Reverse Logistics Network Optimization for Retired BIPV Panels in Smart City Energy Systems

Abstract

Through the energy conversion of building skins, building-integrated photovoltaic (BIPV) technology, the core carrier of the smart city energy system, encourages the conversion of buildings into energy-generating units. However, the decommissioning of the module faces the challenge of physical dismantling and financial environmental damage because of the close coupling with the building itself. As the first tranche of BIPV projects will enter the end of their life cycle, it is urgent to establish a multi-dimensional collaborative recycling mechanism that meets the characteristics of building pv systems. Based on the theory of reverse logistics network, the research focuses on optimizing the reverse logistics network during the decommissioning stage of BIPV modules, and proposes a dual-objective optimization model that considers both cost and carbon emissions for BIPV. Meanwhile, the multi-level recycling network which covers “building points-regional transfer stations-specialized distribution centers” is designed in the research, the Pareto solution set is solved by the improved NSGA-II algorithm, a “1 + 1” du-al-core construction model of distribution center and transfer station is developed, so as to minimize the total cost and life cycle carbon footprint of the logistics network. At the same time, the research also reveals the driving effect of government reward and punishment policies on the collaborative behavior of enterprise recycling, and provides methodological support for the construction of a closed-loop supply chain of “PV-building-environment” symbiosis. The study concludes that in the process of constructing smart city energy system, the systematic control of resource circulation and environmental risks through the optimization of reverse logistics network can provide technical support for the sustainable development of smart city.

1. Introduction

Against the backdrop of the deep integration of global urbanization and low-carbon strategies, the development of smart energy systems has emerged as a critical global challenge for advancing sustainable urban development [1]. This challenge manifests acutely in urban environments worldwide. For instance, studies indicate that traditional building energy systems continue to dominate carbon emission patterns across many regions; as exemplified by China, these systems contribute 50.9% of the national total carbon footprint throughout their lifecycle, with 21.7% originating from direct operational emissions [2]. Amidst the profound global shift in energy structure from fossil fuels to renewables, establishing a new energy ecosystem characterized by multi-energy synergy has become an urgent imperative worldwide [3]. Integral to this transition, BIPV technology is accelerating the transformation of building energy use patterns [4]. However, a significant global challenge looms: BIPV modules will face large-scale decommissioning after 25–30 years of service life, posing a dual threat to the solid waste management systems of future smart cities globally due to the rare metals and potential pollutants they contain [5]. Therefore, building a reverse logistics network suitable for the development of smart cities and achieving efficient recycling and resource utilization of retired PV modules have become the key bottlenecks restricting the sustainability of new energy systems. This is a complex systematic project involving technical and economic assessment, policy and regulation design, and coordination of multiple interests. There is an urgent need to establish a comprehensive solution based on full life cycle management. Technical and economic assessment methods have been widely applied in evaluating the feasibility of energy systems [6]. For instance, in the early research on ground source heat pump systems [7], this method also provides an important methodological basis for optimizing the economic model and cost-benefit analysis of the reverse logistics network of decommissified BIPV panels. It is predicted that the retirement volume of PV modules in China will exceed 1.5 million tons in 2030, and with the proportion of BIPV increasing year by year, the resulting retirement tide will be explosive and concentrated [8]. Different from conventional PV panels, BIPV modules have a deep embedded architectural structure, and with the lack of recycling standards and technical and economic evaluation systems for BIPV construction scenarios currently, which makes the traditional recycling method face the dual dilemma of insufficient module recycling rate and the challenge of fine separation of materials. Furthermore, the relevant policies also explicitly incorporate BIPV into the key tasks of new urbanization construction, requiring the realization of industrial and public buildings with PV roof coverage of over 50% by 2025 as a binding target [9], highlighting the synergistic integration of PV building systems and urban energy networks, and the construction of a reverse logistics system adapted to the new type of solid wastes [10]. As an important component of the low-carbon energy system of smart cities, BIPV decommissioned modules essentially belong to the intersection of environmental system engineering and sustainable supply chain, involving multi-dimensional collaboration such as PV material recycling technology, circular economy model innovation, and policy regulation design [11]. As a means of institutional guarantee, the government reward and punishment mechanism directly affects the location optimization of the recycling network nodes, the participation of stakeholders, and the economic feasibility of the entire chain. The precise design and dynamic adjustment of its policy tools play a key role in the construction of the reverse logistics system. The core of the energy system in smart cities lies in integrating multiple renewable energy technologies, as demonstrated in the practice of using geothermal energy such as solar energy and biomass energy for heating in agricultural facilities [12], and BIPV, as a key form of solar energy application in urban environments, The efficient recovery and recycling network optimization of its decommissioned photovoltaic panels has become an important link in achieving the sustainable development of the system. Most of the existing studies focus on the application technology and dismantling process of traditional PV building materials [13]. However, there is a lack of systematic research on reverse supply chains from the perspective of the whole life cycle. Therefore, the main objective of this study is to establish a dual-objective optimization model for optimizing the spatio-temporal configuration of the retired BIPV panel recycling network, aiming to minimize both the total system cost and the carbon emissions throughout the entire life cycle simultaneously. To achieve this goal, this study has constructed a three-level network architecture of “transit station—regional center” based on the reverse supply chain theory, and the improved NSGA-II algorithm is used to realize the Pareto equilibrium solution set of economic and environmental benefits, and the dynamic coupling mechanism between policy reward and punishment intensity, urban network structure and path planning is revealed. The construction of an energy infrastructure decommissioning management network provides a theoretical basis to help realize the reverse logistics supply chain of “building-PV-environment” symbiosis, and promote the construction of a resource recycling system and the development goal of zero-carbon smart cities in the process of smart urbanization.

2. Literature Review

2.1. Research on Reverse Logistics Networks

The core of a smart city, as a product of the deep integration of new urbanization and informatization, lies in the collaborative optimization of the built environment and energy systems through intelligent technologies. BIPV serves as a vital component of the smart energy system, providing significant benefits for achieving energy self-sufficiency and reducing carbon emissions. However, the decommissioning of BIPV modules at the end of their life cycle poses new challenges to the existing reverse logistics network, making it urgent to build an adaptive recycling system. Existing reverse logistics network studies show that the supply chain network architecture has significant industry heterogeneity [14]. For example, the medical waste recycling system focuses on intelligent monitoring of end-of-line disposal [15], while the automotive industry highlights the core position of remanufacturing enterprises [16]. In contrast, the BIPV decommissioning modules reverse network emphasizes the processing capacity of resource nodes, which is closely related to its unique product material attributes, renewable resource market structure, and economic and environmental constraints.

In the selection of reverse logistics network mode, many decision-making issues are involved, such as facility location [17], production planning [18], and vehicle routing [19], etc., and existing studies have investigated the practice of reverse logistics in different industries and regional settings, but the systematic and holistic reverse logistics design and implementation methods have not been deeply studied. In order to solve the problems above, a multi-dimensional evaluation model of reverse supply chain is constructed in the research, which systematically explained the specific process elements, dynamic constraint mechanisms, and stakeholder collaboration paradigms in the recycling system [20]. To achieve the goal of minimizing the cost of waste treatment and resource utilization under the premise of ensuring ecological security [21]. In view of the problem of insufficient economic benefits of reverse logistics, dynamic game framework with multi-party participation has been established, which is to discusse the influence mechanism of solid waste tax and incentive constraint mechanism on the recycling efficiency and multi-dimensional benefits of retired PV equipment under the framework of extended producer responsibility system, and reveale the optimal tax incentive combination and market pricing scheme under equilibrium conditions [22].

