22-05-2020, 03:30 PM (This post was last modified: 22-05-2020, 03:31 PM by pfortes.)
Hi
I am trying to model the blending of H2 in the natural gas grid. Because the FIXOM and the emission factors of Comm-OUT are linked to the share of H2 I am predefining different blending (5%, 10%, etc), each representing a process.
My problem is that TIMES is choosing different blending at the same time, which from a “reality” point of view may not make sense – you do not have a 100% natural gas grid, and others with different H2 blendings.
Do you know any way to “tell” to the model: if you choose process A you cannot choose B? I guess as a linear optimisation model there is no “if” condition but maybe there is an alternative.
Do you have any suggestion?
Thank you in advance
I concur with the statement that there is no “if” condition in convex programming. If you allow blending of 0%, 5%, 10% etc., the feasible solution space will include any linear combination between the alternative points.
Could you clarify a bit in what aspects it does not make sense in choosing e.g. half of the pipeline flow with 5% and half with 10% blending, and thereby having an average rate of 7.5%? Of course I understand that investments into two smaller pipelines instead of one bigger would not be very sensible, and so I guess I would model the blending options with a single process, with flexible blending between e.g. 0–15%. Increasing variable operating costs along with increasing blending rates could be modelled on the H2 input flow, or even by adding several input flows corresponding to the total H2 input, with share bounds and increasing variable costs. In addition, if higher blending rates should also lead to higher fixed O&M costs, I think that could also be modelled, at least by adding a dummy process for the flows with higher rates. But I think usually the FOM increase could be reasonably well approximated with a variable cost increase. Emissions depending on the blending rate could likewise be modelled on the natural gas / hydrogen flow(s).
25-05-2020, 03:02 PM (This post was last modified: 25-05-2020, 03:03 PM by pfortes.)
Dear Antti
Thank you very much for your fast reply.
The ideia of increasing variable operating costs along with increasing blending rates is good however I have two problems:
1. The relation between the blending and the VAROM is not linear (these values were given by colleagues from the Mechanical Department I need to ask them why, but maybe this can be changed).
2. Different blending shares will also have different CEFF and FIXOM on the end-use sectors that consume those commodities. For example RSDGAS10 (gas with 10% of blending) has a higher CEFF than RSDGAS6 (6% blending), and RSDGAS20 (20% blending) requires the substitution of parts of the technology, so is a kind of new technology with higher FIXOM (this is not totally correct it is just to simplify modeling).
I guess there is no solution for point 2, or do you think there is?
Regarding your sentence “adding several input flows corresponding to the total H2 input, with share bounds”, sorry I am not fully understanding. Are you saying, considering several input commodities representing different H2 blendings?
25-05-2020, 03:29 PM (This post was last modified: 25-05-2020, 03:48 PM by Antti-L.)
Thanks for the useful additional clarifications.
Concerning your point 1), you can model a non-linear relation for variable costs as a function of the blending rate, as long as the costs are increasing with higher rates.
Concerning point 2), this is an important clarification. I guess you then had modelled also different output commodities for the different processes (0%, 5%, 10%, etc), is that correct?
Modeling different output commodities for each blending rate would become cumbersome with the single process approach, and therefore, maybe you original approach was basically reasonable after all, but with the refinement that I think you could add a pipeline process for the total flow, such that the pipeline investment and capacity will be a single process?
So, could you still clarify in what aspects you think it would not make sense in choosing e.g. half of the pipeline flow with 5% and half with 10% blending, and thereby having an average rate of 7.5%?
Concerning your last question, I was thinking of several H2 input flows with share bounds, which would thus correspond to e.g. blending rates 0–5%, 5–10%, 10–15%. But as said, this would become cumbersome if you also need to have a different output for each blend.
Right
I have different blending processes, each with different Comm_OUT.
With this approach I can also put the emissions on end-use sectors and not on the natural gas/H2 blending processes.
"but with the refinement that I think you could add a pipeline process for the total flow, such that the pipeline investment and capacity will be a single process". Are you thinking in a process where the Inputs are for example GASNAT, GASNAT6, GASNAT10, GASNAT20 (each representing different blendings)? Can I make the model only choose one input?
25-05-2020, 04:16 PM (This post was last modified: 25-05-2020, 08:38 PM by Antti-L.)
No, I meant a pipeline capacity and investments for the total flow, because I assume that you can use the same basic pipeline infrastructure for any of those blending rates. But maybe I am still missing something?
Therefore, could you still clarify in what aspects you think it would not make sense in choosing e.g. half of the pipeline flow with 5% and half with 10% blending, and thereby having an average rate of 7.5% in the total flow through the pipeline? The total flow would consist of some mix of the components GASNAT, GASNAT6, GASNAT10, GASNAT20, as chosen by the model. I don't see any option of forcing the model to choose only one of them (without resorting to MIP), because it would not comply with the convexity requirement.
Actually, when all the outputs have only predefined fixed blends, using a single process would seem convenient, like in the example illustration below:
(Sorry Antti I couldn't answer earlier) Thank you very much for the figure. I will add this process. In the distribution network it makes sense to have different blendings but in the transport network if there is continuous production of H2 I think it should be only one blending - assuming that there is only one pipeline (as in Portugal). Thnks again for all your help.
