Include "Pumped Hydro" in Storage

[robertpietzcker]

A discussion started, because @robertpietzcker shifted "Pumped Hydro" from tag_electricity_input_types.yaml (and thereby from the "Secondary Energy|Electricity" and "Capacity|Electricity" variables) to the tag_storage_types.yaml storage subtypes (and thus "Secondary Energy|Electricity|Storage" and "Capacity|Electricity|Storage|XXX") in this PR:
#352

The reason why this shift was proposed is twofold:

1. most databases seem to follow the convention that both capacity and output from pumped hydro storage is NOT counted under hydro, but separately (or not at all), thus comparisons to real-world data are much easier if the same convention is used in model reporting
2. More consistent grouping: pumped hydro storage with no external inflow is a pure storage technology, and thus should be treated as such. It cannot generate electricity by itself, it fundamentally requires electricity to pump water to the higher basin, and one cannot do anything with this water other than reconverting it into electricity. Having it included with the other storages will make the summations and analysis clearer.

on 1: some examples:
AGEB, the German energy statistics work group, reports for 2020 18.7 TWh electricity generation from hydro, explicitly stating "without pure pumped hydro storage", and output from pumped hydro storage at 6.6 TWh.

From this it can be deduced that IRENA hydro numbers also do not included pumped hydro storage, as they state 18.7 TWh from hydro for Germany in 2020 (https://public.tableau.com/views/IRENARETimeSeries/Charts?:embed=y&:showVizHome=no&publish=yes&:toolbar=no)

IEA energy balances also only report generation excluding pumped hydro, as they similarly report 18.3 TWh for Germany in 2020.

The EMBER database also reports only generation excluding pumped hydro, as they also report 18.7 TWh (https://ember-climate.org/data/)

The same is true for capacity, where both IRENA and EMBER reports 5.4 GW hydro power capacity in Germany, which is without pumped hydro storage, as pumped hydro storage capacity ALONE is around 6-7GW (eg https://en.wikipedia.org/wiki/Pumped-storage_hydroelectricity#Worldwide_use )

That being said, I am fully aware that reporting conventions are never perfect, and always open to debate :-)
So I am interested in arguments in favor of having "pumped hydro" within "hydro"

[robertpietzcker]

@sandrinecharousset @erikfilias @danielhuppmann

[sandrinecharousset]

Many thanks @robertpietzcker for this very detailed post.

I just add a few more information:

• We are having this discussion because we need to take a decision about where to include Pumped Storage, as it has in the past been included in the tag electricity, then kicked out and put in the tag storage, then kicked out again and put back to the tag electricity..... and it seemed to me we would like to avoid another round of this :-)

My requests are:

• Avoid to have 2 different variables for the same data (eg Hydro|Pumped Storage and Pumped Hydro).
• Minimise the changes as this means that former datasets / linkage scripts may not work anymore
• Ensure that all the variables that are currently generated with Hydro|Pumped Storage in electricity are still generated if we move Hydro|Pumped Storage to storage (this can be done by adding variables such as Maximum Power|Electricity|Hydro|Pumped Storage to power-plant.yml for instance

For the rest I agree with how @robertpietzcker defined the different sectors of hydropower and I also agree that it makes sense to move it to storage.

[sandrinecharousset]

I include other persons who might be interested in this discussion @arght @loeffko @IngeborgGraabak @phackstock @dc-almeida

[erikfilias]

• same data (eg Hydro|Pumped Storage and Pumped Hydro).
• Minimise the changes as this means that former datasets / linkage scr

Thank you, @robertpietzcker, for the discussion. In Spain, we have both pure and mixed pumped hydro systems. I agree that pure pumped hydro, which only acts as a storage system, should be categorised as storage. However, mixed systems, which also generate electricity from natural inflows, behave more like traditional hydro.

To reflect this, I propose to split pumped hydro into

• Pure pumped hydro (under storage).
• Mixed pumped hydro (under hydro).