In the field of reverse logistics network optimization, scholars at home and abroad have achieved rich research results. In view of the multiple uncertainties in the operating environment of the system, multi-objective decision-making models are generally constructed to take into account the balance between multiple indicators such as transportation costs, environmental benefits, and service efficiency [23]. Stochastic programming model for a multi-product reverse logistics network is proposed, which is to coordinate profits, social benefits, and carbon emission targets under uncertain prices, quantities, and product quality [24]. Besides, based on fuzzy mathematical programming, multi-objective decision-making model also is constructed in view of the multiple constraints of infrastructure construction cost, transportation energy consumption, and pollution emission in the traditional municipal waste treatment system, which focused on the equilibrium between demand response rate and operating cost [25]. Moreover, the improved non-dominated sorting genetic algorithm is chosen to solve the problem with the bee colony algorithm, and the complementary properties of the two algorithms provide methodological innovations for multi-dimensional optimization of reverse logistics networks. With the large-scale expansion of the PV industry and the localization of manufacturing under the incentive of renewable energy policies, the research on the reverse supply chain network of retired PV modules has become a key issue in the field of circular economy. At present, the surge in PV installed capacity and the wave of module scrapping have formed a spatio-temporal mismatch, exposing the systemic shortcomings of the reverse logistics system, such as lack of supervision, fragmentation of recycling networks, and fracture of value streams. After analyzing the challenges of decommissioned PV panels and reverse supply chains, Iakovou [26] proposed future research directions and the cost optimization path driven by the synergy between policy tools and market mechanisms based on the gap between academic research, industry and policy-making challenges,. Despite the 25–30 year cleaner production cycle of solar PV modules, their improper end-of-life disposal can still pose environmental risks, especially in the context of the surge in global PV installations, and the development of economically viable recycling technologies is essential for the sustainable development of the industry. Meanwhile, through technical and economic analysis, it is pointed out that it is necessary to establish a recycling system through policy guidance, optimize the recycling process of high-value materials, promote manufacturers to assume ecological responsibility, and improve the recyclable design of modules, so as to overcome the current dilemma of high module recycling cost, difficult material separation, and poor economy [27].

In summary, in the context of smart city energy system innovation, the large-scale decommissioning of BIPV modules poses new challenges to the reverse logistics network. Although significant progress has been made in the field of node location decision-making model construction, especially the target dynamic programming method based on intelligent algorithms, which has effectively strengthened the system resilience and multi-dimensional benefit synergy of the recovery network, there is still insufficient attention to the response mechanism for the retirement of BIPV modules. Most of the existing academic achievements focus on the process improvement of reverse logistics of general-purpose products, and the physical fusion characteristics of PV building materials and building carriers lead to significant heterogeneity in their recycling paths, and the existing general models are difficult to effectively solve the spatio-temporal matching problem in their reverse logistics networks. Therefore, constructing a reverse logistics optimization model of BIPV modules with multi-objective constraints and exploring the resource recycling path in the dismantling and recycling stage becomes the key research directions to realize the sustainable transformation of urban energy system.

2.2. Research on Construction Waste Recycling Network

With the intensification of global resource shortage and the significant trend of ecological environment deterioration, the research horizon of academia has gradually extended from the forward supply chain management of traditional building products to the construction of reverse logistics system, and rich academic achievements have been accumulated in the theoretical exploration and practical application of the closed-loop system of construction waste recycling economy. Large-scale construction activities have generated huge amounts of construction waste, which have exacerbated problems such as environmental load, resource depletion, and damage to the city [28]. In the research, a multi-objective and multi-period mixed integer linear programming framework has been constructed, which is to minimize the cost of waste treatment for contractors, maximize the economic benefits of recycling enterprises, and optimize the resource recovery rate. The framework integrates the coordination mechanism of the differences in the demands of multiple stakeholders in the construction industry and the dynamic decision-making situation, and provides a quantitative analysis tool for the optimization of resource circulation network in complex environments [29]. Subsequently, urban metabolic system transferred towards a closed-loop circulatory system mode, after the management and cost analysis of construction and demolition waste recycling in developing countries [30,31], and found that construction waste recycling can be promoted through strategies such as recycling, smart technology, government intervention, and economic incentives. Furthermore, existing studies have used game models to analyze the decision-making of multiple stakeholders in the process of construction waste recycling, which provides a reference for solving the problems of the construction waste recycling industry chain [32].

As an important part of the construction of reverse logistics system, the resource disposal of construction waste has become a key research direction in the domestic academic community in recent years. In the research of reverse logistics network under the framework of single-objective optimization, scholars generally focus on the improvement of transportation efficiency or cost control. From the perspective of resource integration, existing studies have demonstrated the quantitative relationship between logistics path reduction and recycling cost control, and proposed a path integration mechanism based on sorting and recycling, which significantly improves the efficiency of waste transfer and provides a theoretical basis for multi-node network optimization [33]. By establishing a multidimensional evaluation system, taking the integrated cost and timeliness optimization as the double constraints, and solving by mixed integer programming method, an operable mathematical modeling paradigm is provided for the cost control and path planning of regional collaborative logistics network [34]. Meanwhile, the carbon benefit model of construction waste management based on system dynamics evaluates the emission reduction effectiveness of multiple scenarios, and verifies that resource utilization has significant emission reduction potential, and the carbon trading mechanism under the baseline policy framework has more environmental benefits than the unconstrained scenario [35]. In response to the environmental pressure caused by the surge of construction waste, the existing studies have also explored the reduction control and resource utilization strategies through the construction of multi-stage models, which provide theoretical support for the comprehensive management of high-density urban waste [36].

In summary, the current research on the reverse supply chain of construction solid waste mainly focuses on the core areas of resource recycling technology and process innovation, multimodal transportation network efficiency optimization and industrialization coordination mechanism. Although a multi-level recycling standard system has been established for traditional building materials, further attention is needed to study the recycling network of new wastes such as retired PV modules for integrated PV buildings. BIPV decommissioning modules have the dual attributes of building materials and e-waste, and their reverse logistics systems need to meet the requirements of building demolition process and PV material recycling characteristics at the same time, and the existing recycling mode faces systemic challenges in terms of material sorting accuracy, environmental risk prevention and control, and economic feasibility. Especially in the context of the transformation of smart city energy system, it is urgent to build a dynamic recycling logistics network model based on life cycle analysis to realize the organic connection between BIPV waste reverse logistics network and urban solid waste management system.

