Comparing the Use of Ant Colony Optimization and Genetic Algorithms to Organize Kitting Systems Within Green Supply Chain Management Practices

Abstract

As product diversity continues to expand in today’s market, there is an increasing demand from customers for unique and varied items. Meeting these demands necessitates the transfer of different sub-product components to the production line, even within the same manufacturing process. Lean manufacturing has addressed these challenges through the development of kitting systems that streamline the handling of diverse components. However, to ensure that these systems contribute to sustainable practices, it is crucial to design and implement them with environmental considerations in mind. The optimization of warehouse layouts and kitting preparation areas is essential for achieving sustainable and efficient logistics. To this end, we propose a comprehensive study aimed at developing the optimal layout, that is, creating warehouse layouts and kitting preparation zones that minimize waste, reduce energy consumption, and improve the flow of materials. The problem of warehouse location assignment is classified as NP-hard, and the complexity increases significantly when both storage and kitting layouts are considered simultaneously. This study aims to address this challenge by employing the genetic algorithm (GA) and Ant Colony Optimization (ACO) methods to design a system that minimizes energy consumption. Through the implementation of genetic algorithms (GAs), a 24% improvement was observed. This enhancement was achieved by simultaneously optimizing both the warehouse layout and the kitting area, demonstrating the effectiveness of integrated operational strategies. This substantial reduction not only contributes to lower operational costs but also aligns with sustainability goals, highlighting the importance of efficient material handling practices in modern logistics operations. This article provides a significant contribution to the field of sustainable logistics by addressing the vital role of kitting systems within green supply chain management practices. By aligning logistics operations with sustainability goals, this study not only offers practical insights but also advances the broader conversation around environmentally conscious supply chain practices.

1. Introduction

Green supply chain management (GSCM) has become a key strategy for incorporating environmental sustainability into supply chain functions. With organizations facing increasing regulatory scrutiny and consumers demanding greener practices, GSCM offers a comprehensive framework to minimize the environmental impact of logistics and production. This concept encompasses various methods of waste reduction, resource conservation, and operational efficiency improvement. It can range anywhere from obtaining supplies from sustainable resources to using eco-friendly methods of transportation and proper waste disposal. The implementation of GSCM will allow firms to not only comply with environmental laws but also to enhance their competitive position and image within the market [1,2].

GSCM also encourages the involvement of all stakeholders in the supply chain to create a transparent and mutually accountable system for sustainability outcomes [3]. Through the use of new technologies and techniques, supply chains can become stronger and not only fulfill their present functions but also help the environment and society [4,5]. To sum it all up, GSCM is a revolutionary method that combines economic objectives with ecological and social concerns, providing a pathway for corporations to engage in “green” business that is not only good for them but good for the Earth. Green logistics is not just about minimizing the environmental impact of logistics operations but includes economic and social concerns in order to achieve sustainable development. The main components of green logistics are route optimization, using energy-efficient vehicles and technologies, using eco-friendly packaging, and supporting material recycling [3]. Additionally, regulatory forces and the consumer preference for “green” products have forced firms to engage in greener logistics practices as part of their corporate social responsibility endeavors [6,7]. While green logistics is a very difficult move for many companies, it also opens up many doors. Although it may be expensive to initially invest in sustainable technologies, in the long run these costs typically pay off through increased efficiency, decreased waste, and an improved brand image [8]. In this intricate web that companies must weave through, there will be a need for innovative ideas and cooperative work in order to successfully promote green logistics. To sum it up, green logistics is an imperative transition in supply chain management because it combines environmental sustainability and economic feasibility. It will be imperative for organizations hoping to survive in a resource-starved world to have a thorough knowledge of the strategies, technologies, and frameworks associated with green logistics as research and practice in this area continue to expand [9].

Green supply chain management (GSCM) represents a confluence of sustainable methodologies and conventional supply chain paradigms; its primary objective is to mitigate environmental impacts while concurrently maintaining economic sustainability. Green storage management, which is a crucial component of GSCM, focuses on diminishing energy consumption, augmenting spatial efficiency, and reducing waste in warehousing and inventory systems. The implementation of effective green storage strategies is not only advantageous but also essential for fulfilling sustainability goals. However, as global supply chains encounter increasing regulatory pressures and consumer demands for eco-friendly practices [10], the importance of these methodologies becomes ever more pronounced [11].

