When planning a new manufacturing center, many questions arise with multiple variables that make finding the optimal solution difficult. During the planning of the desk riser facility, a question proposed by the manufacturing analyst was how many welders and welding machines would be necessary to meet year 5 production of 195,000 units. This question was further complicated by the desire to know how the experience level of the welders and their rate of production and salary would impact the number of welders required. To answer the question, the welding stations along with their operators were modeled in FlexSim to determine the optimal outcome.
System
Before modeling the system in a simulation software, the actual system and its arrangement must be understood. The welders are taking eleven separate parts and combining various pieces to produce three distinct parts. The eleven parts are all being created from raw materials through drilling, cutting, and bending processes. After they are produced, the components making up the three parts are packaged together in three groups. These three groups are then gathered into one kit, which will be delivered near the welding stations. Once delivered, the welder will unpack the kit and begin welding the parts together. The three resulting parts will then be gathered and delivered to a nearby storage area by the welder thus completing the process.
Parameters
The key parameters for the simulation are the processing time for expert and average welders, material supply rate, and salary. Average welders were capable of welding 140 inches per hour, so it was decided that an expert welder must be capable of 200 inches per hour. There are seventeen total inches of weld on the desk riser, so it would take an average welder 437.14 seconds per desk while an expert would only require 306.00 seconds. How often the welders are being supplied with materials was determined through working backwards from the year five production level of 195,000. To produce the 195,000 desks in 250 days, with seven production hours in two daily shifts, the welders would have to be supplied with 55.71 kits per hour. The supply rate was rounded to 56.00 kits per hour for simplification. The hourly pay for average welders in Memphis, Tennessee was found on Indeed to be $23.54 while the expert pay was $35.35.
The model was then created to reflect the physical system while implementing the parameters mentioned above.
System
Before modeling the system in a simulation software, the actual system and its arrangement must be understood. The welders are taking eleven separate parts and combining various pieces to produce three distinct parts. The eleven parts are all being created from raw materials through drilling, cutting, and bending processes. After they are produced, the components making up the three parts are packaged together in three groups. These three groups are then gathered into one kit, which will be delivered near the welding stations. Once delivered, the welder will unpack the kit and begin welding the parts together. The three resulting parts will then be gathered and delivered to a nearby storage area by the welder thus completing the process.
Parameters
The key parameters for the simulation are the processing time for expert and average welders, material supply rate, and salary. Average welders were capable of welding 140 inches per hour, so it was decided that an expert welder must be capable of 200 inches per hour. There are seventeen total inches of weld on the desk riser, so it would take an average welder 437.14 seconds per desk while an expert would only require 306.00 seconds. How often the welders are being supplied with materials was determined through working backwards from the year five production level of 195,000. To produce the 195,000 desks in 250 days, with seven production hours in two daily shifts, the welders would have to be supplied with 55.71 kits per hour. The supply rate was rounded to 56.00 kits per hour for simplification. The hourly pay for average welders in Memphis, Tennessee was found on Indeed to be $23.54 while the expert pay was $35.35.
The model was then created to reflect the physical system while implementing the parameters mentioned above.
Scenarios
After the model was created, ten different scenarios were run in order to determine the proper outcome. Each scenario varied in number of welders and stations from four to eight and experience level of the welder switched between average and expert. For example there would be one scenario with four average welders, and then another scenario with four expert welders. This pattern continued up to a total of eight welders and machines. Each scenario was replicated five times.
The scenarios were also evaluated based on two performance measures: total output of units and worker utilization.
After the model was created, ten different scenarios were run in order to determine the proper outcome. Each scenario varied in number of welders and stations from four to eight and experience level of the welder switched between average and expert. For example there would be one scenario with four average welders, and then another scenario with four expert welders. This pattern continued up to a total of eight welders and machines. Each scenario was replicated five times.
The scenarios were also evaluated based on two performance measures: total output of units and worker utilization.
The general trend showed that the expert worker would have a slightly higher output than the average worker if the number of operators remained below six. Once the parameters for number of operators and workstations reached six, the average and expert workers had the same output. At a parameter value of seven and beyond, the output did not increase and remained consistent for both levels of workers |
The utilization rate was inverse of the throughput rate and decreased as the number of operators increased. The utilization rate was at 99% for both levels of operators up until the parameters relating to quantity increased to six. As the quantities of operators and workstations continued to increase past five, the utilization rate began to decrease with the expert worker’s utilization always being slightly below the average worker. |
Results
Multiple scenarios can immediately be disregarded as they do not meet the required production level for one shift. To meet the year five production level of 195,000 desk risers, 390 products would need to be produced each shift. The first four scenarios with four and five workers fall below that margin, so they are removed from consideration. The remaining scenarios all surpassed the minimum production level, but all remained close to one another in terms of total throughput. Because of the limited number of incoming parts per hour, there was a maximum threshold of parts that could be produced within the shift. Once that amount was reached, the operators would become idle, and their utilization rate would drastically decrease as shown in the last four scenarios in Figure 3. Since scenarios five and six had similar total throughputs, but higher utilization rates they remained in consideration while the rest were eliminated. The only difference between scenarios five and six were the experience level of the operator. The deciding factor between the two scenarios would be the cost of each operator. With a pay of $23.54 per hour for the average worker and $35.35 per hour for the expert worker, the average worker is clearly the more cost-effective option since each scenario shared the same number of operators and similar total throughputs.
The simulation is based on the assumptions that the rate of incoming parts serves as a bottleneck for the throughput of the welders, and that at the higher quantity of welders, the workers would not be cross trained to use their idle time to perform other functions in the facility. The model also assumes that there is not a drastic quality change when switching from average to expert workers but that the only improvement is in processing time. If these factors were to change, then a preference towards a larger quantity of expert workers might be justified, but in the current state, RayDesks recommends that six average welders be hired per shift, and six welding workstations be purchased to meet the year five production level.
The simulation is based on the assumptions that the rate of incoming parts serves as a bottleneck for the throughput of the welders, and that at the higher quantity of welders, the workers would not be cross trained to use their idle time to perform other functions in the facility. The model also assumes that there is not a drastic quality change when switching from average to expert workers but that the only improvement is in processing time. If these factors were to change, then a preference towards a larger quantity of expert workers might be justified, but in the current state, RayDesks recommends that six average welders be hired per shift, and six welding workstations be purchased to meet the year five production level.