New product ramp-up in the aircraft industry is a complex process due to the high number of changes that happen both in product and production processes at the manufacturing stage, which can increase aircraft lead time dramatically. Also, aircrafts are complex, small series products that are highly customized for specific customer needs, which makes launching a new product even more difficult. Today, ramp-ups have become more frequent as the average product lifecycle decreases. All of these issues can make ramp-ups a major challenge for aircraft industry production engineers.
The Airbus Group joined the European Union ARUM (Adaptive Production Management) project, which is focused on creating an IT solution for risk reduction, decision-making, and planning during new product ramp-ups. The project is aimed mainly at aircraft and shipbuilding industries.
Simulation was chosen as a part of the ARUM solution, because it would allow the participants to reproduce the real production facility experience (based on the use case provided by the Airbus Group), and thus provide a benchmark for the ARUM solution testing. AnyLogic was the selected simulation software due to its capability to combine agent-based models with discrete-event approach.
The simulation model included a part of the Hamburg Airbus A350 assembly line where two different pieces of the fuselage were completed. This part of the line consisted of six assembly stations with 30-35 people working at each one and approximately 300 work orders per station. The challenge was to simulate the ramp-up process where general productivity increased over time, with the whole ramp-up period lasting up to two years.
ARUM solution structure.
The agent-based and discrete-event model consisted of three types of elements:
- The flow line, that included working stations, each one with its own labor and physical resources. The stations were modeled as agents.
- The products (sections), going through the flow line parts. Each section required 200-600 work orders assigned to stations. Work orders formed tasks that required specific materials and resources. When a section entered a station, it started going through work processes modeled using the Process Modeling Library, then left and moved to the next station, and finally, to the assembly line in a different city, which was not modeled.
- The control model included plans that were sometimes affected by disturbances. The controller agent modeled the complex behavior of human managers reacting to disturbance events with control strategies.
Among others, the control strategies included open work policy alternatives. This meant that if some of the work could not be done at the moment, it could be delayed until some other point, while the product continued to move beyond this assembly line to the facility in the other city. In this case, Hamburg workers would have to travel to the other facility to complete the work (“traveling work” strategy). Alternatively, the work could be suspended until the disturbance was resolved (“stop and fix” strategy).
The disturbances that occurred during the ramp-up included:
- Unbalanced workload and resource allocation due to the workers’ learning curve and the fact that the same line produced several different products.
- Design non-conformities and changes, as production often began with a not completely prepared product.
- Missing material or material incompatibilities due to late design changes.
The measured model statistics included achieved aircraft lead time, amount of traveling work used, and resource utilization rates (labor, materials, and stations).
The experts created a model that was easy to understand and reuse, and that was integrated to the ARUM solution architecture. It also included the visualization of the assembly line.
The model was run to simulate the impact of the disturbance mitigation strategies currently being applied at the Airbus facility, including “stop-and-fix” and “traveling work” strategies.
The modelers tested multiple ramp-up scenarios with different sets of production plans. They also tried multiple sets of disturbances based on historic data, including extreme scenarios.
The model will be used for comparing plans suggested by the ARUM suite to the current management practices. This will allow development of the best disturbance mitigation strategies for aerospace and shipbuilding manufacturing industries’ ramp-ups.