PRES22_0231.docx DOI: 10.3303/CET2294190 Paper Received: 16 May 2022; Revised: 19 June 2022; Accepted: 20 June 2022 Please cite this article as: Ortenero J.R., Choi A.E.S., Promentilla M.A.B., 2022, A Multi-Agent Decision-Making Model for the Ranking of Energy Storage Technologies, Chemical Engineering Transactions, 94, 1141-1146 DOI:10.3303/CET2294190 CHEMICAL ENGINEERING TRANSACTIONS VOL. 94, 2022 A publication of The Italian Association of Chemical Engineering Online at www.cetjournal.it Guest Editors: Petar S. Varbanov, Yee Van Fan, Jiří J. Klemeš, Sandro Nižetić Copyright © 2022, AIDIC Servizi S.r.l. ISBN 978-88-95608-93-8; ISSN 2283-9216 A Multi-Agent Decision-Making Model for the Ranking of Energy Storage Technologies Joseph R. Ortenero*, Angelo Earvin Sy Choi, Michael Angelo B. Promentilla Department of Chemical Engineering, De La Salle University, 2401 Taft Ave, Manila 0922, Philippines joseph.ortenero@dlsu.edu.ph Energy storage can help in solving the problem of intermittency associated with renewable energy as well as provide a reliable and stable energy supply in the transition to a low carbon society. Energy storage technologies (EST) that are available in the market and in the process of development have their specific strengths and weaknesses. The selection of an EST for a specific application requires the evaluation of its various characteristics. The development of sustainable energy storage necessitates a multi-criteria approach and robust decision support systems. The factors to consider in selecting the best EST from multiple alternatives are energy density, specific energy, cycle efficiency, power density, specific power, technology readiness level (TRL), power/energy capital cost, and lifespan. This study proposes a fuzzy multicriteria decision-making method in a multi-agent environment. A consensus measure is incorporated in the decision model where the evaluation of criteria or alternatives is vague or imprecise. A case study is presented to demonstrate the use of such ranking methodologies which could guide decision-makers in selecting the best EST for stationary power application. 1. Introduction According to the International Energy Agency (IEA, 2021), the contribution of renewable electricity to the power grid is forecast to increase by 60 % between 2020 – 2026, to about 4,800 GW. This is equivalent to the current combined fossil fuel and nuclear power output in the world. The electricity output from renewable energy (RE) is projected to increase further to achieve the pathway to net zero by 2050. To achieve the committed target in COP26, governments around the world should address policy and implementation challenges for renewable and include investment in energy storage technologies (ESTs). RE still has its shortcomings that can be a hindrance to full integration and implementation in the energy and power grid (Chofrey et al., 2019). The major issue of RE is due to its intermittency which affects its dependability as an alternative energy source. It could be addressed through integration of smart systems of energy storage technologies (Ortenero and Tan, 2021). EST could serve as a buffer during peak energy demand and provide a stable and reliable power supply. There have been continual progressive improvements in energy storage systems for integration with