# 150

Dear S. Serna, T. Gerres, R. Cossent

I read your article:

Multi‑Criteria Decision‑Making for Renewable Hydrogen Production Site Selection: A Systematic Literature Review

My comments:

1- In the abstract you say “The review shows that different methodologies and criteria are used depending on the spatial scale of feasible alternatives. Many criteria are related to renewable energy production, such as wind speed or solar irradiance. However, most articles also consider parameters such as hydrogen demand or access to water”

The underlined is strange, all methods should be able to solve a certain problem; there is no reason on the contrary. The scenario is always the same: A set of alternatives subject to a set of criteria. The problem lies on the fact that most methods are unable to MODEL, let alone SOLVE complex scenarios, and in my opinion, this is the biggest problem in most MCDM methods. They are too elemental, since they are not able to represent reality even approximately.

Regarding feasible alternatives, it is meant to indicate that they are possible. That is, if the DM is thinking in say PV, it is feasible but not always, because it may not be enough. In some high latitudes like north Europe or Asia, this method is possible, of course, but it may be not feasible due to cells poor performance, because reduced hours of irradiation, weak irradiation, excessive snow that can cover the cells, etc. Same for tropical areas where the temperature is too high, since it reduces the PV cells power, or because sand covering them, or due to the absence of a nearby electric grid.

Since there is normally a mix of locations that fulfill the minimum requirements, if the main objective is producing energy from de Sun, you can select locations with a large irradiance and/or strong winds, and study the joint location of the two alternatives in some places, but not in others.

If you are aiming at producing hydrogen, you need water, however, you can also consider if some places are also good at the same time for irradiation and wind, because you need electricity for getting hydrogen from the sea water by electrolysis. If some MCDM cannot manage this blend, they cannot be used in a scenario like this. Precisely, MCDM is looking at taken advantage of all opportunities to select something. Therefore, the selection of the MCDM method to use overrules all other considerations.

2- “There is no consensus on how to weigh criteria affecting renewable electricity generation versus those affecting hydrogen production”

I do not see why.

3-Page 2 ”Finally, the results are usually validated by applying another MCDM method and comparing the ranked alternatives. However, this validation is not always done”

Validation is a misconception since no MCDM results can be validated, simply because there is not a yardstick to compare to.

4- “The fourth stage assigns weights to the criteria. In the fifth stage, alternatives are compared based on the weighted criteria. Finally, the results are analysed and validated.”

Regarding assigning weights, not all MCDM methods used weights; they compute the criteria relative importance by other means.

Weights measure criteria significance, which does not mean that they can be used to evaluate alternatives. As a matter of fact, they don’t (Shannon’s entropy do that, not weights).

No validation is possible. Validation is a wrong assertion by many authors. The process of comparing rankings involving different methods is useless. The only possibility of validation is to compare with reality, but since we do not know it, all this exercise is superfluous. And if we know reality, what do we need MCDM for? See the incongruence?

5- In page 2, how do you know if an alternative is feasible? You don’t know it, because feasibility means simultaneous compliance with all criteria, something unknown until a problem is solved mathematically

6- In page 3 “At the same time, classification criteria can be either benefit or cost criteria”

Not all criteria aim at reducing a cost or at increasing a benefit, there are criteria that refer to environment, regarding established pollution limits, as NOx, or to created consumption values, as in the case of water, or to production limits due to machine capacity, which are not connected to cost or benefit.

7- “Some articles also introduce new categories, such as natural resources ([9, 13, 14]) or geographical criteria”

And this is absolutely correct and must be mandatorily considered. There should be no limits for the number of criteria; they can be any, either 8, 50 or 120. The problem is if a MCDM method can manage them, and of course using resources and targets. Otherwise, the problem is flawed because all criteria need resources, and these are finite. The no using of resources in 98% of methods hints that they are unlimited, and then yielding dubious results because, they may be consuming resources that are well below of those needed. Here is an example of weakness of most MCDM methods. Thus, the DM may reach a solution, but he does nor know if it is or not feasible.

