Dealing with variables related to units in different dimensions

[danielhuppmann]

This issue extracts one discussion from the review of #12 (see this comment), which needs to have its own dedicated discussion.

There are variables that may be measured in units in different dimensions, the first example being the storage capacity of a reservoir. This can be measured as volume of water (relevant for hydrological models) or as energy stored (relevant for electricity system models). In these cases, the values can be converted in principle but this requires additional information and depends on the reservoir and turbine characteristics.

[For clarification, this is different from unit conversion from exajoule to megawatthours, which is a straightforward multiplication]

Two options to deal with this:

1. use the same variable e.g. Reservoir Capacity|Electricity|Hydro|Pumped Storage|Power Plant 1 but with different units depending on the model context
   • not having the same (type of) unit for a specific variable makes it difficult/impossible to do automatic unit conversion
2. use distinct variables connected to different units
   • in this case, we could use Storage Capacity for energy and Reservoir Capacity for water volume
   • but this might become complicated if there are many such cases

@arght @sandrinecharousset @erikfilias @tburandt

Question 1 - do you have other examples of such conflicts? Question 2 - any preferences for option 1 or 2?

[arght]

I prefer option 2 because, to my understanding, there is no direct connection between both variables: reservoir capacity and storage capacity. In the first case, the model is dealing with volume units to represent the reservoir management (which will be the case of a hydro scheduling model) and, in the second case, not such detail is considered (because the focus of the model is not on hydro scheduling, e.g., openTEPES or because the model works with aggregated/equivalent reservoirs)

[sandrinecharousset]

We (EDF) usually use the name Volume when it is a real volume, of water, measured in hm3, and storage when it is about energy. Direct conversion is always difficult if not impossible:

• when it is a real reservoir, the efficiency of the turbine depends of the amount of water left in the reservoir; it is even more complex when dealing with complex cascading systems
• same as @arght reservoirs used in models are not always real - eg often we model national reservoirs, here the conversion requires a whole model For me the volume of water in a reservoir is somtthing that is different of the amount of energy that this reservoir is storing, then would deserve different names. Moreover, the data that are using one of the units (eg the volumes in hm3) will never be sed by the models that require storages in GWh, which is another justification for having different names

[erikfilias]

I prefer option 2, too. Differently, I would like to recommend the use of Maximum Capacity and Minimum Capacity to declare the bounds of the electricity production of a power plant.

Normally in some known test cases, it is named maximum power and minimum power, but it could be confused in other models, no?

[sandrinecharousset]

Regarding the question are there others, I only see that proble in the field of hydropower:

• Inflows : same thing you may have inflows in cubicmeters , or MWh and as for the volume/storage it does not represent the same thing. We use the same name Inflows and derive from context but if someone has got an idea of how to rename inflows when dealing with energy (we also have 'inflows' to batteries, in that case the name inflow in particularly not well adapted....)
• Pumping capacity / Turbining capacity should be used only for hm3 ; and other words for MWh....

[sandrinecharousset]

@erikfilias, I am a bit bothered with maximum capacity and minimum capacity for the bounds of the electricity production of a power plant. We usually call them maximum power and minimum power when it refers to a single ower plant (then technical constraints); We usually use the name capacity when refering to one kind of generating power in a region: eg the capacity of OnshorePV in france. Which is I believe also the varable name used for IAMs , where individual plants do not really exist. To be more detailed, when we are working on prospective case studies (2050), we do not know the individual plants but we make asumptions on the capacity (in the sense capacity of a technoogy in a region) and we know that each technlogy has some tehcnical constraints , maximum/minimum power are some of the numerous ones. To go further, Maximum Capacity is sometimes used in investment models to define the upper bound of investment of a given technology, and minimum capacity the lower bound (for models that include phasing out)

[erikfilias]

Well, I only followed the recommended concise example and tried to extend it to discuss a possible confusion in the terms.

• Capacity|Electricity|Hydro|Pumped Storage|Power Plant 1 for maximum generation capacity
• Reservoir Capacity|Electricity|Hydro|Pumped Storage|Power Plant 1 for reservoir capacity

Indeed, I make out and prefer the terms of maximum power and minimum power. If there is not a problem with it. We can use it, I guess.

[sandrinecharousset]

I also add the question of energy: Final Energy , Primary Energy, Secondary Energy variables have Ej/yr unit. When working on electricity we rar more used to GWh/yr There exist a simple conversion (1Wh=3600J) , and I have seen a conversion function was added to pyam but there is still the question of the definition of variables : when I added variables related to eectricity in the yaml files I chose GWh, because this is the usual unit.... Not sure I shouldn't have chosen EJ/yr...

[sandrinecharousset]

I also add the question of energy: Final Energy , Primary Energy, Secondary Energy variables have Ej/yr unit. When working on electricity we rar more used to GWh/yr There exist a simple conversion (1Wh=3600J) , and I have seen a conversion function was added to pyam but there is still the question of the definition of variables : when I added variables related to eectricity in the yaml files I chose GWh, because this is the usual unit.... Not sure I shouldn't have chosen EJ/yr...

There are several options:

allow a list of (recommended) units for each variable (as long as they are the same type, e.g., energy) store the data as uploaded by the teams on the platform (potentially making it difficult to compare across models) convert units during upload to one standard unit (meaning that data displayed and downloaded will have a different unit that what is uploaded by the modelling team) only allow a specific unit for each variable => but how to decide which one? We plan to have a feature for unit-conversion in the Scenario Explorer display, but this will probably not be ready before autumn 2020.

=>My personal feeling is that whenever it is possible to get a compromise on units we should target to have one onIy but it may be difficult not to keep a (short) list of variables with 2 units ( eg the EJ vs GWh). Probably when it is about units that are very close (GW/MW) it will be easy to chose only one, but when it is units that may change the understanding of the data (which shouldn't be numerous), it may be impossible.... The unit-conversion feature above proably solves all this, so if it has to be only temporary maybe we can work with 1 unit for all variables (but which one?)