Picture-Fuzzy Decision-Making Tool for Enhanced Risk Prioritization in Construction and Demolition Waste Management: A Hybrid FMEA–Fine–Kinney–SWARA–TOPSIS Approach

Abstract

Effectively managing Construction and Demolition Waste (CDW) requires prioritizing multi-dimensional risks, a task complicated by the inherent uncertainty and subjectivity of expert judgments. While classical methods like Failure Modes and Effects Analysis (FMEA) and Fine–Kinney (FK) provide a diagnostic structure, they struggle to capture the vagueness in subjective assessments. This study addresses this gap by developing an integrated framework that couples the classical FMEA/FK criteria (Severity, Exposure, Probability, Detectability, Frequency) with Picture-Fuzzy (PiF) multi-criteria decision making. The methodology first elicits criterion importances from 15 experts using PiF Stepwise Weight Assessment Ratio Analysis (PiF-SWARA), which retains approval, indeterminacy, rejection, and refusal degrees to reduce information loss. Subsequently, it ranks 40 risk factors using the PiF Technique for Order Preference by Similarity to Ideal Solution (PiF-TOPSIS). Results show severity is the most influential criterion, followed by exposure and probability. The framework identifies the highest-priority risks as cumulative pollution with rising complaints, groundwater leakage, and insufficient investment/operating budgets. A sensitivity analysis confirms that environmental and financial risks remain consistently prominent across various weighting scenarios. This harmonized FMEA/FK–PiF-SWARA–TOPSIS approach yields a transparent and defensible prioritization, offering a practical tool for managers to allocate resources effectively, focusing on critical environmental controls and addressing core financial deficiencies in CDW systems.

1. Introduction

The construction sector is known for its significant contribution to the development of society. However, it is argued that, alongside its essential contributions, it has negative effects on the environment [1,2]. Examples of these negative effects include increased carbon emissions and greenhouse gases [3,4], high energy and natural resource consumption [5], and intensive water consumption [6]. There are also examples of socio-economic and environmental impacts [4,7]. Therefore, effectively managing construction waste is critical for environmental sustainability, economic efficiency, and social welfare.

In recent years, studies on construction waste management have focused more on waste reduction, reuse, and recycling strategies [8]. According to the European Environment Agency [9], construction and demolition activities account for nearly 35% of total waste generated in the European Union. Despite the growing attention to circular economy practices, the implementation of these strategies is still hindered by numerous risk factors that emerge during waste collection, transportation, storage, and recovery processes. These risks can lead to multidimensional consequences, including workplace accidents, environmental contamination through leachate and dust emissions, cost overruns, and legal non-compliance issues [10]. Therefore, accurately identifying and prioritizing these risks is a critical prerequisite for developing sustainable and safe waste management strategies.

Various studies have applied quantitative risk assessment methods to C&DW processes. Traditional approaches such as FMEA and FK have been employed to evaluate occupational and environmental hazards in the construction industry [11]. However, these conventional techniques often struggle to capture the uncertainty, hesitation, and interdependence among expert judgments. For example, Vargas at al. [10] analyzed the logistics risks in construction waste supply chains using deterministic scoring, while Liu & Tang [12] applied FMEA in a green construction framework without accounting for linguistic vagueness in expert opinions. Consequently, this study integrates FMEA and FK frameworks with PiF MCDM techniques to provide a more flexible and diagnostically comprehensive assessment. The five key criteria, namely probability, severity, detectability, frequency, and exposure, were selected to align with the diagnostic structure of FMEA and FK methodologies.

The motivation behind this study stems from the lack of a holistic, uncertainty-sensitive, and expert-driven framework for risk prioritization in construction waste management. Although earlier studies have examined isolated risk factors, they have not sufficiently integrated diagnostic frequency, exposure assessment, and fuzzy expert reasoning within a unified model. Therefore, the proposed hybrid framework aims to model uncertainty more realistically and enhance strategic decision-making processes that contribute to safer and more sustainable construction waste management. A detailed literature review was first conducted as part of the study, and potential risks related to the subject were identified. The risks addressed were evaluated under criteria derived from traditional risk assessment methods. Within the study’s proposed methodology, criterion weights were calculated using PiF-SWARA, while the assessment of risks was carried out using PiF-TOPSIS. Thus, the aim is to reflect the uncertainties in expert opinions more realistically in the model and to contribute to strategic decision-making processes related to construction waste management.

2. Literature Review

In this study, we conducted a brief systematic review to identify, consolidate, and thematically group risk factors relevant to CDW management. Searches were run in Sco-pus, Web of Science, etc., using Boolean strings combining CDW terms with “risk,” “hazard,” “failure mode,” and “MCDM,” and supplemented by forward–backward citation chaining. Table 1 summarizes the main risk categories identified in the systematic review, along with their concise definitions and supporting references, providing the conceptual foundation for the subsequent PiF–MCDM risk assessment.

Studies indicate that environmental risks occur at multiple scales throughout construction and urban waste management. One study by Singh [13] suggests that the transport of leachate into groundwater, depending on site conditions, is the most critical threat in the construction waste management process, while another example by Negash [14] states that inadequate infrastructure and regulatory gaps trigger serious pollution risks in air, soil, and water resources through illegal dumping and unplanned storage. When these two approaches are considered together, it becomes clear that environmental risks associated with construction waste are fueled by technical factors and institutional and managerial deficiencies. Therefore, site-specific hydrogeological measures and an effective policy and control mechanism must be developed simultaneously for sustainable waste management.

Beyond environmental factors, human-centered risks also emerge prominently. The reviewed articles highlight the decisive role of the human factor at the center of environmental risks. In the case study focused on India, it is shown that uncontrolled waste sites not only cause hydrogeological pollution but also threaten the health and socioeconomic well-being of the communities living in their vicinity [13]. Another study focused on Egypt reveals that the biggest obstacle to sustainable waste management is not technical inadequacies but rather a lack of awareness among workers, weak oversight, and cultural habits [15]. When these two findings are considered together, it is clear that human-induced risks deepen environmental impacts directly and indirectly. Therefore, strengthening education, awareness, and corporate responsibility mechanisms is as essential as technical measures for sustainable construction waste management. Technical deficiencies further compound these challenges also frequently encountered. Furthermore, technical risks pose critical obstacles at the field and sector levels. In one example from the literature, Singh [13], infrastructure deficiencies, limitations of measurement tools, and soil heterogeneity make it difficult to accurately predict pollutant transport; in the case of Egypt, inefficient processes in housing projects, poor material management, and lack of recycling infrastructure hinder the technical control of waste [15]. In an example from India, the non-application of digital technologies, low recycling capacity, and inadequate quality standards are prominent technical risks [16]. These findings show that sustainable construction waste management requires regulatory and cultural reforms, measurement techniques, increased infrastructure investment, and the integration of advanced technologies.

Economic constraints add another layer of vulnerability, have also been identified as barriers in the literature review. On the one hand, the high investment costs, insufficient incentives, and uncertainties in the secondary material market have made circular economy practices globally unattractive from an economic perspective [15]. Furthermore, factors such as a lack of financing in developing countries, the use of low-quality materials, and customer demand indifference to sustainability deepen economic risks [17]. These findings show that environmental benefits in construction waste management can only be realized with strong economic support mechanisms, stable market structures, and long-term financing strategies.

