System Throughput Simulation

In order to increase system throughput and decrease labor expenses for a manufacturing plant, AMT simulated a more cost-effective assembly line. Our solution cost less than 4% of the original proposed revisions and protected our clients from spending unnecessarily.

Background

A plant produced the right and left front control arms, parts which connect the front suspension and frame of a car, for a Detroit-based automaker. The plant struggled to meet production demands without excessive overtime. To increase throughput and decrease labor expenses, they drafted a proposal that required a significant capital expenditure to revise the machine and assembly line and would cost anywhere from $330k to $400k.

AMT Engagement

Prior to making a capital equipment investment and shutting down the plant for installation, the customer chose to consult the experienced team at AMT to validate the assumptions regarding their current line and the proposed revisions. The existing system was a pallet conveyor system with the following stations:

  • Operator load/unload
  • Profile check
  • Camera read data
  • Seventh axis robot load/unload, to/from machine centers
  • Three machine stations
  • Quality check
  • Assembly station 1
  • Assembly station 2
  • Labeler

Rejects occurred at the profile check, camera read, and each machine center. AMT’s proposed revisions were to maintain the position of the machining and assembly stations and increase the pallet conveyor size to route the reject pallets around the bottleneck as well as to increase the buffer size.

We took a holistic approach to the problem by documenting the units per year, takt time, PM schedule, and benchmark availability for the plant operation, etc. From the data, a parameter driven 2D model was developed using a discrete event simulation package. This simulation allowed the team to model various scenarios, and better understand the root cause of the line’s poor performance. The effects of MTBF, MTTR, reject percent, and other factors could be understood by varying the input parameters. Once the simulation was calibrated to a realistic scenario, the model did in fact validate the current throughput, and the proposed revisions were then made to the model.

Once the simulation was run, the model revealed rejects had less impact than downtime, downtime of the assembly stations had more impact than downtime of machining centers, and the top three bottlenecks were so close in cycle time that it would be ineffective to keep repairing all three. Our simulation also indicated that unplanned downtime on the system was so great that even buffering, which is a traditional method of balancing for downtime, would have been space and cost prohibitive. Our team suggested the MTTR’s of the assembly stations and machine centers were too long and engaged an expert to maintain them and make them more reliable. With our recommendation, there would be no investment in capital equipment required.

Afterward

For less than 4% of the proposed revisions cost, the system throughput simulation prevented expenditure for counterproductive capital equipment. Our recommendations brought the different entities (corporate, plant, capital equipment suppliers, and integrator) together to agree, talk, and focus on the problem and solution. We also prioritized which parts of the assembly and machining line had the biggest impact on production when they went down.