Selection of Optimum Renewable Energy Source for Energy Sector in Pakistan by Using MCDM Approach

Abstract

Power demands across the globe have been on rise for quite a time, and they will continue so in the years ahead. Most of the present day, energy requirements are met by non-renewable energy sources, i.e., burning fossil fuels which are not only expensive and polluting our environment but also depleting quite rapidly. A considerable portion of Pakistan import bills consists of petroleum products according to Pakistan Economic Survey 2015–2016. Renewable energy alternatives, i.e., wind, hydel, solar, and biomass, can be considered cheap, reliable, safe, and supply sustainable energy for the ever-increasing population. Renewable energy is getting more and more attention not only abroad but also in Pakistan as well. In order to decide the best renewable energy alternative for investment to meet rising energy demands of Pakistan, several factors such as economic, technological, and environmental have to be taken in consideration by the decision makers. Therefore, the paper discusses three important multiple criteria decision-making techniques, i.e., Analytical Hierarchy Process (AHP), Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), and the VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) method in selection of the best renewable energy source that Pakistan should invest into. The results indicate that hydel power generation is the optimum source for meeting energy requirements followed by wind, biomass, and solar power-generating plants.

Introduction

Energy is a basic necessity for human beings. Currently, the world’s major energy supply comes from the combustion of fossil fuels (coal, hydrocarbon liquids, etc.). However, these resources not only produce harmful byproducts but are also due for depletion over the course of time even if the energy demand of world decreases. Therefore, research to develop renewable energy sources has seen a surge in recent years in order to mollify the consumption of energy and protect the environment (Mardani et al. 2015).

Renewable energy has been proven to be the solution for a sustainable, cost–effective, and environmentally friendly source of energy. It is the future of energy production and is capable of replacing fossil fuels in most of their applications. For this reason, renewable energy exploitation is gaining momentum in many countries across the world.

There are many different sources of renewable energy. The most popular among them which are being used for energy generation are as follows:

• Biomass refers to the biological material, dead or living, which can be used as a fuel.

• Hydel (most widely used form of renewable energy) is the production of energy through falling water under gravity.

• Solar power (one of the fastest growing energy sources) is the harnessing of sun’s energy to produce electricity.

• Wind power is the production of energy by converting wind energy to some useful forms of energy with the help of wind turbines.

Many studies have been conducted for the sustainability of the renewable energy resources. Selection of the most effective and efficient source involves different interacting factors. Traditional single criteria decision-making approaches cannot handle the complexity of this problem (San Cristóbal 2011). Therefore, for the renewable energy decision-making, we have to use multiple criteria decision-making (MCDM) techniques. MCDM methods deal with the process of making decisions in the presence of multiple objectives. A decision maker is required to choose among quantifiable or non-quantifiable and multiple criteria. Several methods have been developed to perform MCDM. Each method has its own characteristics, but they may be combined to give final result. Each methodology shares conflicting criteria, incommensurate units, and difficulty in selection of the alternative. Because of the conflicting criteria, the solution is highly dependent upon decision maker and it is a compromise. The conflicting aspects arise due to increasing complexity of social, technological, environmental, and economic factors (Afgan and Carvalho 2002).

This aim of this study is to use three different MCDM techniques namely VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR), Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), and Analytical Hierarchy Process (AHP) to select optimal renewable energy source for Pakistan to invest, in order to meet its energy demands. Detailed analysis and overview of each above mentioned methods is given in this study.

Literature Review

In recent times, many publications have been published regarding MCDM techniques also showing how it has evolved with time and its applications in various fields. Modern MCDM infrastructure was laid in 1950s and 1960s. Progress in MCDM research increased during the 1980s and early 1990s and continued its accelerated growth (Köksalan et al. 2011; Kabak and Dağdeviren 2014). Table 1 shows the past research conducted on various MCDM methods.

Table 1 Past research carried out in field of MCDM

Over past years, a detailed study of MCDM techniques showed effectiveness in applying these techniques in areas of sustainability and renewable energy (Mardani et al. 2015). Table 2 shows the research contributions of various authors in field of environment and sustainability using MCDM methods.

