Fuzzy Logic–Enhanced PMC Index for Assessing Policies for Decarbonization in Higher Education: Evidence from a Public University

Abstract

Higher education institutions play a critical role in the transition to a low-carbon future due to their research capacity and societal influence. Accordingly, the calculation of greenhouse gas (GHG) emissions and the prioritization of mitigation strategies are of particular importance. In this study, a comprehensive campus-level GHG inventory was prepared for a public university in Türkiye in alignment with the ISO 14064-1:2018 standard, and mitigation strategies were evaluated. To prioritize these strategies, both the classical Policy Modeling Consistency (PMC) index and, for the first time in the literature, a fuzzy extension of the PMC model was applied. The results reveal that the total GHG emissions for 2023 amounted to 4888.63 tCO₂e (1.19 tCO₂e per capita), with the largest shares originating from investments (31%) and purchased electricity (28.38%). While the classical PMC identified only two high-priority actions, the fuzzy PMC reduced score dispersion, resolved ranking ties, and expanded the number of high-priority actions to seven. The top strategies include awareness programs, energy-efficiency measures, virtual meeting practices, advanced electricity monitoring, and improved data management systems. By comparing the classical and fuzzy approaches, the study demonstrates that integrating fuzzy logic enhances the transparency, reproducibility, and robustness of strategy prioritization, thereby offering a practical roadmap for campus decarbonization and sustainability policy in higher education institutions.

1. Introduction

Higher Education Institutions (HEIs) are under mounting pressure to institutionalize sustainability principles due to rising stakeholder expectations and regulations, stricter accreditation requirements, and ambitious net-zero targets. Because campuses function as “compact cities,” emissions associated with energy consumption—encompassing heating, cooling, lighting, and laboratory operations—together with mobility and procurement related to campus services and supplies become particularly salient [1,2]. Given their multi-source character, these emissions call for strategies that privilege pollution prevention at the source rather than end-of-pipe fixes [1,3].

The first step toward carbon neutrality for HEIs is the comprehensive and up-to-date accounting of institutional greenhouse-gas (GHG) emissions [3,4]. In recent years, initiatives such as the American College and University Presidents’ Climate Commitment and the United Nation-backed Race to Zero have accelerated the adoption of campus carbon-footprint accounting [5,6,7,8,9,10] Building on this momentum, a broad literature has reported GHG inventories for HEIs using the GHG Protocol across diverse contexts, including Colombia [11] Chile [12], Mexico [13,14], Saudi Arabia [15], the United States [16], Indonesia [17], New Zealand [18], Korea [19], the United Arab Emirates [20], the United Kingdom [21], and Finland [22].

A smaller set of studies employs the GHG Protocol in combination with International Organization for Standardization (ISO) 14064-1:2018 GHG Standard: often tailoring the methodology for university contexts [23,24,25]. While the GHG Protocol remains the most commonly applied framework for HEIs, applications strictly aligned with ISO 14064—especially the latest ISO 14064-1:2018—are still limited; yet this international standard is critical for producing complete, decision-grade inventories and for ensuring the integrity of mitigation strategy design [26,27].

Beyond identifying decarbonization options, prioritization determines which actions should be implemented first under resource constraints, and the literature predominantly turns to multi-criteria decision analysis (MCDA) for this purpose. The most common line of work applies the Analytic Hierarchy Process (AHP): expert-based weighting weighting to design roadmaps and prioritize sustainability strategies [28,29,30]. When judgments are linguistic and uncertainty is significant, fuzzy/hybrid MCDA is favoured, with studies applying methods such as Fuzzy AHP–VIKOR and Fuzzy AHP and interval TOPSIS [31,32,33,34]. Other contemporary approaches—such as neuromorphic sets, fuzzy credibility graphs, simulation-based AHP, life-cycle lenses, and actor-cantered frameworks further illustrate the breadth of methodological innovation in this field [35,36,37,38,39,40,41,42].

Collectively, these studies position MCDA and its fuzzy derivatives as the contemporary standard for ranking carbon-reduction strategies, and they frame our proposed Fuzzy Policy Modeling Consistency (PMC) as methodologically aligned and complementary under uncertainty.

Proposed by Mario Arturo Ruiz Estrada and grounded in the Omnia Mobilis hypothesis, the PMC index has become a text-based framework for evaluating the internal coherence of policy documents [43] and it is increasingly applied to climate and carbon governance. Recent studies illustrate this focus: Meng and Xu (2023) assessed national and sub-national climate and energy policies [44]. Yiye et al. (2025) examined green development measures in civil aviation [45]. Zhang, Chen, Zheng, and Chen (2023) evaluated maritime decarbonization policies [46], Ekoh, Eneh, and Enyi (2025) and Dai et al. (2021) analysed climate and ecological policies across multiple sectors [47,48]. Taken together, this literature shows that the PMC index provides a decision-oriented lens for the systematic and comparable assessment of climate-related policy texts, enabling researchers to detect design weaknesses, and guide targeted reforms without relying on auxiliary modeling assumptions.

Fuzzy set theory, introduced by Zadeh (1965), addresses the limits of classical set theory in representing uncertainty, and its extensions have been widely applied in contexts with ambiguity, incomplete information, or linguistic judgments [49]. Numerous studies demonstrated that fuzzy approaches outperform deterministic techniques when uncertainty cannot be fully captured [50]. Building on this theoretical foundation, extending the PMC framework to incorporate linguistic uncertainty is a methodological necessity for assessing policy text coherence, yet to our knowledge no fuzzy PMC application has been reported.

This study advances an integrated framework at the intersection of campus carbon accounting and policy prioritization/decision support, addressing several salient gaps in the literature. First, higher-education carbon-footprint assessments are grounded in the GHG Protocol and frequently underrepresent major portions of Scope 3; applications of ISO 14064-1:2018 that cover all emission categories remain rare. In the Turkish context, a comprehensive, campus-scale implementation of this standard has not been documented; our study therefore is among the first systematic applications of ISO 14064-1:2018 at the campus level. Second, although multi-criteria approaches are widely used for ranking mitigation options, campus-level applications that read the internal coherence of policy texts through the PMC lens are lacking. We address this by bringing PMC to the campus scale and by introducing and applying Fuzzy PMC—the first such campus-level application—which incorporates linguistic judgments and uncertainty absent in binary scoring. The resulting approach enables transparent, reproducible, and decision-oriented prioritization of campus policy options, providing a robust methodological foundation for strategy ranking and an evidence-based roadmap for resource allocation higher-education.

2. Materials and Methods

2.1. Campus GHG Inventory

ISO 14064 standards were created in 2006 to calculate, report, and verify GHG emissions and removals [27]. Updated in 2018, ISO 14064-1:2018 remains the most relevant standard, providing a structured methodology for quantifying and reporting organizational GHG emissions. This study applies ISO 14064-1:2018 at campus scale through the following steps:

I.

Establishment of Organizational Boundaries

This step delineates the organizational limits in a manner consistent with the study’s intended purpose. A consolidation approach must be selected—either the control-based or the equity-share approach.

The case application was conducted at Abdullah Gül University (AGU), a public institution in Kayseri, Türkiye; it covered the Sümer Campus—repurposed from the historic Sümerbank Textile Factory and encompassed 3780 students and 586 staff. The control approach was selected, focusing on operations under AGU’s administrative or financial control.

II.

Definition of Reporting (Operational) Boundaries

This phase specifies the operational parameters of the organization, covering both direct and indirect GHG emissions and removals attributable to its activities.

Reporting boundaries were established in line with ISO 14064-1:2018 to include direct and indirect emissions and removals attributable to AGU’s operations. The organizational-level categorization of sources follows the six ISO categories (see Table S1).

III.

Data Collection

Accurate data collection is foundational to robust GHG quantification. Activity data are obtained from enterprise resource planning (ERP) systems, utility invoices, and other verified records. Appropriate emission factors are assigned to each source category using recognized references [51,52]. Detailed conversion factors and parameter values used in this study are listed in the Supplementary Materials Tables S2–S5. Activity data (AD) is presented in Tables S6–S20. To ensure data privacy, commuting and business-travel collected from students and staff were anonymized and aggregated before analysis, and no personal identifiable information was processed in the inventory.