Finally, managerial shortcomings often underpin all these risk domains. Yi et al. [18] proposed an integrated decision model that combines life-cycle assessment and life-cycle costing with AHP–TOPSIS to rank CDW options; recycling for concrete + road construction is best, landfill is worst, with transport optimization recommended. Eghbali-Zarch et al. [19] uses a fuzzy IDOCRIW–WASPAS framework to prioritize strategic CDW management actions under sustainability criteria; implementing robust operational plans/research ranks highest, followed by stricter laws and regulations, and results are validated against several MCDM methods. Kabirifar et al. [20] identifies CE-oriented factors across six life-cycle stages and prioritizes them via an Enhanced Fuzzy Delphi + Cybernetic–Parsimonious Fuzzy AHP; top levers are on-site sorting/reuse/recycling, procurement model choice, and precise enforcement of waste plans/regulations in Tehran case. Weerakoon et al. [21] builds an AHP model to embed CE principles for Sri Lanka; weights favor environmental and economic criteria, and on-site segregation is the most suitable strategy among the alternatives assessed. Chen (Yiwei) et al. [22] develops a hesitant neutrosophic set + GIS MCDM approach with revised DEMATEL-ANP for siting CDW treatment plants; the method handle uncertainty and spatial factors for facilities, shown effective in a Xiamen case study, with a noted limitation on accounting for existing facility coverage.

Table 1. Main criteria of the study.

Main Criteria	Definition	Ref.
Environmental Risks	Risks arising from improper waste handling, storage, or disposal that led to soil, air, or water pollution and ecological degradation.	[13,14]
Social Risks	Risks linked to public opposition, lack of awareness, or adverse community impacts such as reduced quality of life or inequitable burden sharing.	[13,15]
Facility & Operational Risks	Risks related to inadequate infrastructure, poor equipment maintenance, inefficient processes, or technological failures during CDW operations.	[15,16]
Health & Safety Risks	Risks endangering workers’ or residents’ physical well-being due to unsafe practices, exposure to hazardous materials, or poor site controls.	[13,15]
Quality Risks	Risks associated with low-quality or contaminated recycled materials, poor segregation, and weak quality assurance procedures affecting reuse potential.	[16,17]
Financial Risks	Risks stemming from insufficient funding, high investment costs, unstable secondary markets, or inadequate economic incentives.	[15,17]
Management & Regulatory Risks	Risks resulting from weak oversight, lack of coordination, unclear standards, or ineffective enforcement of waste management policies	[14,17,23]

Table 1. Main criteria of the study.

Main Criteria	Definition	Ref.
Environmental Risks	Risks arising from improper waste handling, storage, or disposal that led to soil, air, or water pollution and ecological degradation.	[13,14]
Social Risks	Risks linked to public opposition, lack of awareness, or adverse community impacts such as reduced quality of life or inequitable burden sharing.	[13,15]
Facility & Operational Risks	Risks related to inadequate infrastructure, poor equipment maintenance, inefficient processes, or technological failures during CDW operations.	[15,16]
Health & Safety Risks	Risks endangering workers’ or residents’ physical well-being due to unsafe practices, exposure to hazardous materials, or poor site controls.	[13,15]
Quality Risks	Risks associated with low-quality or contaminated recycled materials, poor segregation, and weak quality assurance procedures affecting reuse potential.	[16,17]
Financial Risks	Risks stemming from insufficient funding, high investment costs, unstable secondary markets, or inadequate economic incentives.	[15,17]
Management & Regulatory Risks	Risks resulting from weak oversight, lack of coordination, unclear standards, or ineffective enforcement of waste management policies	[14,17,23]

As a result of the literature review, it was concluded that managerial risks play a critical role in the sustainable management of construction waste. Whether it is the lack of managerial support, oversight deficiencies, and lack of strategic vision, as seen in the examples of Egypt and Somaliland, or the problems of institutional coordination, lack of standards, and adherence to linear models highlighted in the studies of India and the United Kingdom, it is understood that in all cases, managerial factors trigger environmental, economic, and technical risks [14,15,16,17,23]. In particular, the failure of top management and decision-makers to integrate a sustainability perspective into project planning, tendering, and design phases constitutes a common denominator for these risks. Therefore, sustainable construction waste management requires technical solutions, institutional capacity building, robust oversight mechanisms, and strategic leadership.

This study advances PiF–MCDM research in three ways. First, it develops a city-specific risk ontology for CDW management that spans collection, transport, storage, and recovery, structured around a five-criterion risk model. Second, it conducts a comprehensive robustness analysis, including leave-one-out expert tests and a PiF–VIKOR method swap, to verify rank stability. Finally, it produces actionable processes, criterion heatmaps and control priorities that provide practical guidance for municipal, contractor, and facility stakeholders.

3. Materials and Methods

This study aims to provide a systematic and transparent approach to recognizing and predicting safety, environmental, and operational risks in construction waste management. The method couples the classical scoring of risks with PiF to accommodate the uncertainty embedded in expert judgments and produce an implementable priority list. The risks were identified through an extant literature review on the subject. This study uses the combined application of FMEA and the FK method, which can be considered the methodological contribution of the research, as it merges probability, severity, and detectability with exposure and consequence into a coherent set of criteria. A flowchart of the proposed methodology is shown in Figure 1.

The approach systematically integrates classical diagnostic tools with advanced uncertainty-modeling techniques. We begin by leveraging the established structures FMEA and the FK method to identify a comprehensive set of risk criteria. However, these traditional methods often fail to capture the significant linguistic vagueness and hesitation inherent in subjective expert evaluations. To overcome this limitation, we employ Picture-Fuzzy sets, which allow experts to express their judgments with greater nuance—including degrees of approval, indeterminacy, rejection, and refusal—thus reducing information loss. This fuzzy environment is then operationalized using a two-stage MCDM process: PiF-SWARA is used to elicit and calculate the weights of the risk criteria from experts, and PiF-TOPSIS is applied to rank the final risk factors. This hybrid FMEA/FK–PiF–SWARA–TOPSIS model aims to produce a prioritization that is not only comprehensive and sensitive to uncertainty but also defensible and practically implementable.

We first compiled a long list of candidate risk factors from the literature (database search and screening) and consolidated overlapping items, yielding 40 risks grouped under environmental, human, socio-economic, technical, economic, and managerial themes. To validate wording and ensure contextual relevance, we conducted an expert-elicitation case study with 15 domain experts (construction management, waste management, OHS, and municipal operations; minimum 5 years of experience). Experts did not provide personal information; they only evaluated each risk using a pre-defined linguistic scale mapped to picture-fuzzy numbers for the criteria Probability, Severity, Detectability, Frequency, and Exposure. Individual judgments were aggregated by PFWGO, criteria weights were obtained via PiF-SWARA, and alternatives were ranked with PiF-TOPSIS. The expert step serves to validate and prioritize the literature-derived risks in a realistic urban construction context; it does not constitute human subject research involving personal data. Algorithm 1 presents the pseudocode of the proposed methodology.