Metaheuristic optimization methodologies, particularly Ant Colony Optimization (ACO) and genetic algorithms (GAs), have been extensively utilized to address complex challenges within logistics and supply chain management, including initiatives aimed at sustainable storage solutions. ACO, which originates its principles from the foraging behaviors observed in ants, exhibits significant efficacy in tackling routing and network optimization challenges, employing pheromone-based heuristics [3,12]. However, GAs emulate the process of natural selection, utilizing crossover, mutation, and selection operators to discern the optimal—or near-optimal—solutions for problems such as storage layout and inventory management [13].

Recent academic inquiries underscore that the selection between these algorithms is contingent upon the particular objectives and constraints intrinsic to the storage system. For instance, hybrid models that synthesize ACO and GAs have exhibited enhanced performance in reconciling energy efficiency with operational costs [14]. Although their effectiveness has been validated, further exploration is necessary to thoroughly understand how these algorithms function in real-time contexts and with diverse sustainability goals. Advancements in computational power, coupled with algorithmic innovations, are anticipated to augment their applicability to intricate green storage systems in forthcoming years [9].

This article provides a significant contribution to the field of sustainable logistics by addressing the vital role of kitting systems within green supply chain management practices. As businesses adapt to the increasing demand for diverse and customized products, this study emphasizes the importance of optimizing warehouse layouts and kitting preparation areas to reduce environmental impacts and enhance operational efficiency. Through the application of genetic algorithms (GAs) and Ant Colony Optimization (ACO), the research achieves a notable 24% reduction in energy consumption by optimizing warehouse and kitting layouts simultaneously. The focus on kitting systems is particularly impactful, as it underscores their ability to streamline material handling processes and reduce inefficiencies. By aligning logistics operations with sustainability goals, this study not only offers practical insights but also advances the broader conversation around environmentally conscious supply chain practices [3].

2. Literature Review

Misra and Chakraborty (2024) explored the application of Ant Colony Optimization (ACO) algorithms in supply chain management, focusing on how these algorithms can be enhanced to create more efficient systems. The authors emphasize ACO’s potential to optimize supply chains through distributed optimization methods, highlighting its role in improving efficiency while also reducing operational costs and effectively utilizing resources. They argue that ACO algorithms provide significant contributions to sustainable practices in logistics and production, offering practical solutions that align with environmental sustainability goals [15].

Zhang and Yang (2024) proposes an adaptive ACO to enhance dynamic scheduling in logistics. The research demonstrates that this method effectively reduces transportation costs and improves customer satisfaction by optimizing routing decisions in real-time [16].

Nogareda et al. (2020) investigated a multi-attribute vehicle routing problem that considers time window constraints, a heterogeneous fleet, and multiple depots. To achieve an optimal balance among conflicting objectives in logistics optimization, they proposed a hybrid approach integrating Ant Colony Optimization (ACO) and Genetic Algorithms (GA). Their findings indicate that this combined method surpasses the performance of standalone ACO or GA models in solving complex vehicle routing challenges [17]. Lei et al. (2024) introduced a green supply chain optimization framework that applies a two-stage heuristic algorithm. This framework is designed to align with the priorities of intermediary core enterprises, ensuring the seamless integration of upstream procurement and transportation processes with downstream distribution and logistics operations [18].

Kumar et al. (2023) investigated the performance of ACO and GAs across various aspects of supply chain management. The study emphasize that, while GAs excel in more stationary environments with structured constraints, ACO’s adaptability in dynamic situations offers superior sustainability benefits. The findings indicate that the choice between ACO and GAs ultimately depends on specific objectives and the context of the supply chain [19]. Ren et al. (2024) conducted an optimization model for designing the sea cucumber supply chain [20]. Li et al. (2023) noted that alliances between manufacturers and logistics providers generate shared resources and improved environmental outcomes [21]. This is consistent with the work of Govindan and Bouzon (2024), who emphasized that synergy is key to reaching sustainability targets [22]. Sadeghi and Niaki (2024) introduced a mathematical decision-making model that integrates processes to mitigate environmental impact. By employing a multi-criteria decision-making approach, the study systematically analyzes and compares the effectiveness of the proposed algorithms based on six distinct evaluation metrics [14].

Abbasian et al. (2023) formulated a hybrid optimization approach aimed at constructing a resilient and sustainable supply chain for the perishable food sector. Their method highlights the effectiveness of integrating multiple metaheuristic algorithms to tackle intricate supply chain issues [23]. Asha et al. (2022) conducted a comprehensive analysis of existing research on the use of multi-objective optimization methods in GSCM. Their study highlights essential components of green supply chain structures, examines modeling techniques that accommodate multiple objectives concurrently, and discusses various solution strategies for tackling multi-objective optimization challenges in this field [24].