IV.

Emissions Calculation

Emissions are computed for each category by combining the collected activity data with the selected emission factors. Total organizational emissions are obtained by aggregating category-level results and expressing them as carbon dioxide equivalents (CO₂e). The general calculation formula is:

$$GHG Emission=Activity Data\times Emission Factor$$

(1)

Category 1 includes stationary and mobile combustion, industrial processes, fugitive emissions, and land use; at AGU only combustion and fugitive sources were relevant. Category 2 covers purchased electricity, with other imported energy not applicable. Category 3 comprises commuting and business travel, while product transport was excluded as AGU does not manufacture goods. Category 4 covers products, capital goods, waste, and fuel supply chains, quantified through activity data or expenditure-based proxies. Category 5 was limited to investments, including ongoing campus construction.

V.

Verification and Reporting

Quantification and disclosure adhere to ISO guidance to ensure relevance, completeness, consistency, accuracy, and transparency. Reporting includes data sources, calculation methods, and parameter choices to enable replication. The GHG inventory was independently verified by an accredited third-party assurance firm in accordance with ISO 14064-3:2019 [53].

VI.

Development of a Reduction Strategy

Once the institutional carbon footprint was established, the next step toward net-zero is to identify and implement targeted emission-reduction strategies. Strategy design prioritizes the most significant sources and defines focused mitigation actions. In this study, the mitigation options were compiled from the peer-reviewed literature and international best-practice guidance organized according to their relevance to AGU’s dominant sources.

2.2. Policy Modeling Consistency (PMC)

The PMC-Index, developed by Ruiz Estrada, quantitatively evaluates the design quality and internal coherence of public policies [43,54]. It enables cross-policy comparison by systematically decomposing policy texts into main and sub-variables and evaluating consistency across these dimensions [55,56]. By integrating indicators without arbitrary exclusions and using an equal-weighting scheme, the approach minimizes subjective scoring bias [57,58]. It encourages the inclusion of all relevant variables [59]. In addition, multidimensional visualization via PMC surfaces helps reveal strengths and weaknesses in an intuitive way [60,61]. Owing to its comprehensiveness and operational practicality, the method has been widely used across domains such as environmental–energy governance, economics, and social policy [62,63,64].

In this section, strategies for reducing carbon emissions are prioritized using the PMC-Index. The goal is to assess the consistency of mitigation strategies, highlight strengths and weaknesses, and propose improvements. The application steps are as follows:

Step 1: Identifying strategies.

Prior research shows that understanding and mitigating emission sources is critical for both academic institutions and large organizations [65,66]. Drawing on the literature, a set of GHGs mitigation strategies appropriate for AGU’s Sümer Campus was identified (

Table 1. The Strategies for Reducing GHGs Emissions.

No	Strategy	Explanations	Category	Emissions Share	References
S1	Consumer awareness and stakeholder engagement	Consumer Awareness: It includes organizing awareness campaigns and training about GHG emission sources such as energy saving.	All	100%	[67,68,69]
S2	Improving inventory data	One approach to improving data collection is to create a comprehensive data repository and analyse existing data to identify opportunities for improved data collection methods.	All	100%	[6,70,71]
S3	Increasing the efficiency of the energy management system	The optimization of energy use and the identification of inefficiencies help reduce carbon emissions by implementing a system for monitoring and analysing electricity consumption. Organizations make informed decisions to improve energy efficiency and reduce their overall CF by monitoring real-time electricity usage data.	1.1, 2.1	43.31%	[72,73]
S4	Adjusting the temperatures of heating and cooling systems	Determining ideal indoor temperatures and ensuring compliance with these values through automation systems.	1.1	14.93%	[74,75]
S5	Modernization of Heating and cooling Systems	Replace natural gas boilers with more energy-efficient models and transition to low-carbon or carbon-neutral heating systems. Switching to cleaner alternatives such as electric heat pumps.	1.1	14.93%	[76,77,78,79,80]
S6	Monitoring and Analysing Electricity Consumption	Monitoring and analysing electricity consumption helps identify patterns and inefficiencies in energy use. This strategy enables targeted actions to reduce electricity waste and optimize energy management.	2.1	28.38%	[81]
S7	Use of Renewable Energy Sources	Switching to renewable energy sources such as solar or wind energy for electricity production.	2.1	28.38%	[82,83]
S8	Purchase of Renewable Energy	Reduce emissions from grid electricity by purchasing renewable energy certificates or subscribing to green energy tariffs.	2.1	28.38%	[83,84]
S9	Using electric vehicles	Using electric vehicles reduces GHGs emissions and reliance on fossil fuels. This strategy promotes sustainable transportation and decreases the environmental impact of campus operations.	1.2	0.65%	[85,86]
S10	Public Transport and Shuttle Incentives	Provide transportation allowances or discounts to encourage employees to use public transport or shuttles.	3.4	2.52%	[87]
S11	Bicycle and Pedestrian Infrastructure	Improving bicycle paths and safe pedestrian paths on campus.	3.4	2.52%	[88,89]
S12	Car Sharing Programs	Implementing car sharing programs on campus.	3.3	1.19%	[90,91,92]
S13	Remote Work and education promotion	Increase remote work opportunities.	3.3	1.19%	[93,94,95]
S14	Virtual Meetings	Promote online meetings and conferences as alternatives to physical business travel.	3.5	0.53%	[96,97]
S15	Leak Detection and Repair programs	Implement regular maintenance and leak detection programs for air conditioning and refrigeration systems.	1.4	2.48%	[98,99]
S16	Alternative Refrigerants and other equipment’s	Use of alternative equipment with higher energy performance.	1.4	2.48%	[100]
S17	Green Procurement Policies	Implementing green procurement policies prioritizing products and services with lower carbon footprints reduces supply chain emissions.	4.1, 4.2, 4.4, 4.5, 5.4	46.37%	[101,102]
S18	Waste Reduction and Recycling	Strengthen waste reduction and recycling programs on campus to minimize waste management emissions.	4.3	0.11%	[103]
S19	Water and Wastewater Management	Expansion of the Gray water system in all buildings.	4.3	0.11%	[104,105]
S20	Rainwater Harvesting and Use	Ensuring that this water is used in processes such as irrigation and cleaning by establishing rainwater collection systems.	4.3	0.11%	[106,107]

).

Step 2: Defining PMC criteria.

To evaluate the strategies we adopted nine main variables with binary sub-criteria, guided by Intergovernmental Panel on Climate Change (IPCC), UNFCCC, and prior PMC applications [48,58,63,108,109,110,111]. The full set of criteria is presented in

Table 2. Variable and Sub-variable.

Category Code	Variable	Sub-Variable Code	Sub-Variable
X1	Environmental Impact	X1:1	Carbon reduction potential
		X1:2	Scope coverage
		X1:3	Permanence of reduction
		X1:4	Co-benefits
X2	Economic Viability	X2:1	Cost-effectiveness
		X2:2	Capital expenditure (CAPEX)
		X2:3	Operational expenditure (OPEX)
		X2:4	Payback period
X3	Social Equity	X3:1	Equitable distribution of benefits
		X3:2	Stakeholder acceptance
		X3:3	Inclusiveness in decision-making
X4	Technical and Operational Feasibility	X4:1	Maturity of the solution
		X4:2	Implementation time
		X4:3	Compatibility and scalability
X5	Stakeholder Engagement and Regulatory Compliance	X5:1	Involvement of diverse stakeholders
		X5:2	Alignment with existing regulations
		X5:3	Consistency with industry standards
X6	Risk and Resilience	X6:1	Implementation risk
		X6:2	Supply chain dependency
X7	Implementation and Governance Capacity	X7:1	Clear implementation roadmap
		X7:2	Organizational capacity
		X7:3	Existence of monitoring and feedback mechanisms
X8	Alignment with Long-term Policy and Market Trends	X8:1	Compatibility with national/international climate targets
		X8:2	Potential to attract green finance
X9	Innovation and Knowledge Transfer	X9:1	Degree of innovation
		X9:2	Reproducibility

.

