Project 2: Assembly Test
In addition to evaluating changes to an existing system, simulation is also used to establish the feasibility of a proposed new system. In such cases there may be little or no data available, and the modeling must be done at a high level since details will only be worked out if a “go” decision is made. In this project, a company is trying to decide whether to do some electronic assembly and testing in house; previously this job was subcontracted to another company. The key (ประเด็นสำคัญ) is whether they can attain a desired cycle time with a limited commitment of labor and capital.
The proposed expansion consists of the final operation in production of two different sealed electronic units. The incoming parts are cast metal cases that have already been machined to accept electronic components. The operation involves preparing the cases, inserting the electronic component, sealing the cases, testing the assembly, and reworking those assemblies that fail inspection.
The first units, called type A12117c, are produced in an adjacent department. Historically (แต่เดิม) these parts have been produced at rate of 12 per hour, but the actual flow is not at all steady. (Because time is short, the adjacent department itself will not be modeled.) The proposal calls for these parts to have their own preparation area (หรือเรียกว่ำ “prep area”). At the prep area for A12117c, the mating faces of the case will be machined to insure a good seal, and then deburred and cleaned. A factory worker who used to do this kind of job at another plant thinks it will take about 4 minutes per part, with a range of 1 to 8 minutes. The A12117c parts will then be transferred to the insertion/sealing/inspection operation.
The other units, call B33433x, are produced in different building. To conserve the resource that are required to move the units from the other building to the assembly and test area, the proposal is to bring them in groups of 4 at the time. About 2 group of 4 per hour (กลุ่มละ 4 ชิ้น ด้วยอัตรำ 2 กลุ่มต่อชั่วโมง) will be produced, but again the flow is not all steady. The B33433x parts will also have their own separate prep area, where each batch of 4 cases will be spit and processed individually. The processing steps are similar to the A12117c parts, but in this case data (ใน file “partbprp.dst”) are available on a similar process elsewhere in the plant. Times are in minutes. The B33433x parts will then be transferred to the insertion/sealing/inspection operation also.
At the insertion/sealing/inspection operation (มีสถำนีเดียวสำหรับทั้ง A12117c และ B33433x) the electronic components are inserted, the case is assemble and sealed, and completed unit is tested. The time to accomplish all of this is expected to be 2 minutes for the A12117c assemblies, and 3 minutes for the B33433x assemblies. And average variation of plus-or-minus 20% around these values is considered the standard allowance for variability.
Base on the yields attained by subcontractor, 91% of the parts pass inspection; they will be transferred directly to the shipping department. The subcontractor did not rework the failed assemblies, but the company plans to do so and hopes that 80% can be salvaged; the remainder will be scrapped. Since the time to rework depends very much on the specific problem, a separate rework station will be employed, and a mean time of 15 minutes allocated for rework; the actual rework times are expected to vary greatly.
To be competitive with the subcontractor, the company must be able to achieve an average cycle time 45 minutes to produced 1,200 parts that go directly to shipping (without need of rework). They would also like to know the cycle time for rework parts, scrapped parts, and where bottlenecks are in the proposed design so that if there are additional funds available they will know where to add resources.
Approximation: Use queuing models to approximate the cycle time for products that go directly to shipping. Does it appear that they can achieve the desired cycle time?
Project: Simulate the system to evaluate the performance measures of interest to the company and make a recommendation. Be sure to do the following:
1. Provide point estimates and confidence intervals for all performance measures.
2. Include a table that lists your input model and a justification for choosing each one (this can be in the Appendix).
3. Determine how many replications are needed so that you can make a firm recommendation.
4. Follow the format of the design project report.
Hint for approximation: The queuing formula works only when there is single class of service, i.e., only one service rate. Thus, when there is more than one class of service, we need an average service rate. Suppose there are two types of parts, arriving at a rate of