Sánchez-Flores et al. (2020) provided a comprehensive analysis of existing research on SSCM within emerging economies. The analysis revealed a growing interest in SSCM; however, research focusing specifically on emerging economies remains limited compared to developed countries [1].

Huang et al. (2024) propose a novel two-phase PSO-based Model Predictive Control (MPC) algorithm to address the problem comprehensively. The effectiveness of the proposed algorithm is further evaluated through an empirical study using data from East China [25].

Zhang et al. (2023) presented a novel approach to sustainable supply chain network design by developing a mixed-integer programming model that aims to minimize establishment costs and environmental pollution emissions, while maximizing labor opportunities. The study propose the Chaotic Particle Ant Colony Algorithm (PSCACO) to address premature convergence issues commonly found in particle swarm algorithms [26].

Liu et al. (2023) presented an enhanced Ant Colony Optimization (ACO) algorithm designed to optimize logistics distribution with a focus on reducing carbon emissions is presented by integrating the Internet of Things (IoT), the proposed method improves route planning and resource allocation in logistics networks [27].

To sum up, the literature presents a multidimensional view of green supply chains and logistics, emphasizing the incorporation of technology, agility, regulatory support, circular economy ideologies, cooperation, and performance metrics. While future research should continue to explore these themes, it must not lose sight of the fact that supply chains are dynamic entities that exist and evolve in a regulatory and consumer world that is constantly changing.

Revanna et al. (2022) presented a detailed analysis of the proposed model, demonstrating its superiority over traditional methods through simulations and case studies. The results highlight the potential of hybrid algorithms in tackling complex routing problems, offering practical insights for logistics and supply chain management [28].

The MIT Sustainable Supply Chain Lab highlighted the current state of supply chain sustainability, stressing the need for algorithms like the ACO and GA methods to enhance sustainability efforts and tackle challenges such as scope 3 emissions. They note several applications of these algorithms in developing greener supply chains and propose directions for future research [29].

One major work in this area is the work of Chen et al. (2022) [30], who underlined that digital technologies are the force of change in green supply chain management practices. They stated that blockchain and the Internet of Things (IoT) will create more transparency and efficiency, leading to less environmental impact. This finding is corroborated by Gao et al. [5], who explored how artificial intelligence (AI) could be used to streamline logistics.

Sellito et al. (2024) investigated to evaluate the influence of GSCM dimensions for an agrifood supply chain, specifically focusing on the rice industry in southern Brazil. The findings reveal that green innovation and green operations significantly enhance competitiveness, while green strategy has a minimal impact [31]. Wang and Wang (2023) pointed out that agility is the key to supply chains and that the more resilient the supply chain is, the more sustainable it can be [32]. This is also backed up by the research of Seman et al. (2012) as flexible supply chains not only recover efficiently from disruption but are also more likely to incorporate “green” practices [4].

The concept of the circular economy has started to gain traction in supply chain conversations. Zhou et al. (2021) [17] suggested incorporating the circular economy into logistics for better resource utilization and less waste.

Masruroh et al. (2024) focused on the optimization of warehouse layout and kitting preparation zones, considering two issues of focus related to sustainability: energy consumption and operational efficiency in terms of material flow. It adopts the genetic algorithm and Ant Colony Optimization heuristic methods. This approach aims to enhance lean manufacturing by tackling the NP-hard problem of layout optimization for both storage and kitting areas, with a strong emphasis on environmental considerations [33].

For the first time in this domain, this present study gives prime importance to environmental sustainability in warehouse and kitting optimization problems by trying to reduce energy resource consumption. This aligns with the manufacturing industry’s current demand for green logistics, combining efficiency with ecological sensitivity.

Sustainability has begun to be incorporated into warehouse optimization, particularly in logistics, with a focus on carbon emission considerations. This approach focuses on the simultaneous optimization of warehouse layouts and kitting preparation zones, considering these areas to be interlinked in lean manufacturing. These synergies can improve the overall efficiency of the flows. Most studies either address warehouse optimization or kitting optimization, but rarely both together. A considerable amount of research related to warehouse design and layout has been reviewed, focusing on storage assignment and picking systems, but not kitting systems, such as the research of Gu et al. (2007) [34].

