Poster
Client Context
Wallenius Wilhelmsen is a global shipping and logistics company that serves car manufacturers by installing accessories at four shop areas in its job shop.
Within the job shop, different jobs are done at four shop areas, and not all workers are trained in all jobs. Not all cars go through all shop areas; they skip shop areas in which they have no jobs. Supervisors at the job shop assign workers to shop areas based on the number of cars expected to enter each area.
Project Objective
Because supervisors only consider the number of cars entering each shop area and not the actual amount of work each car needs, some shop areas do not have enough workers assigned while others have too many. This creates bottlenecks and makes the system less efficient, so Wallenius Wilhelmsen has been hiring more temporary workers to support the completion of throughput obligations. This is a major pain point to Wallenius Wilhelmsen because temporary workers are difficult to find, expensive to train, and increase vehicle unit costs.
This project’s objective is to increase the productivity of workers by allocating them more effectively to the four shop areas. This will be done by designing a new labor scheduling system that considers the number and length of each type of accessory installed and prioritizes bottleneck shop areas.
Design Strategy
The team developed a new labor allocation algorithm that uses a mix of deterministic and random selection to assign workers based on relative labor needs expected in the four shop areas. The algorithm also prioritizes the shop area most known to be the bottleneck area based on historical data. Factoring accessory processing times and the number of accessories into the labor scheduling will help Wallenius Wilhelmsen assign workers in proportion to how much actual work needs to be done at the four shop areas, giving employees productive work for longer portions of their shift. Increased employee productivity should lead to lower labor requirements to complete the same amount of work or lead to more work being completed with the same number of employees as before.
The team validated and verified the simulation using both historical data and conversations with the supervisors. Major assumptions were validated or informed by supervisor input, including the use of a static shop flow for each car, simultaneous jobs in Ramp, and employee mobility in Heavy and Light. The team also compared major statistics from simulation runs to actual historical values for 750 historical shifts and consistently found that historical values were within the 95% CI found for that statistic via simulation.
Deliverables
To improve Wallenius Wilhelmsen’s employee throughput, the team delivered an allocation algorithm and a simulation, which is integrated with a user interface on the input side and a dashboard on the output side. The user interface will help supervisors input necessary shift data. The allocation algorithm uses a mix of deterministic and random selection to assign workers based on relative labor needs expected in the four shop areas. The simulation will provide supervisors with key insights on how much work will be completed in each shop area and the entire system. The dashboard will display these insights for interpretation by the supervisors. The team believes this new system will help Wallenius Wilhelmsen supervisors make better informed labor scheduling decisions.
The team’s main recommendation is to use the work-ratio-based algorithm. While the cost is no longer zero due to the training need, the potential savings are greater with even a 1% reduction in labor needs. The lack of software licensing costs coupled with scalability allows Wallenius Wilhelmsen to use this solution at other sites if they choose, further increasing the net benefit.
Value and Impact
Analysis completed by the team has indicated that Wallenius Wilhelmsen can save $43,200 for each percent reduction in labor needs. Testing of the simulation has shown that the new allocation system coupled with training improvements resulted in reduced makespans when compared with the historical allocations. In addition, the team found that the new allocation system results in an increase in work completed per employee of 8%. All software developed for this project is free and open source, meaning that the company will not have to risk upfront investment when implementing the team’s solution. The client would incur training costs, but the team estimated that these costs would total $38,300, less than the projected savings for a 1% reduction in labor needs.
Project Information
Student Team
Meiwen Bi, Sven Rasmus Eriksson, Gianna Fantell, William Fuss, Gabriel Larios, Jiashu Li, Joseph Mao, and Ethan Metcalf
Faculty Advisor
Faculty Evaluator
