Eurystic is an Argentinian consulting company that helps companies continually improve their operations by using high-impact solutions to complex problems through the application of quantitative-based methodologies and tools.
They worked on a project for a large steel manufacturer, that produces a wide range of products, especially metal meshes and drawn wire.
During the manufacturing processes, raw materials come in and go through the rolling process, where the products can be directly prepared as drawn wire and shipped to the clients. Alternatively, it can be used to fit the bending and welding processes to ultimately manufacture metal meshes.
The factory runs mostly under a make-to-stock policy, so it is critical to have a good demand forecast and a balanced production schedule to guarantee a diverse product offering.
Problem
Originally, the production scheduling was carried out manually by the planning team using spreadsheets, so it was difficult to consider relevant factors involved in the production scheduling. As a result, schedules could not be strictly detailed, and overall productivity was quite low.
Given the nature of the products and market specifics, customers usually search for a variety of products. If they can't find the whole range available, they will probably turn to another supplier. Due to the high machine setup time, producing high quantities of the same products for long periods of time is reasonable. But this pattern may cause storage collapse with less product variety, lack of coordination between supply and demand, and possible slowing of sales.
So, the manufacturer sought to find a production planning and scheduling software that could generate an optimized production schedule for each of the machines in the factory, and forecast the systems behavior. The main objective was to maximize service level and fulfillment of the delivery schedule (which requires a balanced product mix), while considering resource efficiency (reducing unproductive setups, buffer saturation, etc.).
To do that, Eurystic engineers had to comply two competing strategies: commercial prioritization and production efficiency. The tool had to produce a schedule with a horizon of at least one month, while considering machine calendar, productivity and efficiency, buffer capacities, transportation means availability, raw material availability, and many more.
Solution
Engineers chose AnyLogic as the manufacturing scheduling software because it provides a great development experience for modeling logics of every component in the system.
While developing the simulation model, Eurystic modelers used an agent-based approach to maximize flexibility and represent the non-linear interaction between the components with their individual constraints. The model could be run at any point of time and captured the system’s current situation.
Most of the inputs had to be flexible enough to allow the client to try different scenarios, so the model included all the relevant parameters:
- Storage capacities
- Transportation availability
- Machine calendar
- Machine productivity and efficiency
- Raw material availability
- Initial stock
- Forecasted and confirmed demand
The greatest part of the input data was taken from SAP system and processed by the model.
The most complex components (e.g., machines) included inner heuristics to optimize certain parts of the process and adapt to changing context conditions.
To find an optimized production schedule, modelers developed a custom optimization engine by applying local search heuristics. This optimization engine runs independently from the simulation model and, in fact, uses the simulation model to test different solutions.
It triggers multiple simulation runs using a heuristic with different production strategies. After initial simulations are finished, the production schedule with the best result is selected and used as a starting point for a local search algorithm that looks for a better solution.
Result
Using AnyLogic as a production scheduling software, the company achieved a 10% increase in production.
The model runs as a standalone application and reads all input data from the Excel files. Additionally, it displays all of the results as KPIs and charts and allows users to export them in Excel files as well. This allows users to carefully analyze and compare the provided solutions.
The exported machine schedules are prepared in a format that can be easily loaded into SAP and launched to the factory floor.
The simulation results helped the company get valuable insights to optimize existing processes:
- Partially automated and greatly reduced the time required for the production scheduling generation process from several days to in a few hours or even less.
- Provided a way to predict and anticipate potential problems, such as warehousing saturation, and take preventive actions to avoid them.
- Allowed them to anticipate the raw material requirement, which can be used to schedule production in processes from other sectors of the factory.
- Enabled them to make important strategic decisions, for example, to assess the operational impact of manufacturing certain products or to discontinue the ones that were not profitable.
- Accurate forecast production and the systems evolution have enabled the planning team to improve the coordination and the communication with other departments.
The factory staff continues to embrace a new way of working that was proposed along with the implementation of the tool, so the company expects further growth.
The case study was presented by Gabriel Goyheix and Maximo Lambruschini, of Eurystic, at the AnyLogic Conference 2021.
The slides are available as a PDF.
