E M 530 Applications in Constraints Management Goldratt Simulator Assignment Notes: I worked with Brennan Pecha on this assignment. Parameter Set 310
In the 310 parameter set, the blue operation was the constraint. It was required for products D and F, and had two lengthier run times of 28 minutes and 14 minutes. In addition, in order to produce product D, two runs of the blue operation were needed. Thus, in my scheduling, I used blue as the constraint factor. For the first scheduled run, I set up the schedule similar to what was recommended in the slides. For the blue constraint, I made 10 of F5, then 15 of E5, then 15 of C5. This was followed by a sec ond ‘batch’ of constraints scheduled as 30 of F5, 35 of E5, and 35 of C5. Product A was scheduled as the free product. A constraint buffer was set to four hours, and a shipping buffer set to 8 hours. Result 1 – The first run turned out relatively well, and is seen in Fig. 1 below:
Figure 1 – Parameter Parameter 310 run with schedule according to recommendation by Dr. Holt
It was noted that the blue operation was in setup or production 97% of the time. It was not, however, at 100% efficiency, meaning that the constraint buffer was penetrated a few times. However, the factory did pretty well and earned a profit of 2405 over the course of the week. One thought that we had was that the setup time of green, which resulted in 46% of green total operation time, could have led to the constraint buffer being penetrated. penetrated. Result 2 – 2 – For the second run, the constraint ‘batches’ were reduced in half. Thus, the constraint batches
were: 10 of F5, 15 E5, 15 C5, 15 F5, 15 E5, 15 C5, 15 F5, 20 E5, and 20 C5. Doing this should give more
time for the prior operations to the buffer to setup and produce WIP, which would improve the constraint buffer. The results of run 2 are seen in Fig. 2 below:
Figure 2 – Parameter 310 run with smaller rotations of the constraint product
This run was significantly more successful, and the plant had no WIP at the end of the week. It was also able to meet all of its demand, resulting in a profit of 5450. Interestingly, the total used time of the blue constraint was reduced, with only 93% of time occupied. The green operation had a significantly improvement on setup time used, with only 34%, compared to the first run in which 46% of the time green was in set up. I concluded that although blue is the constraint, the green operation significantly affects the ability of the plant to keep the constraint supplied with WIP, and as a result was the second most important operation in the plant.
Parameter Set 312
The 312 parameter set proved to be significantly more difficult than I thought. The batch sizes were set to 20, which meant that a process needed to have 20 WIP in front of it before it could start working. As a result, getting a good constraint buffer was next to impossible. In my very first run, I came to the realization that even though the F5 had 10 WIP in front of it, this would not help me at all, because I needed to run in batches of 20. This meant that I had to start the F process chain with 40 raw material, regardless of the 10 WIP already in front of the blue constraint there. In the 2nd run, I set up the blue constraint to work in groups of 20, so 20 of F5, then 20 of E5, 20 C5, and repeat once. The results of run 2 are seen below:
Figure 3 – Parameter 312 run with ‘batches’ of 20 for the constraint
The run overall was pretty bad, with a loss of -6400 after one week. The blue constraint was only producing or in setup 13% of the time. I observed that this was due to the need for a long time before a constraint buffer could be built up. It might be possible if there were already 20 WIP in front of each operation in the plant that the plant could run somewhat efficiently. However, even then the batch sizes of 20 make it difficult to move machines around as needed. Result 2 – In the second run of parameter 312, I ran the entire simulation manually. I decided to neglect
product D altogether, and focus on making only products A and F. I decided this because the plant should be able to make more profit if it doesn’t have much inventory at the end of the week, so my goal was to push all the products through A and F as quickly as possible. This turned out a lot better than the previous scenario, but the plant was still unable to pull a profit, as seen in the figure below.
Figure 4 – Parameter 312 second run with manual control, focus on only products A and F.