Table 2 Use of MCDM techniques in environment and sustainability

MCDM techniques are also being employed for the energy sector of Pakistan. Table 3 shows the contribution of few authors.

Table 3 Use of MCDM techniques in energy sector of Pakistan

Methodology and Analysis

The objective of this study is to find optimum source of renewable energy using VIKOR, TOPSIS, and AHP method. For this purpose, data from various sources were collected for four alternatives: hydel, solar, wind, and biomass energy. All alternatives were evaluated over six criteria discussed below:

• 1. Initial cost

  Initial costs are those which are incurred during the planning, design and construction phase of the project. It takes a huge amount of money to convert a raw renewable energy source to usable energy. Thus, it is one of the most important aspects when considering power generation dynamics. As for the scope of this paper, it is measured in million USD (mUSD).

• 2. Operations and maintenance (O&M) cost

  Equipment in the power producing plants is expensive, so it is uneconomical to replace it before its expected life. Therefore, it is important that the equipment is properly operated and maintained which costs capital. This capital, though not as much as initial cost, plays an important role when considering power generation dynamics. As for the scope of this paper, it is measured in million USD (mUSD).

• 3. Power production capacity

  It is important to know how much power will be produced and added to the national grid by utilizing the raw renewable energy resources. It depends on amount of raw material available, auxiliary equipment, method of energy production, etc. As for the scope of this paper, it is measured in megawatts (MW).

• 4. Efficiency

  Efficiency of a power producing plant tells that how effectively the resources are being used to produce energy. It is one of the most important factors in power generations dynamics and thus considered in the analysis. As for the scope of this paper, it is measured in percentage (%).

• 5. Expected life

  Regardless of the resources used for producing energy (power), every power producing plant has a life expectancy. The expected life largely depends upon the raw resource used to produce energy, therefore, it is an important factor considered in this analysis. As for the scope of this paper, it is measured in years.

• 6. Environmental effects

  With a growing concern over harmful effects of gases and particulate matter released in the atmosphere by fossil fuel based energy sources, it is no doubt that this is the most important factor considered in this paper. The environmental effects considered here tell us that how many million tons of CO₂ can be avoided if we replace the non-renewable energy resources with renewable energy resources considered in this paper. As for the scope of this paper, it is measured in millions tons of CO₂ avoided per year (m tons CO₂ avoided/year).

The first two factors are categorized as economic, next two as technological, while the last one as environmental factor. Table 5 shows the data collected for each renewable energy alternative. The data is collected from various sources including Pakistan Energy Book (2014–2015), annual reports of the power plants in Pakistan, feasibility reports, letters of intent (LoI) of different future plants in Pakistan, newspapers, and other publicly disclosed information by Alternative Energy Development Board (AEDB) of Pakistan.

Assigning weights to criteria is a critical part of any MCDM study. In this paper we use aggregate criteria weights by utilizing empirical rank-weight relationship (Alfares and Duffuaa 2008) in which ordinal rankings of a number of criteria are converted into numerical weights. A linear relationship Eq. 1 specifies the average weight for each rank for an individual decision maker (DM), assuming a weight of 100% for the first-ranked (most important) factor (Alfares 2007). In this method, an m, number of DM’s assign rank r to n, number of criteria, was based on individual DM’s discretion. The individual ranks are then converted into individual weights for each criteria. After this, average weight for each criteria is calculated among all individuals (Alfares 2007). The following three steps are used to calculate the final weight of criteria which is mentioned in Table 4.

• 1. For each individual i, ranks r_i,j are converted into individual weights w_i,j for all n criteria by the following equation:

  $$ {w}_{i,j}=100\hbox{--} {s}_n\left({r}_{i,j}\hbox{--} 1\right)\kern0.5em i=1,\dots, m,\kern0.5em j=1,\dots, n $$

  (1)

  where ${s}_n=3.19514+\frac{37.75756}{n},$ for 1 ≤ n ≤ 21, 1 ≤ r ≤ n, and r and n are integers.

