# 219
Dear Hassan Z. Al Garni, Anjali Awasthi
I read your paper:
Solar PV power plant site selection using a GIS-AHP based approach with
application in Saudi Arabia
1- In the abstract you mention using AHP to determine criteria weights. In my opinion, this is not a good option, because AHP weights are only invented values and without any mathematical support, especially in your problem, where you are using GIS and being able to get the scientific relevance of a criterion using a tool known as Inter and ArcGIS. In addition to
the process of ‘weighting arbitrarily’ a criterion, you are adding them up, which is against Systems Theory conceptualization and produces more uncertainty.
I suggest this procedure:
a) Select the best potential areas of your country, maybe Tabuk and Jawf, in the North and NW, as well as Jedda and Al Makka in the SW
b) Choose the best potential sites within these areas
c) Select criteria, as you did, and add Humidity, to determine suitability
d) Establish a target or limit for each criterion like: (These values are only illustrative)
Max. Slope < 5%,
Max. temperature < 28 C
Max, humidity < 50 %
Etc.
Use as many criteria as you wish and even mini and max values for each target
e) Model the whole scenario in a matrix and fill the spaces with the values you have for each alternative or site and each criterion. Do not worry about transitivity or intransitivity. For subjective criteria collect as much information as possible using surveys and statistics. If one cell cannot be filled, just leave it blank Mix criteria in maximization and minimization in any way. If you have for subjective criteria two reasonably certain values, you can use fuzzy and put the corresponding crisp value in the cell
f) Prepare as many layers as criteria in GIS and use Suitability Modeler from ArcGIS. It will produce for each layer a scientific emergence or combined value, and within the targets or limits imposed.
g) Convert that matrix in binary ( 0 or 1), according the no compliance (0), or compliance (1) of the targets.
h) Use the intersection procedure, that is, a value that considers all criteria simultaneously. It will give you the emergence value of each layer. This is the value for your layer
i) Build a new matrix with the alternatives and these values and use a MCDM method that considers simultaneously all criteria and all alternatives. You will get the score for he best site and the ranking for the others.
As you can see this whole process is completely mathematical and transparent. However, if the DM wants to also input his expertise, know-how and rationality he can do that and run the software again, that is, working scientifically, reasoning, thinking, and applying comm on sense , and the DM can do it now because he will be working on a mathematical solid base
2- In Page 5 you propose weighting sum overlays. In my opinion this is incorrect, because a site is a system with many components like soil, temperature, humidity, slope, etc., and you assume that adding all criteria you get the best result. Do you know that this procedure, very often used unfortunately, is correct in a system? By summing you simply accumulates values, and get a result that does not consider the mutual influence that some criteria have on another. For instance, you can add the value for criteriontemperature to a other criterion humidity as if they were independent issues, when in fact they are very closely related
It is not my idea, but science. I asked AI if there is any relationship between temperature and humidity in a PV farm, and here is the answer:
temperature andtemper humidity together affect the reliability and lifespan of photovoltaic (PV) modules. One widely used model is the Peck model, which estimates the degradation rate of materials (like encapsulants in PV modules) under combined temperature and humidity stress”
Of course, you can check it by yourself.
Do you understand why I say that your approach is incorrect?
By the way, there are not optimal in MCDM, only a balance or a compromise solution
3- High temperature is possibly the most significant factor in Saudi Arbia for PV generation bit, what about humidity, something that you do not even mention?
I suggest investigation of this factor and its effects in scattering and absorption, that reduce solar irradiance, as well as condensation and dew, cloud formation and air transparency.
4- Table 8 shows the criteria importance according to pair-wise comparison where the most significant is C1 (Solar irradiation). One of the several fallacies of AHP is to assume that the preference value of each criterion once ‘computed’ is independent of the alternatives or locations. There is no doubt that in Saudi Arabia the most significant factor is solar irradiance when you consider the whole country, however, it is not when you consider each location.
In Riyadh, located in one of the hottest areas of the country, with a typical desert climate, with temperatures that normally exceeds 45 C, when the max temp for PV cells is 25 C. Therefore, you cannot consider relative importance of at least some criteria by ignoring this fact. Granted, you did not consider Riyadh as a potential site, but it shows that assuming that the most important criterion to evaluate alternatives is irradiation, independently of the characteristics of each location, will lead at a false and misleading result.
Sincerely, as a first approach you not even need mathematical modelling
That indicate that the best areas are on the North and NW of the country, and in the SW near Jeddah and Al Makka, but where temperature and humidity are considerable higher than in the NW, like in Tabut
The GIS shows this very clearly. You, in reality, confirmed that intersection is a must when saying that he Tabuk area offers the best combination of low temperature and low humidity, result that you cannot arrive if you consider both separately.
In other words, it is not valid, whatever the MCD M method you use, to add up the results of criteria or data, you must find the intersection or combination, something that No MCDM method do. You must use other methodologies like MOO of Multi Objective Optimization
As a simple example: Look at your national dish: Kabsa
You enjoy it not by adding rice, chicken, lamb and spices like cardamon, cinnamon cloves, other species. You blend them all, and combining aroma, odors and sconces that they yield in the coking process, and the result is a fragrant delicatesse, not a sum of elements where when eating the cooked dish, you can taste the flavor of each one separately
These are my comments that I hope can help you
Nolberto Munier