2.3. Research on Government Reward and Punishment Policies

In current research on remanufacturing activities driven by government incentive and punishment policies, external policy intervention is considered the key path to achieve circular economy and emission reduction goals. Recycling behavior has a certain external economy, and government subsidy policies can effectively improve the competitive advantage of remanufactured products and the remanufacturing process, and are also conducive to promoting the conclusion of contract coordination mechanisms [37]. It can be seen that the government-enterprise evolutionary game model based on dynamic incentive constraint strategy reveals that the system is non-stationary under static regulation, while the dynamic reward and punishment mechanism can lead to the formation of game equilibrium [38]. The supply chain game model verifies the feasibility of dual-contract coordination under the government circular reward and punishment mechanism [39]. In the supply chain game model including government competition policy and technological innovation factors, based on the differentiated impacts of competition rewards and technological innovation levels on manufacturers’ profits in different contexts, the dual-promoting effects of policy incentives and cooperative cooperation on industrial development are verified [40]. In view of different commodity attributes, existing studies have combined the evolutionary game model [41] and constructed a reverse logistics model to compare and analyze the operational differences among the four types of policy scenarios: no fiscal intervention, subsidies at the production end, incentives in the distribution chain, and third-party support, and have found that there is a significant difference in the impact of different subsidies on supply chain decision-making, with subsidies to retailers effectively enhancing recycling incentives, while subsidies to manufacturers are more conducive to the coordination of the production end. Subsidizing retailers can effectively increase recycling incentives, while subsidizing manufacturers is more conducive to production coordination [42]. Government incentives and punishments regulate the behavior of closed-loop supply chain members through economic levers, but the final effect mainly depends on the design of policies and supporting coordination mechanisms. From the perspective of related research, the government reward and punishment policy regulate the behavior of closed-loop supply chain members through economic levers, and the final effect presented mainly depends on the design of the policy and the supporting coordination mechanism [43]. As well, in the model of subsidizing ladder utilizers, it is necessary to incorporate a cost-sharing contract to improve the efficiency of the supply chain in order to effectively extend the value of the industry chain [44]. Altruistic behavior is the optimal model choice for the supply chain after considering the dynamic characteristics of the recycling rate. The effectiveness of policy instruments is significantly regulated by market structure and stakeholders, and a single reward and punishment mechanism is difficult to form a Pareto improvement, so it is necessary to design a differentiated incentive and constraint system by combining the characteristics of the multi-subject game [45].

In order to reveal the drawbacks of behavioral and regulatory deficiencies under static reward and punishment mechanisms, the combination of evolutionary game and system dynamics is also introduced into the study, and its scenario simulation verifies the role of inhibiting the fluctuation of strategy evolution and optimizing the effectiveness of collaborative regulation [46]. The dynamic regulation strategy can effectively promote the stable evolution of the system [47]. In the study based on the retailer-led reverse supply chain system shows that the rational formulation of government reward and punishment mechanism can effectively promote the structural improvement of resource regeneration efficiency at the manufacturing side [48]. Meanwhile, the dynamic decision-making model based on behavioral science and evolutionary game theory is also introduced into the study to explore the role of dynamic incentive policy on the convergence of strategies of game subjects [49].

In summary, the study of the policy effect of government reward and punishment mechanism in reverse supply chain management has become the focus of academic circles. Most of the existing literature focuses on the field of power battery recycling, and analyzes the guiding role of policy tools on the behavior of different stakeholders from multiple dimensions. The results show that the dynamic subsidy mechanism based on the design of the extended producer responsibility (EPR) system can effectively improve the collaborative efficiency of the recycling network, while the differential setting of reward and punishment parameters significantly affects the stability and sustainability of the recycling channel. Although the existing research results have constructed an analytical model of “policy tools-recycling network-environmental benefits”, there are few papers that compare subsidies and rewards and punishments, and it remains to be further explored which intervention policy is more suitable for new construction wastes with both building materials and energy ambidextrous attributes, such as BIPV modules.

Based on the existing research results, the research compares and studies the impact of four incentive mechanisms, including no reward and punishment from the government, only reward, only punishment, and a combination of government rewards and punishments, on the recycling efficiency of the supply chain, considering the randomness of the recycling volume of retired PV modules and the construction sites of BIPV.

3. Establishment of an Optimization Model for the Recycling Path of Retired PV Panels

3.1. Problem Description

Smart city construction is reconstructing the operation paradigm of cities through digital technology integration. Building a smart energy system with multiple energy complementarities has become a key path to achieve low-carbon transformation, and BIPV technology has significantly improved the energy efficiency level of buildings through the spatial coupling of energy production and consumption. The construction of an efficient reverse logistics network for PV waste in urban agglomerations has become a key issue restricting the sustainable development of smart energy systems. The research focuses on the key links such as site selection and planning of reverse logistics nodes, transportation path optimization, and reward and punishment policies, aiming to solve the multiple game problems of economic cost, environmental efficiency and policy constraints in the process of PV waste recycling, provide decision support for the establishment of a “production-application-recycling” building energy life cycle management system, and help the construction of smart cities to evolve to a closed-loop ecological model of resources.

As presented in Figure 1, the schematic diagram of the operation mechanism of the reverse logistics network of decommissioned components includes construction points, recycling transfer stations, regional distribution centers and specialized processing terminals to form a collaborative system. The decommissioned components implement standardized pre-processing processes and temporary storage scheduling schemes at regional transit nodes, and improve the level of space utilization through the integration of facility functions. Establish an evaluation and classification mechanism in the core processing link of the distribution center: cascade utilization of components with functional potential and directional transfer to reuse plants. For completely failed components, the material dismantling process is entered.

Based on the intelligent algorithm, the dynamic optimization of transportation resource allocation is realized, and the intelligent diversion mechanism of resource circulation path is formed at key nodes. The focus is to build a reverse logistics network optimization model that adapts to the characteristics of BIPV retired PV modules, solve the collaborative optimization problems of multi-level facility site selection, transportation path planning and resource management decision-making, and form a management system that supports the sustainable development of energy infrastructure in smart cities.

3.2. Model Assumptions

BIPV becomes a distributed energy supply network unit through the organic integration of energy conversion units and building envelopes. This technology not only gives the building skin the ability to generate electricity, but also forms a unique logistics transmission model: Firstly, energy transmission and material circulation show spatial synergy. Secondly, there is a time series difference between the decommissioning cycle of PV modules and the service life of buildings. In addition, the distributed layout characteristics lead to significant spatial discretization of the recycling nodes. In view of the above-mentioned multi-dimensional system characteristics, the required basic parameters are presented in

Table 1. Basic parameters.

	Meaning
dij	The transportation distance from the construction point to the transit station
tcij	The cost per unit of transportation from the construction point to the transfer station
sij	The volume of transportation from the construction point to the transfer station
bij	Carbon emission factor per unit of transportation from the building point to the transfer station
DCi	The unit cost of labor that needs to be paid for the dismantling process
PTCj	The cost of operating the unit of the transfer station
ETk	The amount of carbon emissions per unit generated during the operation of the distribution center

.

The following basic assumptions are established when constructing the reverse logistics network model:

Hypothesis 1. 

In the context of smart cities, PV construction waste must be pre-treated by the nodes of the transfer station before entering the core treatment hub for classified disposal, and finally transferred to the recycling plant or dismantling center.

Hypothesis 2. 

The site selection scheme of the candidate transfer station and distribution center is determined by feasibility assessment, and its spatial layout meets the basic requirements of the multi-level processing system.

Hypothesis 3. 

All PV module waste dismantled at the construction site is removed for disposal.

Hypothesis 4. 

The location of the recycling plant and dismantling center in the waste reverse logistics network is known.

Hypothesis 5. 

The operating costs of the reuse plant are external costs and are not under the direct control of the recycling company.

The required decision variables are presented in

Table 2. Decision variables.

Decision Variables	Meaning
xj	Whether or not to choose to open a transit station in an alternate location
zk	Whether to choose to open a distribution center in an alternative location
yab	Whether it is transported between two points of AB (Note: A and B are two nodes in i, j, k, m, and n that are arbitrarily different)

.