Step 3: Creating the multi-input–output table.

Each strategy was evaluated against the first- and second-level criteria. Sub-variables were coded as 1 (evidence present) or 0 (absent). This yielded the PMC index matrix.

Step 4: Calculating the PMC-Index.

The PMC index for each policy was calculated using the binary scores from the matrix, following the established PMC formulation. The normalized score of the ith main variable is computed using Formula (2), and the overall PMC Index is the equal-weighted sum of the main variables as given in Formula (3).

$${\mathrm{X}}_{\mathrm{i}}={\displaystyle \frac{1}{T_i}}{\sum }_{j=1}^{T_i}{X}_{ij} i = 1, 2, \dots , \mathrm{m}.$$

(2)

$$\mathrm{PMC}={\sum }_{i=1}^m{X}_i$$

(3)

In the equations, m denotes the number of main variables; for each main variable i (i = 1, …, m), T_i is the number of its sub-variables and X_ij ∈ {0, 1} (j = 1, …, T_i) is the binary score for sub-variable j (1 = explicit evidence present, 0 = absent), yielding the main-variable score X_i.

Step 5: Plotting the PMC surface.

PMC surface plots are generated from the index matrix, to visualize strengths and weaknesses across dimensions.

2.3. Proposed Method: Fuzzy PMC Extension

Fuzzy set theory, introduced by Zadeh (1965), addresses the limits of classical sets in representing uncertainty [49]. Fuzzy sets capture intermediate shades through membership degrees and enable reasoning with linguistic variables. Fuzzy sets capture these gradations through a membership degree and enable reasoning with linguistic variables. It has been widely applied in engineering, AI, forecasting, and multi-criteria evaluation [112]. The shape of the membership function is selected to suit the problem at hand; in practice, triangular and trapezoidal forms are frequently preferred for their simplicity and transparency [113]. In this study we reference a step (crisp) membership to illustrate strict threshold behaviour (Figure 1a) and a triangular membership to encode graded assessments (Figure 1b).

In policy evaluation, decision environments are uncertain due to multiple stakeholders, limited information, and subjective judgments. The literature shows that fuzzy logic strengthens decision frameworks in such settings by modelling ambiguity and partial evidence [114,115,116,117].

Building on this, we introduce—to the best of our knowledge for the first time in policy evaluation—a fuzzy extension of the PMC index. Unlike classical PMC (binary coding), fuzzy PMC maps expert judgments to linguistic scales represented by triangular fuzzy numbers (TFN), which are then aggregated within the PMC structure. In doing so, the method preserves PMC’s transparency and comparability while systematically reflecting the strength of evidence and evaluator uncertainty. The details and implementation steps follow in the next section.

Step 1: Identifying strategies.

As in the standard PMC workflow, the GHGs mitigation strategy set for AGU’s Sümer Campus was identified from the literature and is listed in Table 1.

Step 2: Defining PMC criteria and TFN linguistic scale.

The previously defined main/sub-criteria structure is retained. Instead of the binary 0/1 coding in the PMC-index, each subvariant is rated according to the linguistic-TFN scale in

Table 3. Linguistic Terms And Corresponding TFNs [123].

Linguistic Term	Abbrev.	TFN
Very Low	VL	(1, 1, 3)
Low	L	(1, 3, 5)
Medium	M	(3, 5, 7)
High	H	(5, 7, 9)
Very High	VH	(7, 9, 10)

. The linguistic term and corresponding TFNs were adopted in line with established practice in the fuzzy MCDM literature, where triangular membership functions are favored for their simplicity, interpretability, and transparency [118,119,120,121,122]. Triangular fuzzy membership has also been emphasized as the most widely used and computationally efficient representation in fuzzy environments, enabling straightforward aggregation and defuzzification while preserving robustness in decision-making. The selected values were subsequently reviewed and validated by the expert panel to ensure their contextual suitability for assessing campus decarbonization strategies, thereby enhancing both methodological consistency and practical relevance.

Step 3: Creating the fuzzy multi-input–output table.

For each main variable i and sub-variable j, the expert’s linguistic judgment is mapped to a TFN $\tilde{X}$_ij = (l_ij, m_ij, u_ij), yielding the fuzzy PMC matrix..

Step 4: Fuzzy Aggregation of Main-Variable Scores

With T_i sub-variables under the main variable i, compute the fuzzy arithmetic mean with Equation (5).

$$\widetilde{X_i}=1/T_i{⨁}_{j=1}^{T_i}\widetilde{X_{ij}}=\left({\displaystyle \frac{{\sum }_{j=1}^{T_i}{l}_{ij}}{T_i}},{\displaystyle \frac{{\sum }_{j=1}^{T_i}{m}_{ij}}{T_i}},{\displaystyle \frac{{\sum }_{j=1}^{T_i}{u}_{ij}}{T_i}}\right), i=1,\dots m.$$

(4)

Step 5: Calculating the fuzzy PMC-Index.

In fuzzy aggregation, the main variables are combined by simple fuzzy addition (⊕) with equal weights.

$$\widetilde{PMC}={⨁}_{i=1}^m\widetilde{X_i}=\left({\sum }_{i=1}^m{l}_i,{\sum }_{i=1}^m{m}_i,{\sum }_{i=1}^m{u}_i\right)$$

(5)

Step 6: Defuzzification

In fuzzy-logic systems, fuzzy results must be converted to crisp values for decision-making. In this study, we adopt the weighted centroid by Kwong & Bai (2003) [124] for TFNs, which emphasizes the modal value (Equation (6)).

$$Defuzz\left(\tilde{a}\right)={\displaystyle \frac{l+4m+u}{6}}$$

(6)

Step 7: Plotting the fuzzy PMC surface.

To visualize strengths and weaknesses, fuzzy PMC surface plots are generated from the index matrix, providing diagnostic information on dimensions that perform well and those that require improvement.

Illustrative Example

After the strategies, the corresponding fuzzy-PMC criteria, and the TFN linguistic scales were defined, the fuzzy scoring procedure was subsequently carried out in the following steps. For illustration, Strategy S1 and the Environmental Impact dimension (X1) are used as an example.

• The expert panel evaluated the sub-variables of X1 (Environmental Impact) for Strategy S1 using the linguistic scale (Very Low–Very High).

  Table 4. Expert Panel’s Linguistic Judgments for Strategy S1.

  Sub-Variable	Description	Expert Panel Judgment
  X1:1	Carbon reduction potential	High (H)
  X1:2	Scope coverage	Very High (VH)
  X1:3	Permanence of reduction	Medium (M)

  presents the raw linguistic judgments.

• These linguistic terms were then mapped to the corresponding TFNs defined in

  Table 5. TFN Representation of Expert Judgments for Strategy S1.

  Sub-Variable	Linguistic Term	TFN (l, m, u)
  X1:1	High (H)	(5, 7, 9)
  X1:2	Very High (VH)	(7, 9, 10)
  X1:3	Medium (M)	(3, 5, 7)

  .

• The TFNs for X1 were aggregated using the fuzzy arithmetic mean, as shown in Equation (5).

$$\widetilde{X_1}=\left({\displaystyle \frac{5+7+3}{3}},{\displaystyle \frac{7+9+5}{3}},{\displaystyle \frac{9+10+7}{3}}\right)=(5.00, 7.00, 8.67)$$

• The aggregated TFN was converted into a crisp score using the centroid method, as shown in Equation (6).

$$\mathrm{D}\mathrm{e}\mathrm{f}\mathrm{u}\mathrm{z}\mathrm{z}\widetilde{X_1}=\left({\displaystyle \frac{l+4m+u}{6}}\right)=\left({\displaystyle \frac{5.00+4 \left(7.00\right)+8.67}{6}}\right) = 6.96$$

The same process was applied to all main variables (X1–X9). The resulting defuzzified scores were summed to obtain the overall fuzzy-PMC score for Strategy S1. This step-by-step illustration confirms how linguistic judgments from the expert panel are systematically transformed into reproducible numerical values.

