One challenge facing the logistics industry is how to most efficiently run various operation bases. This problem is referred to as the "combinatorial optimization problem," and the faster this problem can be solved, the greater business speed will be. In this article, we introduce a decision-making support consulting service that utilizes combinatorial optimization technology alongside a review of logistics base sites. (From Fujitsu Research Institute's information magazine Chisounomori)
- Author Profile
- Chief Senior Consultant, Business Analytics Group, Fujitsu Research Institute
- Takashi Ohta joined Fujitsu in 2001. He was later transferred to Fujitsu Research Institute. He mainly engages in consulting on optimization technology, systemization support, and technical lectures in the fields of logistics and the environment.
How to Quantitatively Evaluate Logistics Costs, a Large Percentage of Administrative Costs
Logistics costs account for a large percentage of distribution costs and general administrative costs, so reducing such costs is one of corporations' most crucial objectives. In particular, transportation and delivery costs account for about 50 to 60 percent of logistics costs, and they are certainly often under the spotlight. Moreover, as the number of corporate mergers and acquisitions has increased recently, reviews of distribution networks after such mergers and acquisitions have become a major challenge, requiring drastic improvements that must consider the big picture of transportation and delivery systems. However, the present reality is that it is difficult to quantitatively judge and evaluate the effectiveness of large-scale measures to reconsider transportation and delivery systems in their entirety.
Fujitsu Research Institute offers a large number of decision-making support consulting services that utilize "combinatorial optimization technology" in the logistics industry. Quantitative evaluations calculated using advanced optimization technology have become the indispensable foundation for decision-making processes in transportation and delivery systems. In addition, measures' effectiveness and the merits of multiple measures become transparent through quantitative evaluations, enabling substantial reductions in decision-making time.
Precise Evaluations by Consulting with a "Logistics Base Review"
As a measure to review the big picture of transportation and delivery systems as well as to create strategic solutions, one element to consider is to review logistics base sites. For example, to improve the warehouse utilization rate and vehicle load capacities, it is effective to reorganize warehouses and distribution centers, but various approaches to reorganizing logistics bases are available. As shown in Figure 1, when consolidating two logistics bases into one, a decision must be made as to which measure will be most effective--whether to abolish one of the bases, or to abolish both bases and build a new one.
By building a logistics base in a strategic location after reviewing the logistics base sites, distribution efficiency is increased while costs and the travel distance for transportation and delivery are reduced. This is why reviewing logistics bases has become a crucial management challenge for many corporations. However, quantitative methods of analyzing and evaluating these factors have not been well established, which is part of the reason that today, most decisions are made without a strong quantifiable basis.
For example, if building a new logistics base by integrating multiple logistics bases, the proposed sites (regions) must first be listed, and then the optimal location for the construction site must be selected out of that list. Moreover, it is common protocol to conduct on-the-spot investigations into matters related to transportation and delivery as well as on-site operations. Construction site selection for logistics bases must consider transportation and delivery costs, the required inventory volume, capacity, and operational efficiency. However, the difficulty of precisely calculating these factors leads to securing needlessly spacious sites or selection of locations that are closer than necessary to city centers, which come at higher prices. The most important factor when evaluating measures to review logistics bases is the evaluation of matters related to transportation and delivery. Inventory volume and capacity can only be calculated once transportation and delivery networks have been decided; if there are any oversights in the transportation and delivery network plans, all calculations that follow will be invalid. However, to evaluate matters related to transportation and delivery, comparative evaluations of various patterns are necessary, making it very difficult to perform precise evaluations.
With an understanding of these realities, the "decision-making support consulting service for reviewing logistics bases" provided by the Fujitsu Research Institute places special focus on evaluations related to transportation and delivery. Precise numerical evaluations are made possible by utilizing combinatorial optimization technology to review multiple proposed sites for bases as well as a large number of transportation and delivery network scenarios.
Using Numerical Simulations to Handle Problems Difficult to Solve by Human Ability Alone
In numerical simulations performed to review sites for logistics bases, highly specific simulation results can be achieved by combining "site simulations for logistics bases" with "delivery route simulations."