Algorithm 1. Picture-Fuzzy SWARA and TOPSIS

Common Inputs and Notation

- Experts E = {e1, …, eℓ} provide linguistic judgments on criteria/alternatives. - Criteria set C = {C1, …, Cm}. - Alternatives A = {A1, …, An}. - Linguistic scale $\mathcal{L}$ with picture-fuzzy mappings to PiFNs (μ, η, ν, π); PFWA = Picture-Fuzzy Weighted Average; S(·) = picture-fuzzy score function.

(1).  Picture-Fuzzy SWARA (PiF-SWARA)—Criterion Weighting.

Input: Evaluated lists of criteria from each expert; linguistic scale $\mathcal{L}$ with PiFN mappings; PFWA and score function S(·).

Output: Normalized weights w1, …, wm with $\sum _j{w}_j=1$

1. Initialize: For each expert e, set $S_1^{\left(e\right)}\leftarrow 0$. Expert e selects a linguistic term ${\mathcal{l}}_j^{\left(e\right)}\in \mathcal{L}$ describing how important Cj is.

2. Map to picture-fuzzy: Convert ${\mathcal{l}}_j^{\left(e\right)}$ to Picture-Fuzzy Number = (μ, η, ν, π).

3. Aggregate experts (PFWA).

4. Score to scalar: sj ← S(Picture-Fuzzy Number) ∈ [0, 1]. Interpretation: sj is the comparative importance of Cj.

5. Recalculation factor: $k_j\leftarrow 1+s_j$ for j = 2, …, m; set $k_1\leftarrow 1$.

6. Intermediate weights: $q_1\leftarrow 1$; for j = 2, …, m, set $q_j\leftarrow q_{j-1}/k_j$.

7. Normalize: $w_j\leftarrow q_j/\sum q_j$, j = 1, …, m.

(2).  Picture-Fuzzy TOPSIS (PiF-TOPSIS)—Ranking Alternatives.

Input: Expert evaluations of alternatives on each criterion (linguistic $\mathcal{L}$ → PiFN), weights w1, …, wm from PiF-SWARA

Output: Scores Ri and ranking of alternatives A1, …, An (higher Ri ⇒ more risky).

1. Build expert decision matrices: For each expert e, record PiFN evaluations ${\mathrm{x}}_{ij}^{\left(e\right)}$ for alternative Ai on criterion Cj using $\mathcal{L}$.

2. Map to picture-fuzzy: Convert ${\mathrm{x}}_{ij}^{\left(e\right)}$ to Picture-Fuzzy Number = (μ, η, ν, π).

3. Aggregate experts (PFWGO).

4. Negative-ideal solutions (NIS) and Positive-ideal solutions (PIS): For each criterion j, determine the NIS and PIS PiFN $t_{ij}$ using the score function S(·).

5. Distances to solutions: For each alternative i, compute: (a) Euclidean (L2) distance to solutions $D\left(X_i,A_j^{\left\{+\right\}}\right) \mathrm{a}\mathrm{n}\mathrm{d} D\left(X_i,A_j^{\left\{+\right\}}\right)$

6. Scores and ranking: For each i, compute $C{C}_i={\displaystyle \frac{D_i^{\left\{-\right\}}}{D_i^{\left\{+\right\}}+ D_i^{\left\{-\right\}}}}$. Rank alternatives in descending order of ${CC}_i$.

End

3.1. Failure Modes and Effects Analysis (FMEA)

FMEA is a structured technique for systematically surfacing potential failures across design, manufacturing, and assembly processes, as well as products and services [24]. Owing to its simplicity and transparency, it has been widely adopted in both research and industrial practice [25,26]. FMEA enumerates plausible risk factors in the system, promotes preventive action, and quantifies risk by assigning a risk priority number (RPN) to each identified failure [27]. In this study, we use the extended RPN defined as the product of probability (P), severity (S), and detectability (D), as shown in Equation (1) [28,29]. The resulting RPN values can be used to classify risks into low, medium, and high categories, with thresholds and scales tailored to the context of the application [28]. Accordingly, our evaluation criteria are derived from the FMEA dimensions, while the scoring and prioritization are performed with MCDM tools. As FMEA has been applied in diverse domains, limitations of the classical form have become apparent; hybrid approaches that explicitly model uncertainty, especially those based on fuzzy numbers, have been proposed to obtain more faithful representations and more robust prioritizations [30]. One of the methods to manage uncertainty effectively is fuzzy numbers. Using fuzzy numbers in risk assessment methods, such as in FMEA, makes it possible to represent reality more accurately, obtain accurate results, and therefore evaluate risks accurately.

$$RPN=\mathrm{P}\times \mathrm{S}\times \mathrm{D}$$

(1)

3.2. Fine–Kinney (FK)

Fine–Kinney assesses risk using P, S, and F. To align with the FMEA notation in Section 3.1, we treat P as equivalent to Occurrence (O), and S retains its meaning in both frameworks. The E component of FMEA is not present in Fine–Kinney, which is consistent with previous comparative studies.

$$RPN=\mathrm{P}\times \mathrm{S}\times \mathrm{E}$$

(2)

Interpretation follows the classical formulation: larger RFK indicates higher risk priority. In this study, FK is reported diagnostically alongside FMEA to preserve comparability for practitioners; the ultimate ranking is produced by the picture-fuzzy MCDM stage, where PiF-SWARA supplies the criterion weights and PiF-TOPSIS yields the final ordering (Section 3.4 and Section 3.5).

3.3. Preliminaries of Picture-Fuzzy Sets

Picture-Fuzzy sets (PiFSs), introduced by Cuong and Kreinovich [31], generalize intuitionistic fuzzy sets (IFSs) by allowing four response degrees—positive (approval), negative (disapproval/rejection), neutral/abstention, and refusal—thereby increasing modeling flexibility for real-world judgments [32]. Compared with classical fuzzy sets (FSs) and IFSs, PiFSs broaden the admissible evaluation space and thus handle uncertainty and imprecision more effectively. The framework also provides operators that extend those of FSs and IFSs, making it well-suited to ambiguous or incomplete information [33]. In practice, PiFSs help reduce information loss by separately retaining positive, neutral/abstention, negative, and refusal tendencies during aggregation with uncertain data [34]. In this context, PiFSs enhance the expressive power of FSs, IFSs, and hesitant fuzzy sets by encompassing a wider range of responses and providing a more detailed and precise representation for complex decision-making problems [35]. The basic definitions employed in this study follow the canonical sources in [31,36,37].

Definition 1.

Given a universe of discourse $U,$ a PiFS ${\tilde{A}}_p=\left\{\left.⟨u,\left({\mu }_{{\tilde{A}}_p}(u),I_{{\tilde{A}}_p}(u),v_{{\tilde{A}}_p}(u)\right)\right| u\in U\right\}.$ Here ${\mu }_{{\tilde{A}}_p}\left(u\right)$, $v_{{\tilde{A}}_p}\left(u\right)$ and $I_{{\tilde{A}}_p}\left(u\right)$ denoted the membership, non-membership, and indeterminacy degrees of $u$ to ${\tilde{A}}_p$, respectively, each taking values in [0, 1] and satisfying [31]

$${\mu }_{{\tilde{A}}_p}\left(u\right)+I_{{\tilde{A}}_p}\left(u\right)+v_{{\tilde{A}}_p}\left(u\right)\le 1$$

(3)

The refusal degree is the

$${\mathsf{\pi }}_{{\tilde{A}}_p}=1-\left({\mu }_{{\tilde{A}}_p}(u)+v_{{\tilde{A}}_p}(u)+I_{{\tilde{A}}_p}(u)\right)$$

(4)

Definition 2.