3. Results

3.1. Campus GHG Inventory

Emissions for all calculated categories, and their shares by category and of total emissions, are reported in

Table 6. GHG Emission Inventory.

Subcategory	GHG Inventory (tCO2e/yr)	% of Scope	% of the Total
1.1 Direct emissions from stationary combustion	730.02	82.67%	14.93%
1.2 Direct emissions from mobile combustion	31.62	3.58%	0.65%
1.4 Direct fugitive/leakage emission from GHG release in anthropogenic systems	121.36	13.74%	2.48%
2.1 Indirect emissions from imported electricity	1387.16	100.00%	28.38%
3.1 Indirect emissions from transportation and distribution of input materials	0.18	0.09%	0.00%
3.3 Indirect emissions from employee commuting	58.08	27.99%	1.19%
3.4 Indirect emissions from transportation of visitors and customers to the facility	123.25	59.39%	2.52%
3.5 Indirect emissions from business travel	26.01	12.53%	0.53%
4.1 Indirect emissions from purchased products	38.74	5.12%	0.79%
4.2 Indirect emissions from capital assets	407.78	53.84%	8.34%
4.3 Indirect emissions from the disposal of solid and liquid waste	5.25	0.69%	0.11%
4.4 Indirect emissions from the use of assets not owned by the entity	26.75	3.53%	0.55%
4.5 Indirect emissions from the use of other services	278.84	36.82%	5.70%
5.4 Indirect emissions from investments	1514.86	100.00%	30.99%
6 Indirect emissions from other sources	138.72	100.00%	2.84%

and Table S21. In 2023, AGU’s activities resulted in 4888.63 tCO₂e in total. Category 1 and Category 2 accounted for 18.06% and 28.38% of total emissions, respectively. Emissions arising from value-chain activities not owned or controlled by the organization were distributed across Category 3, Category 4, Category 5, and Category 6 at 4.24%, 15.49%, 30.99%, and 2.84%, respectively.

The largest contribution was from Category 5 (1514.86 tCO₂e), driven by Subcategory 5.4 (ongoing campus construction). Category 2 (indirect energy) was the second-largest source, totalling 1387.16 tCO₂e.

The third-largest source of emissions is Category 1, totalling 883 tCO₂e and accounting for 18.06% of overall emissions. Within this category, subcategory contributions are 82.67% for 1.1 stationary combustion, 3.58% for 1.2 mobile combustion, and 13.74% for 1.4 fugitive emissions (Figure 2). The activity-based distribution for Category 1 is shown in Figure 2. Heating—using natural gas—is the dominant activity, representing 81.89% of Category 1. Generator fuels constitute 0.79% of total emissions (6.98 tCO₂e) within the stationary combustion subcategory. Emissions from mobile sources represent 3.61% of total emissions (31.87 tCO₂e), with diesel used in on-road vehicles and gasoline in off-road equipment as the principal fuels. The second-largest source within Category 1 is fugitive emissions from fire suppression systems, contributing 8.81% (77.77 tCO₂e). These emissions, counted under subcategory 1.4, arise from the main fire suppression system and portable extinguishers using gases such as CO₂ and HFC-227ea. Additional fugitive emissions from cooling equipment leaks and laboratory gases account for 4.94% of Category 1 (43.60 tCO₂e). Overall, subcategory 1.4 represents 2.48% of total emissions (Table 6).

Indirect emissions from transportation and travel in Category 3 totalled 207.51 tCO₂e, accounting for 4.24% of total emissions. The contribution of Subcategory 3.1 (transportation of purchased goods) to Category 3 emissions was very low at 0.09%, reflecting the university’s purchases being limited to products that meet immediate needs and the absence of any goods production. Subcategory 3.3 (employee commuting) contributed 27.99% to Category 3 emissions. Subcategory 3.4 (student commuting) was the highest source within Category 3, at 59.39%. Business trips and hotel accommodations of employees (Subcategory 3.5) accounted for 12.53% (26.01 tCO₂e) of emissions in this category.

Total emissions in Category 4 were 757.37 tCO₂e, representing 15.49% of total emissions. The contribution of Subcategory 4.1 (purchased products) to Category 4 emissions was 5.12% (38.74 tCO₂e). Subcategory 4.2 (capital assets) had the highest share in this category at 53.84% (407.78 tCO₂e). Emissions from waste disposal (Subcategory 4.3) were the lowest in this category, at 0.69%. Emissions from service procurements (Subcategory 4.4) were also low, totalling 26.75 tCO₂e. Emissions from wastewater treatment calculations (Subcategory 4.5) represented 36.82% of Category 4 (278.84 tCO₂e). Category 6 includes only distribution-line leaks, totalling 138.72 tCO₂e and representing 2.84% of total emissions.

3.2. Mitigation Strategy Results

Implementing effective carbon emissions reduction strategies is crucial in combating climate change and achieving sustainability goals. The most critical steps in the comprehension and management of GHGs emissions are the conduct of GHGs inventories and the evaluation of the CFs of university campuses [65,66,125]. The strategies for reducing GHGs emissions at the AGU Sumer campus are determined in this step and are presented in Table 1.

3.3. PMC Results

In this section, the PMC index was computed for each carbon-mitigation strategy based on the main variable and its sub-variables, and the ordered results are reported in

Table 7. PMC index Results.

Category	Range	n (%)	Strategies
Good consistency	6–7.99	6 (30%)	S1 (7.67), S14 (6.50), S6 (6.42), S3 (6.33), S7 (6.25), S4 (6.25)
Acceptable consistency	4–5.99	12 (60%)	S11 (5.92), S5 (5.58), S12 (5.50), S13 (5.25), S8 (4.92), S10 (4.92), S20 (4.92), S2 (4.75), S19 (4.67), S15 (4.25), S9 (4.17), S18 (4.08)
Low consistency	0–3.99	2 (10%)	S17 (3.83), S16 (3.75)
Perfect consistency	8–9	0 (0%)	—

. To interpret policy consistency, we adopted literature-based evaluation criteria grounded in PMC index scores [43,56,126]; the corresponding threshold ranges are provided in

Table 8. Evaluation Criteria for Policy Consistency Based on PMC Index Scores.

PMC Index	0–3.99	4–5.99	6–7.99	8–9
Evaluation	Low consistency	Acceptable consistency	Good consistency	Perfect consistency

. Additionally, Table S22 PMC Index Matrix determined by the expert panel and the Table S23 PMC Index Values calculated are provided in the Supporting Document.

The PMC and Fuzzy-PMC evaluations were conducted by an expert panel consisting of five members: three from the university’s Sustainability Office and two sustainability professionals from industry. The Sustainability Office representatives included: (i) an academic staff member with 19 years of combined experience in academia and industry, (ii) a sustainability specialist with 5 years of experience focusing on the Sustainable Development Goals, and (iii) an energy expert with 20 years of field experience in energy management and efficiency. The industry representatives comprised: (iv) a carbon management professional with 16 years of expertise, and (v) a senior consultant with 19 years of experience in climate technologies. Panellists were deliberately selected to combine institutional knowledge of campus operations with applied expertise in emission reduction. Their evaluations were informed by their professional experience and domain-specific knowledge, ensuring that the scoring process reflected both practical feasibility and strategic relevance. This composition enhanced the credibility, applicability, and reproducibility of the fuzzy assessments.

Under the classical PMC scoring, the 20 statements have a mean of 5.30 with a standard deviation of 1.06. The highest score is S1 = 7.67, followed by S14 = 6.50, S6 = 6.42, S3 = 6.33, and a tie at S7 = 6.25 and S4 = 6.25. At the lower end are S16 = 3.75, S17 = 3.83, S18 = 4.08, S9 = 4.17, and S15 = 4.25.

