Ports are primarily used for maritime activities such as cruising, maneuvering, and hoteling, which facilitate imports and exports. These activities are crucial for economic growth and development. However, when port activities are scaled up, they become hazardous to the environment and public health due to significant emissions. This exacerbates existing climate change and air quality issues. Despite the economic benefits, it is essential to consider the environmental impacts and strive to mitigate or reduce them to prevent harmful effects. To address this problem, machine learning tools and techniques can be employed to predict emissions and identify the predictors that influence them. The primary goal of this study is to forecast hydrocarbon (HC) emissions resulting from activities like cruising, maneuvering, and hoteling using machine learning models. By implementing these models, we can uncover emission patterns and develop strategies to reduce their impact. Among the models tested, the Random Forest model demonstrated the best performance, followed by Decision Trees. In contrast, our implementation of multiple linear regression models indicated overfitting of the data, making them less suitable for our purposes.
Accurate traffic forecasting has become essential due to increased urban deliveries driven by growing e-commerce and urban expansion. This surge has increased traffic in large cities, resulting in delays and numerous accidents. The convenience of online shopping and home delivery continues to fuel the e-commerce sector. As this sector grows, the complexity of its challenges increases, necessitating quicker solutions. These challenges often extend beyond the control of delivery companies when influenced by external elements such as traffic congestion or adverse weather conditions, particularly in the last-mile delivery. This issue is intensified in regions lacking extensive traffic sensor networks, such as less developed countries. This research contributes by (1) establishing a contextual groundwork for existing traffic prediction frameworks, (2) employing socialmedia and diverse traffic-related data (including weather conditions, significant locations, and event schedules) through social network analysis to refine traffic prediction accuracy, and (3) presenting a method applicable to partially observable traffic situations. The methodology integrates advanced deep learning techniques like Long Short-Term Memory Network. In addition, a future extension aims to develop a robust real-time traffic prediction system, Traffic GPT, utilizing deep learning and transformer methodologies, optimized for diverse urban environments with real-time updates through transfer learning. Key to this approach is the use of sentiment analysis supported by training on datasets like the IMDB sentiment dataset with techniques like AWD-LSTM's attention mechanisms, and Graph Convolutional Networks, along with tools for analyzing social media sentiment.
The port sector is evolving, influencing areas such as operations, infrastructure, governance, technology, partnerships, and commercial strategies. AI can significantly improve the efficiency of intermodal freight transportation by enhancing real-time operational processes and decision-making models. This research evaluates the resilience of machine learning models to various data and architecture threats, focusing on adversarial attacks and data corruption. The study explores a range of attack methods, including white-box attacks, data poisoning, and model evasion, and tests defensive strategies like adversarial training, resilient optimization, and anomaly detection on a developed DNN model. The findings highlight the vulnerabilities of current models and emphasize the need for robust security measures. The paper concludes with recommendations for enhancing model robustness and suggests directions for future research to bolster AI systems against evolving threats, ensuring their reliability and security in unpredictable environments.
The impact of emerging operations and marketing management trends on retail goods returns is significant. While customers are experiencing increasingly lenient return and purchase procedures, retailers face more difficult and expensive processes. The main objective of this paper is to examine the effects of new trends and the COVID-19 pandemic on supply chain expenses, customer behavior, and the pricing and refund strategies of retailers in managing product returns. This study also considers the significant economic consequences of product returns. To achieve this goal, an inclusive analytical model is developed that considers the endowment effect, customer heterogeneity, return leniency, and other widely investigated factors in the literature. The analytical framework and a series of numerical simulations with a wide range of parameter values that reflect practical situations are used to demonstrate the effects of increased supply chain costs on customer purchase and return behavior and the best pricing and refund strategies for retailers. The sensitivity analysis shows that higher supply chain costs may reduce market demand while increasing the optimum price and level of reimbursement. However, the frequency of returns may fluctuate up or down. The results also highlight the significance of the endowment effect in addressing the issue which reduces market demand while raising the optimum price and degree of reimbursement; return frequency, however, may fluctuate up or down. The results further show the importance of the endowment effect in resolving the problem.