Let ${\tilde{A}}_p=\left({\mu }_{{\tilde{A}}_p},I_{{\tilde{A}}_p},v_{{\tilde{A}}_p}\right)$ and ${\tilde{B}}_p=\left({\mu }_{{\tilde{B}}_p},I_{{\tilde{B}}_p},v_{{\tilde{B}}_p}\right)$ be two Picture-Fuzzy numbers (PiFNs). Basic operations between them [36,38]:

$${\tilde{A}}_p\oplus {\tilde{B}}_p=\left\{{\mu }_{{\tilde{A}}_p}+{\mu }_{{\tilde{B}}_p}-{\mu }_{{\tilde{A}}_p}{\mu }_{{\tilde{B}}_p},I_{{\tilde{A}}_p}{I}_{{\tilde{B}}_p},v_{{\tilde{A}}_p}{v}_{{\tilde{B}}_p}\right\}$$

(5)

$${\tilde{A}}_p\otimes {\tilde{B}}_p=\left\{{\mu }_{{\tilde{A}}_p}{\mu }_{{\tilde{B}}_p},I_{{\tilde{A}}_p}{I}_{{\tilde{B}}_p},v_{{\tilde{A}}_p}+v_{{\tilde{B}}_p}-v_{{\tilde{A}}_p}{v}_{{\tilde{B}}_p}\right\}$$

(6)

$${\tilde{A}}_p^{\lambda }=\left\{{\mu }_{{\tilde{A}}_p}^{\lambda },I_{{\tilde{A}}_p}^{\lambda },\left(1-{\left(1-v_{{\tilde{A}}_p}\right)}^{\lambda }\right)\right\}$$

(7)

$$\lambda \cdot {\tilde{A}}_p=\left\{\left(1-{\left(1-{\mu }_{{\tilde{A}}_p}\right)}^{\lambda }\right),I_{{\tilde{A}}_p}^{\lambda },v_{{\tilde{A}}_p}^{\lambda }\right\}$$

(8)

Definition 3.

Picture-Fuzzy Weighted Geometric Operator (PFWGO) for multiple Picture-Fuzzy numbers $w=\left(w_1,w_2,\dots ,w_n\right)$; ${0\le w}_j\le 1$; $\sum _{j=1}^n w_j=1$ [38,39]. We aggregate using the algebraic t-norm/t-conorm:

$${\mathrm{P}\mathrm{F}\mathrm{W}\mathrm{G}\mathrm{O}}_w\left({\tilde{A}}_1,\dots ,{\tilde{A}}_n\right)=\left\{\prod _{j=1}^n {\mu }_{{\tilde{A}}_j}^{w_j},\prod _{j=1}^n I_{{\tilde{A}}_j}^{w_j},1-\prod _{j=1}^n {\left(1-v_{{\tilde{A}}_j}\right)}^{w_j}\right\}$$

(9)

Definition 4.

Let ${\tilde{A}}_p=\left({\mu }_{{\tilde{A}}_p},I_{{\tilde{A}}_p},v_{{\tilde{A}}_p}\right)$ is a PiFN. The score function for a ${\tilde{A}}_p$ [40].

$$\mathrm{S}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}\left({\tilde{A}}_p\right)=\left(2{\mu }_{{\tilde{A}}_p}-v_{{\tilde{A}}_p}-{\displaystyle \frac{I_{{\tilde{A}}_p}}{2}}\right)$$

(10)

We rank picture-fuzzy evaluations using an affine score. This choice belongs to the standard linear/weighted family of PiFS scores used for ordering alternatives. In the PiFS literature, additive forms such as $\mu +\eta -\nu$ (combining positive/neutral and penalizing negative) are commonly employed for ranking; weighted variants are used to reflect problem-specific priorities. In our urban risk setting, we (i) up-weight positive evidence (coefficient 2 on $\mu$) to reward strong support, (ii) penalize explicit contradiction (−1 on $\nu$) more than uncertainty, and (iii) apply a lighter penalty to indeterminacy (−$I/2$), acknowledging that “unknown” degrades confidence but less than known opposition. This yields a monotone, interpretable score aligned with the PiFS ranking practice.

Definition 5.

Let ${\tilde{A}}_p=\left({\mu }_{{\tilde{A}}_p},I_{{\tilde{A}}_p},v_{{\tilde{A}}_p}\right)$ and ${\tilde{B}}_p=\left({\mu }_{{\tilde{B}}_p},I_{{\tilde{B}}_p},v_{{\tilde{B}}_p}\right)$ are two PiFNs. The distance between ${\tilde{A}}_p$ and ${\tilde{B}}_p$ [41]. We use the Wang–Xin mixed distance (average absolute gap plus a worst-gap term), extended from IFS to PiFS by including the refusal component $\pi$, and normalized to [0, 1] as

$$\begin{array}{c}d\left({\tilde{A}}_p,{\tilde{B}}_p\right)={\displaystyle \frac{1}{4}}\left(\left|{\mu }_{{\tilde{A}}_p}|-|{\mu }_{{\tilde{B}}_p}\right||+|\left|I_{{\tilde{A}}_p}-I_{{\tilde{B}}_p}\right||+|\left|v_{{\tilde{A}}_p}-v_{{\tilde{B}}_p}\right||+|\left|{\pi }_{{\tilde{A}}_p}-{\pi }_{{\tilde{B}}_p}\right|\right)\\ +{\displaystyle \frac{1}{2}}max\left(\left|{\mu }_{{\tilde{A}}_p}-{\mu }_{{\tilde{B}}_p}\right|,\left|I_{{\tilde{A}}_p}-I_{{\tilde{B}}_p}\right|,\left|v_{{\tilde{A}}_p}-v_{{\tilde{B}}_p}\right|,\left|{\pi }_{{\tilde{A}}_p}-{\pi }_{{\tilde{B}}_p}\right|\right)\end{array}$$

(11)

3.4. PiF—SWARA

In this study, the criterion weights were determined using PiF-SWARA. SWARA is a decision-maker-oriented, subjective weighting approach that has been widely used to elicit and rank criteria that are important in MCDM settings. The method explicitly relies on expert judgments; hence, the role and credibility of decision-makers directly influence the resulting weights [42]. In this research, the PiF-SWARA method is applied to determine the significance weights of the main and sub-criteria across all hierarchical levels for evaluating construction waste risks. The experts’ evaluations, formulated as PiFNs, are first combined using the PFWGO operator, followed by the application of SWARA’s step-by-step procedure to obtain the final weights. The procedure is:

Step 1

Experts assess each criterion using the linguistic terms listed in

Table 2. Linguistic terms for PiF-SWARA [43].

Linguistic Terms	$\mathbf{PiFNs} \left(\mathit{\mu },\mathit{I},\mathit{v}\right)$
VL-Very Low Importance	(0.1, 0.01, 0.85)
L-Low Importance	(0.25, 0.05, 0.6)
F-Fair Importance	(0.5, 0.1, 0.4)
H-High Importance	(0.75, 0.05, 0.1)
VH-Very High Importance	(0.9, 0.01, 0.05)

Note: μ = membership, I = indeterminacy, v = non-membership.