By evaluation category, strategies distributed as follows: Good consistency (6–7.99) included six strategies (30%; S1, S14, S6, S3, S7, S4); Acceptable consistency (4–5.99) included twelve strategies (60%; S11, S5, S12, S13, S8, S10, S20, S2, S19, S15, S9, S18); Low consistency (0–3.99) included two strategies (10%; S17, S16); and no strategy attained Perfect consistency (8–9). This profile indicated that the portfolio was concentrated at the Acceptable level, with a smaller set reaching Good and two strategies remaining in the Low class, signaling clear opportunities for improvement.

The ties observed at the top (S7/S4 = 6.25) and in the middle–lower band (S8/S10/S20 = 4.92) suggest that the classical (0/1) PMC scheme may have limited discriminatory power among strategies that are similar in coverage but differ in intensity or quality. Moreover, the upper quartile (≥6.25) contains six strategies, whereas the lower quartile (≤4.56) contains five. In the mid-range, six strategies meet or exceed 6.0, four fall within 5.0–5.99, and ten are below 5.0 (Q1 = 4.56, Q3 = 6.25). This pattern reveals a small leadership cluster, a crowded middle band, and a short yet meaningful lower tail.

Overall, the classical PMC results delineate a leader group (S1, S14, S6, S3, S7/S4), a middle segment with notable ties (especially S8/S10/S20), and a lagging lower end (S16–S17). This structure is typical of a coverage-oriented, count-based approach: it recognizes breadth but does not always sharply differentiate among strategies with comparable coverage yet differing quality or intensity—an analytical limitation addressed in the next section via the fuzzy (linguistic/TFN-based) PMC.

In this study, each policy was coded across nine main variables (X1–X9) and arranged into a 3 × 3 PMC matrix to generate surface plots. The surface provides an intuitive visualization of the components underpinning the PMC index: convex peaks indicate strong areas, whereas concave troughs denote relative weaknesses. As the values of the main variables converge, the surfaces become more similar; greater dispersion produces more differentiated shapes. The surfaces for S1–S20 are presented in Figure 3.

PMC surfaces simultaneously visualize the contributions of the nine main variables (X1–X9), thereby revealing where policy components are concentrated and where they remain underdeveloped. In our dataset, the surfaces consistently indicate three recurrent strengths and three recurrent weaknesses.

• Strengths:

(i)

Implementation and Governance Capacity (X7) is pronounced: the persistent convex ridge in the lower-right mid-zone signals the widespread presence of clear implementation roadmaps (X7:1), organizational capacity in people/equipment/budget (X7:2), and monitoring–feedback mechanisms (X7:3).

(ii)

Technical and Operational Feasibility (X4) is strong; many policies articulate solution maturity (X4:1), reasonable implementation timelines (X4:2), and solid compatibility/scalability (X4:3), yielding operationally actionable designs.

(iii)

Risk and Resilience (X6) is comparatively robust: provisions addressing implementation risk (X6:1) and supply-chain dependency (X6:2) align with the elevated middle band observed in the surfaces.

• Weaknesses:

(i)

Innovation and Knowledge Transfer (X9) is the most fragile dimension: concavities in the lower-right corner reflect that degree of innovation (X9:1) and reproducibility (X9:2) are often insufficiently specified.

(ii)

Alignment with Long-term Policy and Market Trends (X8) is underdeveloped; compatibility with national/international climate targets (X8:1) and the potential to attract green finance (X8:2) are commonly stated at a programmatic level rather than operationalized with evidence.

(iii)

Economic Viability (X2) shows uneven evidentiary support: standard analyses of cost-effectiveness (X2:1), CAPEX/OPEX decomposition (X2:2–X2:3), and payback period (X2:4) are not reported systematically or in a comparable manner across policies.

3.4. Fuzzy PMC Results

In this section, we present the results of the fuzzy PMC evaluation, which extends the classical PMC framework by incorporating linguistic scales and expert-based judgments to better capture uncertainty in policy consistency assessment. Table S24 Fuzzy PMC Matrix determined by the expert panel and the Table S25 Fuzzy PMC Index values calculated are provided in the Supporting Document.

Extending the scoring with fuzzy membership (triangular linguistic scales) yields a higher central tendency and lower dispersion than the classical crisp scheme. Across the 20 strategies, the mean fuzzy PMC score is 5.74 (vs. 5.30 under the classical PMC) and the sample standard deviation is 0.75 (vs. 1.06), corresponding to an 8.4% increase in the mean and an approximately 29.0% reduction in dispersion. The highest values are S1 = 7.63, followed by S3 = 6.88, S6 = 6.65, S14 = 6.55, and S13 = 6.20, with a minor tie at S4 = 6.04 and S7 = 6.04. At the lower end, scores rise to S10 = 4.81 and S17 = 4.66; notably, all strategies remain at or above the “acceptable” band. By evaluation category, 7 strategies (35%) fell into Good consistency (6–7.99) and 13 strategies (65%) into Acceptable consistency (4–5.99); no strategy attained Perfect consistency (8–9) or Low consistency (0–3.99). The complete set of fuzzy PMC scores and category assignments is reported in

Table 9. Fuzzy PMC Index Results.

Category	Range	n (%)	Strategies
Good consistency	6–7.99	7 (35%)	S1 (7.63), S3 (6.88), S6 (6.65), S14 (6.55), S13 (6.20), S4 (6.04), S7 (6.04)
Acceptable consistency	4–5.99	13 (65%)	S2 (5.63), S5 (5.77), S8 (5.39), S9 (5.06), S10 (4.81), S11 (5.66), S12 (5.33), S15 (5.08), S16 (5.66), S17 (4.66), S18 (5.40), S19 (5.12), S20 (5.27)
Low consistency	0–3.99	0 (0%)	—
Perfect consistency	8–9	0 (0%)	—

.

Using fuzzy scores for the nine main variables (X1–X9), a 3 × 3 Fuzzy PMC matrix was constructed for each statement, and the corresponding surfaces were plotted for every strategy. For interpretation, convex peaks indicate stronger performance, whereas concave troughs denote relative weaknesses—analogous to the classical PMC visualization. The Fuzzy PMC surfaces for S1–S20 are shown in Figure 4.

• Strengths:

(i)

Technical and Operational Feasibility (X4) is consistently high across many statements; for example, S1, S4, S8, and S14 reach 8.83, and several others lie in the 6–7 band, indicating mature solutions (X4:1), feasible implementation timelines (X4:2), and solid compatibility/scalability (X4:3).

(ii)

Implementation and Governance Capacity (X7) forms another prominent ridge (typically 6.3–8.2; S1–S3 and S6 at 8.22), reflecting the presence of clear implementation roadmaps (X7:1), organizational capacity (X7:2), and monitoring/feedback mechanisms (X7:3).

(iii)

Stakeholder Engagement and Regulatory Compliance (X5) is generally strong (often ≥6.3, peaking at 7.61–8.22 for S11–S14), capturing involvement of diverse stakeholders (X5:1), alignment with existing regulations (X5:2), and consistency with industry standards (X5:3).

• Weaknesses:

(i)

Innovation and Knowledge Transfer (X9) remains the most fragile dimension: many statements cluster around 3.0–5.0 (e.g., S10–S15 at 3.0–4.0, S9 at 2.17), indicating limited specification of degree of innovation (X9:1) and reproducibility (X9:2); only a few cases are high (S1 = 7.92, S18 = 7.00).

(ii)

Alignment with Long-term Policy and Market Trends (X8) is underdeveloped, with values concentrated in the 4.2–6.9 range (e.g., 4.17 for S10–S13). This suggests that compatibility with national/international climate targets (X8:1) and the potential to attract green finance (X8:2) are not yet fully operationalized.

(iii)

Economic Viability (X2) exhibits marked heterogeneity—from very low values (1.33 for S8; 2.17 for S11) to high scores (7.92 for S5–S6 and 7.42 for S4/S14)—indicating uneven evidence on cost-effectiveness (X2:1), CAPEX/OPEX structure (X2:2–X2:3), and payback period (X2:4).