8- In table 1 you make all criteria quantitative. What about qualitative, like people opinion, people acceptance and others? If a method does not accept a mix of quantitative and qualitative criteria, if it does not take into account resources and targets, if it does not allow a mix on criteria calling for maximization, minimization or equalization, they cannot be used in serious problems.

9- It is interesting that you considered 45 criteria, a very reasonable number not often seen, to involve all class or criteria that can influence the site selection, and your concept of exclusivity is not often considered either. These criteria involve technical, economic, engineering, social criteria, as it must be.

However, I do not understand why different authors treat them separately, when the 45 can perfectly be integrated in a sole matrix. If there are say, 3 potential locations, we could build a matrix with these 3 locations in columns and 45 criteria in rows or vice versa. However, it would be more productive if we consider simultaneously in a unique matrix the 3 locations, the 45 criteria, and the 2 different sources, that is: Photovoltaic and wind, to generate only a product, hydrogen.

It can be done, if each location is subdivided in 2 columns, corresponding to the electricity sources, and placing in the intersecting columns cells, a value corresponding to each one. As an example, assume for instance 2 criteria out of 45, and the 2 sources.

In the intersection of each sub column and a criterion we have these options:

Let’s analyze PV

In general, closeness to sea and wind force are irrelevant, therefore, these boxes are left blank.

The only significant columns for PV are those corresponding to irradiation. Assume that we have measured irradiation in the three locations (A, B and C), as follows:

For A: 1250 kWr/m2-year, we put that value for site A. Perhaps for site B there is on 850 kWh/m2-year, and we put that value in the corresponding intersection. And site C, in other latitude, has a low 450 kWh/m2-year. Thus, we have the values for the three sites, and we establish say that it must be maximized. The irradiation criterion will look as:

For site A: 1250 0

For site B: 850 0

For site C: 450 0

Therefore, for PV we have: 1250, 850, 450

Wind

Obviously, the criterion wind force is very significant here. Assume that we have measured these values:

For site A: 0 2.5 0 m/s

For site B: 0 8.5 0 m/s

For site C: 0 5.1 0 m/s

Therefore, strong wind criterion will be: 2.5 8.5 5.1. We must indicate the software that the result must be greater than 5 m/s, which is the estimated minimum wind speed

In this way it is possible to determine the best location as well as the best source of energyfor each one, to produce hydrogen

Is it possible to consider that for a same location there could be two or more inclusive sources. For instance, in a desertic area we can have that in some location, both PV and wind can be installed, if a site possesses the two conditions, like high irradiation and enough wind speed, provided that there is enough space for both.

10- Page 7 “This method is based on assessing the variability of each criterion among the different alternatives by assigning higher weights to those with greater variation”

There is a mistake here, because weights are not related to variation, it is related to the dispersion of the criterion performance values, not to the relationship between criteria.

Performance values variation is defined by entropy. High entropy values mean small variation, with maximum entropy equal 1, with no dispersion or variation. The minimum entropy is 0, which corresponds to lowest entropy value, and with the largest dispersion or variation, and the largest amount of information,which is what really matters, because information is what evaluates alternatives.

11- “The subjective weights of water availability and annual hydrogen production are 0.06 and 0.03, while their objective weights are 0.16 and 0.005. In the frst case, the objective weight is lower than the subjective weight assigned by a group of experts”

To make 1 kg of hydrogen normally you need 9 litres or 9 Kg of water. Therefore, the relation is 0.11

According to your numbers the relation between both is 0.50. It appears that subjective values are wrong as well as the objective weights where you need 150 grams of water to get 5 milligrams of H2 or 30 tons of water per kgr of H2. I suggest to check your figures.

12- “This method focuses on interactive decision-making, where the criteria and preferences of the decision-makers can change during the process”

Your definition of iteration is correct but you do not need a DM for that. The Simplex algorithm, which is an Excel add-in in your computer, under the ’Solver’ name, does it automatically and very very fast and without errors, using Linear Programming

I hope that my comments may help

Nolberto Munier