.

Step 2

Component-wise fusion of the experts’ PiFNs is performed via the PFWGO operator to generate the aggregated evaluation of each criterion.

Step 3

The aggregated PiFN of each criterion is transformed into a crisp score for ordering.

Step 4

Criteria are arranged from highest to lowest according to their score values.

Step 5

The score gap between each criterion $\left(j\right)$ and the one ranked just before it $(j-1)$ is used to determine their relative significance.

Step 6

Based on these differences, the comparative coefficients are computed.

$$k_j=\left\{\begin{array}{c}1, j=1\hfill \\ s_j+1, j>1\end{array}\right.$$

(12)

Step 7

The updated weights are computed.

$$q_j=\left\{\begin{array}{c}1, j=1\\ {\displaystyle \frac{q_{(j-1)}}{k_j}}, j>1\end{array}\right.$$

(13)

Step 8

The final weights of the criteria are computed.

$$w_j={\displaystyle \frac{q_j}{\sum _{j=1}^n q_j}}$$

(14)

where $n$ is the criteria number.

3.5. PiF—TOPSIS

Decision-makers evaluate every defined risk factor for each alternative. To prioritize the tasks, we apply TOPSIS under PiFS. PiF-TOPSIS represents expert judgments with membership, indeterminacy, non-membership, and implied refusal degrees. PiF-TOPSIS explicitly considers both the positive-ideal and negative-ideal solutions, while remaining computationally transparent and easy to integrate with PiF-SWARA for criterion weighting. The procedure is:

Step 1: Experts assess each alternative using the linguistic terms listed in

Table 3. Linguistic terms for PiF-TOPSİS.

Linguistic Terms	$\mathbf{PiFNs} \left(\mathit{\mu },\mathit{I},\mathit{v}\right)$
VL—Very Low	(0.05, 0.45, 0.5)
L—Low	(0.1, 0.4, 0.45)
ML—Medium Low	(0.15, 0.35, 0.4)
F—Fair	(0.3, 0.35, 0.35)
MH—Medium High	(0.4, 0.2, 0.15)
H—High	(0.45, 0.15, 0.1)
VH—Very High	(0.5, 0.1, 0.05)

, and their judgments are aggregated via PFWGO to form the decision matrix $X_{ij}$ (risk factor i under criterion j); with PiF-SWARA weights $w_j$.

Step 2: The ideal profiles for each criterion j are constructed by the score-based extremum, based on criteria type.

For benefit criteria:

$$A_j^{\left\{+\right\}}=\left\{{\tilde{X}}_{ij} where \underset{i}{\max }{\tilde{x}}_{ij}\left|i=\mathrm{1,2},\dots ,m\right.\right\}$$

(15)

$$A_j^{\left\{-\right\}}=\left\{{\tilde{X}}_{ij} where \underset{i}{\min }{\tilde{x}}_{ij}\left|i=\mathrm{1,2},\dots ,m\right.\right\}$$

(16)

For cost criteria:

$$A_j^{\left\{+\right\}}=\left\{{\tilde{X}}_{ij} where \underset{i}{\min }{\tilde{x}}_{ij}\left|i=\mathrm{1,2},\dots ,m\right.\right\}$$

(17)

$$A^{A_j^{\left\{-\right\}}}=\left\{{\tilde{X}}_{ij} where \underset{i}{\max }{\tilde{x}}_{ij}\left|i=\mathrm{1,2},\dots ,m\right.\right\}$$

(18)

In this context, $A_j^{\left\{+\right\}}$ and $A_j^{\left\{+\right\}}$ correspond to the positive-ideal and negative-ideal solutions. Ideal solutions are denoted as $x_j^{+}=({\mu }_j^{*},v_j^{*},I_j^{*})$, $x_j^{*}=({\mu }_j^{-},v_j^{-},I_j^{-})$, respectively.

Step 3: For each alternative i, compute the weighted separations to the positive and negative ideal, $D_i^{\left\{+\right\}}$ and $D_i^{\left\{-\right\}}$ respectively.

$$\begin{array}{c}D\left(X_i,A_j^{\left\{+\right\}}\right)\\ =w_j{\displaystyle \sum _{j=1}^n} \left[\begin{array}{c}{\displaystyle \frac{1}{4}}\left(\left|{\mu }_{{\tilde{x}}_{ij}}-{\mu }_j^{*}\right|+\left|I_{{\tilde{x}}_{ij}}-I_j^{*}\right|+\left|v_{{\tilde{x}}_{ij}}-v_j^{*}\right|+\left|{\pi }_{{\tilde{x}}_{ij}}-{\pi }_j^{*}\right|\right)\\ +{\displaystyle \frac{1}{2}}max\left(\left|{\mu }_{{\tilde{x}}_{ij}}-{\mu }_j^{*}\right|+\left|I_{{\tilde{x}}_{ij}}-I_j^{*}\right|+\left|v_{{\tilde{x}}_{ij}}-v_j^{*}\right|+\left|{\pi }_{{\tilde{x}}_{ij}}-{\pi }_j^{*}\right|\right)\end{array}\right]\end{array}$$

(19)

$$\begin{array}{c}D\left(X_i,A_j^{\left\{+\right\}}\right)\\ =w_j{\displaystyle \sum _{j=1}^n} \left[\begin{array}{c}{\displaystyle \frac{1}{4}}\left(\left|{\mu }_{{\tilde{x}}_{ij}}-{\mu }_j^{-}\right|+\left|I_{{\tilde{x}}_{ij}}-I_j^{-}\right|+\left|v_{{\tilde{x}}_{ij}}-v_j^{-}\right|+\left|{\pi }_{{\tilde{x}}_{ij}}-{\pi }_j^{-}\right|\right)\\ +{\displaystyle \frac{1}{2}}max\left(\left|{\mu }_{{\tilde{x}}_{ij}}-{\mu }_j^{-}\right|+\left|I_{{\tilde{x}}_{ij}}-I_j^{-}\right|+\left|v_{{\tilde{x}}_{ij}}-v_j^{-}\right|+\left|{\pi }_{{\tilde{x}}_{ij}}-{\pi }_j^{-}\right|\right)\end{array}\right]\end{array}$$

(20)

Step 4: Ranking can be reported in descending order $R{I}_i$ (higher indicates higher risk priority), which aligns with the FMEA/FK interpretation of “larger = riskier”.

$$C{C}_i={\displaystyle \frac{D_i^{\left\{-\right\}}}{D_i^{\left\{+\right\}}+D_i^{\left\{-\right\}}}}$$

(21)

Step 5: The alternatives are ordered from highest to lowest.

4. Case Study

This research utilizes a case study focused on the CDW management system of Istanbul, Turkey. This specific case was selected for several key reasons. First, as a megacity with intense and continuous urban transformation, Istanbul generates massive volumes of CDW. Its management system is characterized by high scale and complexity, involving multiple municipal authorities, numerous private contractors, and various transport and recovery facilities, which presents the exact multi-dimensional risk environment this study aims to address. Second, the system’s established processes and regulatory scrutiny allowed for the identification and validation of the 40 distinct risk factors spanning environmental, social, financial, and operational categories. Finally, this specific urban context provided crucial access to the required panel of 15 experienced professionals and academics, whose diverse judgments were essential for populating the Picture-Fuzzy MCDM model. This combination of features made Istanbul an ideal setting to test the practical applicability and robustness of the proposed framework.

