• Types

• Tools

• Selection of the MCDA technique

• MCDA in RES projects: attributes under which the methods should be evaluated

• Case Studies

There exist a variety of methods available and there are no better or worse techniques, rather some techniques better fit some kinds of situations. The current software tool includes four (4) techniques to be used, namely the ELECTRE III, the PROMETHEE II, the NAIADE, and the REGIME method.

• Types

• Regime – Flag (Original version)
• ELECTRE III (Demo version)
• NAIADE (Original version)
• Promethee I, II

(right click on the above desired link and select save target as to download the file to your computer)

• Tools

• Data requirements on criteria
MONETARY (CBA)
NON-MONETARY (MCA-qualitative, weights, thresholds)

• Uncertainty on weights score and thresholds
• Decision-tree

• Selection of the MCDA technique

The use of Multi-Criteria Decision Analysis (MCDA) techniques provides a reliable methodology to rank alternative RES projects in the presence of numerous objectives and constraints. Despite however, the large number of available MCDA methods none of them is considered the best for all kind of decision-making situations. There are no better or worse techniques, only techniques that fit better to a certain situation or not. Nevertheless different methods, when applied to the same problem using similar data, often produce differing results. The main question is, therefore, how to choose the appropriate MCDA methodology in RES decision-making.

The 4 multi-criteria software tools used by this Tool kit are:

REGIME – FLAG (original version)
ELECTRE III (demo version)
NAIADE (original version)
PROMETHEE II (to be decided)

This STEP No.6 of the MCDA-RES Tool kit aids the user in selecting the best MCDA software for his case. Next, the main attributes under which these methods should be evaluated are presented and a comparative matrix with the 4 multi-criteria techniques for RES projects and their performance is created.

 

• MCDA in RES projects: attributes under which the methods should be evaluated

1. Degree of compensation allowed between the criteria
2. Modeling of DM’s (and social actors’) preferences
3. Ease of use
4. Ability to support a large number of Decision-Makers - capacity of taking into account the conflicting point of view of different social actors
5. Capacity to handle many criteria and alternatives
6. Treatment of inaccurate or uncertain criteria - Treatment of uncertainty - Allowing the use of mixed information
7. Interpretation of software parameters - Transparency

Quite a lot of MCDA methods realize some of the above listed requirements but no method is capable to incorporate all at the same time. In the next paragraphs a tutorial is presented to help the user to choose the appropriate MCDA technique for his particular situation.

1. Degree of compensation allowed between the criteria

The degree of compensation allowed between the evaluation criteria has important consequences for RES planning. If we accept that there exist some ecological thresholds that cannot be exceeded, then we have to apply a decision-making method that limits compensation between the criteria selected. This is usually the case with areas that are about to host RES projects (remote areas, mountainous, with wild life and rich flora). Therefore the inclusion of non-compensatory or partial compensatory methods approaches seems to be a wiser approach for energy and environmental decision-making.

The PROMETHEE and ELECTRE methods typically do not authorize for complete compensation between different dimensions-criteria. The use of veto thresholds in the analysis explicitly limits the compensation between the criteria and the indifference and preference thresholds do not render all differences between the criteria accountable for the overall ranking.

In NAIADE the degree of compensability can be chosen, by means of the three possible operators of the criteria aggregation procedure, which are:

● Minimum operator: No compensability at all

● Zimmerman-Zisno operator: It allows to choose the degree of compensability γ, which ranks from 0 (no compensability at all), and 1 (complete compensability)

● Simple product: Complete compensability

Also, the “α” parameter (minimum fuzzy relation requirements) influences the degree of compensability. In fact, when “α” raises, the model includes in the aggregation procedure only the criteria with larger differences between alternatives. In this case, a criterion which goes against an alternative is not taken into account if the difference between alternatives is not sufficiently high, so that it can be said that it is compensated by the criteria in favor of the alternative that have high distances between alternatives.

REGIME: since REGIME belongs to the class of outranking methods it does not allow for complete compensation.

FLAG: the flag model does not allow compensation.