With growing product proliferation and various demand effects, retailers face the challenge of finding an effective and efficient product assortment and shelf-space allocation. These two planning problems are interrelated, particularly when shelf space is limited. Therefore, it is essential to consider these planning problems jointly to maximize revenue. Moreover, many existing models consider shelf space as one-dimensional. However, since the dimensions of both shelves and products vary in reality, solutions obtained from these one-dimensional models are unable to cope with real-life situations. To approximate the actual shelf dimensions, this paper presents an integrated assortment and shelf-space allocation model for two-dimensional shelves. To solve the model, we proposed three heuristics and tested them on randomly generated data sets. Among these three heuristics, the mixed heuristic proves to be the best approach, efficiently yielding near-optimal solutions within a very short runtime. Comparisons between the heuristics and the optimization model demonstrate that the mixed heuristic is competitive with other heuristics and the optimization model, especially for large-scale problems. We expect that our heuristics will assist retailers in making decisions on product assortment and shelf-space allocation.
This paper presents a sustainability-oriented model of urban planning developed to support the decision-making process for regional urban planning. The methodology employed in this approach and solution utilizes mathematical modeling. The model encompasses various systems within a city, including land use, public space, population, housing, economic sectors, facilities, employment, and transportation, emphasizing their interrelationships. The primary objective of the model is to identify strategies that promote the development of a harmonious and economically sustainable city, while ensuring equitable distribution of space. By simulating the intricate functioning of the city and analyzing the interplay between different zones, the model provides valuable insights. The findings and conclusions derived from this model can serve as a reference and guide for municipal authorities in formulating regulations, incentives, and simulating specific scenarios. Furthermore, it offers valuable advice to the city council on achieving an optimized distribution of territory. Ultimately, this model helps raise governmental awareness of the manifold benefits associated with the implementation of an optimized spatial zoning plan, thereby fostering sustainable development in the region.
Microstores contribute to the economic welfare of the specific regions where they are positioned. Due to this fact, the government must seek economic and social programs to promote them. In this paper, we study a problem that aims to enhance the microstores' welfare through government intervention. We consider a government authority implementing nearshoring strategies by opening community stores to shorten supply times for microstores, thereby aiding their economy. On the other hand, microstore owners decide whether to source their products from the community stores or wholesalers. Due to the inherent hierarchy among decision-makers, a bilevel optimization framework is used to model the problem. Because of the specific characteristics of the bilevel problem, it can be reformulated as an equivalent single-level mixed-integer problem. This solution methodology is employed to address a case study from the city of Pachuca, Hidalgo, Mexico. The applicability of our approach is validated through computational experimentation. Additionally, a sensitivity analysis is conducted to provide interesting managerial insights.
The increasing complexity of supply chain (SC) networks and the associated risks have captured global attention, leading to the emergence of the concept of supply chain resilience (SCRES). Over the past two decades, SCRES has been a focal point of research, explored through various perspectives, approaches, and tools. Among these, the discrete event simulation (DES) technique stands out for its effectiveness in modeling SCRES. While DES models offer multiple advantages and have been widely used in the literature, they lack the capability to measure a crucial element of SCRES: the cumulative learning of a SC network as it experiences risk events over time. The absence of this attribute renders attempts to operationalize SCRES incomplete. This research aims to address this methodological gap by proposing–from a theoretical standpoint–the integration of artificial intelligence (AI) algorithms into DES models. The research delves into several categories of AI algorithms that can learn from successive iterations of DES models. Based on this exploratory analysis, it is suggested that neural networks, particularly backpropagation, Kolmogorov-Arnold, and reinforcement learning algorithms, are the most suitable to address this gap in the literature. Additionally, a novel definition of SCRES is proposed, emphasizing the importance of learning within supply chain networks.
Buffer zones for moveable shelves or pallets with intermediate products are ubiquitous in production systems to decouple production stages. These buffer zones, usually organized by humans in dense grids, pose a challenge for direct access to all unit loads. To access the retrieval unit load, it often becomes necessary to reshuffle blocking unit loads. When autonomous mobile robots supply production machines without human intervention, their responsibilities encompass both the reshuffling of buffer zones and the transportation of unit loads to production cells adhering to the sequence prescribed by the production plan. This paper introduces the buffer reshuffling and retrieval problem, which is a variant of the block relocation problem known from container terminals at shipyards. We propose an integer programming formulation designed for the static variant of the problem. We created an instance generator with customizable layout sizes and found optimal solutions to solve these instances.
This work aims to generate a planning model using knowledge representation techniques from artificial intelligence to plan the transfer routes and the sequence of visits to multiple points of interest for a tourist, addressing this as a Tourist Trip Design Problem. Our model considers user preferences as variables to be optimized or as constraints, which may include time, walking distances, and economic costs. To facilitate movement between different points of interest, we propose to use the public transportation network. The user specifies his starting point, his destination, and the points of interest he wants to visit, allowing for multi-day planning. The methodology used is an artificial intelligence planning model where actions, preferences, and spatial topology are modeled. Planning algorithms (planners) are used to solve the tourist trip design problem. A comparison of the results obtained by different planners is performed, demonstrating the effectiveness of using Artificial Intelligence Planning to solve such complex problems.
