areleu
commented
1 year ago I have been reading further in the topic and I got the impression that the problems related to example 2 go deeper than what I thought. I came across this paper https://content.iospress.com/articles/applied-ontology/ao200234 which handles ontologies of the physical world and their experiments using, as I understood, some kind of modal mereology. I am picking this concept very carefuly and I am probably misunderstanding something but so far I have failed to find other way of describing this usage of "sister" ontologies in composition to use mathematical formalizations to describe truths of the physical world.
In general I think that, in order to have a solid semantics that describe the workflow of ask question -> design modelling experiment -> answer question , the OEO would eventually need some kind of complemnting ontology to implement something like what is presented in the paper I just posted.
This might be unrelated to the main topic of this Issue, but it is so fundamental that I don't think that makes sense to make an Issue in the ontology itself.
l-emele
Description of the issue
I was considering writing this comment as part of #839 but after thinking about it for a bit I think it derserves its own thread. This is one of the series of issues that I will rise based on the Energy systems cost workshop from KU Leuven.
In short I think that, althought the definition of levelised cost of electricity is fitting, its position in the hierarchy is not appropiate. Despite of it being called a cost, I think a better parent class for this and other similar concepts should be metric. Since the theory behind this is rather dense and proably deserves a well thought explanation in itself, I will try to prove my point with two application examples of usage of the metric from two points of view.
Example 1: A renewable energy integration report
Let's say we have an analysis project where some kind of model is used to calculate the installed capacity and costs of the energy infrastructure of a particular region. The analysis is performed by having model runs holding a N number of data variables constant (For example demand, existing capacity) and a single variable P (wind energy penetration for example) is changed across these runs. From these runs one gets C (total discounted cost) and G (generation) of the generator i associated with wind energy. In table 1 we can see how such an output would look like in a tabular structure.
[Here is the code for the plot](https://gist.github.com/areleu/88916220ea21be8b3c52ad287f9c3531) In this example the LCOE is calculated based exclusively on model values. And it is used to do a comparision of different scenarios of the same base case. In this sense, it does not represent Cost of something because for something to have a cost there has to be buying and paying parties. By this I don't mean that using this metric is right or wrong ([Although one can argue that is not the best for doing these analyses](https://www.aeaweb.org/articles?id=10.1257/aer.101.3.238)) but that it is not used as a cost ever. And it should not be used as a cost. To make this last point more clear I will use the next example. ### Example 2: An agent based simulation excercise. First of all, what I am writing here can be lacking. I am not a ABS expert and I migth have misunderstandings, please prove me wrong if that is the case. In this example we have a scientst or group of scientists using an agent based simulator model that has four agents: a producer, a broker and two consumers. The broker then will maximize its profit by leveraging the sell price to the consumers. Due to problem scope, computing demands and data availiblity our scientists decide that they want to model the consumers demands in high resolution but the producer will provide the electricity at a fixed price. To get the fixed electricity price, our scientists rely on the results of an optimization model. This model calculates values like in example 1 and from that they calculate the LCOE and decide to use it as the elctricity price. The input file for our producer agent might look like this: `{ "electricity_price": 100, "capacity": 20 }` Where the price is in €/MWh and capacity is MW. In this case the entity refered, even if it is being simplified is the electricity price/cost. In this regard, if the value of LCOE is being used it is done as proxy to what the electricity cost is. Proxy values are not uncommon in the field but I think the property `models `covers them. I know this example is rather clunky and I doubt people use this kind of boundary conditions in their models but I see no better way of making this point. ## Ideas of solution In example one, one would use the ontology concept of LCOE to annotate the data of a table behind the plot. In this case the value is being used as a metric to compare several scenarios. In example two, our researchers have no dat availible so they use this as a proxy value of electricity cost. In neither case it is being used directly as a cost. There is three possible solutions I would suggest. ### Solution 1: Add a subclass of `quantity value` called `indicator`, `metric` or something like that. The correct word can be up to discussion. Then I would say that LCOE and its cousins like system LCOE, vaLCOE etc. should be instances of this class. For the right definition I would need a little help. I think this is stepping on the abstact side, and as I understand BFO does not deal with abstract things, but being practical these values are being used in reports so I think this is relevant. One could also add a subclass of metric and call it 'economic value metric' I am not really sure if this is necessary though. ### Solution 2: Add a property to quantity value that makes it a metric. Or have something that would aid in differentiating things that may have similar units to some quantity value but have a different meaning than it. ### Solution 3: Do nothing. ## Workflow checklist - [x] I am aware of the [workflow](https://github.com/OpenEnergyPlatform/ontology/blob/dev/CONTRIBUTING.md) for this repository