In Table 1, MCDA methods are compared according with the degree of compensation they allow between the evaluation criteria. The scale “+++/---’’ is ordinal in nature (‘+’ is more desirable than ‘-’ ‘+++’ is more desirable than ‘+’ and ‘---’ is less desirable than ‘-’) and reflects the previous discussion and the experience gained from our direct involvement in a number of real case studies.

Table 1. Comparison of methods by criterion 1:
Degree of compensation allowed between the criteria

2. Modeling of DM’s preferences

In most multi-criteria methods a weight is assigned to each criterion, expressing its relative importance. The use and the analysis of the weights depend on the selected decision model.

PROMETHEE II, ELECTRE III, REGIME and FLAG, use weights while NAIADE does not.

On the other hand, Naiade allows performing a coalition analysis, showing the ranking of the alternatives for each group of actor

The use of weights provides the Decision Makers with a way for indicating their preferences. In RES planning, however, many times the DM’s are not willing to express their weights between the different dimensions-criteria due to political reasons. In fact, giving weights to the criteria means to give weight to the group of social actors. It is recommended, however, to include weights in the analysis to enhance the transparency of the decision-making process.

In order to increase transparency, it might be useful to show the different outcomes resulting from various sets of weights, making clear that the choice of weights is a political decision, and reflects the values of the user. In different sets of weights produce
In the next Table 2 MCDA methods are juxtaposed with the ability to include weights in the analysis they provide.

Table 2. Comparison of methods by criterion 2:
Modeling of DM’s preferences

3. Ease of use

Ease of use is considered an important parameter of the applicability of a software program. It is highly depended on the structuring of the software and the interaction of the user with it.

ELECTRE techniques demand the estimation of thresholds (three kinds in the general case), and weights. These factors however, help sometimes the DM to fully understand the problem and form his preferences consistently. Nevertheless, these features represent some abstract meaning.

PROMETHEE techniques seem to be easy to use. The two thresholds to be appointed by the DM have some concrete meaning and the user can follow the calculation easily.

In the NAIADE method, the use of preference and indifference thresholds can be complicated. Moreover, the determination of four degrees of preferences (much better, better, approximately equal and equal) is a difficult task. Also, dealing with fuzzy numbers can be a hard task, although this feature gives an important value added to the model.

REGIME:. In the REGIME methods criteria scores and weights have to be defined. This should be carefully guided since it might be a difficult task. However, the inclusion of the SAATY methods provides good assistance in the calculation of criteria weights.

FLAG: Once the meaning of CTVmin, CTVmax and CTV are explained to the user the model is very intuitive and easy to use.

In the next Table 3 MCDA methods are compared according to their ease of use.

Table 3. Comparison of methods by criterion 3:
Ease of use

4. Ability to support a large number of Decision-Makers

In public (renewable) energy and environmental planning the number of people involved in the decision-making process is usually big (in some cases around 50). It is, therefore, necessary for the chosen MCDA technique to be able to support a large number of DM’s.

ELECTRE III and PROMETHEE II models can handle a great number of Decision-Makers, provided that they are willing to offer their weights and determine thresholds and other operational parameters.


NAIADE does not have the possibility of using weights in the evaluation. However, in addition to the use of preference thresholds, NAIADE method has an additional feature which is the Conflict Analysis approach. Through this module it is possible to obtain the Degree of credibility of coalition formation between the different social actors. Also it is possible to get a set of rankings of the alternatives for every social actor and coalition.
The combination of the results of the technical evaluation and the social evaluation provides very useful information to reach to a Compromise Solution of the problem at hand.

REGIME can handle a great number of Decision-Makers, provided that they are willing to offer their weights and determine thresholds and other operational parameters.

FLAG process is different than the above mentioned techniques. A great number of Decision-Makers can be included in the evaluation process. However, this is not directly facilitated by the Flag Model. Decision-makers have an important role in the selection of criteria and the establishment of critical threshold values. The inclusion of a great number of decision-makers should be focus on reaching consensus regarding the above mentioned inputs of the model.

In the next Table 4 MCDA methods are juxtaposed with their ability to support a large number of Decision-Makers.

