A New Simulation Technique Using a Holistic Approach and Methodology to Assess Productivity of the New Plant for Manufacturing the Boeing 787 2007-01-3908

Alenia Aereonautica is manufacturing two fuselage sections of the new Boeing 787 airplane in a new plant in Grottaglie, Italy. The manufacturing line of the plant consists in 24 production cells and three Automatic Guided Vehicles (AGVs). The plant has been in production since the beginning of 2007. The shop floor information system controls both the cells and the AGVs using a Real Time software environment that allows three production shifts. The design of the most important cells is completely new and there are no statistics available to evaluate the Mean Time Between Failure (MTBF) and the effective productivity of the cell and of the entire moving line.

Alenia Aereonautica made the request to study an adaptive Real Time Scheduler to control the yield of the plant and to allow the management to support the highly demanding yields from Boeing.

The authors had two issues:

1. 1
   The design of the most important cells is completely new and there were no statistics available to evaluate the Mean Time Between Failure (MTBF) and the effective productivity of the cell and of the entire moving line.
2. 2
   The design of the shop floor information system was designed as a custom system. The Real Time Scheduler and the connections to the higher level information system were untested in a real production status and they could not tested with a traditional technique.

To solve the issues, the authors designed a holistic methodology (i.e. evaluating both the estimated cells model behavior and the real control software subsystems) based on a Quasi Monte Carlo Analysis (QMC). A delay line is the basic element that models the manufacturing cell. The delay lines are connected in parallel or in series according to the layout of the plant. The input signal is the actual fuselage section - or “One Piece Barrel” (OPB). All the delays can be modified within a specific range. Failure events or simple production stops are simulated by random events.

The manufacturing plant yield can be tested with many simulation runs where the results are analyzed in order to have the best guess of the yield.

Using the simulation run, the user can test the resilience of the entire plant to modification of the estimated standard delays in the cells and he/she can control the capability of the shop floor real time scheduler to recover from such unforeseen events as a failed quality test or any technical problems in the cells' hardware. A Quasi Monte Carlo Analysis is the best effort methodology to obtain the measure of risk and to assess the ability of the shop floor, in real time, to re-schedule the production and maintenance tasks. The goal of the Grottaglie plant is to achieve the highest possible yield in order to follow the strict requirements from Boeing. The simulation software program has a structure similar to microprocessor technology where single instructions (the micro-activity in a cell) are executed for a time assigned by the production instruction and altered by a random delay or by a random event. Many other environments can reuse this approach.