3.3. Model Establishment

As presented in Figure 2, for the sake of understanding the analytical method and step-by-step logic presented in this paper, the flowchart of the analytical steps is as follows. This flowchart Outlines the core content of the research: Firstly, two objective functions integrating government incentives and penalties are established; Then define the various constraint conditions that the reverse logistics network needs to meet; On this basis, the multi-objective optimization of the reverse logistics network of retired photovoltaic panels is carried out by using MATLAB R2023b software. Finally, a sensitivity analysis is conducted on the optimization results to evaluate the robustness of the model. This figure clearly presents the complete analytical framework from model construction to result verification.

3.3.1. Cost Objective Function

Minimize the total cost of the logistics network with the goal of economic efficiency. The total cost of BIPV’s retired PV module reverse logistics network includes dismantling cost (DC), facility set-up cost (FC), facility operating cost (VC), transportation cost (TC), and government incentive and punishment cost f(w).

$$MinC=DC+FC+VC+TC+f(w)$$

(1)

On-site dismantling cost: The main consideration is the manual dismantling cost required when dismantling the PV module at the construction site. It is expressed as the product of the disassembly weight and the labor unit cost, as detailed in Equation (2).

$$DC={\displaystyle \sum _{\mathrm{i}=1}^I{\displaystyle \sum _{\mathrm{j}=1}^{J}D}}{C}_{\mathrm{i}}{s}_{\mathrm{ij}}$$

(2)

Among them, i represent the construction point, j represent the transfer station. Additionally, DC_i is the unit cost of labor that needs to be paid for the dismantling process; s_ij is the volume of transportation from the transfer station to the distribution center.

Facility opening costs: mainly the site construction costs and equipment purchase costs in the early stage of transfer stations and distribution centers. This part is only related to the number of transfer stations and distribution centers. See Equation (3) for details of fixed costs.

$$FC={\displaystyle \sum _{j\in J}{F}_j}{x}_j+{\displaystyle \sum _{k\in K}{F}_k}{z}_k$$

(3)

Among them, k represent the distribution center, and F_j is the fixed cost of opening a transfer station at location j; Similarly, F_k is the fixed cost of opening a transfer station at location k.

Facility operating costs: mainly the equipment operating costs of transfer stations, distribution centers, and dismantling centers. The Facility Operating Cost (TC) is expressed as the product of the unit operating cost at each point and the actual amount processed, as detailed in Equation (4).

$$VC={\displaystyle \sum _{i\in I}{\displaystyle \sum _{j\in J}P}}T{C}_j{s}_{ij}{x}_j+{\displaystyle \sum _{j\in J}{\displaystyle \sum _{k\in K}PR{C}_K{s}_{jk}{z}_k}}+{\displaystyle \sum _{m\in M}{\displaystyle \sum _{k\in K}P}}M{C}_m{s}_{km}{z}_K$$

(4)

Among them, PTC_j, PRC_k, and PMC_m respectively represent the operating cost per unit at the transfer station, the unit operating cost for sorting and packing at the distribution center, and the unit operating cost at the dismantling center. s_jk represents the transportation volume from the transfer station to the distribution center, and s_km denotes the transportation volume from the distribution center to the dismantling center. Other parameters are defined earlier.

Shipping Costs: BIPV retired PV modules have the characteristics of regional dismantling and remanufacturing due to the use of composite materials and modular structure. In view of the characteristics of its application in the urban environment, combined with the spatial relationship between the distributed PV network and the density of the urban road network, the research determines that road transportation is the optimal transfer mode. The transportation cost mainly considers the fuel consumption cost and the distribution distance of the transport vehicle in the process of transporting the goods [50], and does not consider the additional influence of complex road conditions on the transportation process, and the transportation cost (TC) is expressed as the product of the transportation cost per unit distance and the transportation distance, see Equation (5) for details.

$$\begin{array}{c}TC={\displaystyle \sum _{i\in I}{\displaystyle \sum _{\mathrm{i}\in j}{y}_{ij}}}{s}_{ij}{d}_{ij}t{c}_{ij}+{\displaystyle \sum _{k\in K}{\displaystyle \sum _{j\in J}{y}_{jk}}}{s}_{jk}{d}_{jk}t{c}_{jk}+{\displaystyle \sum _{m\in M}{\displaystyle \sum _{k\in K}{y}_{km}}}{s}_{km}{d}_{km}t{c}_{km}\\ +{\displaystyle \sum _{n\in N}{\displaystyle \sum _{k\in K}{y}_{kn}}}{s}_{kn}{d}_{kn}t{c}_{kn}\end{array}$$

(5)

Among them, n represent the reuse plant and d_ij denotes the distance from the construction point to the transfer station, d_jk indicates the distance from the transfer station to the distribution center, and d_km represents the distance from the distribution center to the dismantling center. Additionally, d_kn signifies the distance from the hub to the recycling plant. Furthermore, the research define the transportation costs as follows: tc_ij is the cost per unit of transportation from the construction point to the transfer station, tc_jk is the cost per unit of transportation from the transfer station to the distribution center, tc_km represents the unit transportation cost from the hub to the dismantling center, and tc_kn indicates the unit transportation cost from the hub to the reuse plant.

Cost of Government Incentives and Punishments: In the research, the government’s two-way incentive policy is considered to punish the distribution center with a low recycling and sorting rate, and reward the reverse [51]. If the sorting rate of the distribution center reaches the limit of the reward point, it means that the degree of recovery of resources by the distribution center has reached the expectation of the government, so the government will reward the distribution center accordingly; If the sorting rate of the distribution center reaches the penalty point boundary, it means that the distribution center does not fully sort the retired PV materials and thus causes a waste of resources, so the penalty fee needs to be paid, and the reward and punishment cost is presented in Equation (6).

$$f(w)=\left\{\begin{array}{l}{a}_2(R_2-w)Q_{\mathrm{i}}\\ 0\\ a_1(w-R_1)Q_{\mathrm{i}}\end{array}\right.\begin{array}{c}w>R_2\\ R_1\le w\le R_2\\ w<R_1\end{array}$$

(6)

Among them, the government specifies a recovery penalty factor a₁ and a reward factor a₂. Additionally, the government sets the lower limits for the rewards and penalties associated with the recycling sorting rate, denoted as R₁ and R₂. Furthermore, Q_i represents the amount of scrap that appears at point i, while w indicates the recycling sorting rate.

3.3.2. Carbon Emission Target Function

Based on the life cycle carbon footprint analysis, the reverse logistics network of BIPV decommissioned PV modules is integrated into smart energy system optimization, with a carbon emission minimization objective function. By quantifying the carbon emissions of transportation, processing and other links in the recovery path, combined with the government’s carbon tax incentive and pollution punishment mechanism, the carbon collaborative control of the new building energy system and the reverse logistics network can be realized.