The efficiency of inbound supply chains can be increased by consolidating shipments from suppliers to manufacturing plants. We address a key challenge in this area involving a combined truck loading and supplier to plant assignment problem. Determining whether a truck can be loaded with a particular set of items is challenging and time consuming, thus we introduce a deep neural network (DNN) that predicts the feasibility of packing a truck load. We integrate the DNN into an existing heuristic framework to tackle the combined problem of item to truck assignment and subsequent truck loading. The DNN is used to ignore truck packing problems that are likely infeasible, saving runtime by avoiding unproductive computation steps. We evaluate our approach on a real-world problem provided by Renault and show that our learning-based approach finds better solutions faster than the existing state-of-the-art heuristic, resulting in an improvement of the objective function value by 3.83
Efficient management of financial resources is vital for the sustainable operation of banks, particularly in optimizing policy acquisition to ensure liquidity and enhance financial planning. This paper introduces an intelligent decision support system utilizing machine learning (ML) to predict policy acceptance, aiming to reduce acquisition costs, improve customer retention, and support small and medium-sized enterprises (SMEs) with more favorable loan terms. Our approach integrates the CRISP-DM methodology with Continuous Integration and Continuous Deployment (CI/CD) processes, leveraging agile Scrum practices to ensure iterative development and rapid deployment. Ensemble learning techniques (combining Neural Networks, Random Forest, and Support Vector Machines) are employed to achieve high predictive accuracy. The system's effectiveness is demonstrated through experiments using the Bank Marketing Dataset, with results validated by standard quality metrics. Additionally, a user-friendly dashboard and REST API have been developed to facilitate efficient client identification and model deployment. The V-model is applied to ensure rigorous testing and validation throughout the project lifecycle. This comprehensive approach enhances internal logistics, optimizes resource management, and supports SME growth, thus fostering economic development and job creation. Our work provides a robust framework for integrating ML solutions and agile methodologies in banking and similar sectors.
Gasoline distribution depends of a complex infrastructure which consists of refineries, storage depots, gasoline stations, roads and pipelines. In Mexico, recent research has supported the need to increase storage capability to ensure high service level in the event of disruption of pipeline distribution. The present work extends on this scenario by proposing alternatives focused on: (a) determining the balanced number of stations to be served by the storage depots to reduce their overload and thus, reduce the negative impact if such depots are closed, and (b) determining the outcomes of increasing the number of storage depots to reduce shortage and service distance. For this work, data regarding locations of gasoline depots and stations were considered. Also, due to the large size of this data, the adaptation of deterministic algorithms for assignment and location of facilities were considered. As result, it was determined that limited storage depots can affect the efficiency of distribution through tanker trucks. Thus, efforts should be aimed to increase the number of these facilities.
Efficient management of berth and quay crane assignment is critical for optimising operations at maritime container terminals. This paper addresses the integration of berth and quay crane assignment, recognising the essential role it plays in terminal efficiency. The deterministic nature of assigning quay cranes is challenged by uncertainties, primarily due to their potential breakdowns but also influenced by factors such as maintenance schedules and workforce availability. To address this complex problem, we formulate it as a mixed integer mathematical model, where the uncertainties appear on the right-hand side (RHS) of a block of constraints. This is a special stochastic formulation, which is much worth investigating. Due to the NP-hardness of the problem, we propose a matheuristic approach that combines variable neighbourhood search (VNS) and fixed set search (FSS) with mathematical optimisation of sub-problems with a standard solver. The approach aims to find robust solutions that account for uncertainties in the number of available quay cranes at each time period based on some possible scenarios. We conduct extensive experiments using two alternative algorithms. One is a method from the literature and the other is our methodology without mathematical optimisation. Results from testing on a set of randomly generated instances of varying sizes demonstrate the effectiveness of our proposed matheuristic. On average, our approach outperforms alternative methods, providing superior results in optimising berth and quay crane assignment at maritime container terminals under uncertain conditions.
