With ever increasing pressure for transportation and distribution companies to reduce costs while improving performance current techniques for load density, route and warehouse planning optimization are falling short. Although these methods have improved in the past decade there remains a gap between the needs of suppliers and available optimization systems.
In the recent past powerful optimization tools have been researched and refined in the academic world. These methods of optimization have their origins in evolution, both physical and social. The main concept of this paradigm is each new generation looks to improve itself based upon the success, and failures, of the past generation. Because these tools are computer based a generation’s life span is very short and thousands of generations can be run with little time. This gives us the ability to search a very large solution space quickly to find “best” solutions. Additionally, because of the speed and simplicity of these tools, constraints can be changed with relative ease allowing management teams to check a range of circumstances. These tools have only recently made their way to industry, primarily in the area of defense. One of the greatest advantages of these tools is their ability to quickly search a very large solution space. Another is their simplicity. A few innovative industry leaders have begun using these planning and optimization tools with great success.
A Fortune 100 transportation company has recently employed a Particle Swarm Optimization (PSO) version of an asset allocation planning tool. This tool has shown the potential for a reduction of 2.5 FTEs in a 8.5 FTE operation, 29%. The United States Air Force has also moved to a PSO version of mission planning for the unmanned aerial vehicle (UAV). The previous planning tool used was a Mixed Integer Programming approach, common in Supply Chain optimization. The PSO has proven to reduce the planning time by orders of magnitude with improved results. Delphi used neural networks to refine their electric car battery’s state of charge reading so that the battery became a viable solution. Genetic algorithms have been used to improve dock yard performance.
The number of uses for the computational intelligence tools within the logistics industry is very large. Several scenarios are listed below.
1. Optimization of Inventories
Resent studies have shown it is extremely beneficial to share information between manufactures and suppliers. The benefits of this information sharing are a higher profits, lower inventory and manufacturing costs. The main question to be answered is what information to share to maximize profits. A computational intelligence tool can be applied to this scenario to determine exactly which information is of greatest importance to optimize the supply chain
2. 3PL Planning
As is well known a 3PL handles the distribution for other companies that would rather focus their energy on their core competencies. Hence a 3PL has the unique requirement, or opportunity, to deal with freight from multiple vendors and or customers simultaneously. This situation affords the 3PL much opportunity for optimization. These opportunities are in a. pick-up route planning, b. consolidation planning, c. long-haul route planning, d. unload planning, e. put-away planning, f. order fill planning and g. delivery planning.
a. Facility Layout
Storage space within a warehouse is one of the greatest costs to a warehousing facility. Before storage racks are put into place a facility planning team could plan optimal space, checking every possible configuration, utilization using a computational intelligence tool.
Picking and replenishment locations in a warehouse have a great impact
on the productivity. These locations are generally chosen based upon
velocity of the product being shipped. The velocity comes from the
vendor/customer order history. A computational tool can be applied to
the order history, along with any other constraints important to the 3PL,
to find the optimal storage layout. The end result is a more efficient
put-away, replenishment and pick which would reduce labor and asset
4. Asset Scheduling
In this scenario an asset is a vehicle such as a tug on an airport ramp or any device which holds or moves freight pulled by a vehicle such as a dolly or unit load device. Other examples are forklifts, trailers and yard-mules. Generally when optimizing an asset scheduling problem there are multiple concerns: number of assets available, time an asset spends waiting to start a task, time spent waiting for an available asset and shift start and end times are a few. It has been shown that the Particle Swarm Optimization can search a very large problem space with relative quickness to find optimal solutions of this multi-objective optimization problem.
5. Route Planning & Cube Utilization
The cost of transportation is another of those costs that is necessary and high. It is in the best interest of any company to maximize cube utilization while at the same time minimizing the cost of inventory. There are many choices to achieve this goal and those choices are dynamic. Computational intelligence planning tools allow a Transportation department to plan and re-plan quickly throughout the course of a shift routes that will both minimize transportation costs while keeping inventories at desired levels.
The approach used to employ these tools is to investigate current practices, both perceived and actual, by means of observation and interview. From the information collected an operational simulation is developed. A computational intelligence tool is then chosen and applied to the simulation. The results will show the Executive Management team where potential savings exist as well as the potential for performance enhancement.
May 26, 2013