This section presents a case study conducted to analyze the risks encountered in construction waste management processes and to illustrate the practicality of the methodology outlined in the study. The comprehensive analysis of the risks encountered in construction waste management processes aimed to minimize their impact. For the case study, insights were obtained from fifteen professionals experienced in waste management and occupational health and safety. The research comprises two primary stages. Initially, the significance levels of the criteria were evaluated using the PiF-SWARA approach, and subsequently, the risk factors were prioritized using the PiF-TOPSIS approach. This integrated approach overcomes the limitations of classical methods by allowing for the uncertainty and hesitation found in expert opinions. Construction waste management is not merely a technical process; it involves environmental, economic, and social risks. Improperly managed waste processes can lead to environmental threats, occupational health and safety issues, increased costs, and legal non-compliance. A literature review and expert interviews identified 40 critical risk factors in this context. These factors are classified as environmental, social, facility and operations, occupational health and safety, quality, financial, and management and legislation.

4.1. Determination of Criteria Weights with PiF-SWARA

This subsection presents the calculation of criteria weights through the PiF-SWARA technique. The criteria were identified with reference to several established risk assessment methodologies. The empirical data for this study was collected from a panel of 15 experts. These professionals were selected based on having significant experience in waste management and occupational health and safety. The panel was intentionally balanced to capture both theoretical and practical perspectives, comprising 7 academics (including professors and senior researchers specializing in civil engineering, environmental engineering, industrial engineering/risk analysis, and construction management) and 8 industry professionals. The industry experts included senior project managers, environmental consultants, OHS managers, waste facility operators, and municipal-level regulatory officers, all with extensive practical experience in C&D waste operations. The data collection involved two distinct stages: first, experts provided linguistic assessments of the five criteria using the PiF-SWARA scale (Table 2); second, they evaluated all 40 risk factors against these criteria using the PiF-TOPSIS linguistic scale (Table 3). The corresponding evaluation results for criteria weighting are shown in

Table 4. Evaluation matrix of criteria.

	E1	E2	E3	E4	E5	E6	E7	E8	E9	E10	E11	E12	E13	E14	E15
Probability	F	MH	F	H	MH	MH	MH	H	H	VH	H	H	H	H	H
Severity	VH	H	H	VH	VH	VH	H	VH	VH	H	VH	VH	VH	VH	VH
Detectability	VH	MH	MH	MH	H	MH	F	MH	MH	VH	H	MH	F	F	MH
Frequency	H	F	F	H	F	F	F	MH	H	H	VH	F	ML	ML	F
Exposure	VH	H	H	F	H	MH	MH	VH	H	H	VH	H	H	H	H

.

Linguistic evaluations were first converted into PiFNs, and the resulting matrices are presented in Table S1 in the Supplementary Materials. The matrices obtained from fifteen experts were then combined using Equation (9). The combined matrix is presented in Table S2 in the Supplementary Materials. Since a predefined scale was used during the creation of the decision matrices, the normalization step was not necessary. Subsequently, exact values were calculated based on the pooled matrix using Equation (10). These exact values were used to determine the ranking of the criteria, where severity received the highest priority and frequency received the lowest priority. The exact values obtained from the calculation and the ranking of the criteria are given in Table S3 in the Supplementary Materials. The comparative importance coefficients ($k_j$), and $p_j$ values corresponding to this ranking are given in

Table 5. Criterion weights.

Criterion	Score	${\mathit{s}}_{\mathit{j}}$	${\mathit{k}}_{\mathit{j}}$	${\mathit{P}}_{\mathit{j}}$	Weight
Probability	0.7686		1	1	0.208
Severity	0.7455	0.02	1.02	0.98	0.203
Detectability	0.7321	0.01	1.01	0.96	0.201
Frequency	0.7205	0.01	1.01	0.95	0.198
Exposure	0.6785	0.04	1.04	0.92	0.190

and Equations (12)–(14) were used.

The final weights of the criteria are as shown in Figure 2.

The high weight of the severity criterion demonstrates that it plays a pivotal role in risk assessment. This shows that experts consider the potential level of harm that risks can cause to be the most critical factor. The exposure and probability criteria, which have very similar values, rank second and third. This indicates that the probability of the risk occurring and the degree of exposure to this risk play an important role in the assessment. It should be noted that the detectability criterion has a relatively lower weighting, but the controllability of risks should still be considered. Despite having the lowest weight, the frequency criterion is not entirely negligible; rather, it plays a comparatively minor role relative to the other criteria.

4.2. Evaluation of Risk Factors with PiF-TOPSIS

This section evaluates the expert-defined risk factors (Table 1) through the PiF-TOPSIS approach, with the assessed factors depicted in Figure 3. Each risk was evaluated by experts on five criteria—Probability, Severity, Detectability, Frequency, and Exposure—using concise operational definitions and a 7-level linguistic scale $\{\mathrm{V}\mathrm{L}, \mathrm{L}, \mathrm{M}\mathrm{L}, \mathrm{F}, \mathrm{M}\mathrm{H}, \mathrm{H}, \mathrm{V}\mathrm{H}\}$. Responses were collected independently and anonymously; items were randomized and a short calibration with examples was provided beforehand. Each linguistic term was converted to a picture-fuzzy number via a monotone mapping (Table 3). Individual judgments were aggregated by PFWGO to form the decision matrix. Bias was mitigated through calibration, randomized order, no group discussion, attention/consistency checks, and robustness analyses.

The expert inputs obtained earlier were revisited to develop the risk factor evaluation matrix, shown in Table S4 in the Supplementary Materials. The linguistic expressions within this matrix were then translated into PiFNs, as defined in Table 3. Subsequently, the evaluation matrices were combined using Equation (9). In this step, the weights of the experts were considered equal. In the combined matrix, positive ideal and negative ideal solution values were determined for each criterion. The five criteria are treated as mixed-type: Probability, Severity, Frequency, and Exposure are benefit-type, and Detectability is cost-type. Each criterion is first mapped to the PiF domain and normalized with the appropriate rule. Benefit-type criteria are scaled so that larger values imply higher risk. The cost-type criterion is scaled so that larger values imply lower risk. Next, Equations (17) and (18) were applied to calculate the separation of each risk factor from the positive and negative ideal solutions. The criterion-wise distance values were aggregated to determine the overall distance for each factor.

Then, a score for each risk factor is determined by considering the average of these totals, which are shown in Figure 4. Risks are ranked according to their scores. Based on the scores and rankings obtained from the calculation, it was observed that environmental and health-related risks ranked high. R6, which has the highest score and ranks first, is the risk of cumulative pollution and increased complaints, indicating that construction waste causes long-term environmental pollution and thus cumulative effects, and stands out as the most critical risk. This situation shows that environmental damage is a major pressure factor both ecologically and in terms of social acceptance.