The combination of the GA method with ACO is new; the GA method performs well when the solution space is immense, while ACO fine-tunes path-based problems, such as in logistics, for minimizing material handling distances and reducing energy consumption.

The GA method has been widely applied in warehouse layout and material handling optimization, while ACO is rare in the related literature. ACO has been more commonly applied to vehicle routing problems, where path optimization is critical. ACO for minimizing material flow paths is innovative and not usually seen in kitting and warehouse layouts. This combination of the GA method for structural optimization and ACO for dynamic path optimization could provide more efficient solutions than applying either algorithm independently.

The optimization of the warehouse layout and kitting preparation zones, considering two issues of focus related to sustainability (energy consumption and operational efficiency in terms of material flow), constitutes the focus of this study. It adopts the GA and ACO methods. In such a way, this approach aims at enhancing lean manufacturing by taking into consideration the NP-hard problem of the layout optimization of both storage and kitting areas, with full attention being paid to environmental considerations.

Sustainability Emphasis—Warehouse and Kitting Optimization: For the first time, prime importance is given to environmental sustainability in warehouse and kitting optimization problems by trying to reduce energy resource consumption by reducing the forklift travel distance from goods acceptance to warehouse and from warehouse to kitting area simultaneously. This meets the contemporary demands of the manufacturing industry for green logistics, combining aspects of efficiency with ecological sensitivity.

This approach emphasizes the importance of simultaneously optimizing warehouse layouts and kitting preparation zones, recognizing these areas in lean manufacturing as interlinked. Because of these reasons, synergies would be drawn on to lower the overall efficiency of the flows.

Most studies either deal with warehouse optimization or kitting optimization, seldom together. For instance, Gu et al. (2007) reviewed a considerable amount of research related to the problems of warehouse design and layout. This work considered storage assignment and picking systems, while it did not take kitting systems into account [34].

Lei et al. (2024) introduced a hybrid framework that integrates ACO with brainstorming optimization techniques to enhance the efficiency of green storage within supply chains. That study demonstrated that the approach markedly reduces carbon emissions and transportation costs. Furthermore, the research elucidated the adaptability of the hybrid model in responding to dynamic green storage requirements. Although it exhibits superiority when compared to traditional GA methods, it particularly excels in real-time optimization contexts (due to its innovative design). However, one should take into account the possible limitations that such a hybrid model may impose [18].

Pakdel, He, and Pakdel (2024) concentrated on a multi-objective methodology utilizing NSGA-II (a variant of genetic algorithms) to enhance storage strategies for perishable goods within green supply chains. Their investigation revealed that this algorithm efficiently minimizes greenhouse gas emissions, while also balancing cost and efficiency; it surpasses particle swarm optimization techniques in both speed and accuracy. However, the authors underscored its practicality in conditions of uncertain demand, because such environments often complicate strategic decision-making. Although the results are promising, the applicability of the findings may vary depending on specific contextual factors [35].

Kaoud et al. (2024) investigated a bi-objective optimization model that employs both ACO and GAs within the context of green closed-loop supply chains, especially concerning storage and transportation. Their findings illuminated ACO’s efficacy in dynamically adapting to varying storage requirements; however, the GA method demonstrated its superiority in yielding swift computational outcomes in less complex situations. That study emphasized the inherent trade-offs that exist between efficiency and the minimization of carbon footprints. Although the results are promising, they also raise questions about scalability and long-term applicability [36].

The GA method has already gained wide applications in warehouse layout and material handling optimization. However, ACO can rarely be seen in the related literature. ACO has been more commonly applied to vehicle routing problems, where path optimization is critical. This work explicitly acknowledges the complexity and size of the given problem and hence focuses on heuristics for near-optimal solutions within reasonable time. The related literature is as follows:

Golmohammadi (2024) developed a model using mixed-integer linear programming that minimizes supply chain costs’ total and the overall negative environmental impacts. A multi-objective dragonfly algorithm (MODA) is used to solve this model. The results are compared with those obtained using the Non-Dominated Sorting Genetic Algorithm and the Epsilon Constraint Method (EPC) [37]. De Koster et al. (2007) identified the characteristics of storage assignment to be NP-hard in nature, but most solutions isolate either warehouse design or picking system optimization [38].

The simultaneous optimization approach of this study, along with the use of GA and ACO heuristics, test this multi-faceted complexity in a more comprehensive manner.