In the process of constructing the carbon emission minimization objective function, there are two main parts of carbon emissions that can be calculated. Part of it is the carbon emission generated by fuel consumption in the process of vehicle transportation, and its emission intensity is closely related to the transportation distance, load factor and fuel type, and this part of the carbon emission is defined as direct carbon emissions (DCE), and the calculation process is presented in Equation (7); At the facility operation level, the carbon emissions of the distribution center and the dismantling center are derived from the operating energy consumption of the treatment equipment, and are dynamically calculated according to the power parameters of different types of machinery, which is defined as indirect carbon emissions (ICE) in the research, and the calculation process is presented in Equation (8).

$$\begin{array}{c}DCE={\displaystyle \sum _{i\in I}{\displaystyle \sum _{j\in J}{y}_{ij}}}{s}_{ij}{d}_{ij}{b}_{ij}+{\displaystyle \sum _{k\in K}{\displaystyle \sum _{j\in j}{y}_{jk}}}{s}_{jk}{d}_{jk}{b}_{jk}+{\displaystyle \sum _{m\in M}{\displaystyle \sum _{k\in K}{y}_{km}}}{s}_{kn}{d}_{km}{b}_{km}\\ +{\displaystyle \sum _{n\in N}{\displaystyle \sum _{k\in K}{y}_{kn}}}{s}_{kn}{d}_{kn}{b}_{kn}\end{array}$$

(7)

$$ICE={\displaystyle \sum _{j\in J}{\displaystyle \sum _{k\in K}{s}_{jk}E{T}_k{y}_{jk}}}+{\displaystyle \sum _{k\in K}{\displaystyle \sum _{m\in D}{s}_{km}}}E{T}_m{y}_{km}$$

(8)

Among them, b_ij represents the carbon emission factor per unit of transportation from the construction point to the transfer station. Similarly, b_jk denotes the carbon emission factor for transportation from the transfer station to the distribution center, while b_km indicates the carbon emission factor for transportation from the distribution center to the dismantling center. Additionally, b_kn reflects the carbon emission factor per unit of transportation from the hub to the reuse plant. Moreover, E_Tk signifies the amount of carbon emissions generated per unit during the operation of the distribution center, and E_Tm represents the amount of carbon emissions produced per unit during the operation of the dismantling center.

Combining the above two parts of carbon emissions, the carbon emission target function can be expressed as Equation (9).

$$\begin{array}{l}MinE=\min (DCE+ICE)=({\displaystyle \sum _{i\in I}{\displaystyle \sum _{j\in J}{y}_{ij}}}{s}_{ij}{d}_{ij}{b}_{ij}+\\ {\displaystyle \sum _{k\in K}{\displaystyle \sum _{j\in j}{y}_{jk}}}{s}_{jk}{d}_{jk}{b}_{jk}+{\displaystyle \sum _{m\in M}{\displaystyle \sum _{k\in K}{y}_{km}}}{s}_{kn}{d}_{km}{b}_{km}+\\ {\displaystyle \sum _{n\in N}{\displaystyle \sum _{k\in K}{y}_{kn}}}{s}_{kn}{d}_{kn}{b}_{kn})+({\displaystyle \sum _{j\in J}{\displaystyle \sum _{k\in K}{s}_{jk}E{T}_k{y}_{jk}}}+{\displaystyle \sum _{k\in K}{\displaystyle \sum _{m\in D}{s}_{km}}}E{T}_m{y}_{km})\end{array}$$

(9)

The constraints of the model are presented in Equation (10), which constrain the model from multiple perspectives. First of all, it ensured the balance of the inflow and outflow of the transfer station and the distribution center to avoid the situation of material accumulation or shortage. Secondly, the maximum processing capacity of the transfer stations and distribution centers was constrained to prevent them from operating under overload, ensuring the stability and reliability of the model. Secondly, the number limit of operational transfer stations and distribution centers has been clearly defined to rationally allocate resources. In addition, to prevent unselected transfer stations from operating and accepting retired PV modules from other distribution centers, regulations have been made on the quantity of retired PV modules recovered by transfer stations and distribution centers, ensuring the orderly flow of reverse logistics. Meanwhile, considering the processing capacity of the distribution center, constraints were imposed on its recycling and sorting rate to give full play to its processing level. Due to the saturation of each node, the model also imposes restrictions on the transportation capacity of each facility node to ensure the smoothness of reverse logistics transportation. Finally, the operation of the model is inseparable from the values of the decision variables. Therefore, this model also imposes constraints on the range of the decision variables to ensure the feasibility and validity of the model.

S.t.

$$\left\{\begin{array}{l}{\displaystyle \sum _{i\in I}{s}_{ij}}{y}_{ij}{x}_j={\displaystyle \sum _{k\in K}{s}_{ij}}{y}_{jk} \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\forall j\in J\\ {\displaystyle \sum _{j\in I}{s}_{ij}}{y}_{ij}{z}_k={\displaystyle \sum _{m\in M}{s}_{km}}{y}_{km}+{\displaystyle \sum _{n\in N}{s}_{kn}}{y}_{kn}\hspace{1em} \forall k\in K\\ {\displaystyle \sum _{j\in J}{s}_{jk}}{z}_k{{\displaystyle w}}_k={\displaystyle \sum _{n\in N}{s}_{kn}}{y}_{kn} \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\forall k\in K\\ {\displaystyle \sum s_{ij}}\le p_j{x}_j \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \hspace{1em}\hspace{1em}\hspace{1em}\forall j\in J\\ {\displaystyle \sum s_{jk}}\le p_k{z}_k\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \hspace{1em}\hspace{1em}\hspace{1em}\forall k\in K\\ {\displaystyle \sum s_{ij}}{x}_j\ge x_j \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \hspace{1em}\hspace{1em}\hspace{1em}\forall j\in J\\ {\displaystyle \sum s_{jk}}{z}_k\ge x_j \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} \hspace{1em}\hspace{1em}\hspace{1em}\forall k\in K\\ q_1\le {\displaystyle \sum _{j\in J}{x}_j}\le q_2,q_3\le {\displaystyle \sum _{k\in K}{z}_k}\le q_4\\ x_i,z_k,y_{ij},y_{ik},y_{km},y_{kn}\in (0,1)\end{array}\right.$$

(10)

Among them, the minimum number of transfer stations is represented by q₁, while the maximum number is denoted as q₂. Similarly, the minimum number of distribution centers is indicated by q₃, and the maximum number is represented by q₄. Other parameters have been explained in the previous sections and will not be elaborated upon here to maintain clarity and focus.

4. BIPV Simulation Analysis of Retired PV Module Recycling

4.1. Initial Parameters

Based on the background of smart city energy system, the research focuses on the spatial optimization of the reverse logistics network of BIPV decommissioned PV panels. Firstly, a square planning area with a side length of 100 units is constructed to simulate the geospatial framework of the urban building energy system, which is characterized by a typical urban form of central centralization and peripheral expansion [52]. For the planning area, it is necessary to solve the contradiction between the uncertainty of the spatiotemporal distribution of PV retired modules and the efficiency of the reverse logistics network.

Two spatial coordinate point generation methods are adopted: First, a two-dimensional normal distribution model with μ = 50 and δ = 10 is used to simulate the spatial agglomeration effect of PV buildings in the urban core due to the market mechanism. Second, based on the circular constrained random distribution model with the center O of the circle (50,50) and the radius R = 50, the planning characteristics of uniform diffusion in the emerging urban area are characterized, and the spatial parameter design of the two models is in line with the distribution law of real building PV. On this basis, the normal distribution function N(20,2²) is integrated to predict the recycling amount at the building point, reflecting the dynamic fluctuation in the recycling volume of decommissioned components. A dataset of 100 sample points is generated using Monte Carlo simulation, but due to space limitations, only the spatial coordinate parameters and recovery prediction parameters of 20 samples are presented in the research, as presented in

Table 3. Generate point coordinates and yield data.

