Let me start by presenting a simple example.
Consider that I live with my family in a small farm with few chickens that produce eggs, which are pretty much for our own self-consumption. The main reason we decide to do so is that we would like to be 100% sure that our eggs come from a clean source (i.e. not genetically modified, not heavily using antibiotics, etc.).
Although our goal is to be self-sustained, our egg production and consumption have some daily variations: some days my family consumes more than we produce, while other days we have some eggs left (in fact only few days our production and consumption are indeed the same). We also keep track of our daily production-consumption.
Looking at this situation, I feel the need to answer the following questions: How should I manage my eggs so that I can always have eggs when I need them and what should I do when I overproduce?
In my family we have identified few options: store the egg surpluses when it happen, go to the local market or big supermarkets to buy/sell, exchange/share eggs with our neighbor farms (either by buying/selling or borrowing or giving away), and exchange/share eggs within a small egg producers’ association that we are part.
This is the inventory management problem.
Now, I would like to make my solution efficient. But, I asked myself: “efficient in terms of what”? As mentioned before, my goal is basically self-consumption so in my own perspective: Efficiency means minimizing the days that I need to go elsewhere to get eggs. In this case, my efficient policy would be store my egg surpluses, and only go to other places when I need to. In this case, I use the historical data to check my efficiency (for example, by evaluating how many eggs not from my farm I consumed during the last year) and to make (better) predictions.
This approach, however, has also many drawbacks as, for instance: strong dependency of the conditions of my own chickens, local weather, space for egg storage, good balance between production and consumption, and how long a egg is still good after store it. To cope with those issues, I could constrain my problem (for example, I can only store 20 eggs) or I could also change my goal by preferring to trust in the producers’ association or in the markets.
We could also focus on maximizing profits from my egg production as suggested by a friend. But, by doing so, the optimal policy could change as well. For example, I may decide to sell my eggs to a small market (with higher prices) and buy more eggs than I produced in the supermarket since their prices are lower. This would maximize my profit. But, for my family, this seems unfair and we prefer to keep our approach.
In any case, this example shows that inventory management is tricky: it depends on the thing you are looking at (in our example, eggs), how you relate to them (for example, if the primary goal is using or exchanging), what the constraints are (storage constraint), who is involved in the problem (for example, neighbors or market participants), how I am related to the others (for example, sharing or selling my surpluses) and this could go on and on.
Besides, other things related to long-term planning can arise: if our own egg consumption is growing or if we decide to produce for the others, should we invest in more storage facilities and/or acquire more chickens?
Now think about a scenario where, instead of eggs, electricity is locally produced and consumed at my farm. How should my electricity inventory be managed? Are there storage capabilities (batteries)? What kind of data is needed to evaluate my efficiency metric and guide my decisions? What are the possible options when over- or under-producing? What is the best policy towards a given efficiency metric? What are the constraints (normally imposed by the power grid and the electricity market)? What kind of objective (for example, being self-sustained or maximizing profit) is considered?
Surely these are far from being the only questions, but give a good feeling of the problems we are dealing with.
Pedro H. J. Nardelli (and many suggestions from Florian Kühnlenz and Mauricio de Castro Tomé)