Overall, the Fuzzy PMC surfaces describe a portfolio that is operationally well prepared (X4, X7, X5) but would benefit from deeper innovation and reproducibility, stronger long-term alignment and green-finance linkages, and more consistent economic viability documentation.

The identified persistent weaknesses are not merely methodological artifacts but also reflections of the structural barriers faced by HEIs. Institutional culture is a critical determinant of sustainability adoption; without a shared vision and awareness among faculty, staff, and students, carbon reduction strategies remain fragmented. Luo et al. (2018) demonstrate how cultural environments shape organizational responses to climate action, highlighting why HEIs often struggle to embed environmental objectives into their strategic missions [127]. A second barrier is the limited focus on research and development (R&D) dedicated to sustainable technologies and practices. As Daddi et al. (2018) argue, the underutilization of organizational and management theories in climate studies hinders institutions from leveraging innovative frameworks, while Abeydeera et al. (2019) emphasize the lack of novel mitigation approaches—such as carbon capture—in developing contexts [128,129]. Furthermore, financial constraints remain one of the most pressing challenges [130]. Many universities, particularly in the public sector, operate under severe budgetary restrictions, limiting their ability to finance capital-intensive infrastructure upgrades or sustainability-oriented R&D. Restricted access to green finance and lengthy approval processes exacerbate these challenges [131].

Addressing these gaps requires a multi-faceted approach: fostering a sustainability-oriented institutional culture through capacity-building and governance reforms, such as integrating sustainability into mission statements and providing training for faculty and students [132]; strengthening collaborative R&D through the establishment of dedicated sustainability research centres and joint initiatives with industry and government bodies [133]; and leveraging diverse financing mechanisms such as green bonds, climate grants, and sustainability-focused corporate sponsorships [134].

Innovation is a key determinant of successful campus decarbonization. Without continuous research, development, and deployment of low-carbon solutions, universities risk relying on outdated or fragmented measures. Empirical studies show that institutions often underutilize innovative approaches such as carbon capture or disruptive low-carbon technologies, especially in developing contexts where resources are scarce [128,129]. To address this gap, HEIs should establish dedicated sustainability research centres, strengthen collaboration with industry and government, and create supportive environments that encourage experimentation with emerging technologies. Such efforts would enable universities not only to improve their carbon performance but also to contribute to broader knowledge transfer and societal innovation.

Green finance provides the essential foundation for transforming innovative ideas into tangible carbon-reduction outcomes. By mobilizing targeted investment, universities can bridge the financial gap between strategy design and implementation. Recent evidence highlights that structured green finance mechanisms, such as green bonds, climate grants, and low-carbon credit, significantly enhance the feasibility of energy efficiency and renewable energy projects in the education sector [131,135]. Furthermore, the interaction between green finance and innovation is critical: financial instruments are most effective when paired with institutional capacity to adopt and scale new technologies [136,137]. Integrating these mechanisms into governance structures, aligned with the SDGs, can ensure that financing not only supports environmental targets but also engages campus stakeholders in building a culture of sustainability [138,139,140].

These actionable pathways underscore that effective decarbonization in HEIs depends not only on prioritizing strategies but also on enabling institutional capacity, financial resources, and supportive policy frameworks.

4. Discussion

4.1. Campus GHG Inventory

Leading HEIs are increasingly quantifying their CFs and implementing climate measures to reduce emissions.

Table 10. GHG Emissions Studies of Different HEIs.

Reference	University	Country	Year	Annual GHG Inventory (tCO2e)	CF per Capita (tCO2e/Person)	Action Plan	Standard
This Study	Abdullah Gul University	Türkiye	2023	4888.63	1.19	✓	ISO 14064-1:2018/IPCC
[141]	Çankırı Karatekin University	Türkiye	2019	5633.13	4.54	✗	IPCC
[142]	Muğla Sıtkı Koçman University	Türkiye	2022	10,093.96 (IPCC)/7652.29 (DEFRA)	N/A	✗	IPCC/DEFRA
[143]	Mehmet Akif Ersoy University	Türkiye	2017	217.5	0.57	✗	DEFRA
[144]	Sakarya University	Türkiye	2015	12,330.73	0.15	✗	IPCC, WRI/WBCSD
[145]	Harran University	Türkiye	2021	11,840.00	0.44	✗	IPCC
[146]	Erzincan Binali Yıldırım University	Türkiye	2020	2383.70 (IPCC)/1826.50 (DEFRA)	0.09	✗	IPCC/DEFRA
[147]	University of Cape Town	Africa	2007	84,925.50	4.01	✗	IPCC
[148]	University of Sydney	Australia	2008	20,100.00	N/A	✓	N/A
[149]	University of Leuven	Belgium	2010	7085.00	0.93	✗	N/A
[102]	Talca University	Chile	2012	1568.60	1	✓	GHGP
[12]	Talca University	Chile	2012–2016	5472.89	0.72	✓	GHGP
[150]	Tongji University	China	2009–2010	NA	3.84	✓	N/A
[23]	Universidad Nacional de Colombia (UNAL)	Colombia	2019	7250.52	0.43	✗	ISO 14064-1:2018/GHGP
[11]	Technical University of Pereira	Colombia	2017–2018	8969.00	0.4	✓	ISO 14064-1:2018/GHGP
[151]	Escuela Superior Politécnica del Litoral	Ecuador	2017	5009.22	0.356	✓	ISO 14064-1:2018/GHGP
[152]	Birla Inst of Technology & Science Pilani	India	2014/2015	16,500.00	3.70	✓	ISO 14064-1:2018/IPCC
[153]	University of Diponegoro	Indonesia	2018	16,345.83	N/A	✓	IPCC
[154]	University of Diponegoro	Indonesia	2019	13,945.55	5.55	✗	PAS 2050 [155]
[17]	Universitas Pertamina	Indonesia	2018–2019	1351.98	0.52	✓	N/A
[156]	Trisakti University	Indonesia	2018	11,994.86	N/A	✓	N/A
[157]	University Technology Malaysia	Malaysia	2011	57,576.00	2.1	✗	GHGP
[158]	University Technology Malaysia	Malaysia	2016	9.30	N/A	✓	GHGP
[159]	Autonomous Baja California University	Mexico	2013	706.52	4.74	–	N/A
[13]	National Autonomous University	Mexico	2010	1577.00	1.48	✓	GHGP
[14]	University Autonomous Metropolitan	Mexico	2016	2956.28	1.07	✓	GHGP
[18]	Massey University	New Zeland	2004	26,696.00	N/A	✓	IPCC
[160]	Fed University of Agriculture Abeokuta	Nigeria	2011–2012	5935.00	N/A	✓	GHGP
[161]	Norwegian University of Technology & Science	Norway	2009	92,000.00	3.61	✓	GHGP/EEIO
[162,163]	University of Haripur	Pakistan	2016–2017	578.90	0.14	✓	IPCC
[15]	Qassim University	Saudi Arabia	–	123,997.47	N/A	✓	GHGP
[164]	University of Maribor	Slovenia	–	974.00	N/A	N/A	N/A
[19]	Pusan National University	South Korea	2007–2011	33,629.83	0.99	✓	IPCC
[165]	Polytechnic University of Madrid	Spain	2010	2147.00	1.55	–	GHGP
[166]	University of Castilla-La Mancha	Spain	2005–2013/2013	23,000.00	2.13	✓	GHGP
[167]	University of the Basque Country	Spain	11/12—15/16	597.15	0.558	✓	ISO 14064-1:2018
[168]	Valaya Alongkorn Rajabhat University	Thailand	2016/2017	663.60	0.064	N/A	N/A
[169]	Mea Fah Luang University	Thailand	2011–2014	7330.72	0.52	✓	IPCC
[20]	The American University of Sharjah	UAE	2018–2020	94,553.30	15.7	✓	GHGP
[170]	Leeds University	UK	2010/2011	161,819.00	2.36	✗	GHGP-EEIO
[21]	De Montfort University	UK	2005–2009	51,080.00	2.00	✓	GHGP-HLCA
[171]	Bournemouth University	UK	2019	2139.60	1.41	✗	GHGP
[16]	St. Edward’s University	USA	2008–2013	18,541.70	3.7	✓	GHGP
[172]	Clemson University	USA	2014	95,418.00	3.57	✓	GHGP
[173]	University of Illinois	USA	2004–2008	275,000.00	10.9	✓	GHGP
[174]	Louisiana State University	USA	2007–2008	162,742.00	N/A	✓	GHGP
[175]	Yale University	USA	2003–2008	817,000	N/A	✗	GHGP/BSI
[65]	Rowan University	USA	2007	38,000.00	4.0	✗	N/A

N/A—not applicable.

summarizes published CF assessments. However, few studies comprehensively account for all attributable emissions—including both direct and indirect sources—across Scopes 1–3. While calculations for Scopes 1 and 2 are commonly reported, Scope 3 is frequently omitted or limited to a narrow set of sources. Most studies apply established frameworks such as the GHG Protocol, IPCC guidelines, and ISO 14064 Standards, although some adopt bespoke methodologies. Notably, many lack explicit, actionable emission-reduction roadmaps, despite CF accounting’s central purpose of informing climate policy and target setting.