[sandrinecharousset]

@erikfilias I fully agree with you, it is the same in France. But I think the difficulty we have here is that we both want to categorize hydro when aggregated and hydro when disaggregated. The typical definition of Pumped Storage as detailed by Robert refers to the aggregated capacity at regional (or country) level. In this case, usually the mixed pumped hydro capacity is not detailed as it is, but in the end the pumping goes to pumped storage and the turbining goes to either reservoir (when there is an aggregated modelling of a valley and this turbining capacity is directly linked to the global flow in the valley) or to 'run of river ' (in french 'fil de l'eau et éclusées' , don't know how to translate 'éclusées', but it is related to the turbining capacity of short term storage reservoirs, usually downstream of a valley)
We could define Pure Pumped Storage (or closed loop pumped storage) and mixed pumped storage (I would suggest Hydro|Pumped storage|Closed loop), but this will only be used for disaggregated modelling (as you do in openTEPES), but I fear it wouldn't solve our issue......

[robertpietzcker]

Thank you, @erikfilias, for the comment!

One question: wouldn't the "mixed pumped hydro" fit well in the "Reservoir" category? If not, I am perfectly happy to have a "mixed pumped hydro" category under "Hydro"

[robertpietzcker]

@sandrinecharousset could you explain once more /rephrase the use case/challenge you see with either using "reservoir" for all pumped hydro with relevant natural inflow, or adding this "mixed pumped hydro" category under hydro? I have to admit I didn't fully get the problem above.

Is it that some statistics that you want to compare to use different aggregations? if yes it would be helpful to know which ones these are.
Or is it that in your model you have a technology where you have one turbine that generates electricity, but the total output is the sum of the natural inflow and pumping? but then it would be easy to split the energy amounts in the reporting, I think.

[sandrinecharousset]

Thank you, @erikfilias, for the comment!

One question: wouldn't the "mixed pumped hydro" fit well in the "Reservoir" category? If not, I am perfectly happy to have a "mixed pumped hydro" category under "Hydro"

I think it is far more complex than that as it refers to the way hydro valleys are modelled at different granularity levels. As I mentioned above, from wht I know, thise mixed pumped storage units are sometimes included in the Reservoir, sometimes in run of river and the pumping capacity usually is in pumped storage......

This being said, we could go back to what we need from this nomenclature:

• we need to have all the variables used by the different models
• we need to avoid misunderstandings (ie when using data from a model in another model, by different teams, we have to be sure of what it is, from the variable name and thus the definition)
• we need to not have 2 variables with different names representing the same thing
• we need consistency in the 'levels' of variables (eg sum of Capacity|Electricity|Hydro|* = Capacity|Electricity|Hydro)

And we have this difficulty with aggregated/disaggregated modelling......

Sorry for this comment with only questions and not answers..... will try to think about it but at the present minute I cannot find sth I find ok.....

[erikfilias]

Thank you, @sandrinecharousset, for your detailed input. I agree that the classification of mixed pumped hydro is more complex, especially given how hydro valleys are modelled at different levels of granularity. As you mentioned, these systems can sometimes be classified as reservoir or run-of-river, while their pumping capacity is often classified as pumped storage. This variability adds to the challenge of creating a nomenclature that works for both aggregated and disaggregated models.

To meet the needs you have outlined:

1. We need to cover all variables used by different models and ensure compatibility across teams and use cases.
2. Clarity is key - each variable name and definition must avoid misunderstandings, especially when data is shared between models.
3. We need to avoid duplicate variables with different names representing the same thing.
4. Consistency is key - summaries such as Capacity|Electricity|Hydro must be consistent across all sub-levels.

With these points in mind, I suggest the following approach:

1. For tag_electricity_input_types.yaml:

• Hydro|Pumped Storage: Keep this for systems with short-term storage, including pumped storage, in line with ENTSO-E. This allows it to be used for both aggregated (where pumping capacity is included in pumped storage) and disaggregated models.
• Hydro|Reservoir: For systems with seasonal storage capacity, possibly including some pumping capacity.
• Hydro|Run of river: For schemes without storage capacity, based purely on natural flows.
• If necessary, for more detailed models, we can introduce a subcategory such as Hydro|Pumped Storage|Mixed to capture the hybrid nature of these systems, where they generate electricity from both inflows and pumping.

