Route planning: Witron and TU Dresden are developing a new code – route planning | News | LOGISTICS TODAY

Dr. Stefan Bauer of the intralogistics company Witron in Parkstein is responsible for the CPMS corrugated board business in the Upper Palatinate. He and his team are also involved in the development and further development of various platform solutions that will later be installed in food retail logistics centers around the world. As Witron recently informed, this also applies to the latest project, for which Dr. Bauer: Automation of route planning with integrated 3D loading optimization as part of a development project with TU Dresden.

“We already had an innovative solution for this that, based on a savings algorithm, was in great demand in the sectors in the early 2000s and was highly appreciated,” explains Bauer. Problem: Over the years, developers have introduced more and more restrictions. Uptime was no longer satisfactory for large supply pools, and code coverage grew.

On relationships in the Upper Palatinate

“The maintenance was getting harder and harder,” recalls a man from the Upper Palatinate. “Both technology and science have evolved, we knew we had to apply a new technique to continue to set industry standards in the future.” Bauer and his team sought outside support. Leopold Kuttner learned about the needs in Witron through contacts in the Upper Palatinate. Kuttner is an assistant professor at the Department of Industrial Economy of prof. dr. Udo Buscher at TU Dresden. “We are not afraid of real problems,” jokes scientist Buscher.

Time window, order split and packing problem

The aim of the project was therefore to improve Witron’s existing trip planning solution, taking into account complex transport tariffs and various practical constraints. For example, restrictions on route planning include multiple time windows, splitting orders and a heterogeneous fleet of vehicles, as well as packaging restrictions such as LIFO stacking and unloading. The proven tool for load optimization by Witron and the packaging algorithm developed by the department are used to solve the packaging problem.

“In order to achieve good quality feasible solutions, the problem of route planning and packaging has to be solved in an integrated manner. The difficulty of the task lies in the combination of these two demanding optimization problems. As an additional challenge, there are numerous practical limitations to be considered, ”explains Kuttner, who supervised the project for TU.

Divided into sub-projects

Scientists developed a proof of concept and added increasingly important practical constraints to it. “We have broken down the overall task into sub-projects. For the sake of simplicity, most of the restrictions were initially neglected. Following the successful validation of the subsection by WITRON, additional conditions were adapted and more were gradually added until all planning requirements were met, ”reports Kuttner.

“The biggest challenge was first of all to understand Witron’s client, because science and business sometimes speak different languages. Additionally: We do not plan to build from scratch. We were allowed to optimize the existing system, ”explains Buscher. This slows down some progress but also simplifies the work: “Some in the industry still swear by Excel, but thank God Witron was on,” adds the university lecturer.

save computing time

In order to solve the problem of route planning, and above all speed, the implementation of classic design and improvement processes was used. Thanks to integrated scheduling, frequent calls to the load optimization tool are a computational time bottleneck. To keep these calls to a minimum, a first packet-route-second approach was used. The basic idea here is to approximate the time-consuming three-dimensional problem with a one-dimensional quantity, the load counter, and thus save computing time.

Algorithmic decision support solution

The project resulted in an algorithmic decision support solution that generates alternative planning proposals to better support dispatchers in operational planning. As announced by Witron, this ensures compliance with all planning specifications for route planning and loading. Using the parameterizable cost function, dispatchers can define their planning preferences and get some cost-effective suggestions for route plans generated in seconds. By saving time on planning, dispatchers can devote more attention to other value-added tasks.

Bauer praises the speed of the new solution. “We looked at completely new methods, learned a lot from e-commerce, and then added our heuristics to it.” Project sections were validated against live production data. ‘Scientists also had to get involved with our data models. But it also illustrates the success of the project because we have a common language and work together with different tools. ” But seasoned IT expert Bauer also knows: “We now have to continue to maintain and develop the algorithm. It is not always easy. We are fully focused on the design business. Work very hard in the present. ”

Implementation as the next step

And then Bauer looks to the future again: “The next step is to implement the solution in the food retail sector as it is a key building block in an end-to-end platform.”

Leave a Comment