The Boeing Co. (Seattle) is one of the biggest airframe manufacturers in the world. In any given year, its facility produces 300 to 600 commercial airplanes, ranging from 717s to 747s. In addition, the company has evolved to become a major defense contractor and the one of the biggest aerospace manufacturers in the world.

One of Boeing’s core engineering technologies is airplane wings. Early in the manufacturing process of wings, two critical components are built: wing spars, which are large beams that support the bulk of the wing structure; and wing skins, which are the metal outer coverings of the wing structure. To create these parts, plates of aluminum are cut down in a milling process.

The shot peening process is one major step in manufacturing wing parts. Shot peening involves hurling streams of tiny, forged steel balls at the parts. This shapes the parts. It also improves fatigue life and prevents corrosion. An area called the forming corridor contains five large machines that perform shot peening operations. An overhead material handling system consisting of a rail network, load bars, cranes, storage locations and a transfer bridge ties the five machines together.

Senior management considered the forming corridor to be a space-limiting constraint in the skin and spar manufacturing process. Excessively long flow Arial were a symptom that management wanted to eliminate. The underlying goals were shorter lead and cycle Arial, and lower work-in-process inventories. Management also wanted to see if more space could be made for additional work.

People acquainted with this area felt that it could run more efficiently. But they didn’t have a solution. Because the shot peening area was so complex, it was difficult to look at the area with a normal process-improvement approach. Fifty different kinds of parts, each with different run and setup Arial on each machine, share five different routings through the area.

Management didn’t want to risk experimenting with the system for fear of causing additional production slowdowns. Research and development employees concluded that the area was a good candidate for ProModel’s (Orem, UT) simulation modeling.

A model was developed that management agreed was valid. From there, additional code was written to permit experimenting with parameters of interest, including the number of load bars in the system, prioritization of transfer bridge usage, machine operation schedules, and how machine downArial affected the system.

One key finding involved the number of load bars in the system. Model analysis showed that as the number of load bars was reduced, system performance improved. Further analysis helped establish a safe lower bound for cutting load bars. This was based on minimum throughput requirements and practical limitations, such as storage availability. Modeling also showed that cutting the number of load bars from 22 to 14 was practical and would reduce average flow time by about 7 percent.

Decreasing load bars would reduce the number of parts that could be in the system at any given time, so work-in-process inventory would also drop by one-third. In addition, the model predicted that delivery performance would improve as a result of this change.

The model also helped management understand the effects of operating policies. Increasing one machine’s usage from two to three shifts per day had a noticeable impact on performance. Increasing utilization on another machine also helped.

In addition, modeling clearly showed the impacts of machine failures on system performance for the first time. This data can be used in the future to help justify machine reliability improvement efforts.

After seeing the model results, senior manager immediately decided to implement these changes.

For more information on software modeling, call 801-223-4600, visit www.promodel.com.