STEP-BY-STEP CRITICAL CHAIN IMPLEMENTATION
Sequence of Events
Day 1—The first day was devoted to preparation for onsite activity to commence with the H-46 Rotary Wing Team on Tuesday, September 7,1999. Introductions and discussions took place between Goldratt Institute personnel, the Depot Commanding Officer, his staff program coordination manager, the project manager, and the external LMI observer/facilitator. This meeting was held at a nearby Holiday Inn on Labor Day 1999. The depot personnel who would be participants were defined, a course of action for the week was discussed, and administrative issues were handled. A tentative objective was discussed for a pilot implementation and study of the H-46 rotary wing (helicopter) to achieve a dramatic 50 percent reduction in repair throughput time from an average of 185 days to 90 days.
Day 2 The H-46 Team met with the Goldratt Institute personnel for project startup. This session eased into education on the use of TOC and critical chain concepts while applying them to worthy depot issues. The team and facilitators focused on defining the project, surfacing obstacles, conducting an exercise to demonstrate the effects of multitasking, reviewing simulations of project planning probabilities, and introducing the concept of buffers. The basic questions dealt with were as follows:
• What to change—identifying the core conflict
• What to change to—constructing a solution
• How to cause the change—devising an implementation plan.
Sequencing was important in how the above questions were handled. The class gave feedback on obstacles in their repair processes and associated planning during this session.
Day 3—TOC and critical chain concepts education continued. The core of the training message on this day was related to probability and buffer construction. For example, assume human nature tends to lead to estimated task times that have a 90 percent probability of success. If we pull out 50 percent as the safety time for tasks along the critical path and place at the end of the project as a buffer, then we can focus on fully utilizing resources up front to stay on track, yet the protection is there and visible if we eat into the buffer. We can also protect the longest path from variability in connected shorter paths by creating feeder buffers at their intersection with the critical path (critical chain). Then we may be able to cut the buffers in half, as they probably overstate the risk, in that most likely only half of the 50:50 time estimates will be late. The result is a reasonably aggressively timed project plan with a safety net of buffers. Several of the basic concepts are described in more detail below as presented in this session:
The Student Syndrome
The student syndrome is characteristic of the typical work pattern of many people. They tend to do less than a third of the work on an activity during the first two-thirds of the activity duration. Then they do two-thirds of the work during the last third of the activity duration. Just think of how many students addressed term papers during their school years by waiting until the day before, and consider that how this behavior tends to spill over into their adult lives. After consuming much of their available time without sizeable progress, even if they then work at 100 percent of their capacity to complete two-thirds of the work in one-third of the time remaining, there is no chance to keep to the planned project activity duration with a quality result. There is an opportunity to significantly accelerate a project, with no additional cost or complexity, by just getting people to start early and work at 100 percent until complete. This approach positively impacts quality, as there does not have to be a rush at the end to make a schedule at the expense of covering all the details.
Managers tend to use an early start schedule for non-critical path activities as well and earlier than is necessary to meet the schedule date. People working on such activities may realize that there is slack in their activity and this may well encourage the student syndrome effect. This can also lead to confusion over what the real priorities should be and jam the process with too much work-in-process.
During one discussion, die H-46 Team agreed that people probably give task durations they are likely to meet 90 percent of the time or more. The thought is that most people think the task duration they are asked for is expected to be met—period, so they give times that they believe to be in the range of 95+ percent probability. Typically, they may give the longest time they can remember it taking them in the past, but certainly not one that is expected to be exceeded 50 percent of the time. The team also perceived that the existing operational standards might be based on a similar perspective. Certainly the standards developers had not purposely set standards diat would only be achieved 50 percent of the time. This then led to a mapping of current related repair processes and whether they could be rearranged to improve flow and utilization of constrained resources. This led to challenging of previously constructed routers for MRP.
If we take a conservative case, 50 percent probability activity duration would be about half of the duration we normally experience. Therefore, we have at least a 100 percent contingency built into the initial activity duration estimates or even more.
To Be Continued