renewable energy (RE). Various energy systems have a vital role in energy harvesting derived from different sources. This in turn converts these forms of energy to the required functions of application associated with the building, industry, transportation (Chuah et al., 2021), and utility. In thermal power plants, energy generated from fossil fuel can be readily used according to customer demand. However, other RE sources that include wind and solar require harvesting and proper storage until the demand is needed. The application of energy storage has multiple benefits in its energy systems. This includes the following: 1) good economic performance, 2) allowing substantial penetration of RE, 3) essential towards electrical systems for damping energy oscillations, proper peak shaving and load leveling, regulated frequency, 4) and refined power reliability and quality (Gielen et al., 2019). Various forms of energy storage systems have been introduced and they are categorized according to the type of energy transformation from electricity generated from RE to storage. Among these classifications are mechanical energy storage (MES), electrochemical energy storage (EES), chemical energy storage (CES), electrical and magnetic energy storage (EMES), and thermal energy storage (TES). EST can be further 1141 classified based on the chemical composition of the active materials, mode of storage, and type of chemical storage (Koohi-Fayegh and Rosen, 2020). The performance of ESTs varies widely depending on the type and level of development. One way to evaluate ESTs is to check its performance with respect to characteristics such as power density, energy density, specific energy/power, life cycle, efficiency, technology readiness level, environmental impact, and power and energy capital cost. The type of application is a major consideration such that a specific criterion could play a critical role in the selection. For instance, for space-constrained applications in urban areas, compact, light, and modular type are most preferable, so high specific and volumetric energy/power density are dominant factors to be considered. The storage energy density and power density are important aspects to consider to appropriately evaluate the energy accumulation and energy transfer per unit mass/volume. In areas of abundant land and with the right geographical features, specific energy/power capital cost could be the dominant criteria. This study is carried out to use a proposed innovative method in ranking energy storage technologies as illustrative case study. Three experts have been involved in the decision-making process that entail value judgments and uncertain data for the complex problem in the case study. The novel application in the fuzzy multi-criteria decision making under a multi-agent environment is to rank several ESTs based on the rating of these experts. This then highlights the spherical fuzzy extension of simple additive weighting to rank the alternatives with the simultaneous methods of the spherical analytic hierarchy process (SFAHP) and spherical fuzzy additive weighting (SFAW) to derive the criteria weights for weighted performances and the fuzzy evaluation matrix (Gündoğdu and Kahraman, 2019). Spherical fuzzy set is an essential component to model uncertainty in human opinion, which is