2. For tag_storage_types.yaml, pure pumped storage would still be categorised as storage, separate from electricity generation.

I think it can provide flexibility for both aggregated and disaggregated modelling, avoids duplicate variables and maintains consistency across the classification hierarchy.

One additional thing is about using the complete term Hydropower as I mentioned in #352.

[sandrinecharousset]

@sandrinecharousset could you explain once more /rephrase the use case/challenge you see with either using "reservoir" for all pumped hydro with relevant natural inflow, or adding this "mixed pumped hydro" category under hydro? I have to admit I didn't fully get the problem above.

Is it that some statistics that you want to compare to use different aggregations? if yes it would be helpful to know which ones these are. Or is it that in your model you have a technology where you have one turbine that generates electricity, but the total output is the sum of the natural inflow and pumping? but then it would be easy to split the energy amounts in the reporting, I think.

@robertpietzcker , the typical use case that I am familiar with, and have implemented in openENTRANCE and is implementing in other projects now is that we are using the results of energy system models (typically GENeSYS-MOD) as inputs to a power system model, with an aggregated representation of hydropower (with the typical categories Reservoir=no pumping, seasonal management, pumped storage,= dayly/weekly storages, run of river= evrything else, including not only pure run of river but also a part of the downstream capacities of some hydro valleys). For feeding my model correctly I need to exactly know what each of the variables really is. Pumping, even when directly connected to a big lake, is not managed as 'seasonal storage' so it is usually accounted for in Pumped storage and this is how we have interpreted the data we have been using.

But the same model that I am using (plan4res) could be used with a disaggregated representation of hydro (cascading valleys). In that case I would use different data, and the mixed pumped storage would be represented in details.

Anyway, the thing is that what we want is to know exactly what each variable means..... and we need to have variables to host the data that do exist (eg the IEA data, the GENeSYS-MOD results.......).

[arght]

Just a minor question. The names of the mixed pumped storage hydro I recommend to be open-loop, while the name of the pure should be closed-loop

[robertpietzcker]

another short question, @erikfilias, on "Keep this for systems with short-term storage, including pumped storage, in line with ENTSO-E" - could you point me to the place in the ENTSO-E documentation where this split is described, so I can understand how they are doing the split?

Edit: ah, maybe I know understand - do you simply mean that because ENTSO-E has the names "Hydro Pumped Storage", "Hydro Run-of-river and poundage" and "Hydro Water Reservoir", we should have the same 3 categories under "hydro"?

[robertpietzcker]

With these points in mind, I suggest the following approach:

if I understand your proposal correctly, I would then simply add "Closed-loop Pumped Hydro" in the "Storage" tags, but keep the three subcategories of "Hydro" in the electricity-input tags, correct?

[robertpietzcker]

on the renaming to "Hydropower" I am skeptical - I think this is a change (with all its complications) with limited benefit. Theoretically, it could be confused with hydrogen, but most other sources I know of also use "hydro" when reporting hydropower.
Also, I have always seen "hydrogen" either spelled out as "hydrogen" or put as "H2", so I don't think this can be confused for hydro.

[sandrinecharousset]

on the renaming to "Hydropower" I am skeptical - I think this is a change (with all its complications) with limited benefit. Theoretically, it could be confused with hydrogen, but most other sources I know of also use "hydro" when reporting hydropower. Also, I have always seen "hydrogen" either spelled out as "hydrogen" or put as "H2", so I don't think this can be confused for hydro.

I also would keep Hydro, also for the reason that Hydro can be seen as a fuel (as Gas) for producing electric ity and the variables with Hydro always have Electricity before, Electricity|Hydropower is redundant for me.

