EM 530 Applications In Constraints Management

Hilbert Robinson

Critical Chain Project Management/DBR Assignment

I want to submit something I am in the process of adopting/developing for the shop floor. Have you ever heard the term "Rabbit Chasing?" It is used in our context by the lean manufacturing folks to describe a particular group behavior that is encouraged as an antidote to the batch scheduling of common assembly operations.

Here is an example of how it would work. Say you have a process, which takes several hours to be completed by several mechanics, but they are all doing essentially the same five or six steps. The conventional practice is to assign each mechanic hes share of the work, determined by location usually, and apportioned to ensure that the work is balanced across the crew of mechanics. Lets say the process involves drilling 6 holes per square foot in a 12 SF plate of aluminum using four mechanics. The hole drilling process itself consists of five steps per hole, mark, pilot, full size, ream and deburr. The plate would be divided into 3 SF segments and each mechanic would proceed to mark out all their holes. The mechanic would then pick a starting location and pilot all the marked holes. Next would come the full size drilling of all the holes. Eventually all the holes would be deburred.

In the Rabbit Chase concept, one mechanic would start the marking out process and as soon as there was room to work behind the first mechanic, the second mechanic would start piloting the first set of holes marked. As soon as there was room, a third mechanic would now start at the same starting location and bring the first set of piloted holes up to full size. Then would come the reaming mechanic and the deburring mechanic. Once all five mechanics had something to do, they would be chasing each other across the plate like rabbits, eventually drilling all 72 holes together as a team.

This concept is easy to implement when the five steps in the process take about the same time. In the case I am working on, they do not. It also assumes that one team is sufficient to meet your overall process completion time objectives. It is not. Just to make matters a little more interesting, the sheet of aluminum gets thicker as you move from one edge to the next and the number of holes per square feet is different for each square foot.

I decided to attack this problem as a Job Shop Drum Buffer Rope scheduling problem, in which each square foot represents a different product type and each product type follows the same routing sequence through each of five machines in the shop. If I need more than one team, that will be treated as similar to installing another identical production line (or cell). I considered defining a unit of production as each individual hole produced. Then I realized that this still did not give me the lowest common denominator, since the holes also differed in diameter. This meant that I would have to go to a finished hole surface area or volume if I really wanted a common unit of measure for pacing purposes.

Since striving for a balanced pace was promising to become such a complicated exercise, I knew DBR would have to be the answer. Why move mountains to balance everything out when we know how to maximize throughput when we have a bottleneck? Sometimes I even have to remind myself that a bottleneck is a good thing. This is so easily lost in an environment where balanced lines are promoted as the ideal situation to strive for.

The most interesting part of this DBR version of the "Rabbit Chase" concept is that the resource at each step is now the actual tool, which accomplishes the task, and not the mechanic who operates the tool. What this means is that for a five step process, depending on the relative speed of each process compared to the bottleneck operation, you can accomplish it with less than five mechanics. This is similar to the traditional cell concept where a few machine operators keep several machines going by rotating through all the machines performing set ups and change overs as needed to keep the various products flowing through the cell. In this case, the "machines" are hand tools, which can not operate by itself without an operator holding on to the tool. However, when a machine is standing idle (which it will in an unbalanced line) the operator does not have to stand idle. Also, we will know if we have enough operators to go around if the slowest operation is never starved or blocked. Here we will institute a policy to create "virtual machine blockage" conditions when the bottleneck has produce several "parts" only to have them sit in a queue waiting to be processed by the next machine in the five step process. In the event of a blockage condition, the operator will have to leave the bottleneck machine and go attend the next machine downstream to clear the backlog before returning to the bottleneck machine for further processing.

Now that we have established a pace setter machine for each team, the idea is to manage this machine closely so that it is never sitting idle. This now becomes the main job of the shop manager for that day. Every minute lost on this machine is lost to the entire shop, while every minute gained here is gained to the entire shop. Here is where we will focus process improvement and quality improvement efforts. Similarly, the processes which take less time, we will continue to do as fast as we can and then use the excess capacity if and when we can to help speed up the constraint machine.

In a nut shell, here is what we will be doing in the next few days.

1. Identify the process step that is the slowest.

We have detailed time study data, which can tell us the time each of the 72 products take through each step. This should allow us to quickly zero in on the step that is the slowest across all products. (Create a spreadsheet with products across the top and process steps down the side and plug in from the time study reports the time required for each of the five steps or each product listed in the order of the steps and with the product listed from inboard to outboard.)

2. Get all the production time we can get out of this step by making sure it is running around the clock, or as close as possible. We will make sure extra drills are set up and ready for the next sized hole and with sharpened drill bits so no time is lost to tool change. This set up will be the responsibility of the other operators while they are waiting on the pace setter since they have faster processing times for their steps in the process.
3. Monitor all other steps in the process to make sure they do not over whelm the pace setter by getting to far ahead or too far behind.
4. The extra processing capacity that would be used to get ahead should be used to assist the bottleneck step instead.

5 Once the process is well understood, return to step one.

The process described above is actually a sub process that needs to be completed before another sub process consisting of a similar number of smaller activities/operations can start. In fact, there are five such sub processes in total, which together constitutes the construction of a major subassembly. The overall construction process is currently scheduled and managed as a traditional milestone driven construction project. We will eventually link all the precedence diagrams to create a large network of entire process. This will give us the opportunity to schedule the whole process as a critical chain project schedule while running a nested drum buffer rope schedule and the micro process level. I think this would make a great demonstration project of the concepts discussed in class on how to schedule the whole Boeing Everett Facility using a series of nested schedules tied to a virtual drum.