The inventory left in the system was much lower, and had a value of only 1750, compared to the earlier run of 8050. Thus, my goal of reducing inventory in the system was successful. I managed to ship 35 units of product A and 40 units of product F. The cyan process ran out of time at the end and could not finish the last 5 units of product A. If all of product A had shipped, the plant would have been a good bit closer to breaking even. Parameter Set 350
In this scenario, I first noted that there was already a large buildup of WIP in front of the blue operations, which indicates that the constraint should be the blue operation. Product A is a free product, and it looks like products D and F both use similar amounts of the blue constraint time. Looking at the octane calculation for D and F, it was found that product D has an octane of $7.80/min and F has an octane of $8.38/min, which was only slightly higher. Only 10 of product A is needed, and there is already WIP ready to make it, so I scheduled product A to be finished in the first 8 hours. This was followed by blocks of 10 production for each blue constraint, and then two sets of 20 for each constraint. The CCR buffer was set to 6 hours and the shipping buffer to 12 hours. This resulted in a successful run, as seen in the figure below:
Figure 5 – Parameter 350 successful run
The blue constraint was working or in setup 91% of the time. Almost all of the product demand was met, except for 6 units of product D, and this resulted in a profit of 2130 for the week. If the blue constraint had been a bit more efficient, then it may have been possible for all of the demand for product D to have been met.
Parameter set 360
In this scenario, I originally thought that the cyan should be the constraint, since there is only one machine available for cyan with a setup time of 60 minutes. Working according to this, I set up the constraint as cyan, and started working from left to right on the products. Although this worked adequately, much of the product demand could not be met, as seen below:
Figure 6 – Parameter 360 run with cyan set as constraint
I saw that cyan was not active much of the time, with only 31% of time in setup or production. On the other hand, the blue operation was working much more, with a production and setup time of 81%. I also noticed that more variability had been introduced in this scenario, with the machines breaking between 3-7% of the time. At this point, I did two revisions to the scheduling. Firstly, I decided that the blue operation is more constrained, since it has a massive setup time of 240 minutes. Even though there are 2 blue machines, the long setup time and long process time needed for each blue operation means that blue requires more process time than cyan, as can also be seen by the fact that cyan was not in operation much. Next, I scheduled the blue constraint as follows:
This was chosen so that the blue pathways that did not have branches were worked on before those that had branches. This setup worked significantly better, with the results seen below:
Figure 7 – Parameter 250 run with blue constraint
Nearly all of the product demand was met, with only product G missing demand by 1 (19 sold out of 20 needed). This resulted in a great profit of 4440 and a low inventory value of 100. The cyan was actually used more this time, with 67% total operating time, and the blue constraint was operating (or broken) 95% of the time. From this scenario, it became apparent that although the cyan process initially appeared to be constrained, the very long setup time of the blue process makes it the real constraint. Since the operating time needed to finish the product (cyan process) ranged from only 3-7 minutes, it did not need much time to create the finished product. On the other hand, the blue operation needed between 15 and 45 minutes. Parameter Set 390
This scenario was interesting, and looked like 2 constraints could be possible. I did scheduled runs with either the blue process or the red process as a constraint, and they both worked well, although scheduling the blue process as a constraint resulted in a slightly better performance, with all products being completed by the end of day 4. The octane calculation using the blue operation as the constraint showed that: A= $30/3 = $10 B = $277/35 = $7.91 C = $30/4 = 7.5 D = $203/30 = 6.76 E = $42/5 = 8.4
F = $186/42 = 4.43 G = $44/2 = $22 H = $226/30 = 7.53 The blue constraint was scheduled accordingly, and this resulted in:
Figure 8 – Parameter 390 with blue scheduled as the constraint
Actually, in this system, it did not appear that there was any ‘real’ constra int, as blue was only in
production or break 59% of the time. The factory could probably have handled 30-40% more weekly demand before a constraint appeared. Another run with red scheduled as the constraint gave the following results:
Again, all of the demand was met, although it took a little longer time (4 days 2 hours instead of just less than 4 days).