The results of AGU were contextualized through comparison with both Turkish and global HEIs that have reported campus-level GHG inventories (Table 10). AGU’s 2023 emissions totalled 4888.63 tCO₂e, corresponding to 1.19 tCO₂e per capita, verified under ISO 14064-3:2019 GHG standard. This per capita figure was substantially lower than the reported global mean of 2.65 tCO₂e per capita and well below the global institution-level average of 54,638.7 tCO₂e, highlighting AGU’s relatively small institutional footprint. Within Türkiye, AGU’s per capita emissions are higher than those of Sakarya University (0.15 tCO₂e/person) and Erzincan Binali Yıldırım University (0.09 tCO₂e/person), but considerably lower than Çankırı Karatekin University (4.54 tCO₂e/person). Compared with Mehmet Akif Ersoy University (0.57 tCO₂e/person) and Harran University (0.44 tCO₂e/person), AGU occupied an intermediate position.

These variations partly reflect differences in campus size, activity data coverage, system boundaries, and emission factors, as well as the extent to which Scope 3 categories (e.g., commuting, investments) are included. Importantly, AGU’s application of ISO 14064-1:2018 offers one of the most comprehensive inventories among Turkish HEIs, providing decision-grade transparency across all scopes. Nonetheless, the heterogeneity in methodological choices complicates direct comparability, underscoring the need for harmonized standards for GHG reporting in HEIs both nationally and internationally.

4.2. PMC and Fuzzy PMC Results

Classical PMC and its fuzzy extension yield broadly consistent leadership signals while differing in their ability to discriminate among mid-ranked strategies and to reflect partial evidence. Under classical PMC, the portfolio centres in the “acceptable” band (mean = 5.30; SD = 1.06), with six strategies in “good” consistency and two in “low.” The fuzzy formulation—based on triangular linguistic terms and centroid defuzzification—shifts the distribution upward and tightens it (mean = 5.74; SD = 0.75), i.e., an 8.4% increase in central tendency and 29% reduction in dispersion. The “good” set expands to seven and the “low” class disappears, indicating that graded membership captures strengths that binary scoring under-recognizes (Table 6 and Table 7).

Discrimination and tie-breaking improve materially with fuzzy scoring. Classical PMC exhibits tie at both the upper and mid–lower bands (e.g., S7/S4 = 6.25; S8/S10/S20 = 4.92), showing limited ability to distinguish strategies when coverage is similar, but intensity differs. The fuzzy scheme resolves these ties and slightly reorders the upper middle: besides preserving S1 at the top (7.63), it elevates strategies such as S13 into the “good” tier (6.20). At the lower end, strategies such as S10 (4.81) and S17 (4.66) exceed the acceptability threshold, reflecting that partial compliance and credible implementation pathways deserve non-zero credit instead of a strict 0/1 evaluation.

Surface diagnostics are directionally consistent across methods but become more interpretable with fuzzification. Both approaches reveal recurrent strengths in Technical & Operational Feasibility (X4) and Implementation & Governance Capacity (X7), and persistent weaknesses in Innovation/Knowledge Transfer (X9) and long-term alignment/green-finance linkages (X8). Fuzzy aggregation accentuates these patterns by damping binary edge effects, yielding smoother, more contrasted ridges and troughs that better correspond to evaluator confidence and evidence strength.

From a decision-support perspective, fuzzy PMC offers three practical advantages. First, it reduces false negatives among “soft” yet impactful strategies (e.g., behavioural/organizational measures), which classical PMC may undervalue when evidence is qualitative or evolving. Second, it expands the set of ready-to-implement options without inflating top ranks, thereby supporting robust near-term action portfolios under resource constraints. Third, lower dispersion suggests greater stability to small scoring perturbations (e.g., when additional documentation marginally shifts a linguistic rating).

Two caveats remain. Outcomes still depend on (i) the quality and calibration of linguistic scales/membership functions and (ii) the equal-weighting assumption inherited from PMC; both should be stress-tested where stakes are high (e.g., via alternative defuzzification rules, expert re-elicitation, or complementary MCDA with cost/effectiveness lenses). Even so, the empirical pattern here—higher central tendency, reduced spread, resolved ties, and consistent surface morphology—supports the conclusion that the fuzzy extension retains PMC’s transparency while better encoding uncertainty and partial compliance, thus providing a threshold-robust basis for prioritizing campus decarbonization strategies.

As an extension of the fuzzy PMC prioritization, we developed a roadmap (Table S26 Roadmap) that specifies indicative short-, medium-, and long-term timelines, co-benefits, and measurable Key Performance Indicators (KPIs) for each strategy. This roadmap translates the prioritization results into actionable guidance, balancing cost considerations with broader sustainability and feasibility factors. The full roadmap is provided in the Supporting Information due to space constraints.

4.3. Sensitivity Analysis

In this study, a two-level sensitivity analysis was conducted to evaluate the robustness of the proposed Fuzzy-PMC results. First, sensitivity to the defuzzification method was examined by comparing the conventional approach with the Mean of Maxima method [176]. The results are presented in Figure 5.

The findings indicate that, although minor numerical differences exist between the two approaches, the overall ranking of strategies remained largely consistent. High-priority strategies such as S1, S3, S6, S13, and S14 consistently achieved the top scores under both methods, thereby confirming their robustness as leading strategies. Similarly, low-priority strategies such as S17, S10, and S9 remained at the bottom of the ranking in both defuzzification techniques.

Notably, S16 exhibited a significant downward shift, moving from the 10th rank under the Centroid method to the 19th rank under the Mean of Maxima approach, suggesting a sensitivity of this strategy’s prioritization to the choice of defuzzification technique. In contrast, strategies such as S2, S8, S12, S15, S19, and S20 demonstrated only marginal rank adjustments, which did not alter their classification within the mid-priority group.

The Mean of Maxima method produced slightly lower values for the highest-ranked strategies, reflecting a tendency toward more conservative outputs by reducing the influence of extreme values. These findings demonstrate that while the choice of defuzzification method can introduce shifts in individual rankings—most notably in the case of S16—the proposed Fuzzy-PMC model yields generally stable and robust prioritization results, underscoring its suitability as a decision-support tool for campus decarbonization planning.

In the second step, the calculations were repeated using the TFN sets proposed by Lamata (2004) in the literature [177]. The TFNs employed are presented in

Table 11. Linguistic Terms and Corresponding TFNs [177].

Linguistic Term	Abbrev.	TFN
Very Low	VL	(1, 1, 1)
Low	L	(2, 3, 4)
Medium	M	(4, 5, 6)
High	H	(6, 7, 8)
Very High	VH	(9, 9, 9)

, and the resulting outcomes are summarized in Figure 6.