Table 4. Comparison of methods by criterion 4:
Ability to support a large number of Decision-Makers

5. Capacity to handle many criteria and alternatives

In public (renewable) energy and environmental planning the number of criteria and alternatives to be considered varies between case studies. In most of the cases 4-8 alternatives are considered enough. The criteria, however, should embrace all important aspects that need to be considered. This Tool-Kit proposes a set of 18 criteria to be used for an integrated evaluation of RES plants.

ELECTRE III and PROMETHEE II models can handle a big number of criteria and alternatives, provided that the means to evaluate consistently the alternatives according to the selected criteria are available.

The NAIADE method as well allows dealing with a big number of criteria and alternatives

REGIME as well allows dealing with a big number of criteria and alternatives.

In FLAG Model criteria are divided into three classes, economic, social and environmental. The model allows the inclusion of a great number of criteria in those classes.

In the next Table 5 MCDA methods are juxtaposed with their capacity to handle many criteria and alternatives. The scale “+++/---’’ is ordinal in nature (‘+’ is more desirable than ‘-’ ‘+++’ is more desirable than ‘+’ and ‘---’ is less desirable than ‘-’) and reflects the previous discussion and the experience gained from our direct involvement in a number of real case studies (I would not repeat it for each table, once is enough)

Table 5. Comparison of methods by criterion 5:
Capacity to handle many criteria and alternatives

6. Treatment of inaccurate or uncertain criteria (treatment of uncertainty)

In RES projects some of the information required is rather qualitative and other is just uncertain, like the future unfolding of the liberalization in the EU and its effects upon prices and subsidies.

In the ELECTRE III and the PROMETHEE II approaches, the inaccuracy of the criteria values can be modeled through indifference and preference thresholds. These intervals not only try to model the DM’s preferences but also capture the uncertainty of the data between the). Of course, threshold values must be assessed separately for each criterion and for each problem. It seems that this approach is better suited for energy and environmental problems.

One of the features that makes NAIADE different is its ability of dealing with mixed information. There is also the possibility of using stochastic valuation of criteria when probabilistic uncertainty exists, and qualitative valuation when trends are known (or expected) in a narrative way.

REGIME is able to deal with mixed information. There is also the possibility of using stochastic valuation of weights when probabilistic uncertainty exists.

FLAG: The inaccuracy of the criteria values can be accounted for by adjusting the minimum and maximum values for the critical threshold values. In this case the bandwidth effects have to comply is narrowed or widened.

In the next Table 6 MCDA methods are juxtaposed with their treatment of inaccurate or uncertain criteria. The scale “+++/---’’ is ordinal in nature (‘+’ is more desirable than ‘-’ ‘+++’ is more desirable than ‘+’ and ‘---’ is less desirable than ‘-’) and reflects the previous discussion and the experience gained from our direct involvement in a number of real case studies

Table 6. Comparison of methods by criterion 6:
Treatment of inaccurate or uncertain criteria

7. Interpretation of software parameters (Transparency)

Each MCDA technique is characterized by some parameters. These parameters include weights, thresholds, aspiration-reservation values, etc depending on the methodology applied. The direct interpretation of these parameters represents a desirable feature for the corresponding software to be implemented, since the DM can understand better how a decision is taken and can perform easier a sensitivity analysis.

In the ELECTRE III method the weights, and cutting values, represent some kind of abstract meaning, while the PROMETHEE framework utilizes parameters (indifference and preference threshold) which have a more concrete meaning for the DM.

In the NAIADE method, the parameters used by the model, namely “α” and “γ”, are not so intuitive and the aggregation procedure is quite complicated to understand.

REGIME: The parameters used by the model, score, alternative, and weights are intuitive. However, the aggregation procedure is quite complicated to understand

FLAG: The parameters used by the model, CTVmin, CTVmax, and CTV and criterion score are intuitive. However, the aggregation procedure is easy to understand

In the next Table 7 MCDA methods are juxtaposed with the way they interpret the software tool parameters.

Table 7. Comparison of methods by criterion 7:
Interpretation of software parameters

The choice of the appropriate software is depends on the features of the problem at hand, the particular case study and Decision-maker. It is thought that the insights and comparative evaluation of the software tools presented here, have given essential guidelines in selecting the most appropriate MCDA software tool for a specific case study.