the most recent extension of Zadeh’s fuzzy set by expressing fuzziness into three components namely membership degree, non-membership degree and indeterminacy degree under the condition that the sum of squares of these components is less than one (Mahmood et al., 2019). A generated defuzzification of the spherical fuzzy score is then used to rank the alternative that leads to the computation of the consensus. A group-aggregated score to rank the alternatives can then be achieved by the experts through a quick and efficient method in arriving with a linguistic rating. 2. Methodology 2.1 Preliminaries This section introduces the definitions related to spherical fuzzy set and its generalization, T-spherical fuzzy set. Definition 1. Let X be in a finite domain and 𝑥𝑥 ∈ 𝑋𝑋. T-spherical fuzzy set (TSFS) is defined as: 𝑇𝑇 = {𝑥𝑥, 𝜇𝜇(𝑥𝑥), 𝜈𝜈(𝑥𝑥), 𝜋𝜋(𝑥𝑥) | 𝑥𝑥 ∈ 𝑋𝑋} with the condition that 0 ≤ 𝑆𝑆𝑆𝑆𝑆𝑆(𝜇𝜇𝑡𝑡, 𝜈𝜈𝑡𝑡, 𝜋𝜋𝑡𝑡) ≤ 𝑟𝑟𝑡𝑡 ∀ 𝑡𝑡 ∈ 𝑍𝑍 ≥ 1 . Here three components 𝜇𝜇, 𝜈𝜈, 𝜋𝜋: 𝑋𝑋 → [0,1] represents the degree of membership, degree of non-membership, and degree of indeterminacy. 𝑍𝑍 refers to positive integers and 𝑟𝑟𝑡𝑡 → �1, 31/𝑡𝑡�. Note that 𝑟𝑟𝑡𝑡 = 1 is a particular case of T in X, for example is a spherical fuzzy set (SFS) at t = 2 with the condition of 0 ≤ 𝑆𝑆𝑆𝑆𝑆𝑆(𝜇𝜇2, 𝜈𝜈2, 𝜋𝜋2) ≤ 1, i.e., 0 ≤ 𝜇𝜇2 + 𝜈𝜈2 + 𝜋𝜋2 ≤ 1. For ease of computation, T-spherical fuzzy number is designated as an ordered triple: 𝑇𝑇�𝑠𝑠 = (𝜇𝜇𝑇𝑇�𝑠𝑠, 𝜈𝜈𝑇𝑇�𝑠𝑠, 𝜋𝜋𝑇𝑇�𝑠𝑠). Definition 2. T-spherical weighted geometric mean (TSWGM) is an aggregation operator for n T-spherical fuzzy numbers using weighted geometric mean such that the weight vector 𝑤𝑤𝑖𝑖 ∈ [0,1]; ∑ 𝑤𝑤𝑖𝑖 = 1𝑛𝑛𝑖𝑖=1 . 𝑇𝑇𝑆𝑆𝑇𝑇𝑇𝑇𝑇𝑇�𝑇𝑇�𝑠𝑠1 … . . 𝑇𝑇�𝑠𝑠𝑛𝑛� = Π𝑖𝑖=1 𝑚𝑚 (𝑇𝑇�𝑠𝑠𝑛𝑛)𝑤𝑤𝑖𝑖 = �∏ 𝜇𝜇𝑇𝑇�𝑠𝑠𝑖𝑖 𝑤𝑤𝑖𝑖𝑛𝑛 𝑖𝑖=1 , �1 − ∏ �1 − 𝜈𝜈𝑇𝑇�𝑠𝑠𝑖𝑖 𝑡𝑡 � 𝑤𝑤𝑖𝑖𝑛𝑛 𝑖𝑖−1 � 1 𝑡𝑡 , ∏ 𝜋𝜋𝑇𝑇�𝑠𝑠𝑖𝑖 𝑤𝑤𝑖𝑖 ,𝑛𝑛𝑖𝑖=1 � (1) Definition 3. T-spherical weighted arithmetic mean (TSWAM) is an aggregation operator for n T-spherical fuzzy numbers using weighted arithmetic mean such that the weights 𝑤𝑤𝑖𝑖 ∈ [0,1]; ∑ 𝑤𝑤𝑖𝑖 = 1𝑛𝑛𝑖𝑖=1 . 𝑇𝑇𝑆𝑆𝑇𝑇𝑇𝑇𝑇𝑇𝑤𝑤�𝑇𝑇�𝑠𝑠1 … . . 𝑇𝑇�𝑠𝑠𝑛𝑛� = 𝑤𝑤1𝑇𝑇�𝑠𝑠1 + 𝑤𝑤2𝑇𝑇�𝑠𝑠2 + ⋯ + 𝑤𝑤𝑛𝑛𝑇𝑇�𝑠𝑠𝑛𝑛 = ∑ 𝑤𝑤𝑖𝑖𝑇𝑇�𝑆𝑆𝑖𝑖𝑛𝑛𝑖𝑖=1 = ��1 − ∏ �1 − 𝜇𝜇𝑇𝑇�𝑠𝑠𝑖𝑖 𝑡𝑡 � 𝑤𝑤𝑖𝑖𝑛𝑛 𝑖𝑖−1 � 1 𝑡𝑡 , ∏ 𝑣𝑣𝑇𝑇�𝑠𝑠𝑖𝑖 𝑤𝑤𝑖𝑖 , ∏ 𝜋𝜋𝑇𝑇�𝑠𝑠𝑖𝑖 𝑤𝑤𝑖𝑖 𝑛𝑛𝑖𝑖=1 𝑛𝑛 𝑖𝑖=1 � (2) Definition 4. Defuzzification of T-spherical fuzzy number is defined as follows: 𝑆𝑆𝑆𝑆𝑆𝑆𝑟𝑟𝑆𝑆�𝑇𝑇�� = 1 − �1 3 �(1 − 𝜇𝜇𝑡𝑡)𝛽𝛽 + (𝜈𝜈𝑡𝑡)𝛽𝛽 + (𝜋𝜋𝑡𝑡)𝛽𝛽�� 1 𝛽𝛽� (3) where 𝛽𝛽 ≥ 1 is the distance parameter. Here the 𝑆𝑆𝑆𝑆𝑆𝑆𝑟𝑟𝑆𝑆(𝑇𝑇�) → [0,1]. 1142 2.2 Fuzzy Multiple Criteria Decision-Making (MCDM) with consensus measure The proposed method in a step-by-step procedure is depicted in Figure 1. Figure 1. Procedural flow of the proposed fuzzy MCDM under multiple agents The procedures are as follows: Step 1: Create the evaluation matrix by defining m alternatives and n criteria. Step 2: Elicit from each expert the relative importance (see Table 1) of n criteria to populate the spherical fuzzy pairwise comparison matrix. The 𝑎𝑎𝑖𝑖𝑖𝑖 represent the relative importance rating of row criteria i over the column criteria j. The 𝑎𝑎𝑖𝑖𝑖𝑖 is the inverse of 𝑎𝑎𝑖𝑖𝑖𝑖. For example, if criteria i is strongly more important (STM) than criteria j, then criteria j is strongly less important (STL) than criteria i. Just like the classic AHP, the matrix requires n(n- 1)/2 pairwise comparative judgments. Note that the entries in the diagonal (𝑎𝑎𝑖𝑖𝑖𝑖) is automatically set to EQ rating, i.e, criteria i is equally important to criteria i. To compute the weights, the importance rating is transformed to the spherical fuzzy number described in Table 1. Details of computation is described elsewhere (Kuok and Promentila, 2021). Step 3: Elicit from each expert the performance rating of the alternatives to populate the individual fuzzy evaluation matrix 𝑍𝑍 = �𝑧𝑧𝑖𝑖𝑖𝑖�𝑚𝑚𝑚𝑚𝑛𝑛 . In this matrix, the number of rows corresponds the number of alternatives whereas the number of columns corresponds the number of criteria. Use the linguistic scale provided in Table 1 for the performance rating. Let 𝑇𝑇�𝑖𝑖𝑖𝑖 as the entry to 𝑧𝑧𝑖𝑖𝑖𝑖 to describe the performance rating of alternative i with respect to criteria j as represented by a spherical fuzzy number. Step 4: Calculate the composite score of the alternatives from evaluation matrix using SFAW which is analogous to simple additive weighting (SAW) method. This can be done by finding the sum of the weights of the performance rating on each alternative on all criteria so that it can determine the best alternative. Since the performance rating is in a form spherical fuzzy number, the spherical fuzzy set extension of SAW is defined by the following equation: �̃�𝑆𝑖𝑖 = ∑ 𝑤𝑤𝑖𝑖𝑇𝑇�𝑖𝑖𝑖𝑖 ,𝑚𝑚𝑖𝑖=1 ∀ 𝑖𝑖 = 1 … 𝑆𝑆 (4) where �̃�𝑆𝑖𝑖 is the total spherical fuzzy score of alternative i and the criteria weight vector 𝑤𝑤𝑖𝑖 ∈ [0,1]; ∑ 𝑤𝑤𝑖𝑖 = 1𝑛𝑛𝑖𝑖=1 . ��̃�𝑆𝑖𝑖 = (𝜇𝜇�̃�𝑆𝑖𝑖, 𝜈𝜈�̃�𝑆𝑖𝑖, 𝜋𝜋𝑆𝑆𝑖𝑖) = �1 − ��1 − 𝜇𝜇𝑇𝑇�𝑖𝑖𝑖𝑖 2 � 𝑤𝑤𝑖𝑖 𝑛𝑛 𝑖𝑖−1 � 1 2 , �𝑣𝑣𝑇𝑇�𝑖𝑖𝑖𝑖 𝑤𝑤𝑖𝑖 , �𝜋𝜋𝑇𝑇�𝑖𝑖𝑖𝑖 𝑤𝑤𝑖𝑖 𝑛𝑛 𝑖𝑖=1 𝑛𝑛 𝑖𝑖=1 � (5) Eq(5) uses the aggregation operation described in Eq(2) at t = 2 for spherical fuzzy set. Step 5: Defuzzification of the spherical fuzzy score to rank the alternatives. Eq(3) at 𝑡𝑡 = 2 is used to compute the crisp score (𝑆𝑆�̅�𝑖) of the alternatives. The distance parameter 𝛽𝛽 is set to 19/8 to give a score of 0.50 to a satisfactory rating with a spherical fuzzy number (0.50, 0.50, 0.50). The higher the score, the closer the alternative to the ideal alternative. Thus, the highest score is ranked 1st, followed by the second highest score, and so on. Step 6: Compute the consensus measure using the Kendall’s W or coefficient of concordance (Kendall and Smith, 1939). The ranking of alternatives from each expert are used as input in the calculation to measure the degree of agreement among experts. A Kendall’s W of 0.60 or higher suggests strong consensus while a Kendall’s W of lower than 0.30 suggests weak consensus. 1143 Step 7: Compute the group-aggregated score from k experts ranking of alternatives using the following equation: 𝑆𝑆𝑖𝑖 = ∑ 𝑤𝑤ℎ𝑆𝑆�̅�𝑖 ,𝑘𝑘ℎ=1 ∀ 𝑖𝑖 = 1 … 𝑆𝑆 and expert weight vector 𝑤𝑤ℎ ∈ [0,1]; ∑ 𝑤𝑤ℎ = 1 𝑘𝑘 ℎ=1 (6) The highest group-aggregated score is ranked 1st. Table 1: Linguistic scale used in this study with the corresponding spherical fuzzy numbers Performance Rating (μ, ν, π) Relative Importance Rating (μ, ν, π) Excellent (EX) (0.90, 0.10, 0.10) Very highly more important (VSM) (0.90, 0.10, 0.10) Very good (VG) (0.80, 0.20, 0.25) Strongly more important (STM) (0.80, 0.20, 0.25) Good (GD) (0.70, 0.30, 0.35) Moderately more important (MM) (0.70, 0.30, 0.35) Slightly good/Above satisfactory (AS) (0.60, 0.40, 0.40) Slightly more important (SM) (0.60, 0.40, 0.40) Moderate/Satisfactory (S) (0.50, 0.50, 0.50) Equally important (EQ) (0.50, 0.40, 0.40) Slightly bad/Below Satisfactory (BS) (0.40, 0.60, 0.40) Slightly less important (SL) (0.40, 0.60, 0.40) Bad (BD) (0.30, 0.70, 0.35) Moderately less important (ML) (0.30, 0.70, 0.35) Very bad (VB) (0.20, 0.80, 0.25) Strongly less important (STL) (0.20, 0.80, 0.25) Worst (WO) (0.10, 0.90, 0.10) Very strongly less important (VSL) (0.10, 0.90, 0.10) 3. Case Study Energy storage is seen as the solution to provide a buffer between supply and demand by storing the excess energy generated during lean season and providing additional capacity during peak energy consumption. Several energy storage technologies are available that can be used to provide frequency regulation, stability in energy supply from RE, and damping energy oscillations. The ESTs that are considered in this study are flywheel energy storage (FWES) and pump hydro energy storage (PHES), which are forms of MES. The other ESTs are lithium-ion battery (LIB) and sodium-sulfur battery (SSB), which are under EES classification. There are other types of ESTs, but these are considered the most popular, display superior properties in storing energy, and technological maturity, which are important for effective storage of energy from RE. Table 2: Evaluation matrix of the four alternative ESTs by four criteria E1 Energy Storage Technology (EST) Specific Energy Density Efficiency Cycle Life Energy Capital Cost FWES S GD VG VB PHES BD BS BD EX LIB GD GD WO BS SSB EX GD VB GD E2 Energy Storage Technology Specific Energy Density Efficiency Cycle Life Energy Capital Cost FWES AS EX EX WO PHES VB GD EX VG LIB EX EX GD S SSB VG GD AS AS E3 Energy Storage Technology Specific Energy Density Efficiency Cycle Life Energy Capital Cost FWES GD VG VG BD PHES S GD VG VG LIB VG EX AS S SSB VG GD GD AS These ESTs are compared to determine the best option for a specific application, for instance in stationary power application. The attributes that are commonly checked for comparison are power density, energy density, specific energy/power, life cycle, efficiency, technology readiness level, environmental impact, and power and energy capital cost. For demonstration of the proposed fuzzy MCDM, the attributes that are considered are specific energy density (Wh/kg), efficiency (%), cycle life (h), and energy capital cost ($/kWh). The rating 1144 provided, for example, by three experts or decision makers noted as Expert 1 (E1), Expert 2 (E2), and Expert 3 (E3) are shown in Table 2. Note that the performance rating is based on the linguistic scale provided in Table 1. For example, E3 rates performance of flywheel energy storage (FWES) as good (GD), very good (VG), very good (VG), and bad (BD) with respect to specific energy density, efficiency, cycle life and energy capital cost, respectively. To rank these alternative ESTs, these ratings will be aggregated through a simple additive weighting to yield a composite score for each alternative. In real multi-agent decision making, each expert will have their own set of weight vectors that will reflect the relative importance of the criteria used for evaluation. Table 3 provides an example of the pairwise comparison matrix that describe the value judgments of experts in determining the weight vectors of the criteria. For example, E1 thinks that specific energy density is moderately more (MM) important than efficiency. This judgment is reflected from the first row, second column of the pairwise comparison matrix shown in Table 3. Note that these verbal judgments are transformed to spherical fuzzy numbers (µ, v, π) according to Table 1. The weights are then computed using the algorithm of spherical fuzzy AHP (SFAHP) described in Kuok and Promentilla (2021). Result for the criteria weight vectors is summarized in Table 4. Table 3: Sample pairwise comparison matrix from three decision makers for SFAHP weights E1 EST Classification Specific Energy Density Efficiency Cycle Life Energy Capital Cost Specific Energy Density EQ MM STM SM Efficiency ML EQ STM MM Cycle Life STL STL EQ SL Energy Capital Cost SL ML SM EQ E2 EST Classification Specific Energy Density Efficiency Cycle Life Energy Capital Cost Specific Energy Density EQ STM ML SL Efficiency STL EQ MM SL Cycle Life MM ML EQ STM Energy Capital Cost SM SM STL EQ E3 EST Classification Specific Energy Density Efficiency Cycle Life Energy Capital Cost Specific Energy Density EQ ML EQ ML Efficiency MM EQ SM SL Cycle Life EQ SL EQ EQ Energy Capital Cost MM SM EQ EQ Table 4: Criteria weights for the specific energy density, efficiency, cycle life, and energy capital cost using fuzzy AHP Criteria E1 E2 E3 Specific Energy Density 0.324 0.259 0.197 Efficiency 0.305 0.223 0.283 Cycle Life 0.158 0.292 0.229 Energy Capital Cost 0.213 0.226 0.291 It can be observed in Table 4 that there are differences in the way each expert puts importance on the different criteria, E1 values specific energy density, E2 prefers cycle life, and E3 puts premium on energy capital cost. Thus, the ranking of the alternatives could differ even though the experts may have comparable performance ratings for this set of alternatives. Table 5 shows the scores obtained from the defuzzification of the spherical fuzzy numbers (µ, v, π) resulted from simple additive weighting of rating for each alternative. The value closest to 1 is the best alternative while the value closest to zero is the worst. It can be observed that there is an agreement between E2 and E3 when it comes to the best (LIB) and worst (SSB) performing EST technology while E1 prefers the fourth EST (SSB) as the best option. Assuming equal weights for the experts, the group-aggregated score is also shown in Table 1145 5. Kendall consensus index with a value W = 0.222 shows that there is weak consensus in the ordinal ranking of the alternatives from the three experts. The consensus among the experts is that the two battery energy storage technologies are the top options for stationary power applications while the two MES lag in performance. It should be noted that the two battery storage technologies have superior properties when it comes to volumetric and specific energy density and efficiency. Lithium-ion battery has the highest volumetric power density which is about sixty times that of second ranked SSB on average. Sodium sulfur battery has slightly higher volumetric and mass energy density and it is more affordable than lithium-ion battery. However, sodium sulfur is not yet widely used compared with lithium-ion but there is great potential for sodium-sulfur to compete with lithium-ion in the future. PHES is the worst technology due to its low volumetric power and energy density while FWES suffers