The second risk factor with the highest score is R1. The risk of groundwater leakage ranks high due to its potential to cause irreversible effects from a hydrogeological perspective. R28. Insufficient investment and operating budgets are critical because they hinder the sustainability of waste management infrastructure, making it difficult to control other risks. On the other hand, it was observed that the risk factors with the lowest scores were R40. Lack of inter-institutional cooperation and coordination, R32. Inadequate performance specifications, contract monitoring, and R31. Lack of management support and oversight is a risk stemming from management and organizational issues. Although these risks are not considered critical in the short term, such managerial weaknesses should not be overlooked. They may lead to systematic inconsistencies in the long run, again triggering environmental and operational risks. These results highlight that decision-makers should focus on urgent environmental and financial risks and long-term measures to strengthen institutional capacity.

Table 6. Risks by PiF–SWARA–TOPSIS with $C{C}_i$ (Method) and ranks under FMEA and FK.

Risk	Score	Rank			Risk	Score	Rank
		Method	FMEA	FK			Method	FMEA	FK
R1	0.161	2	1	2	R21	0.144	35	4	10
R2	0.145	31	7	12	R22	0.146	28	36	36
R3	0.155	17	2	3	R23	0.145	32	24	33
R4	0.144	33	16	22	R24	0.148	26	30	34
R5	0.156	9	6	5	R25	0.152	19	5	13
R6	0.162	1	19	7	R26	0.146	30	39	39
R7	0.155	14	40	27	R27	0.143	37	20	16
R8	0.143	36	27	29	R28	0.159	3	23	4
R9	0.157	6	38	19	R29	0.155	15	34	15
R10	0.150	22	26	31	R30	0.152	18	18	17
R11	0.149	24	31	30	R31	0.140	38	10	25
R12	0.156	7	33	35	R32	0.140	39	11	32
R13	0.148	27	20	24	R33	0.156	8	32	40
R14	0.149	23	35	38	R34	0.155	12	28	21
R15	0.151	21	3	1	R35	0.151	20	29	20
R16	0.157	5	13	9	R36	0.144	34	22	23
R17	0.158	4	8	26	R37	0.155	10	17	11
R18	0.155	13	9	6	R38	0.155	16	25	18
R19	0.148	25	15	8	R39	0.155	11	14	14
R20	0.146	28	36	36	R40	0.138	40	12	28

summarizes the top 10 risks according to our PiF–SWARA–TOPSIS results. For each item, we report its closeness coefficient $C{C}_i$ (higher = riskier), the rank in our method, and the corresponding FMEA and FK ranks. Overall, there is clear alignment for R1 and R5, which remain high priority across all three approaches. Differences appear for items such as R9, R12, and R33, which our method ranks higher than FMEA/FK; this reflects the integrated weighting of Detectability and Exposure in the picture-fuzzy framework, capturing effects that single-formula indices may understate. No ties occurred in $C{C}_i$ values.

5. Sensitivity Analysis

In this part of the study, a sensitivity analysis was carried out to evaluate the robustness of the proposed methodology. The sensitivity analysis investigated the extent to which changes in criterion weights affected the final ranking results. For the sensitivity analysis, the initial weights obtained via the PiF-SWARA method were restructured to form six alternative weighting scenarios. In each scenario, one criterion was emphasized by assigning it a weight of 0.5, while the remaining criteria were each allocated a weight of 0.125. This approach allowed every criterion to be highlighted individually across different sets. Additionally, a final scenario was introduced in which all criteria were assigned equal weights of 0.2. The weight distributions for all scenarios are listed in

Table 7. Criteria weight sets for sensitivity analysis.

	Set-0	Set-1	Set-2	Set-3	Set-4	Set-5	Set-6
Probability	0.200	0.5	0.125	0.125	0.125	0.125	0.2
Severity	0.208	0.125	0.5	0.125	0.125	0.125	0.2
Detectability	0.198	0.125	0.125	0.5	0.125	0.125	0.2
Frequency	0.190	0.125	0.125	0.125	0.5	0.125	0.2
Exposure	0.203	0.125	0.125	0.125	0.125	0.5	0.2

.

Figure 5 presents a heat map of rank positions for the top 15 risks across Set-0 to Set-6, where darker cells indicate better ranks (lower values), enabling direct visual comparison. When considered together with criterion weight sets and risk rankings, sensitivity analysis results provide important insights into the model’s consistency and flexibility. In Set-1, the highlighted probability criterion with a weight of 0.5 created a ranking focused on the likelihood of risks occurring, and in this case, R6. Cumulative pollution and increased complaints risk ranked first, as in the original ranking. In Set-2, with severity receiving a weight of 0.5, the impact of risks became decisive; in this scenario, R1. Risk of groundwater leakage ranked first, while R6. Risk of cumulative pollution and increased complaints fell to 7th place. In Set-3, with priority given to detectability, technically uncontrollable risks moved to the top ranks, R9. Infrastructure inadequacy and R33. Data and traceability errors came to the fore. When the frequency criterion became dominant in Set-4, more frequently recurring risks came to the fore, and occupational health and safety risks such as fire and dust dispersion became critical. In Set-5, with the exposure criterion receiving the highest weighting, risks directly affecting humans and the environment rose to the top. Finally, in Set-6, all criteria were given equal weight, and in this case, the model produced a more balanced ranking; however, fundamental environmental and financial risks such as R6, R1, and R28 retained their most critical positions. Numerically, these environmental/financial exemplars remain highly stable underweight perturbations: R6 stays in the top 10 in 7/7 scenarios and in the top 3 in 6/7 scenarios (median rank = 2; range = 1–7). R1 stays in the top 10 in 6/7 and in the top 3 in 5/7 scenarios (median = 2; range = 1–14). R28 stays in the top 10 in 6/7 and in the top 5 in 4/7 scenarios (median = 4; range = 1–13). By contrast, operational items fluctuate more strongly. For example, R33 varies between ranks 3 and 24, and R9 between 4 and 21 showing that stability concentrates on core environmental/financial risks rather than operational ones.

6. Discussion

This study proposes a risk-prioritization tool that couples the diagnostic clarity of FMEA/Fine–Kinney with PiF-SWARA and PiF-TOPSIS to handle indeterminacy and hesitation in expert judgements. Across forty CDW risks, severity emerged as the dominant criterion, followed by exposure and probability; the top risks were cumulative pollution with public complaints, groundwater leakage, and insufficient investment/operating budgets. Sensitivity checks showed these environmental and financial risks remain prominent under plausible weight perturbations, supporting the robustness of prioritization.

At a strategic level, fuzzy MCDM frameworks have been used to rank policies and management tactics for CDW. Eghbali-Zarch et al. [19] integrated fuzzy IDOCRIW and WASPAS to prioritize strategies; they found the most effective levers were (i) implementing operational plans and research and (ii) enacting stronger rules to curb CDW generation. Our highest-priority risks map onto those levers: mitigating groundwater leakage and diffuse pollution demands near-term operational controls, while the budgetary deficit risk (insufficient investment/operating funds) points to contract and regulatory revisions that strengthen financial accountability precisely the governance emphasis highlighted in that study.

