Modeling of pipeline network and H2 feedin

[fwitte]

This issue will be updated with different possible solutions. The final decision should be aimed for the end of January.

Issue

The H2 feedin into the CH4 grid should be restricted to a maximum share in the eGon2035 scenario.

Possible Solutions

Solution 1

Parallel "virtual" CH4/H2 mixture with fixed mass fractions pipeline network:

• For every CH4 bus in the system, there will be a CH4_H2_mixture bus.
• CH4 buses and the corresponding H2 buses are connected to the CH4_H2_mixture bus with unidirectional links.
• The flow of H2 into the mixture is at a fixed ratio to the flow of CH4.
• Both pipelines (CH4 and CH4_H2_mixture) share a common capacity value (link_p of CH4 + link_p of CH4_H2_mixture <= p_nom).
• Every load, OCGT, CHP and gas boiler is disconnected from the CH4 grid and connected to an own load bus instead.
• The load bus is connected with unidirectional links to both the CH4 pipeline and the CH4_H2_mixture pipeline.

Open questions:

• How to deal with storage?

Benefits

• It works :)
• The H2 share can take any value between 0 and the maximum allowed share. Locally higher shares are not possible.

Downsides

• Needs quite a lot of adjustments to the data generated by the data model.
• Introduces many new constraints.

Solution 2

Restrict the feedin of H2 relative to the production and feedin of CH4 at each CH4 bus.

• The H2 feedin cannot be higher than the maximum allowed share based on the CH4 production by Methanation (minus SMR) and CH4 import from generators for each timestep at each CH4 node.

Benefits

• It somewhat works (see downsides).
• Only one additional constraint, no modifications to the data model.

Solution 3

Restrict the feedin of H2 relative to the total production and feedin of CH4 in the systems.

• The H2 feedin cannot be higher than the maximum allowed share based on the overall CH4 production for each timestep within the whole system.

Benefits

• It somewhat works (see downsides).
• Only one additional constraint, no modifications to the data model.

Downsides

• Locally, higher shares of H2 (even pure H2) are possible.

Implications

Data model

Depending on the method that will be implemented to model H2 feedin into the CH4 pipeline grid, additional data are required for the optimization.

Solution 1

There are a lot of modifications to make based on the current state of the data exported from egon-data. However, all modifications can be made in eTraGo and do not require the changes to be made in the egon-data process.

Solution 2

No data model adjustments required.

Optimization model

Depending on the method that will be implemented to model H2 feedin into the CH4 pipeline grid, additional constraints are required for the optimization.

Solution 1

• Two corresponding pipelines share a common total capacity.
• Pipelines must be unidirectional for this to work, and therefore, each pipeline must be splitted into two.
• A mixture bus is required for every CH4 bus.
• A bus is required for every CH4 bus, which is connected to any number of loads (industrial load, CHP, OCGT, gas boilers).

Solution 2

• Add constraint to limit the H2 feedin with respect to the CH4 generation at that node.

Decisions

• What solution to implement?
• Where to put what part of the code?
  • i.e. egon-data, eTraGo before/after clustering and aggregation, ...?
• ...?

[ClaraBuettner]

I had another idea for the H2 feedin limitation, I hope it's fine to add this here:

There are fixed gas loads per node (industry and rural heat buildings without a heat pump), so we already know in advance that at least this amount of CH4 has to be available at each node and time step. We could assume that the H2 feedin at each node and snapshot can't be more than X% (e.g. 15% vol.) than the fixed gas load. That way we ensure, that each fixed CH4 load can be supplied by maximum x% of H2. It would be a conservative assumption, because there are actually also other gas demands which are part of the optimization (CHP, OCGT) and not considered in this limitation, but the resulting share would be always smaller than the selected limit. And they are most likely smaller that the rural heat loads.

To model this, we can set a p_max_pu time series for the H2 feedin links which are calculated based on the CH4 loads. This could be easily done in eTraGo and might not effect the calculation time that much, we also use that for modelling other technologies.