4 ms·

90% is pretty close to the 50 days of empty battery. Keep in mind that on empty battery days, you still produce some energy as well. My approach was very very

90% is pretty close to the 50 days of empty battery. Keep in mind that on empty battery days, you still produce some energy as well. My approach was very very simple: if production is higher than consumption, store energy. Energy storage is limited, so when the stroage is full, energy is lost. If consumption is higher then production, take the difference from the battery. I have not considered any smart usage patterns or any advanced concepts at all. Also keep in mind that this ignores transport an heat. Pure electricity was what I looked at and that is somewhat around one third of the total energy used in Germany.

jhgb 5y ago
> 90% is pretty close to the 50 days of empty battery. That's not quite how I understood "50 days of empty battery", but fair enough. This is not something that can't be solved, though. Even the 90% estimate is the worst case estimate, since there's actually a high likelihood of 3 TWh of storage being sufficient for 100% of renewables in any given German year. > My approach was very very simple: if production is higher than consumption, store energy. Energy storage is limited, so when the stroage is full, energy is lost. If consumption is higher then production, take the difference from the battery. That's Sinn's approach ("waste no energy and store everything, no matter the cost"), which yields significantly higher total costs (and storage capacities) than Zerrahn's approach which actually takes costs into consideration and leads to much smaller storage capacities on the basis of throwing energy away or using it opportunistically being actually cheaper than storing it for later grid use. > Also keep in mind that this ignores transport an heat. Transport and heat actually simplify things significantly for the grid since charging of EVs in Europe can be shifted around by as much as several days while heat will have been largely solved by the EU's 2010 and 2012 energy efficiency directives.
- turbinerneiter 5y ago
  I'm sure that there is a lot of smart things that can reduce the amount of storage needed. I'm not a researcher in this field, I wanted to do simple but valid calculations that represent a realistic scenario, accepting being more pessimistic to keep it simple. I also found that increasing the production by 20% does more than increasing the stroage by 10x. I think at 200% renewable production and 2 TWh I ended up with 0 empty battery days in my simple model. So I believe you when you say that the storage can be smaller. I hope I can also run these numbers myself if I find some time. I am however also a bit sceptical about all these smart optimizations. It's hard for me to judge hoe much behavioural changes that would need. I'd rather overbuild production and storage than have people reject renewables, because they feel reminded of the DDR when they get their charge-time allotment to be consumed during a certain time only. Thanks for the names for the approaches, will check that out! I didn't do any research on existing material, I wanted to figure it out on my own first, to get a better feeling for the problem. Often when is start with the papers, I don't actually end up really studying the problem, because I hunt trough (often hard to understand) papers and end up giving up being overwhelmed.
  - jhgb 5y ago
    I'm pretty sure that all you need to start with is linear programming and the Aggregate Production Planning model -- which is usually being used for physical manufacturing, but can be adapted to electricity production. Ideally you'd want a MILP solver to make automated decisions to assign certain binary variables (for example whether or not to build certain large-scale storage projects of very specific sizes -- typically PHES, since they're all different), but you can often get by by simply generating multiple models for every combination of binary variables and picking the best one. Similarly you can handle the integration of nuclear power plants, which would be integer variables (for example the number of 1 GWe blocks in operation) -- just generate N models for 0..N-1 nuclear reactors on the grid. I did one such model for the Czech grid some time ago but I'd have to dig it up. The basics are not difficult, though -- just split up the time period into intervals of a suitable size and observe in each time interval that Production_t + Storage_t - Consumption_t - Storage_t+1 = 0. That's what I did, having known production models but having read no papers on the renewable storage subject either. The logic is the same, though. I just got my grid information from https://open-power-system-data.org/ https://open-power-system-data.org/ and ran with it.

jhgb 5y ago

> 90% is pretty close to the 50 days of empty battery. That's not quite how I understood "50 days of empty battery", but fair enough. This is not something that can't be solved, though. Even the 90% estimate is the worst case estimate, since there's actually a high likelihood of 3 TWh of storage being sufficient for 100% of renewables in any given German year. > My approach was very very simple: if production is higher than consumption, store energy. Energy storage is limited, so when the stroage is full, energy is lost. If consumption is higher then production, take the difference from the battery. That's Sinn's approach ("waste no energy and store everything, no matter the cost"), which yields significantly higher total costs (and storage capacities) than Zerrahn's approach which actually takes costs into consideration and leads to much smaller storage capacities on the basis of throwing energy away or using it opportunistically being actually cheaper than storing it for later grid use. > Also keep in mind that this ignores transport an heat. Transport and heat actually simplify things significantly for the grid since charging of EVs in Europe can be shifted around by as much as several days while heat will have been largely solved by the EU's 2010 and 2012 energy efficiency directives.

turbinerneiter 5y ago
I'm sure that there is a lot of smart things that can reduce the amount of storage needed. I'm not a researcher in this field, I wanted to do simple but valid calculations that represent a realistic scenario, accepting being more pessimistic to keep it simple. I also found that increasing the production by 20% does more than increasing the stroage by 10x. I think at 200% renewable production and 2 TWh I ended up with 0 empty battery days in my simple model. So I believe you when you say that the storage can be smaller. I hope I can also run these numbers myself if I find some time. I am however also a bit sceptical about all these smart optimizations. It's hard for me to judge hoe much behavioural changes that would need. I'd rather overbuild production and storage than have people reject renewables, because they feel reminded of the DDR when they get their charge-time allotment to be consumed during a certain time only. Thanks for the names for the approaches, will check that out! I didn't do any research on existing material, I wanted to figure it out on my own first, to get a better feeling for the problem. Often when is start with the papers, I don't actually end up really studying the problem, because I hunt trough (often hard to understand) papers and end up giving up being overwhelmed.
- jhgb 5y ago
  I'm pretty sure that all you need to start with is linear programming and the Aggregate Production Planning model -- which is usually being used for physical manufacturing, but can be adapted to electricity production. Ideally you'd want a MILP solver to make automated decisions to assign certain binary variables (for example whether or not to build certain large-scale storage projects of very specific sizes -- typically PHES, since they're all different), but you can often get by by simply generating multiple models for every combination of binary variables and picking the best one. Similarly you can handle the integration of nuclear power plants, which would be integer variables (for example the number of 1 GWe blocks in operation) -- just generate N models for 0..N-1 nuclear reactors on the grid. I did one such model for the Czech grid some time ago but I'd have to dig it up. The basics are not difficult, though -- just split up the time period into intervals of a suitable size and observe in each time interval that Production_t + Storage_t - Consumption_t - Storage_t+1 = 0. That's what I did, having known production models but having read no papers on the renewable storage subject either. The logic is the same, though. I just got my grid information from https://open-power-system-data.org/ https://open-power-system-data.org/ and ran with it.