[sandrinecharousset]

With these points in mind, I suggest the following approach:

if I understand your proposal correctly, I would then simply add "Closed-loop Pumped Hydro" in the "Storage" tags, but keep the three subcategories of "Hydro" in the electricity-input tags, correct?

Wouldn't it be Hydro|Pumped storage|Closed-loop ?

[sandrinecharousset]

@robertpietzcker see https://eepublicdownloads.entsoe.eu/clean-documents/sdc-documents/MAF/2020/Hydropower_Modelling_New_database_and_methodology_V1_0.pdf#:~:text=In%20previous%20market%20modelling%20studies%20performed%20at%20ENTSO-E%20(e.g.%20Ten page 6 (this is from 2019, probably there is a new version attached to the TYNPD 2024?)

[arght]

Let me add my viewpoint.
Closed-loop pumped storage corresponds to daily/weekly management pumped storage because the upper reservoir usually has a limited size, sometimes because it is an artificial one. No natural hydro inflows
Run of the river corresponds to very small reservoirs with less than daily storage capacity and, consequently, almost no flexibility
Reservoir hydro corresponds to a hydro power plant with a reservoir. No pumping capacity
Open-loop pumped storage corresponds to a reversible hydro power plant (turbining and pumping) with a (tipically large) reservoir with daily/weekly/seasonal management.

This classification is being used in the TYNDP 2024. These are the data for Spain from this TYNDP

Hydro PEMMDB 3.4

Run of River - Total turbining capacity (MW) | 3424.665616
Pondage - Total reservoir capacity (GWh) | 0
Pondage - Total turbining capacity (MW) | 0
Reservoir - Total reservoir capacity (GWh) | 14775.0725
Reservoir - Total turbining capacity (MW) | 11414.19
Pump Storage (open loop, with natural inflows) - Total reservoir capacity (GWh) | 5763.86
Pump Storage (open loop, with natural inflows) - Total turbining capacity (MW) | 4223.32
Pump Storage (open loop, with natural inflows) - Total pumping capacity (MW) | -3964.005633
Pure Pump Storage (closed loop, no natural inflows) - Total reservoir capacity (GWh) | 101.221
Pure Pump Storage (closed loop, no natural inflows) - Total turbining capacity (MW) | 5304.485333
Pure Pump Storage (closed loop, no natural inflows) - Total pumping capacity (MW) | -5204.59

and for France

Hydro PEMMDB 3.4 |  

Run of River - Total turbining capacity (MW) | 3424.66562
Pondage - Total reservoir capacity (GWh) | 0
Pondage - Total turbining capacity (MW) | 0
Reservoir - Total reservoir capacity (GWh) | 14775.0725
Reservoir - Total turbining capacity (MW) | 11414.19
Pump Storage (open loop, with natural inflows) - Total reservoir capacity (GWh) | 5763.86
Pump Storage (open loop, with natural inflows) - Total turbining capacity (MW) | 4223.32
Pump Storage (open loop, with natural inflows) - Total pumping capacity (MW) | -3964.00563
Pure Pump Storage (closed loop, no natural inflows) - Total reservoir capacity (GWh) | 101.221
Pure Pump Storage (closed loop, no natural inflows) - Total turbining capacity (MW) | 5304.48533
Pure Pump Storage (closed loop, no natural inflows) - Total pumping capacity (MW) | -5204.59

This classification I think is valid for aggregated representation of the hydro subsystem. If you want to represent real vally and hydro cascades then the topological waterways must be considered. From my viewpoint this detail can't be considered in energy or integrated assessment models

[robertpietzcker]

@robertpietzcker see https://eepublicdownloads.entsoe.eu/clean-documents/sdc-documents/MAF/2020/Hydropower_Modelling_New_database_and_methodology_V1_0.pdf#:~:text=In%20previous%20market%20modelling%20studies%20performed%20at%20ENTSO-E%20(e.g.%20Ten page 6 (this is from 2019, probably there is a new version attached to the TYNPD 2024?)

awesome, thanks!
very interesting to see the detailed split for the different countries!