from high cost, about three times that of lithium-ion, while the specific energy density is about one-sixth that of lithium-ion. Table 5: Ranking of the alternative ESTs Energy Storage Technology E1 E2 E3 Group-aggregated Score Rank Score Rank Score Rank Score Rank FWES 0.596 2 0.761 2 0.677 3 0.678 3 PHES 0.588 4 0.736 3 0.706 2 0.677 4 LIB 0.595 3 0.781 1 0.730 1 0.702 1 SSB 0.741 1 0.663 4 0.674 4 0.692 2 4. Conclusions This work applied the fuzzy multi-criteria decision analysis under a multi-agent environment to rank the energy storage technologies based on the following four criteria: specific energy density, efficiency, cycle life, and energy capital cost. The relative importance of the criteria was made explicit using the process of spherical fuzzy AHP. Ranking is based on the composite scores obtained from the spherical fuzzy evaluation matrix. Though the group consensus measure indicates a weak consensus, the group-aggregated score indicates that lithium- ion battery is the best alternative followed closely by sodium-sulfur battery. The worst technology based on fuzzy MCDA methodology is PHES due to its low volumetric power and energy density. Future work includes extending the methodology to more EST alternatives, consider more criteria to gauge the overall performance of the technologies, and include more experts in the decision-making process. This methodology could serve as a guide to decision-makers in the selection of appropriate energy storage technology that could be integrated to existing or future installation of renewables. References Chofreh A.G., Goni F.A., Davoudi M., Klemeš J.J., Moosavi S., 2019, Strategic plan method for future renewable energy technologies. Chemical Engineering Transactions, 76, 1159–1164. Chuah L.F., Salleh N.H.M., Osnin N.A., Alcaide J.I., Majid M.H.A., Abdullah A.A., Bokhari A., Jalil E.E.A., Klemes J.J., 2021, Profiling Malaysian ship registration and seafarers for streamlining future Malaysian shipping governance, Australian Journal of Maritime & Ocean Affairs, 13, 225–261. Kendall M.G., Smith B.B., 1939, The problem of rankings, Annals of Mathematical Statistics, 10, 275-287. Mahmood T., Ullah K., Khan Q., 2019, An approach toward decision-making and medical diagnosis problems using the concept of spherical fuzzy sets, Neural Computing & Applications, 31, 7041–7053. Gielen D., Boshell F., Saygin D., Bazilian M.D., Wagner N., Gorini R., 2019, The role of renewable energy in the global energy transformation, Energy Strategy Reviews, 24, 38–50. Gündoğdu F.K., Kahraman C., 2019, A novel spherical fuzzy analytic hierarchy process and its renewable energy application, Soft Computing, 24, 4607–4621. International Energy Agency, 2021, Report Extract: Executive Summary accessed 22.02.2022 Koohi-Fayegh S., Rosen M.A., 2020, A review of energy storage types, applications and recent developments, Journal of Energy Storage, 27, 101047. Kuok F., Promentilla M.A.B., 2021, Problem analysis on public-private partnership for small and medium enterprises: A case study in Cambodia, Chemical Egineering Transactions, 88, 841 – 846. Ortenero J.R., Tan R.R., 2021, Ranking energy storage technologies with VIKOR, Chemical Engineering Transactions, 88, 151–156. 1146 PRES22_0397.pdf A Multi-Agent Decision-Making Model for the Ranking of Energy Storage Technologies