Serial Number	Abscissa	Ordinate	Recycling/t	Serial Number	Abscissa	Ordinate	Recycling/t
1	57	62	18	11	63	51	19
2	63	53	24	12	47	46	22
3	52	58	22	13	56	38	23
4	52	49	18	14	44	33	19
5	35	53	17	15	59	44	21
6	62	53	22	16	74	31	20
7	62	59	20	17	36	36	26
8	37	53	19	18	58	61	16
9	57	56	22	19	48	65	20
10	56	54	19	20	39	44	22

. By accurately simulating the randomness and regularity of the recycling network, it provides decision support for the resource circulation system of smart cities.

Based on the objective optimization framework of the research, the location of the transfer station and the distribution center is carried out, and the transportation cost, service coverage and carbon emission constraints are comprehensively considered. The research identifies 5 alternative transfer stations and 3 alternative distribution centers, with their numbers determined by logistics data cluster analysis and the regional demand coverage model. The number of transfer stations reflects the density gradient of urban administrative zoning and PV buildings, with 2, 2, and 1 nodes in high, medium, and low density areas, respectively, and the number of distribution centers according to the threshold of economies of scale in waste treatment [53]. By improving the NSGA-II algorithm, the adaptive crossover probability and elite retention strategy are introduced to solve the optimal solution of the Pareto front of the facility combination. The parameters of the alternative transfer station and distribution center are detailed in

Table 4. Alternate select the transfer-related parameters.

Node	Abscissa	Ordinate	Transit Limit/t	Opening Cost/10,000 yuan	Unit Operating Cost/(10,000 yuan/t)
Transit Station 1	66	69	3800	1180	25
Transit Station 2	73	57	4500	1500	32
Transit Station 3	78	67	2700	865	48
Transit Station 4	56	70	3200	950	65
Transit Station 5	52	84	4300	135	53

and

Table 5. Parameters related to alternative hubs.

Node	Abscissa	Ordinate	Sorting Upper Limit/t	Opening Cost/10,000 yuan	Unit Operating Cost/(10,000 yuan/t)	Carbon Emissions per Unit (kg/t)	Recycling Sorting Rate
Distribution center 1	69	54	121,500	2850	90	15	0.8
Distribution center 2	77	63	9500	2570	70	12	0.75
Distribution center 3	58	54	8700	2330	54	11	0.6

. The existing dismantling center and reuse plant implement a strategy of sharing idle facility capacity to save on investment costs for new facilities. This means that no additional construction is required, and the operating parameters are detailed in

Table 6. Other relevant parameter values.

Node	Abscissa	Ordinate	Unit Operating Cost/(10,000 yuan/t)	Carbon Emissions per Unit (kg/t)
Dismantling center	79	65	20	3.58
Recycling plant	74	75	/	/

.

In the dismantling stage of BIPV decommissioned PV modules, since multiple processes and different carbon sources are involved, it is appropriate to choose the carbon emission factor method for calculation [54]. In terms of carbon emission databases, domestic scholars have also conducted a lot of research [55], in order to reasonably calculate the carbon emissions of the recycling process of retired PV modules, the research selects the carbon emission factor data in the “Building Carbon Emission Calculation Standard” (GB/T51366-2019) [56], and conducts a comprehensive analysis based on the existing domestic research results, and lists the carbon emission factors that may be involved, as presented in

Table 7. List of relevant carbon emission factors [56].

Carbon Emission Sources	Carbon Emission Factor
The average emission factor of the national power grid	0.5703 kg CO2/(kWh)
Provincial Grid Average Emission Factor (Chongqing)	0.1031 kg CO2/(kWh)
diesel fuel	3.096 kg CO2/kg
gasoline	2.925 kg CO2/kg
Trucks (unit mass 3 t)	73.45 kg CO2/shift
Crane (lifting mass 10 t)	100.51 kg CO2/shift
Rebar cutting machine (diameter 40 mm)	18.62 kg CO2/shift
Spot welding machine (capacity 50 kv·A)	59.87 kg CO2/shift
Diesel generator set (power 60 kw)	226.63 kg CO2/shift

.

4.2. Analysis of Results

The research constructs a new “transfer station-distribution center” address model for the BIPV retired PV module recycling network, by improving the NSGA-II algorithm to dynamically adjust congestion parameters and utilizing the MATLAB R2023b platform to achieve objective optimization. In this research, two different strategies for generating coordinates for building points are compared: The first is clustering based on two-dimensional normal distribution, which can effectively capture the distribution characteristics of urban BIPV clusters. The second is a uniform distribution model with a preset service radius, which ensures the accessibility of distributed PV modules. The results show that compared with the traditional scattered layout, the “1 + 1” site selection mode has significant advantages in transportation cost and processing efficiency, and its two-node coordination mechanism can effectively cope with the tide of BIPV decommissioning. The “1 + 1” site selection model embeds PV recycling nodes into the smart city digital twin system, realizes the whole process scheduling of retired modules from dismantling to recycling, and provides key infrastructure support for the closed-loop of the new power system. The following is a detailed analysis of the experimental results.

4.2.1. Normal Distribution Simulates the Effect of Urban Agglomeration

Based on the framework of smart city construction, the probability model is used to characterize the spatial agglomeration characteristics of the retirement of PV modules in urban areas, and the geographical distribution of urban PV module retirement is simulated by constructing an N(50,10²) distribution model. The improved NSGA-II algorithm is used to simulate the scenario of the reverse logistics network, and only a set of Pareto optimal solution sets are presented due to space limitations, as presented in Figure 3, where Figure 4 shows the specific location of the transfer station and the distribution center.

The data results show that when only one transfer station and distribution center are set up in the corresponding area, the synergistic optimal solution can be realized, the carbon emission is reduced to 76,171.8 tons compared with the traditional model, and the economic cost is controlled at 344 million yuan. The combination improves the efficiency of reverse logistics through dynamic path optimization, and verifies the spatial decision-making value of “1 + 1” layout in building PV recycling network.

4.2.2. Random Distribution in the Preset Circle to Simulate the Urban Agglomeration Effect

A circular study area, centered on geographic coordinates (50,50) with a radius of 50 units, is set to simulate the optimal layout of the BIPV decommissioning and recycling network in smart city construction. Figure 5 shows the optimal solution set of Pareto, and the corresponding transfer station and distribution center location scheme are presented in Figure 6.

The data results show that in the reverse logistics network planning, the “1 + 1” collaborative configuration of the transfer station and the distribution center can achieve the optimal solution of the system, the total carbon emission is reduced to 88,010.2 tons, and the total cost is optimized to 269 million yuan. The dynamic path al gorithm is used to schedule the transportation network, which significantly improves the operational efficiency of the reverse logistics system, and proves the decision-making effectiveness of the “1 + 1” dual-node spatial layout strategy in the BIPV recycling system.