This directly relates to how a contribution was made to the optimization of the warehouse and kitting systems for the goal of avoiding waste, the basic principle of lean manufacturing. Another important contribution has been integrating into the lean process the problem of environmental sustainability.

• Most logistics optimization projects are based on the principles of lean manufacturing. However, most research, such as Womack and Jones (1996), addresses the issue strictly from a production waste minimization perspective rather than that of logistics [7].
• This study fits well within the literature in which green manufacturing is integrated with lean manufacturing—a field of study that is still in its developmental stage.

The novelty of this study lies in its combined focus on warehouse and kitting system optimization problems using heuristic methods, with a particular emphasis on sustainability. To the authors’ knowledge, very few studies have been conducted in this field using ACO. Moreover, considering the growing emphasis on environmental sustainability, this research holds significant relevance to the fields of green logistics and sustainable lean manufacturing. While the importance of these themes continues to rise, they remain relatively underexplored in the context of logistics optimization. By addressing critical research gaps in these areas, this study provides a meaningful contribution to the optimization of lean and green manufacturing systems, offering insights that are valuable both theoretically and in practical applications.

3. Materials and Methods

3.1. Ant Colony Optimization (ACO)

Ant Colony Optimization (ACO) represents a metaheuristic inspired by natural phenomena, specifically emulating the intricate behavior of ants as they seek the optimal pathways between their colonies and food sources. Initially posited by Marco Dorigo during the early 1990s, ACO has found extensive application in a variety of combinatorial optimization challenges, including the traveling salesman problem, vehicle routing, and network design [22]. The core principle underpinning ACO hinges on the manner in which ants deposit pheromones along the paths they navigate. These pheromones serve as a medium of indirect communication, thus directing fellow ants towards more efficient routes. In a computational framework, a colony of artificial ants engages in a collective search for solutions within a designated problem space. Each ant progressively constructs a solution predicated on probabilistic choices that are influenced by both pheromone concentrations and heuristic data [39]. The algorithm functions iteratively, encompassing several primary phases. Initialization occurs when the pheromone trails are set with uniform values; however, the subsequent processes depend on the dynamic interplay of pheromone deposition and evaporation, which adjusts the trail strengths over time, allowing the system to converge toward optimal solutions.

Solution Construction: Each ant constructs a viable solution by navigating the problem space, a process that is governed by a transition probability formula that accounts for both pheromone intensity and heuristic desirability [22]. Pheromone Update: Following the evaluation of solutions, the pheromones undergo an update process. The paths that demonstrate success receive elevated pheromone levels, thereby enhancing their appeal; however, pheromone evaporation serves to thwart premature convergence by diminishing trail intensity over time [6]. Optimization Cycle: These procedural steps are reiterated until a termination criterion is satisfied, whether that be a predetermined number of iterations or convergence to an acceptable solution [12]. The strengths of ACO reside in its remarkable adaptability, positive feedback mechanism, and distributed computational architecture, making it particularly effective for tackling intricate optimization challenges. However, its performance can be significantly impacted by the parameter configurations, such as the rate of pheromone evaporation and the equilibrium between exploration and exploitation [12].

The applications of Ant Colony Optimization (ACO) are remarkably diverse; they encompass areas such as logistics optimization (which includes routing and scheduling), network traffic management, and even the selection of features in machine learning. Recent advancements, however, have seen ACO being integrated with other methodologies: for instance, genetic algorithms and machine learning techniques. This integration aims to enhance its efficiency—an objective that has gained considerable attention in contemporary research [6]. Although there are challenges inherent in this process, the potential benefits are substantial.

3.2. Genetic Algorithms

Genetic algorithms (GAs) represent sophisticated computational frameworks, fundamentally drawing from the tenets of natural selection and genetic inheritance—a concept first articulated by John Holland during the 1970s [13]. As members of the broader category of evolutionary algorithms, GAs endeavor to address optimization conundrums by emulating the processes of biological evolution. They exhibit notable efficacy in resolving intricate NP-hard challenges across diverse domains, spanning from engineering design to artificial intelligence [39]. The algorithm begins by generating an initial population, where each entity (referred to as a chromosome) represents a potential solution to the given problem. These chromosomes are systematically encoded as strings, frequently utilizing a binary format. A fitness function serves to assess the quality of each chromosome, thereby indicating its effectiveness in addressing the problem at hand. Based on the fitness scores, a selection mechanism determines which individuals are poised to contribute to the subsequent generation [40]. This iterative process not only enhances the probability of finding optimal solutions but also underscores the dynamic nature of evolution itself.

