter the config or snakefile to remove un-necessary jobs?

Estimating renewable potential takes a long time to run but I don't think I need it for my  purposes.

Thank you so much for your help!

Best,
Hye Min Park (Hey-Meen)
Doctoral Fellow | Assistant Policy Researcher
Pardee RAND Graduate School | RAND Corporation

[Fabian Neumann]

Hi,

to run an operational study with PyPSA-Eur, you'll have to exit the workflow after the prepare_network rule:

Look for a file named according to scheme "networks/elec_s{simpl}_{clusters}_ec_l{ll}_{opts}.nc"

This one you can load this with n = pypsa.Network() and then set p_nom_extendable / e_nom_extendable for all components to false.

The build_renewable_profiles rule cannot be skipped since it also computes the wind and solar capacity factor time series. The potential calculation takes only a fraction of the time spent on that rule.

Best wishes,

Fabian

[Adrian Odenweller]

Hi Fabian,

just a quick follow-up question: So it wouldn't be sufficient to just remove every generator/storage unit/store/link from the "extendable_carriers" section and set the ll wildcard to v1?

Thanks!
Adrian

[Max Parzen]

@Adrian, good suggestion that might also work. Please report if it does and in case not, why it doesn't.

Fabian N. solutions is one that gurantees dispatch only &

could be also integrated to a workflow option in case a use case demands it.

Best wishes,

Max

[Fabian Neumann]

Hi Adrian,

This is actually a great solution! Much better than what I proposed, and it doesn't even require exiting the workflow!

Best wishes,

Fabian

[Hyemin Park]

Thanks Fabian, Adrian, and Max,

I tried the solution but the runtime didn't change. It ran for all the jobs just like having "extendable_carriers" and "copt".

Sorry I have few more questions on the output.

I am currently using cbc solver. So I made the config snapshot to be a week and clustering for 38 nodes to make sure it can handle the problem. I was able to run "snakemake -j 1 solve_all_networks" without any error.

However, I get empty dataframes for below components:

network.buses_t.marginal_price (This is the clearing price at the node?)

network.generators_t (This should be the actual dispatch?)

My colleague is also running in parallel (using wildcard "copt" with empty extendable_carriers), he was able to get values for "network.generators_t". I thought this is the actual generation but solar has the same value for all hours while wind have 0s for some hours.

I am looking for clearing prices to estimate electricity prices (euro/MWh) and delivered generation (excluding curtailment but including charging energy storage) by source (MWh for solar, wind, etc).

Could you confirm these are the right components to look at?

network.buses_t.marginal_price

network.generators_t

And do you have any clue why I would get empty dataframes? Is it because didn't allow capacity expansion and couldn't solve the problem?

In fact, "n.buses_t" had all empty data frames..

Thank you so much for your help!

Hye Min Park (Hey-Meen)
Doctoral Fellow | Assistant Policy Researcher
Pardee RAND Graduate School | RAND Corporation

[Fabian Neumann]

Hi,

yes, n.buses_t.marginal_price will show clearing price in €/MWh for successfully solved models.

And n.generators_t.p will give you optimised dispatch (which might be lower than available power due to curtailment).

These fields will be empty, if the optimisation did not complete successfully.

Best wishes,

Fabian N

[Hyemin Park]

Thank you so much Fabian for your quick response!

So something went wrong when my colleague got the same generation values for solar around all hours from "network.generators_t"?

Ideally if we get to run correctly we should be able to see solar generation based on min(generation capability, deliverability) when we output "network.generators_t"?

Thank you!

Hye Min