In this research, the Inventory Routing Problem (IRP) with priority customers under a green measure to reduce emissions is studied. In general, the IRP involves deciding which clients to visit, how much product to deliver, and the order of visits. The IRP presented here is inspired by a real-world problem faced by a company that produces and distributes gases. In this case, we focus on the distribution of oxygen. A mixed-integer linear programming model is presented to minimize greenhouse gas emissions and include priority for clients in the health sector. The need to incorporate priority for this set of clients is to fulfill the requirements of Mexican law to maintain product quality. The model utilizes a Comprehensive Modal Emission Model (CMEM) formulation to calculate the fuel consumption of heavy-duty vehicles and then, through a factor, obtain the emissions per liter of diesel consumed. The model was implemented using CPLEX, and a set of adapted benchmark instances were solved. The main results show that the minimal average emissions are obtained with instances having 0
AI is fundamentally changing logistics worldwide, and Mexico is no exception. This analysis explores the impact of AI on Mexico’s logistics industry, examining both opportunities and challenges. Job displacement, skill development, and effective human-machine collaboration are key areas of concern. AI technologies like machine learning and robotics revolutionize logistics by enabling smarter routes, predictive maintenance, and automated warehouses. Additionally, AI-powered platforms enhance customer service. For instance, AI optimizes delivery routes, reducing fuel consumption and delivery times. However, automation of repetitive tasks may displace warehouse workers, drivers, and administrative staff. While job displacement is a concern, AI also unlocks opportunities. New skill sets are in demand, such as AI, robotics, and data analysis, along with digital literacy for collaborating with AI systems. Soft skills like communication, teamwork, and adaptability become increasingly important. Educational institutions and companies must invest in training programs to equip workers with these essential skills. A critical aspect is fostering human-machine collaboration. AI excels at repetitive tasks, while humans focus on creative, judgment-based, and interpersonal functions. Redesigning workflows to incorporate AI tools ensures seamless collaboration and maximizes benefits. Continuous feedback between humans and AI systems is crucial for optimal performance. Mexico’s strategic location, labour force, and competitive costs make it an ideal partner for nearshoring under the USMCA. By investing in AI and developing human capital, Mexico can enhance its logistics capabilities and create new jobs. However, the current state of higher education, with only 20
During this era of digital transformation, applications in Logistics and Supply Chain based on Operations Research (OR) and Machine Learning (ML) techniques has catalyzed the development of innovative approaches that redefine industry standards. Particularly, transportation dispatching – a critical aspect of logistics – has seen significant advancements through the application of Reinforcement Learning (RL), achieving notable enhancements in operational efficiency. Despite these advancements, current research predominantly focuses on ride-sharing and on-demand delivery, with limited attention to fair dispatch practices by the workers point of view. This research addresses this gap by proposing a fair truck dispatch system designed to equitably distribute loads from a shipping company to carriers without compromise service levels. Utilizing real-world data characterized by uncertain demand and a dynamic fleet size our empirical results demonstrate the effectiveness of the proposed dispatch strategy, confirming its capability to ensure equitable load distribution across different operational scenarios reaching 86
Electric car sharing systems provide users with short-term access to a variety of shared electric vehicles, with payments linked to usage. It is an innovative transport strategy that enables users to find the adequate type of electric vehicle according to their needs and only when required. The configuration of these systems has a major impact on the attractiveness of the car sharing operators, hence, their revenues and their ability of serving as many customers as possible. This configuration consists of several parameters such as the location and capacity of stations for recharging and parking electric cars, the number and types of electric cars, as well as their battery capacities. In this paper, we investigate how to design an efficient one-way electric car sharing system. The aim is to select a set of locations, from a set of possibilities, to establish the stations and to determine the initial number of electric cars at each station. Our main contribution is a ruin and recreate neighborhood heuristic that we propose to generate new solutions. We evaluated our heuristic on scenario data from Manhattan, USA.
This work addresses a proposal for the process of container unloading from ships to terminal tractor through reach stacker cranes inside of a container terminal. The aim is to analyze the process using a simulation system through the Arena program, in which all activities have a consignee time, to get a productivity level of the resources involved, in accordance with the set operations up in a container terminal, it is also complying with the client requirements. The design of this model has as the purpose to analysis and improvement the unloading process of 1000 containers from a ship to the different zones in a terminal and identify the resources required for the efficient development of the operation corresponding to the maximum use of resources and compliance with the 16 h required by the operation. For the development of this operation, the program is expected to perform an analysis of the unloading process. By designing this simulation system, the terminal seeks to know the number of resources to be assigned for the complete maneuver, as well as their time of use, without the need to make an investment, due to the model's capacity to predict the real operation.