• 2. Average the weights obtained from all individuals m for all criteria to obtain aggregate weight w̅ _i for all criteria.

  $$ {w}_i=\frac{1}{m}{\sum}_{i=1}^m{w}_{i,j}\kern0.5em j=1,\dots, n $$

  (2)

• 3. Final weight w_fi for each criteria n is then calculated as

  $$ {w}_{fi}=\frac{{\overline{w}}_i}{\sum_{i=1}^n{\overline{w}}_n} $$

  (3)

  where i = 1, 2…n, j = 1, 2…n.

Table 4 Calculation of weights by three DM’s

Ranking is done for each criterion by researchers’ discretion and engineering knowledge about power plants. The weights assigned are 0.2 for initial cost, 0.12 for O&M cost, 0.17 for capacity, 0.17 for efficiency, 0.13 for expected life, and 0.21 for environmental effects. Environmental effects have highest aggregate weight as one of the main reasons to move towards renewable energy is to minimize or perhaps even eliminate the harmful effects of fossil fuel based energy on the environment. Initial cost has the second highest aggregate weight because in order to convert the raw resources of renewable energy to usable energy, a huge amount of capital is required. It is also important to know how much and how efficiently electricity is being produced; thus, a high weightage for capacity and efficiency is observed (Table 5). Also, all of the power-producing plant statistics (capacity, revenue, breakeven period, etc.) depend heavily on the overall efficiency of the plant. Capacity and efficiency are followed by expected life with second lowest weight and O&M cost with lowest weight because a much smaller amount of capital and effort is required for this as compared to others.

Table 5 Data for various renewable energy sources

VIKOR Method

The VIKOR method was proposed by Opricovic and Tzeng in 2004 (Opricovic and Tzeng 2004). It is used to solve set of multiple criteria decision-making (MCDM) problems in which there are distinctive or conflicting units of criteria. The VIKOR method consists of an aggregating function that represents distance from the ideal solution. It takes the relative importance of all criteria into consideration as well as balance between total and individual satisfaction (San Cristóbal 2011). Compromise solution is the feasible solution based on “closeness to the ideal” concept. The compromise ranking is developed from the Lp-metric used as an aggregating function in a compromise programming method.

This method consists of following four steps (San Cristóbal 2011).

• Step 1: Calculate best ideal value ideal value f _i ^* and the worst ideal value f _i ^¯ for all criteria.

where f _i ^* = max fij if criteria represents a benefit and fi^* = min fij if criteria represents a loss and f _i ^¯ = min fij if criteria represents a benefit, and f _i ^¯ = max f _ij if criteria represents a loss and j represents the alternatives. Table 6 shows results of step 1.

• Step 2: Calculate the value of S _j using Eq. 4 which is weighted and normalized Manhattan (taxi cab) distance and R _j using Eq. 5 which is weighted and normalized Chebyshev distance.

  $$ {S}_j={\sum}_{i=1}^n\frac{w_{fi}\left({f}_i^{\ast }-{f}_{ij}\right)}{\left({f}_i^{\ast }-{f}_i^{-}\right)} $$

  (4)

  where w _fi represents weight of a criteria

  $$ {R}_{\mathrm{j}}=\max\ {w}_{fi}\frac{\left({f}_i^{\ast }-{f}_{ij}\right)}{\left({f}_i^{\ast }-{f}_i^{-}\right)} $$

  (5)

  where w _fi represents weight of a criteria and j = 1, 2…J

• Step 3: Calculate the values of Q _j using Eq. 6 by relation

Table 6 Best ideal value f _i ^* and the worst ideal value f _i ^¯ for all criteria are based on Table 5

$$ {Q}_{\mathrm{j}}=\frac{v\left({S}_{\mathrm{j}}-{S}^{\ast}\right)}{\left({S}^{-}-{S}^{\ast}\right)}+\frac{\left(1-v\right)\left({R}_{\mathrm{j}}-{R}^{\ast}\right)}{\left({R}^{-}-{R}^{\ast}\right)} $$

(6)

where S^∗ = min S_j, S⁻ = max S_j and R^∗ = min R_j, R⁻ = max R_j..

