Simulating Production in Factory Assembly Lines

This is a case study meant to simulate the production and assembly of parts in a factory. Steel Nuggets Inc. manufactures three types of part called A, B, and C and they wish to maximize the capability of their production facility.

Problem: The manufacturing process includes sending the three types of parts through separate conveyors to a Trim operation, then transporting them to be stacked, palletized and shipped to customers. At the factory’s current sub-optimal state, the weekly output averages at 2546 parts per week. The company wants to determine ways of maximizing the production efficiency.

Goal: The objective is to construct a valid Witness model of the factory’s current production process, and to use this model to analyze the factory’s options for improving its efficiency.

Tools: Lanner Witness 14, Microsoft Excel, Queuing Theory, Simulation


The factory is run 24 hours a day, 7 days a week. Each position involving people is manned for the entire duration of production. Each of the three parts A,B and C being manufactured has a first operation performed on a dedicated machine (MakeA, MakeB and MakeC).

These machines are never starved of the raw material (metal blocks) needed to manufacture the parts. A single part is manufactured on each machine at a time.

The parts are delivered to a Trim operation on three separate conveyors that can each take 3 parts only. The Trim operation takes parts from the conveyors in sequence – but will not wait if there is no part present on a particular conveyor – it will continue to the next conveyor in the sequence.

The Trim operation outputs the part onto another conveyor that delivers the parts to a stacking operation. The conveyor is long enough to hold 10 parts. The stacking operation is actually a person who takes the components and stacks them in three separate feeds to the palletizer depending on part type.

When a stack reaches 12 high there is no more room for additional parts of that type until the stack is palletized. The palletizer takes the stack and loads it onto a pallet which has been taken from the Pallet Store. The company currently owns just 12 pallets. All conveyors in this facility are of the accumulating type. Each machine has its own Mean time to fail and repair.

A witness model was created. When the system was run for different range from 0 to 100 weeks. The good optimal warm-up period was observed to be 50 weeks. The average value of output was found to be 2538 units, that is very close to the actual value.

After running simulation for various scenarios, these were determined to be the best ways to optimize the factory:

  • Limit the number of pallets to 15.

  • Having faster delivery

  • Increased trim machine reliability

  • Improved repair time for palletizer

  • Hiring 1 new employee to repair the broken machines

  • Giving higher maintenance priority to the Palletizer than the trim machine