In this research, two different spatial coordinate generation methods are compared: spatial clustering based on probability density can effectively characterize the agglomeration characteristics of urban BIPV, while the grid layout model with service radius constraints ensures the service coverage balance of distributed nodes. Simulation results show that compared with the traditional empirical discrete layout, the “1 + 1” two-node collaboration mode between the transfer station and the distribution center shows significant advantages in terms of transportation cost and carbon emissions. The model embeds recycling nodes into the smart city infrastructure management system, forming a regulation mechanism from component dismantling to resource recycling. Although there is heterogeneity between the two spatial layout strategies, the optimization results finally show the same trend, which verifies the spatial adaptability and decision-making robustness of the site selection model, and provides a replicable construction paradigm for the low-carbon transformation of smart city infrastructure.

4.3. Parameter Sensitivity Analysis

The government’s regulation and control measures act on the decision-making behavior of recycling enterprises through the leverage of economic interests, and its core impact is in two dimensions: node site selection optimization and logistics path reconstruction. A multi-scenario dynamic simulation model is constructed for the “1 + 1” dual-core network site selection model of the transfer station and distribution center in the BIPV retired PV module recycling system to reveal the mechanism of policy parameters. In terms of spatial modeling, a two-modal comparative analysis method is adopted: one is based on the two-dimensional normal distribution to describe the urban agglomeration effect, and its spatial coordinate parameters meet the distribution characteristics of N (50,10²), the mean value is set as the geographic center coordinate of the city, and the standard deviation is set according to the scale of the built-up area; Second, a uniform random distribution model under circular constraints is constructed, and the influence of geographical radius limitation on network toughness is verified by Monte Carlo simulation.

A scenario simulation model is constructed to analyze the sensitivity of government reward and punishment parameters, examine the two-way mechanism of penalty coefficients, quantify their economic sensitivity, and verify the resilience of reverse logistics through carbon footprint. This analysis establishes a dynamic policy regulation threshold for the PV recycling system, enabling the optimal alignment of the spatial flow of construction waste and renewable resources. To investigate the most effective reward and punishment methods, the sensitivity analysis is categorized into the four scenarios detailed below.

4.3.1. There Are No Rewards and Punishments from the Government

Firstly, the benchmark model of zero policy intervention is constructed, and when the reward and punishment parameters a1 = a2 = 0 are set, the benchmark data of the reverse logistics system are presented in

Table 8. Data related to logistics networks in the absence of government incentives and punishments.

Experimental Group	a1	a2	Cost/RMB	Carbon Emissions/t
(a)	0	0	344,887,000	71,470.9
(b)	0	0	344,411,000	76,089.9

. The experimental group (a) shows the normal spatial distribution of building nodes, and the (b) group adopts the circular random configuration pattern, and the data distribution is presented in Figure 7.

In the process of dynamic adjustment of policy parameters, the follow-up research will use the benchmark value as a reference system to systematically evaluate the heterogeneous impact mechanism of differentiated policy combinations on the carbon emission intensity and total operating cost of logistics networks through multi-dimensional policy scenario simulation.

4.3.2. The Government Only Punishes the Situation

Under the smart city management framework that only implements the penalty policy, the government does not reward the distribution center that overcompletes the task, but punishes the distribution center that does not complete the recycling requirements, that is, when the recycling penalty system a1 < 0 and the recycling reward coefficient a2 = 0, the relevant data of the reverse logistics network are obtained as presented in

Table 9. Logistics network data table in penalty case only.

Experimental Group	a1	a2	Cost/RMB	Carbon Emissions/t	Experimental Group	a1	a2	Cost/RMB	Carbon Emissions/t
(a)	−2	0	344,201,000	71,605.6	(e)	−2	0	344,184,000	75,449.8
(b)	−4	0	344,172,000	70,513.0	(f)	−4	0	344,149,000	75,319.1
(c)	−6	0	344,253,000	70,936.6	(g)	−6	0	344,329,000	75,507.5
(d)	−8	0	344,428,000	71,798.2	(h)	−8	0	344,408,000	75,719.3

. Based on the comparative study of the generation mechanism of spatial distribution, the layout of building points in the experimental groups (a)–(d) showed normal diffusion characteristics, and the (e)–(h) groups conformed to the random distribution law of circular area, which can be quantitatively analyzed by Figure 8 and Figure 9.

According to the analysis of Figure 7 and Figure 8, the results show that when the value of the recovery penalty coefficient a1 is relatively small, that is, in the stage of low-intensity policy regulation, the recovery rate is increased through the innovation of sorting technology, forming a synergistic effect of economic cost reduction and environmental benefit gain. When the regulatory parameters exceed the threshold, the marginal benefit of technological innovation decreases significantly, resulting in the increase in penalty costs exceeding the benefits of environmental governance, and triggering the simultaneous increase in total operating costs and carbon emission intensity. However, this evolution path is not related to the initial spatial distribution model, and the normal layout and random distribution system show the convergence response characteristics of environmental and economic indicators in the parameter domain.

4.3.3. The Government Only Awards the Situation

Under the smart city management framework that only implements incentive policies, the government rewards distribution centers that exceed the completion of tasks, and does not penalize distribution centers that fail to complete the recycling requirements, that is, when the recycling penalty coefficient a1 = 0 and the recycling reward coefficient a2 > 0, the relevant data of the BIPV decommissioned PV panel reverse logistics network are obtained, as presented in

Table 10. Only in the case of incentives, data related to the logistics network.

Experimental Group	a1	a2	Cost/RMB	Carbon Emissions/t	Experimental Group	a1	a2	Cost/RMB	Carbon Emissions/t
(a)	0	2	344,050,000	70,797.0	(e)	0	2	344,100,000	75,488.3
(b)	0	4	344,036,000	70,676.7	(f)	0	4	344,087,000	75,300.6
(c)	0	6	344,032,000	70,469.7	(g)	0	6	344,088,000	75,815.6
(d)	0	8	344,013,000	70,792.2	(h)	0	8	344,089,000	75,993.7

. Based on the comparative analysis of the spatial coordinate generation mechanism, the layout of building points in groups (a)–(d) presents a normal spatial distribution pattern, while the (e)–(h) group adopts a uniform random distribution strategy in a circular area, and its characteristics can be quantified by Figure 10 and Figure 11.

The data show that in the process of increasing the intensity of financial incentives, the total cost of the logistics system gradually shows a steady-state characteristic, while the carbon emission index experiences a nonlinear response relationship of first decreasing and then rising. This phenomenon reveals that the behavior pattern of enterprises obtaining subsidies by expanding the scale of reverse logistics can reduce unit costs in the short term, but it will lead to a rebound effect of carbon footprint and a diminishing marginal benefit of fiscal expenditure. Although there are differences in the spatial distribution mechanism of initial construction sites, the two types of systems show the convergence of policy intervention effects.

4.3.4. A Combination of Government Rewards and Punishments

Under the smart city management framework of the implementation of the policy of combining rewards and punishments, the government will reward those who exceed the quota and punish the distribution centers that fail to complete the recycling requirements. That is, when the recycling penalty coefficient is a1 < 0 and the recycling reward coefficient is a2 > 0, the relevant data of the reverse logistics network of retired PV panels are obtained, as presented in

Table 11. Data related to the logistics network in the case of reward combination.

Experimental Group	a1	a2	Cost/RMB	Carbon Emissions/t	Experimental Group	a1	a2	Cost/RMB	Carbon Emissions/t
(a)	−2	2	344,204,000	71,056.9	(e)	−2	2	344,234,000	76,171.8
(b)	−4	4	344,182,000	70,859.6	(f)	−4	4	344,193,000	75,445.0
(c)	−6	6	344,241,000	70,951.0	(g)	−6	6	344,252,000	75,517.2
(d)	−8	8	344,365,000	71,393.9	(h)	−8	8	344,362,000	75,656.7

. The data of groups (a)–(d) conform to the normal distribution law of building points, and the data of groups (e)–(h) follow the principle of uniform random distribution of building points in a circular area, and their characteristics can be quantitatively characterized by Figure 12 and Figure 13.