The fundamental operations inherent in genetic algorithms (GAs) encompass selection, crossover, and mutation. Selection serves to ensure that the individuals exhibiting greater fitness possess an augmented probability of transmitting their advantageous traits to subsequent generations. Crossover, alternatively referred to as recombination, amalgamates the segments from two parental chromosomes to engender offspring, thereby fostering diversity and the exploration of the solution landscape. Mutation, on the other hand, introduces stochastic alterations to individual genes, which are essential for thwarting premature convergence and sustaining genetic diversity [41]. This iterative process persists as the population undergoes evolution across generations. With every iteration, the algorithm converges toward an optimal or near-optimal solution. By judiciously balancing exploration and exploitation, GAs adeptly traverse extensive and intricate search domains, thereby circumventing the pitfalls associated with local optima [42]. GAs have been effectively employed in a plethora of real-world challenges, including, but not limited to, machine learning, scheduling, routing, and resource optimization. Their inherent adaptability and robustness render them an indispensable instrument in the realm of computational problem-solving [43].

4. Implementation via Genetic Algorithms and Ant Colony Optimization

4.1. Problem Definition

In lean manufacturing systems, kitting systems are frequently employed on assembly lines. Inspired by the Toyota executives who observed supermarkets in the United States, kitting has become a dominant approach with the growing variety of products. As product diversity expands, there is no longer sufficient space around production lines for packaging. Operators are also relieved of non-value-adding tasks such as selecting materials and walking to the packaging area. Other advantages include preventing forklifts from entering the production area and freeing the assembly line from packaging clutter.

However, establishing a dedicated kitting area becomes essential. Consequently, the first level of internal logistics flow involves two separate transport processes: the first from the receipt of material to the warehouse and the second from the warehouse to the kitting area. Therefore, when planning both the warehouse layout and the kitting area, the primary objective is to minimize the forklift travel distances. While the warehouse layout is an NP-hard problem, addressing it concurrently with the kitting area layout represents an even more challenging optimization problem.

The layout of the assembly workshop is shown in Figure 1.

In the layout of the assembly workshop shown, trucks unload their packaging at the truck unloading area. After the material acceptance, the packaging is transported by forklift to the relevant warehouse. Forklifts can operate in the green areas, while they are prohibited from entering the grey zone. Once the packaging is emptied in the kitting area, the forklift retrieves the corresponding product from the warehouse and transports it to the kitting area.

In kitting, operators fill the kitting carts according to the production sequence. Each kitting cart corresponds to a specific product on the assembly line. The kitting trolley is transported to the assembly line by an Automated Guided Vehicle (AGV). Each product can only be placed in one designated location in both the warehouse and the kitting area.

The average hourly rate of subcomponent parts are shown as

Table 1. The average hourly rate of subcomponent parts.

Subcomponent	Quantity in Packages	Average Consumption per Hour	Subcomponent	Quantity in Packages	Average Consumption per Hour	Subcomponent	Quantity in Packages	Average Consumption per Hour
X1001	8	60	X1031	19	24	X1061	10	15
X1002	19	12	X1032	9	20	X1062	6	18
X1003	9	10	X1033	8	14	X1063	6	12
X1004	8	12	X1034	10	20	X1064	5	20
X1005	15	60	X1035	6	10	X1065	16	8
X1006	10	15	X1036	6	5	X1066	18	14
X1007	6	12	X1037	5	10	X1067	9	8
X1008	6	10	X1038	16	12	X1068	14	7
X1009	5	15	X1039	18	10	X1069	19	5
X1010	16	10	X1040	6	20	X1070	9	16
X1011	18	15	X1041	9	7	X1071	13	10
X1012	6	40	X1042	14	16	X1072	19	7
X1013	9	10	X1043	19	4	X1073	6	18
X1014	14	5	X1044	9	14	X1074	11	8
X1015	19	10	X1045	13	30	X1075	7	18
X1016	9	12	X1046	20	20	X1076	8	8
X1017	13	20	X1047	19	16	X1077	7	7
X1018	20	40	X1048	6	10	X1078	16	18
X1019	19	5	X1049	11	7	X1079	10	8
X1020	6	12	X1050	7	10	X1080	20	8
X1021	11	6	X1051	8	7	X1081	15	8
X1022	18	60	X1052	7	16	X1082	19	8
X1023	7	12	X1053	16	10	X1083	19	10
X1024	8	10	X1054	10	7	X1084	9	5
X1025	7	5	X1055	20	7	X1085	8	12
X1026	16	12	X1056	15	7	X1086	10	10
X1027	10	6	X1057	19	7	X1087	6	14
X1028	20	6	X1058	19	12	X1088	6	14
X1029	15	6	X1059	9	5	X1089	5	15
X1030	19	6	X1060	8	13	X1090	16	14

.