The comparison between the two approaches revealed only marginal numerical deviations across most strategies. For instance, Strategy S1 showed a slight increase from 7.63 to 7.72, while S3, S6, and S14 exhibited modest improvements of approximately 0.04–0.06. Similarly, incremental gains were observed for S7, S11, S12, S13, S18, S19, and S20. Conversely, small decreases were recorded for S8, S9, and S10, with S16 experiencing the most pronounced decline, dropping from 5.66 to 5.06 and shifting from 10th to 17th place in the ranking.

Despite these quantitative fluctuations, the overall ranking of strategies remained largely stable, with S1 consistently achieving the highest score under both TFN configurations, thereby confirming its robustness and dominance within the prioritization framework. Moreover, strategies such as S17 and S10 consistently occupied the lowest ranks, reinforcing their low-priority status regardless of configuration.

These findings indicate that while the choice of fuzzy number configuration can produce minor numerical and positional variations—particularly in mid-ranked strategies such as S16 and S20—the proposed Fuzzy-PMC model delivers stable and reliable prioritization results, underscoring its credibility and robustness as a decision-support tool for campus decarbonization planning.

These findings indicate that the applied methodology delivers stable and reliable rankings even under alternative fuzzy number configurations, thereby reinforcing the credibility and robustness of the analysis.

4.4. Limitations

This study has several limitations that should be acknowledged. First limitation relates to the economic dimension of the analysis. While comprehensive quantitative estimates for CAPEX, OPEX, and payback periods were not available for all strategies, this gap was addressed through structured expert evaluations within the fuzzy PMC framework. The reliance on expert judgment ensured that the economic dimension could still be meaningfully incorporated, providing a more nuanced perspective than the binary assessments of the classical PMC method. Nonetheless, the absence of detailed cost data remains a constraint, and future studies should aim to integrate more precise economic metrics.

Second, both PMC and fuzzy PMC rely on the equal-weighting assumption across main variables. This choice ensured methodological transparency and comparability with prior studies, as well as alignment with the intrinsic logic of the classical PMC framework. However, it does not reflect potential differences in the relative importance of dimensions such as economic viability or technical feasibility.

Third, the fuzzy PMC results were not externally validated against real-world implementation outcomes. The prioritization relied exclusively on the judgments of an expert panel, ensuring methodological robustness but not providing empirical confirmation of actual performance. This constraint should be considered when interpreting the findings, as the absence of external validation limits the extent to which the rankings can be generalized beyond the present case.

4.5. Future Work

Building on these limitations, several directions for future research are recommended.

• Future studies should incorporate more precise economic data—such as CAPEX, OPEX, and payback periods—into the prioritization framework. This would strengthen the quantitative foundation of the analysis and reduce reliance on expert judgment.
• Future studies should also explore the application of unequal weighting schemes in fuzzy PMC to reflect the varying importance of different variables. Such weights could be derived from expert elicitation or empirical data, enabling more context-sensitive prioritization. In addition, further research could examine potential overlaps between strategies by applying clustering or grouping techniques, which would help to systematically capture complementarities while minimizing redundancy.
• Future research should incorporate external validation of fuzzy PMC results by linking them with empirical data from implemented strategies, such as observed reductions in energy consumption or GHGs. Additionally, cross-institutional studies could compare fuzzy PMC-based rankings with real-world outcomes in multi-campus universities or other sectors. Such validation would strengthen the reliability of the method and further demonstrate its applicability beyond expert-based evaluations.
• Future research could extend the fuzzy PMC framework through integration with other MCDA methods to enhance robustness and comparability. For instance, (AHP could be employed to derive context-specific weights for the main variables, while methods such as TOPSIS could be used to cross-validate the prioritization outcomes. Exploring such hybrid approaches would not only address weighting concerns but also strengthen the applicability of fuzzy PMC across diverse institutional and sectoral contexts.
• Longitudinal inventories would allow universities to capture changes in emissions patterns over time and assess the long-term impacts of mitigation strategies. Also, AI-enhanced fuzzy models offer a promising avenue to refine the prioritization of strategies, as they can incorporate larger datasets, handle dynamic weighting schemes, and improve the precision of decision-making under uncertainty. Together, these directions can strengthen the robustness, transparency, and policy relevance of campus decarbonization assessments.
• Finally, broader replication is needed. Comparative studies across universities, countries, and sectors would consolidate the external validity of the fuzzy PMC framework. The method is not limited to a single campus; it can also be applied to multi-campus universities and non-HEI sectors, provided that the strategy set is reviewed and redefined according to sector-specific emission sources. This flexibility underscores the framework’s transferability while ensuring that prioritization remains tailored to contextual emission drivers.

5. Conclusions

This study delivers an end-to-end GHG accounting and decision-support framework for a public university and applies it to AGU. The campus-wide inventory totals 4888.63 tCO₂e, with emissions dominated by Scope-3 investments (Category 5.4: 1514.86 tCO₂e; 30.99%) and grid electricity (Category 2.1: 1387.16 tCO₂e; 28.38%). Scope-1 sources—primarily stationary combustion and fugitive refrigerants—jointly account for ~18.06%, while purchased goods, capital assets and services (Category 4) contribute 15.49%. Mobility-related Scope-3 sources (employee/student commuting and business travel) add 4.25%, and other minor sources represent 2.84%. These patterns indicate that carbon management at HEIs must simultaneously address financed/embodied emissions from investments and capital projects, electricity-related emissions, and on-site thermal and refrigerant losses, with targeted actions on mobility where feasible.

Using a structured portfolio of 20 carbon-mitigation strategies mapped to inventory categories, we compare classical PMC with its fuzzy extension. Both methods produce consistent leadership signals, yet Fuzzy PMC better captures partial evidence and evaluator uncertainty, increasing the portfolio’s mean consistency from 5.30 to 5.74 and reducing dispersion (SD 1.06 → 0.75). Tie-breaking and mid-rank discrimination improve, elevating strategies whose effects are diffuse but material. Priorities converge on: energy-management system upgrades (S3); modernization and electrification of heating/cooling (S4–S5); electricity-side measures—monitoring/analytics (S6), on-site renewables (S7) and green power procurement (S8); virtualization of travel (S14) and remote work/education (S13); and, critically, green procurement and investment policies (S17) that target the largest source (Category 5.4) and major Category 4 drivers (capital assets and services).

For governance, the results imply a phased but integrated decarbonization program: (i) procurement & investment decarbonization (EPD-backed materials, low-carbon construction, financed-emission screens) to tackle Categories 5.4 and 4.x; (ii) power-sector measures (on-site PV, PPAs/green tariffs) coupled with demand-side efficiency; (iii) thermal system retrofits and refrigerant management to curb Scope-1; and (iv) mobility management (public transport incentives, active mobility, travel substitution). Methodologically, Fuzzy PMC preserves PMC’s transparency while yielding more threshold-robust rankings for carbon decisions under uncertainty. As an early application in the decarbonization context, further work should include sensitivity analyses for membership functions/defuzzification, alternative weighting schemes, and cross-case validation against ex-post carbon performance to consolidate external validity.

The classical PMC method is a practical and widely used tool for policy evaluation; however, it suffers from limitations such as tied results and high score dispersion, which make it difficult for decision-makers to obtain a clear and reliable ranking of strategies. The fuzzy PMC approach proposed in this study addresses these limitations by reducing uncertainty and producing more consistent outcomes. While PMC applications have been employed in various policy analyses in the literature, a fuzzy extension that tackles issues of uncertainty and ranking ties in the prioritization of sustainability strategies has not previously been introduced. In this respect, the study advances the theoretical framework of PMC, provides a methodological innovation, and offers a more reliable tool for supporting decision-making processes.

In conclusion, while the proposed strategies provide a robust prioritization framework for campus decarbonization, their practical implementation remains constrained by structural barriers, including institutional culture, limited R&D capacity, and financial restrictions. Overcoming these challenges will require the integration of sustainability into institutional missions, the establishment of collaborative research initiatives, and the mobilization of diverse green financing mechanisms. These considerations emphasize that successful decarbonization in higher education is contingent not only on strategy prioritization but also on strengthening institutional capacity and support structures.