In "site simulations for logistics bases," the number of efficient logistics base buildings and their positions, as well as the assigned areas for each base, are decided using data (e.g., delivery destination distribution situations and demand rates). This differs from methods that rely on experience or guesswork by those in charge, or simple calculations that rely on selected conditions and information. Instead, highly reliable numerical results can be achieved by conducting precise simulations. With "delivery route simulations," in which multiple transport vehicles depart from a logistics base to travel to multiple destinations to deliver shipments, the objective is to determine how efficiently each transport vehicle's movements can be improved to make each shipment to each destination and in what order. By combining these simulations, the required number of transport vehicles for logistics bases' scenarios can be calculated to achieve more exact simulation results that include the details of daily operations. Moreover, besides presenting these simulation results in numerical form, they can also be displayed on maps, facilitating sensory understanding through visualization to contribute to customers' decision-making processes.
Figures 2 and 3 show the results of site simulations for logistics bases and delivery route simulations. Site simulations for logistics bases simulate the optimal construction sites for bases (e.g., distribution centers and hubs).
Various forms of logistics simulations can be conducted, such as simulations of distribution center sites in the case of simply delivering shipments from distribution centers to delivery destinations, and even simulations of distribution hub sites in the case of delivering shipments from factories to delivery destinations by way of distribution hubs. To solve such challenges, many factors must be considered, such as logistics bases' fixed costs, logistics bases' capacities, delivery destinations' rates of demand, the road distances between logistics bases and delivery destinations, and variable costs incurred in transportation. It is extremely difficult to consider all these conditions and evaluate the vast number of patterns by human calculation.
For delivery routes, simulations of the most efficient delivery routes in the region are conducted (Figure 4). Fujitsu's technology can simulate not only delivery routes but also transport models from cargo collection (milk runs) or cargo collection destinations to delivery destinations. In such scenarios, vehicles' maximum carrying capacities, cargo loads, designated delivery times, travel distances and travel times on roads to delivery destinations, and cargo loading/unloading times must be considered. As the number of vehicles and delivery destinations increase, the number of combinations increases exponentially, making it exceedingly difficult to make decisions by human calculation alone.
Fujitsu Research Institute develops efficient solutions for combinatorial optimization problems with an optimization technology called the "metaheuristic solution." The metaheuristic solution is a well-established method for efficiently obtaining effective combinations from among a vast amount of combination patterns.
Clear Results of Measures: An 11% Reduction in Transport Vehicles
In one case study, after a corporate merger, simulations were conducted to review the assigned areas where each company's logistics bases made deliveries. The results enabled the merged company to reduce the number of transport vehicles by 11 percent. In another case study, as a customer grappled with the administrative challenge of whether to secure an external distribution hub for assortment tasks, the solution prompted the customer to decide that using their existing logistics base and building a new distribution hub would reduce transportation and delivery costs by 6 percent compared to their previous estimate. They achieved substantial cost reductions even after accounting for the new distribution hub's maintenance costs. In another example, a measure that was at first considered to be advantageous later turned out not to reduce costs as much as expected, while another measure thought unlikely to be effective yielded better results. Subsequently, demonstration experiments and implementation reviews were conducted for the latter measure.
In this way, the results of measures that customers had been reluctant to enact were made clear, which not only aided the decision-making process but also shed light on realities far from what was initially expected.
Pursuing Simulation Results Close to "Human-devised Answers"
Above, we explained the decision-making support consulting service for the logistics industry, which utilizes logistics base site simulations and delivery route simulations that employ combinatorial optimization technology. With combinatorial optimization technology, problems prevalent in various fields can be solved not only in the logistics industry but also more broadly--in supply chain management, production schedule formulation and daily scheduling, work shift chart creation, and elsewhere.
Moreover, against the backdrop of big data, IoT, and AI, advancements in new technologies are being made in both hardware and software. Fujitsu Research Institute endeavors to combine these technologies through optimization technology. By these efforts, we not only pursue the optimal numerical results but also strive to meet the needs of customers who pursue simulation results that are closer to human-devised answers.