The average hourly rate of subcomponent parts are shown as

Table 2. The average hourly rate of subcomponent parts.

Subcomponent	Quantity in Packages	Average Consumption per Hour	Subcomponent	Quantity in Packages	Average Consumption per Hour	Subcomponent	Quantity in Packages	Average Consumption per Hour
X1091	18	21	X1121	8	6	X1151	11	11
X1092	9	9	X1122	7	9	X1152	10	5
X1093	14	8	X1123	16	6	X1153	20	5
X1094	19	7	X1124	10	5	X1154	15	11
X1095	9	8	X1125	20	5	X1155	19	5
X1096	19	8	X1126	15	5	X1156	10	6
X1097	6	14	X1127	19	5	X1157	20	6
X1098	11	9	X1128	9	10	X1158	19	6
X1099	7	14	X1129	6	7	X1159	11	15
X1100	8	9	X1130	16	10	X1160	11	15
X1101	7	8	X1131	9	12	X1161	11	10
X1102	16	14	X1132	14	15	X1162	11	20
X1103	10	9	X1133	19	6	X1163	30	5
X1104	20	9	X1134	19	15	X1164	30	6
X1105	15	9	X1135	11	10	X1165	30	7
X1106	19	9	X1136	8	12	X1166	30	8
X1107	19	2	X1137	7	15	X1167	30	9
X1108	9	10	X1138	10	10	X1168	30	10
X1109	8	9	X1139	20	10	X1169	30	11
X1110	6	6	X1140	15	10	X1170	30	4
X1111	6	12	X1141	19	10	X1171	19	10
X1112	16	6	X1142	9	8	X1172	19	20
X1113	9	6	X1143	19	9	X1173	19	12
X1114	14	9	X1144	11	4	X1174	19	18
X1115	19	8	X1145	8	8	X1175	22	20
X1116	9	10	X1146	10	4	X1176	22	10
X1117	19	9	X1147	20	4	X1177	22	30
X1118	6	6	X1148	15	4	X1178	17	20
X1119	11	5	X1149	19	4	X1179	17	10
X1120	7	6	X1150	19	11	X1180	17	30

.

4.2. GA Implementation via Python

The algorithm for the GA method is shown in Algorithm 1.

Algorithm 1. GA Algorithm Steps

1. Initialize parameters (population size, mutation rate, crossover rate, etc.) 2. Generate initial population of random solutions (assignments of items to storage locations). 3. For each generation: 4. Evaluate the fitness of each individual in the population. 5. Select parents based on their fitness using a selection method (e.g., roulette wheel). 6. For each pair of parents: 7. Perform crossover to create offspring (recombine the solutions of parents). 8. If the crossover rate is met: o Create offspring; otherwise, retain parents. 9. Mutate offspring with a given mutation rate (randomly change one or more item-location assignments). 10. Replace the old population with the new population. 11. Return the best solution after all generations.

A flowchart of the genetic algorithm method is shown in Figure 2.

4.3. ACO Implementation via Python

The algorithm for ACO is shown in Algorithm 2.

Algorithm 2. ACO Algorithm Steps

1. Initialize parameters (number of ants, iterations, pheromone matrix, heuristic matrix, etc.) 2. Initialize pheromone levels to a constant value (e.g., 1). 3. For each iteration: 4. For each ant: 5. Build a solution by selecting a storage location for each item based on pheromone levels and heuristic information. 6. Calculate the fitness (quality) of the solution. 7. If the fitness of the solution is better than the best solution so far: o Update the best solution. 8. Add the solution’s pheromone to the pheromone matrix. 9. Evaporate pheromones by reducing their levels according to a given evaporation rate. 10. Update pheromone levels based on the best solutions found by the ants. 11. Return the best solution after all iterations.

A flowchart of ACO is shown in Figure 3.

4.4. Results

The problem was addressed using a program developed in Python, with the solutions generated independently through the genetic algorithm and Ant Colony Optimization methods. When the programs were run on the computer, it was observed that the genetic algorithm method performed better in terms of both speed and finding a more optimal solution compared to Ant Colony Optimization.