With the increase of the regulation intensity of the reward and punishment mechanism, the carbon emissions show a gradual convergence trend, while the total cost shows a nonlinear characteristics of first decreasing and then increasing. Within the adjustment range of reward and punishment parameters a1 and a2, the comprehensive cost of the system is always significantly lower than the benchmark level in the state of zero intervention (a1 = 0, a2 = 0).

In view of the particularity of BIPV module recycling, this research constructs four policy scenarios of government no intervention, single incentive, single constraint and mixed incentive constraint, explores the impact of government incentive and punishment mechanism on carbon emission intensity and total cost of logistics network, and reveals the dynamic relationship between policy tools and environmental economic system. The results show that the total cost and carbon emissions of the logistics network are always better than those in the baseline scenario without policy intervention (a1 = 0, a2 = 0) under the effect of the reward and punishment mechanism. In the process of dynamic adjustment of policy reward and punishment parameters, although there are differences in the normal layout and random distribution of the initial spatial generation mechanism of the system, the carbon emission intensity and total cost indicators show convergent characteristics. This conclusion provides important implications for policymakers: in the process of reverse logistics network optimization, priority should be given to the parameter calibration of the reward and punishment mechanism, rather than over-investing in the initial planning of the spatial layout.

5. Conclusions

In order to solve the problem of the management of the retirement tide of BIPV photovoltaic panels in the construction of smart city energy system, this research uses the improved NSGA-II algorithm to integrate the urban node grading system (building point-transfer station-distribution center-reuse factory/dismantling center) and the government’s reward and punishment policy to construct a reverse logistics network optimization model, and systematically demonstrates the decision-making value of resource processing point location and government reward and punishment mechanism on resource allocation and environmental governance in complex urban networks. It is found that the reasonable setting of resource treatment points and government reward and punishment thresholds can effectively regulate the behavior of stakeholders, and form an optimal balance between the spatiotemporal transfer of PV construction waste and the control of environmental externalities. The research transforms traditional empirical decision-making into a scientific paradigm based on goal optimization, providing an economically feasible and ecologically sustainable solution for energy infrastructure decommissioning in smart city construction, and laying a foundation for the “energy-PV-environment” closed-loop smart city resource circulation system. The main conclusions of this research are as follows:

(1) Considering the significant diversity of urban building layouts, this research compares two distinct geospatial analysis models: the urban center agglomeration model (simulating spatial autocorrelation in built-up areas) and the regional equilibrium distribution model (evaluating coverage efficiency under radiation radius constraints). This comparison reveals a fundamental challenge for traditional node layouts, the high-density concentration in the agglomeration model risks overwhelming collection and overburdening distant facilities, while the dispersed equilibrium model can suffer from inefficient coverage and low facility utilization. To address this spatial heterogeneity inherent in BIPV decommissioning, we propose a novel reverse logistics site selection model based on a “1 + 1” dual-core linkage system between Transfer Stations and Distribution Centers. Simulation results demonstrate the clear superiority of the “1 + 1” dual-core approach over traditional models in both operating costs and environmental benefits. Crucially, the Transfer Station acts as a localized aggregation hub, to minimize initial collection distances, enable rapid consolidation of small batches, and provide preliminary sorting. Conversely, the Distribution Center functions as a centralized processing and dispatch hub, often near transportation arteries or end-processing facilities, optimizing transportation economies, deep sorting, resource allocation, and interfacing with government reward-penalty mechanisms. The power of the “1 + 1” model lies in the synergistic linkage between Transfer Station and Distribution Center. This structural synergy significantly reduces per-unit logistics costs, directly translating to lower operational costs and environmental impact. A highly significant finding is the convergence of facility spatial layouts after iterative multi-objective optimization using the improved NSGA-II algorithm, despite starting from the two divergent initial geographic strategies (agglomeration vs. equilibrium). This convergence is not incidental, it fundamentally demonstrates that the “Transfer Station-Distribution Center “ structure emerges as the optimal or near-optimal configuration under the economic and environmental objectives. This robust convergence powerfully validates the inherent technical superiority and general applicability of the “1 + 1” dual-core system. Therefore, the “1 + 1” site selection model possesses profound engineering application value, particularly for the large-scale decommissioning of BIPV modules characterized by distributed, building-attached sources and potential surge retirements. It provides a scientifically grounded, adaptable blueprint for planners to determine the optimal location and scale of T and DC facilities. This ensures the economically viable and environmentally sound management of end-of-life building-integrated PV resources, enabling efficient recovery and regeneration to feed back into the construction sector. Consequently, the model offers critical decision support for establishing a true closed-loop “resource production—energy consumption—material regeneration” system within the built environment. (2) This research, through the “1 + 1 dual-core collaborative management”mechanism, dynamically couples the government’s reward and punishment regulation with the operational efficiency of sorting facilities, significantly enhancing the comprehensive benefits of the system. Its internal mechanism is manifested in two core levels: Firstly, the nonlinear reward and punishment parameters reconstruct the economic decision-making boundary. Stepwise subsidies encourage sorting centers to optimize equipment load rates, while government penalties force dismantling centers to adopt low-carbon transportation solutions, directly reducing the space for illegal disposal. Secondly, the initial differences of different coordinate generation methods in the spatial dimension show a converging trend under policy intervention. The internalization of the cost of transportation pollution drives the reorganization of logistics routes, and inefficient nodes are eliminated through the parameter sensitivity screening mechanism. Through the iterative optimization of reward and punishment parameters, the operating cost of the system has been reduced from the benchmark state of 344,887,000 yuan to a minimum of 344,013,000 yuan (a decrease of 0.25%). The total carbon emissions were compressed from 76,171.8 tons to 70,469.7 tons (a reduction of 7.5%), prompting the network layout to contract towards the Pareto front.

The digital governance framework thus constructed has the ability to migrate to smart cities. Its universality stems from the collaborative design of the agent response model (the transmission of policy signals—economic incentives—behavioral changes) and the adaptive iterative module of the reward and punishment coefficient based on real-time carbon emission data of the Internet of Things algorithm interface. This framework transforms the traditional end-of-pipe treatment into a full-process closed-loop regulation and control, providing an extensible decision-making paradigm for the urban solid waste system. Although this study provides a theoretical foundation and practical insights for optimizing the reverse supply chain of BIPV (Building-Integrated Photovoltaics), further expansion of research dimensions is necessary to meet the dynamic development needs of smart cities. First, existing models remain limited in analyzing stochastic variables such as demand volatility and facility elasticity coefficients. Future research could incorporate stochastic programming methods to enhance the network’s adaptability to market fluctuations. Second, regarding spatial optimization, while current generation methods based on randomized layouts can capture the scenario-specific characteristics of building points, they lack deep integration with urban digital twin systems. In conclusion, promoting the transformation of reverse supply chain optimization from static planning to dynamic governance will help promote the organic integration of building PV formation recycling system and smart city infrastructure, as well as the evolution of smart city resource recycling system in the direction of intelligence and refinement.