The solution obtained by min S _j is with a maximum group utility (“majority” rule), and the solution obtained by min R _j is with a minimum individual regret of the “opponent.” Here v is the weight for the strategy of maximum group utility, and 1 − v is the weight of the individual regret. Although value of v varies between 0 and 1 based on decision maker (individual or group) choice, it is usually taken as 0.5 (San Cristóbal 2011).

• Step 4: Rank all the alternatives by sorting Q, R, and S values in decreasing order as shown in Table 7. The compromise solution is decided as A⁽¹⁾ which is minimum value by Q ranking if following two conditions are fulfilled (Opricovic and Tzeng 2007).

• a. Acceptable advantage

  Q [A⁽²⁾ − A⁽¹⁾] ≥ DQ, where DQ = 1 / (J − 1), A⁽¹⁾ and A⁽²⁾ are the minimum and second minimum values in ranking by Q while J is the total number of alternatives.

• b. Acceptable stability in the decision-making.

  The minimum value of S or/and R ranking should also indicate the alternative A⁽¹⁾ to be the best solution.

Table 7 Values of S, R, and Q with v = 0.5 in decreasing order

It can be seen that both the abovementioned conditions (a and b) are true so hydel is the compromise solution by VIKOR method.

However, if one of the two conditions is not fulfilled, then a set of more than one compromise solutions are proposed.

• c. Alternative A⁽¹⁾ and A⁽²⁾ are both compromise solutions if only condition b is not fulfilled.

• d. Alternative A⁽¹⁾, A⁽²⁾,…A^(M) are all compromise solutions if condition a is not fulfilled. A^(M) can be calculated by the equation Q(A^(M) − A⁽¹⁾) < DQ for maximum value of M (position of mentioned alternatives are “in closeness”).

In this case, however, both the conditions are satisfied, so there is only one compromise solution.

TOPSIS Method

TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) is a multiple criteria decision analysis technique which was presented by Hwang and Yoon in 1981. The basic concept of this method is that the selected alternative should have the shortest distance from the positive ideal solution (PIS) or longest distance from the negative ideal solution (NIS) in geometrical sense. It assumes that criteria are monotonically (linearly) increasing or decreasing. It is a compensatory aggregation in to which certain weightage is assigned to each alternative according its importance and coherence with the required criterion. An evaluation matrix X _mn is created which composes of m number of alternatives (rows) and n number of criteria (columns). A preference is assigned to each criterion between 1 and 9 for every alternative as shown in Table 18. This preference is based on decision makers’ (DM’s) discretion and data collected. The matrix is then normalized (N _ij ) using Eq. 7, and weighted decision matrix (W _ij ) is formed by multiplying normalized matrix columns with respective weight of the criteria using Eq. 8. N _ij and W _ij are shown in Tables 19 and 20, respectively. Positive ideal solution P _i and negative ideal solution N _i are calculated for each criteria, such as “P _i  = max (W _ij )” if it represents a benefit and “P _i  = min (W _ij )” if it represents a loss. Similarly, “N _i  = min (W_ij)” if it represents a benefit and “N _i  = max (W _ij )” if it represents a loss. Table 8 gives the values P _i and N _i from weighted decision matrix based on data from Table 5.

$$ {N}_{ij}=\frac{X_{ij}}{\sqrt{\sum_{i=1}^m{\left({X}_{ij}\right)}^2}} $$

(7)

Table 8 Positive ideal solution P _i and negative ideal solution N _i for each criteria

where i = 1, 2…4 and j = 1, 2…6

$$ {W}_{ij}={N}_{ij}\times {W}_j $$

(8)

where W _j represents weight of the criteria

Geometric distance between an alternative from the negative ideal solution S _i and positive ideal solution S _i * is calculated using Eqs. 11 and 12, respectively. This is followed by a separation measure S _n given by Eq. 9 which describes distance from negative ideal solution. Its value varies between 0 and 1 indicating worst alternative best alternative, respectively. It should be noted that separation measure can be used to measure distance from positive alternative using Eq. 10. In that case, best alternative will be the one with minimum value. The result in both the cases will be same. The result in both the cases will be same. The values of S _i ′, S _i *, and S _n are given the Table 9 in decreasing order and are based on Table 8.