The system specifications used in solving the problem were as follows:

Processor: Intel i5

RAM: 16 GB

Operating system: Windows 10 Pro

The parameters used for Ant Colony Optimization in solving the problem were as follows:

Node: 180

Ant count: 50

Alpha: 1

Beta: 5

Evaporation rate: 0.5

Q: 100

Iteration: 100

Optimal solution: 29,224.17

The solution was found using Ant Colony Optimization in 5 min and 57 s.

The parameters used for the genetic algorithm method in solving the problem were as follows:

Node: 180

Population size: 100

Mutant rate: 0.01

Generation: 100

Optimal solution: 28,470.64

The solution was found using the genetic algorithm method in 11.8 s.

The initial average path length for the electric forklift was 37,600 m per hour. Through the implementation of genetic algorithms (GAs), this value was successfully reduced to an average of 28,470 m per hour, resulting in a significant 24% improvement. This enhancement was achieved by simultaneously optimizing both the warehouse layout and the kitting area, demonstrating the effectiveness of integrated operational strategies.

5. Discussion

Companies increasingly prefer electric forklifts over diesel counterparts as part of their commitment to green logistics. In this analysis, a typical electric forklift consumes approximately 1.7 kWh per kilometer, which is relevant for mid-tonnage models. Each hour, a reduction of 9130 (37600–28470) meters in trajectory is realized by this approach. If this value is multiplied by the energy consumption (17 kWh/km), the optimization efforts in this study yield a notable energy saving of 15.13 kWh per hour of operation. When extrapolated over the course of one year, this translates to a total energy saving of 99,858 kWh.

This substantial reduction not only contributes to lower operational costs but also aligns with sustainability goals, highlighting the importance of efficient material handling practices in modern logistics operations.

6. Conclusions

The GSCM practices in the kitting system was addressed using a Python-based program, and the solutions were independently generated through the application of the genetic algorithm (GA) and Ant Colony Optimization (ACO) techniques. Upon execution of the programs, it was observed that the genetic algorithm method demonstrated a superior performance compared to Ant Colony Optimization in terms of both speed and the ability to identify more optimal solutions.

Initially, the average travel distance for an electric forklift was 37,600 m per hour. By applying genetic algorithms, this value was successfully reduced to an average of 28,470 m per hour, resulting in a substantial 24% improvement. This enhancement was achieved through the simultaneous optimization of both the warehouse layout and kitting layout, thereby illustrating the effectiveness of integrated operational strategies.

In the context of this analysis, a typical electric forklift consumes approximately 1.7 kWh per kilometer, which is significant for medium-ton models. This optimization led to a reduction of 9130 m (37,600–28,470) in travel distance per hour, which, when multiplied by the energy consumption rate of 17 kWh/km, results in a considerable energy saving of 15.13 kWh per hour. When extrapolated over the course of a year, this equates to a total energy saving of 99,858 kWh.

Logistics, frequently regarded as a primary contributor to carbon emissions, must evolve toward green logistics through the adoption of innovative strategies. In this context, artificial intelligence applications can support decision-makers in making tactical decisions. The genetic algorithm and Artificial Ant Colony Algorithm, frequently employed in solving NP-hard problems, have been considered and applied in this area.

Kitting systems and warehouse layout design present complex optimization challenges. Recently, the application of genetic algorithms (GAs) has gained significant attention for addressing these challenges, as GAs are particularly suited for solving large-scale, NP-hard problems involving numerous variables.

Warehouse optimization primarily focuses on efficient storage location assignments, material handling, and minimizing the travel distances of forklifts. Artificial intelligence techniques, such as GAs, are employed to determine the optimal travel distances for forklifts. By optimizing storage layouts, these methods reduce the time and energy consumed during the movements between storage locations and production areas.

Studies have shown that genetic algorithms are especially effective in terms of performance and computational efficiency, making them a highly appropriate method for such optimization problems. Using this technique, a 24% improvement is reached.

One notable advantage of genetic algorithms is their relatively faster computation time compared to Ant Colony Optimization (ACO). Genetic algorithms can produce results in approximately 11 s, whereas ACO may require up to 6 min to reach a solution. These factors make genetic algorithms a more suitable choice for addressing this problem. In line with lean principles, transportation—considered a non-value-adding activity—can thus be carried out with minimal energy consumption. Further research could involve developing an AGV scenario within the created model that minimizes the energy consumption of AGVs.