Littlefield Simulation Analysis, Littlefield, Initial Strategy PDF

Preview

Summary

Homework assignment...

Description

Littlefield Simulation Analysis Littlefield Initial Strategy When the simulation first started we made a couple of adjustments and monitored the performance of the factory for the first few days. Initially we set the lot size to 3x20, attempting to take advantage of what we had learned from the goal about reducing the lead-time and WIP. We also changed the priority of station 2 from FIFO to step 4. After viewing the queues and the capacity utilization at each station and finding all measures to be relatively low, we decided that we could easily move to contract 3 immediately. Except for one night early on in the simulation where we reduced it to contract 2 because we wouldn’t be able to monitor the factory for demand spikes, we operated on contract 3 almost the entire time. This proved to be the most beneficial contract as long as we made sure that we had the machines necessary to accommodate the increasing demand through day 150. Machine Purchases The first time our revenues dropped at all, we found that the capacity utilization at station 2 was much higher than at any of the other stations. So we purchased a machine at station 2 first. Station 2 never required another machine throughout the simulation. After all of our other purchases, utilization capacity and queuing at station 2 were still very manageable. As demand began to rise we saw that capacity utilization was now highest at station 1. We nearly bought a machine there, but this would have been a mistake. A huge spike in demand caused a very large queue at station 3 and caused our revenues to drop significantly. Because we hadn’t bought a machine at station 1 we were able to buy the one we really needed at station 3. This taught us to monitor the performance of the machines at the times of very high order quantities when considering machine purchases. After making enough money, we bought another machine at station 1 to accommodate the growing demand average by reducing lead-

time average and stabilizing our revenue average closer to the contract agreement mark of $1250. Right before demand stopped growing at day 150, we bought machines at station 3 and station 1 again to account for incoming order growth up until that point in time. At this point we knew that demand average would stabilize and if we could make sure our revenue stayed close to the contract mark we wouldn’t need any more machines. Our assumption proved to be true. Station 2 Priority We found the inventory process rate at stations 1 and 3 to be very similar. Thus we adopted a relatively simple method for selecting priority at station 2. For most of the time, step 4 was selected as the step to process first. This is because we had more machines at station 1 than at station 3 for most of the simulation. When this was the case, station 1 would feed station 2 at a faster rate than station 3. When demand spiked station 3 developed queues if the priority was set to FIFO because station 1 could process the inventory quicker. If priority was set to step 4, station 2 would process the output of station 3 first, and inventory would reach station 3 from station 1 at a slower rate. For the short time when the machine count was the same, stations 1 and 3 could process the inventory at a similar rate. Thus we wanted the inventory from station 1 to reach station 3 at a rate to effectively utilize all of the capability of the machines. To accomplish this we changed the priority at station 2 back to FIFO. We never saw a reason to set the priority to step 2 because we never had more machines at station 3 than at station 1. Setup Times and Batch Size Before purchasing our final two machines, we attempted to drop the batch size from 3x20 to 5x12. We thought because of our new capacity that we would be able to accommodate this batch size and reduce our lead-time. However, this in fact hurt us because of long setup times at station 1 and 3. However, we wrongly attributed our increased lead times to growing demand. At this point we purchased our final two machines. When this didn’t improve lead-time at the level we expected we realized that the increased lead-time was our fault.

We changed the batch size back to 3x20 and saw immediate results. The average queues at stations 1 and 3 were reduced. At this point we realized that long setup times at both stations were to blame. (It also helped when we noticed the sentence in bold in the homework description about making sure to account for setup times at each of the stations.) We further reduced batch size to 2x30 and witnessed slightly better results. We left batch size at 2x30 for the remainder of the simulation. Inventory Purchases Initially we didn’t worry much about inventory purchasing. We spent money that we made on machines to build capacity quickly, and we spent whatever we had left over on inventory. We didn’t consider the cost of paying $1000 a purchase versus the lost interest cost on the payment until demand stabilized after day 150 and we had resolved our problem with batch size and setup times. It also never mattered much because we never kept the money necessary to make an efficient purchase until this point. We did calculate reorder points throughout the process, but instead of calculating the reorder point as average daily demand multiplied by the 4 days required for shipment we used average daily demand multiplied by 5 days to make sure we always had enough inventory to accommodate orders. This was necessary because daily demand was not constant and had a high degree of variability. We came very close to stocking out several times, but never actually suffered the losses associated with not being able to fill orders. The cost of not receiving inventory in time with a promised lead-time of 0.5 days was way too high. When demand stabilized we calculated Qopt with the following parameters: D (annual demand) = 365 days * 12.5 orders/day * 60 units/order = 273,750 units S (order or setup cost) = $1000 H (annual holding cost per unit) = $10/unit * 10% interest = $1 Then Qopt = 2DSH = 23,400 units

We also set up financial calculations in a spreadsheet to compare losses on payment sizes due to the interest lost on the payment during the time until the next purchase was required. This method verified the earlier calculation by coming out very close at 22,600 units. We attributed the difference to daily compounding interest but were unsure. We set the purchase for 22,500 units because we often had units left over due to our safe reorder point. Thus our inventory would often increase to a point between our two calculated optimal purchase quantities. Final 100 Days Our final inventory purchase occurred shortly after day 447. This meant that there were about 111 days left in the simulation. Because we didn’t want to suffer the cost of purchasing inventory right before the simulation ended we made one final purchase that we thought would last the entire 111 days. We then set the reorder quantity and reorder point to 0. The costs of holding inventory at the end were approximately the same as running out of inventory. Cunder = $600/order Cover = $1200 (average revenue) - $600 = $600/order Qnecessary = 111 days * 13 orders/day * 60 units/order = 86,580 units This is the inventory quantity that we purchased and it is the reason we didn’t finish the simulation in first. In our final purchase we forgot to account for the inventory we already had when the purchase was made. Thus we spent $39,000 too much. Qpurchase = Qnecessary – Qreorder = 86,580 – 3,900 = 82,680 units...