Table 9 Geometric distance from positive ideal solution S _i *, negative ideal solution S _i ′, and separation measure S _n based on Table 8

$$ {S}_n={S_i}^{\prime }/\left({S_i}^{\prime }+{S_i}^{\ast}\right) $$

(9)

$$ {S_n}^{\hbox{'}}={S_i}^{\ast }/\left({S_i}^{\prime }+{S_i}^{\ast}\right) $$

(10)

$$ {S_i}^{\prime }=\sqrt{\sum_{j=1}^n{\left({W}_{ij}-{N}_j\right)}^2} $$

(11)

$$ {S_i}^{\ast }=\sqrt{\sum_{j=1}^n{\left({W}_{ij}-{P}_j\right)}^2} $$

(12)

In Eqs. 11 and 12, i and n represent alternatives and criteria respectively while N _i and P _j represent negative ideal solution and positive ideal solution, respectively, for each criteria.

Since maximum S _n value is for hydel, it is the optimal alternative followed by biomass, solar, and wind energy.

AHP Method

The Analytic Hierarchy Process (AHP) introduced first by Prof. Thomas L. Saaty. It is a multiple criteria decision-making approach in which a complex problem is broken down into hierarchy with objective/goal at top level, criteria, and sub-criteria at mid-level while alternatives are placed at the bottom level of hierarchical process. Starting from the bottom, alternatives are compared with each other to find their relative importance with respect to each element in the above level (criteria or sub-criteria). The verbal terms of the Saaty’s fundamental scale of 1–9 are used to assess the intensity of preference between two elements. The value of 1 indicates equal importance, 3 moderately more, 5 strongly more, and 7 very strongly, and 9 indicates extremely more importance. The values of 2, 4, 6, and 8 are allotted to indicate compromise values of importance (Pohekar and Ramachandran 2004). AHP methods takes into account the consistency ration (CR) which is ratio of inconsistency index (CI) to randomly generated index (RI) to assure the decision maker that his judgments were correct and final answer is consistent. It is calculated for each pairwise comparison matrix at each level, and its value should be less than 0.1. If it is not, then the pairwise comparison matrix needs to be re-evaluated. Method for checking consistency is discussed in detail in Saatys’ original AHP paper (give ref. to AHP original paper)

In this case, the alternatives are compared with each other for initial cost, O&M cost, etc. Thus, we have six matrices denoted by M as shown in Table 14. The preference assigned to the alternatives is based on researchers’ discretion; however, assigned preferences appeal common sense and are justified according to the data in Table 5. The value of preferences is iterated many times to reduce the consistency ratio (CR) below 0.1.

After this, a normalized matrix is formed by dividing column elements with the column sum for each criterion as shown in Table 15. An equation representing the normalized matrix can be written as

$$ {N}_{ij}=\frac{Mij}{\sum_{i=1}^j Mij} $$

(13)

where i and j represent row and column of the each matrix, respectively.

Eigenvectors of each level are found by taking average of rows in normalized matrix for each criteria (total six eigenvectors). A matrix is created by combining all the eigenvectors obtained at a level shown in Table 10.

Table 10 Combined eigenvector matrix for alternatives

Then, same procedure is done for upper levels and a matrix of eigenvectors is formed at each level until there is only one eigenvector formed at criteria level. Since what we have are not considering any sub-criteria, we will have only one more upper level (criteria) and hence only one eigenvector after comparing the criteria with each other shown by pairwise comparison matrix. The pairwise comparison matrix of criteria is shown in Table 16, normalized comparison matrix of criteria is shown in Table 17, while eigenvector for criteria is shown in Table 11. Preference to criteria is given purely on the basis of weights of criteria. Equation 13 can also be used here to represent the normalized comparison matrix for criteria. However, the matrices M and N will be replaced by M^c and N^c respectively where “c” denotes criteria.