In a circular-economy framing, Kabirifar et al. [20] used a hybrid fuzzy approach to identify life-cycle factors; “on-site sorting, reusing and recycling of wasted material” and diversified procurement mechanisms were pivotal. Our results explain why those factors repeatedly surface: the environmental risk cluster we rank highest (dust/runoff, leakage, cumulative pollution) is exactly what on-site sorting and process control are designed to abate, and the prominence of financial risk underscores the role of procurement and budgeting in enabling those controls. Thus, our analysis provides a mechanism-level justification for strategy rankings reported in earlier work. Complementary findings also appear in national case applications using classical AHP: in Sri Lanka, environmental criteria carried the greatest salience (≈41.6%), ahead of economic, technical and social factors; on-site segregation ranked best among alternatives [21]. The dominance of environmental (and closely trailing economic) considerations in our weights—and the policy/operational nature of our top risks aligns with that pattern while offering a finer-grained, risk-centric view.

Recent siting and planning studies in CDW incorporate richer uncertainty models, e.g., hesitant neutrosophic sets (HNS) fused with GIS and DANP, arguing that classical hesitant-fuzzy and single-valued neutrosophic forms inadequately express neutral/opposing information and struggle under multivariate uncertainty [22]. Our design follows the same trajectory toward expressive uncertainty representation but in an operational risk context: PiF preserves approval, indeterminacy, rejection, and refusal degrees at once, reducing information loss when aggregating expert judgements and making the weighting (PiF-SWARA) and ranking (PiF-TOPSIS) steps more faithful to what experts actually mean. Relative to frameworks that require spatial data pipelines or bespoke influence-network estimation, our pipeline remains lightweight and auditable, while still capturing richer epistemic structure.

Compared with strategy ranking studies that stop at policy prescriptions [19,20,21], our analysis identifies concrete failure modes environmental leakages and funding shortfalls as the practical bottlenecks, and links them to implementable controls (storage integrity, source separation, runoff abatement, preventive maintenance) and governance fixes (risk-based contract clauses, oversight, and budget thresholds). This bridges the gap between “what to prioritize” (strategies) and “what to fix first” (risk mechanisms). It also preserves semantic continuity with legacy risk indices by treating all criteria as cost-type and aligning the closeness coefficient with the “larger = riskier” convention, which aids uptake in organizations already using FMEA/FK.

Beyond methodological novelty, the proposed PiF–SWARA/TOPSIS risk model advances risk assessment along three practice-relevant dimensions: (i) Information fidelity: PiF captures four judgment states, addressing the neutrality/ambiguity limitations noted for other fuzzy families [22]. (ii) Interpretability: By harmonizing FMEA/FK with PiF-SWARA/TOPSIS, the model remains transparent to engineers and auditors while upgrading uncertainty handling. (iii) Actionability and robustness: The sensitivity-stable elevation of environmental and financial risks connects cleanly to enforceable controls and budget reforms, echoing and extending strategic priorities identified by earlier fuzzy MCDM work [19].

Finally, the findings both corroborate and sharpen the literature; where prior studies rank strategies or design siting frameworks, we show, with higher-resolution risk mechanics, which environmental exposures and financial constraints demand immediate attention and provide a defensible, uncertainty-aware pathway to act on them.

7. Managerial Implications

Translating the risk-prioritization results into practice points to two layers of action. For cumulative pollution with escalating complaints (R6) and groundwater leakage (R1), the principal agents are the contractor and site manager, supported by the resident engineer and the OHS (Occupational Health and Safety) manager. Immediate measures focus on (i) storage integrity—lined, sealed and covered areas for temporary stockpiles; (ii) source separation—physically demarcated bays, clear signage and contamination checks at hand-off; (iii) runoff and seepage abatement—perimeter drains, silt traps and protected transfer zones designed with basic hydrological input; and (iv) preventive maintenance—scheduled inspection and upkeep of crushers, conveyors and pumps. The expected effect is to suppress diffuse emissions and leachate pathways, improve regulatory compliance and community acceptance, raise recyclate purity (and thus substitution potential), and reduce unplanned downtime.

For budgetary shortfalls (R28), action is required above the site level. Responsibility should rest with the project owner and the municipal sponsor, with the procurement and finance functions implementing ring-fenced allocations for storm-water controls, storage integrity and maintenance, under proportionate oversight by the regulator (or an independent engineer). Contracting should incorporate risk-based clauses that make separation standards, maintenance and runoff controls enforceable obligations with routine audits and corrective-action triggers; owners should also set minimum budget thresholds so these controls cannot be value-engineered out during execution. Siting and logistics decisions should exploit spatial analysis to shorten haul distances and reduce community exposure, thereby stabilizing operating conditions and sustaining public acceptance.

8. Conclusions

This study develops and tests an interpretable decision-support framework that couples classical risk diagnostics (FMEA/FK) with picture-fuzzy SWARA–TOPSIS to prioritize construction-and-demolition-waste (CDW) risks under expert uncertainty. Using a case application with 15 experts and 40 risk factors from a large urban CDW system, the method retains approval, indeterminacy, rejection, and refusal degrees in both weighting and ranking, thereby reducing information loss while preserving the semantics familiar to practitioners who use FMEA/FK. The model treats all criteria consistently and aligns the TOPSIS closeness coefficient to the “larger = riskier” convention to facilitate adoption in practice.

Methodologically, the contribution lies in harmonizing the transparency of FMEA/FK with the expressive uncertainty handling of picture-fuzzy sets, which capture four judgment states and thereby address neutrality and ambiguity limitations of other fuzzy families. The design bridges the gap between high-level strategy studies and on-the-ground failure modes by linking ranked risks to enforceable controls and governance levers, while remaining lightweight and auditable. Relative to strategy-ranking work in CDW management—which highlights operational plans and regulatory tightening this mechanism-level viewpoints directly to what must be fixed first in day-to-day operations. Criterion importance follows Severity > Exposure ≈ Probability > Detectability > Frequency, indicating that consequence magnitude and pathways of contact dominate recurrence or ease-of-detection in expert judgments. The three highest-priority risks are cumulative pollution with rising complaints (R6), groundwater leakage (R1), and insufficient investment/operating budgets (R28). Sensitivity analyses across seven weighting scenarios show these environmental and financial risks remain prominent (e.g., R6 stays in the top 10 in 7/7 scenarios; R1 in 6/7; R28 in 6/7), supporting robustness of the ranking. These results echo but also sharpen the literature’s strategic emphasis [19] on operational planning and regulation by isolating the specific environmental leakages and funding shortfalls that constitute practical bottlenecks.

The prioritized risks point to a staged program: (i) near-term consequence and exposure controls (leak-proof temporary storage, source separation, dust/runoff abatement, preventive maintenance); (ii) medium-term data and traceability upgrades; and (iii) governance fixes risk-based contract clauses, oversight intensity, and budget thresholds so that environmental controls are financially durable. These prescriptions complement spatial planning and siting frameworks in the CDW literature [22] by providing an upstream risk screen that can inform buffer choices and layout trade-offs.

Findings reflect one metropolitan context and an equal-weight expert panel; larger and more diverse panels, alternative expert-weighting schemes, and cross-city replications are needed. Longitudinal data would allow tracking risk migration over time and calibrating action thresholds to real outcomes. Coupling the present risk module with spatially explicit, GIS-enabled siting and network models would support end-to-end decisions from risk abatement to facility placement. Extending comparison to other expressive uncertainty is a further avenue. Overall, the integrated FMEA/FK–PiF SWARA–TOPSIS framework supplies a transparent and defensible basis for allocating resources to the highest-impact environmental and financial risks in CDW systems.