Table 11 Eigenvector for criteria

Then, eigenvector matrices of two successive levels are multiplied with each other from the bottom until final composite vector of weight coefficients (weight of an alternative with respect to the goal) is obtained. In this case, eigenvector form criteria level as shown in Table 11 is multiplied with matrix of eigenvectors from alternatives level as shown in Table 10. Values in the final vector represent the importance of each alternative to the goal with highest value being the most important as shown in Table 12.

Table 12 Ranking of alternatives in decreasing order by AHP

Table 12 shows hydel energy to be optimal solution as it has largest weight coefficient with respect to the goal. It is followed by wind, biomass, and solar energy.

Discussion

The study used three multi-criteria decision analysis tools, VIKOR, TOPSIS, and AHP, and all of them resulted in hydel being the optimal renewable source of energy as an alternative to the non-renewable sources currently used in Pakistan for energy production. The renewable energy sources were evaluated under six criteria including initial cost, operations and management cost, capacity, efficiency, expected life, and environment. By comparing the results of the three different techniques, it can be concluded that all the three tools settle for the same alternative to be the optimal one. However, TOPSIS and AHP show differences for other sources.

The results of this research relate to that of the study of Amer et al. in context of Pakistan (Amer and Daim 2011). However, the difference lies within the evaluated alternatives. Hydel was not considered to be evaluated by them so the optimal alternative by their research was biomass. This study ranks biomass as second best alternative by TOPSIS while third by AHP due to the consideration of hydel energy source.

Among the criteria selected to evaluate the four renewable sources of energy in this study, high capacity, longer life expectancy, and minimum environmental damage acted positively for hydel thus making it the optimal choice of energy production despite of its high initial cost.

Conclusions and Recommendations

In this study, three MCDM methodologies, VIKOR, TOPSIS, and AHP, were used to determine the optimum source of renewable energy in Pakistan by evaluating them over mentioned six criteria. All three of them conclude that hydel energy is the optimal source of renewable energy for Pakistan to invest in order to meet its energy demands. Comparison of the results obtained by all three methods is shown in Table 13.

Table 13 Ranking results as calculated by VIKOR, TOPSIS, and AHP

Pakistan has an abundance of natural hydel resources; therefore, Pakistan needs to gain maximum benefit by utilizing all its natural water supplies and construct small and large hydel energy projects in order decrease the supply demand gap, hence reducing energy crisis. The results also show biomass to be the second and third best alternative according to TOPSIS and AHP, respectively. There are too few projects of biomass working in Pakistan right now as compared to other alternatives. Since biomass energy generation greatly depends upon amount of raw material available (wood chips, crop wastes, etc.) and Pakistan is an agricultural country, a great deal of work can be done in this field. Wind energy, despite its high efficiency, is second and fourth according to AHP and TOPSIS results, respectively. This is justifiable as seen by its high initial cost and short-life expectancy. Also, production of wind energy greatly depends upon climate conditions such as humidity, wind speed, and air density which are not always reliable. However, if the wind corridors in Pakistan are utilized optimally, wind energy can add a considerable amount of energy in the national grid. Similarly, solar energy is ranked as fourth and third according to TOPSIS and AHP results, respectively. It is also justifiable as seen by its low power capacity/cost ratio as compared to other alternatives and its short-life expectancy. It is recommended to not to invest in this field as of yet; however, its potential may be utilized later in the future.

Appendix

Table 14 Pairwise comparison matrices of alternatives for all six criteria

Table 15 Normalized pairwise comparison matrices of alternatives for all six criteria

Table 16 Pairwise comparison matrix of criteria

Table 17 Normalized pairwise comparison matrix of criteria

Table 18 Evaluation matrix (X _mn ) of TOPSIS method

Table 19 Normalized matrix (N _ij ) of TOPSIS method

Table 20 shows weighted normalized matrix (W _ij ) of TOPSIS method