Critical Chain - Eliyahu M. Goldratt [83]
"Correct."
"Ruth is talking about a situation where we are working on a step, call it step N, which is on a path we considered noncritical. Nevertheless, right now step N is delayed to the extent that it is causing the biggest penetration into the project-buffer. Doesn't this mean that right now, the longest chain of steps, longest in time, is starting at step N?"
"What are you saying?" Ted jumps in. "That midway through the project, we'll change the critical path? That's crazy." "Why?" I ask. I know why. I also see why it is necessary. But I have to gain time to think it over.
Ted seems stuck. His intuition is excellent, his verbalization lags behind. Fred answers for him; it's apparent to me that the think tank really thought it over. "We put feeding buffers only where a noncritical path merges into the critical path. Changing the critical path will necessitate changing the location of many feeding buffers."
"And that," Ted concludes, "will mess up the whole project. We can't do that."
"Agreed," Ruth says calmly. "But on the other hand, don't you see what will happen if we don't do it?"
"No!" Ted responds too hastily. The thought of rearranging everything every time we face a serious delay on a noncritical path scares him. It scares me too.
"If we don't do it," Ruth continues, "we're ignoring reality. Let's face it, whether or not we like it, right now the critical path does start at step N. And this path is not protected from disruptions in other paths by feeding buffers. It is also not protected by resource buffers. So the chance of recovery is reduced. On the contrary, there is a good chance that the delay will intensify. Don't you realize that we must rearrange the project?"
"We're doomed if we do, doomed if we don't," Ted concludes.
"That's our conceptual problem," Ruth says.
I control the wave of panic starting to build in me. Conceptual problem? Conceptual problem! What an understatement. It may ruin everything we have done!
Why haven't we seen it in reality? Four projects did finish much earlier than expected, so our method does work. Maybe it's because all these projects were already well along their way when we started. But in all cases we had feeding buffers that were exhausted; we had the situation Ruth described, so the problem cannot be as big as it seems. Where is the erroneous assumption the think tank is making?
I don't have a clue. I also don't have time. The class is waiting. Not just the class, Jim as well. I turn to the board and start to write the cloud. The objective is to finish the project on time. One necessary condition is what Ted said. We cannot afford to rearrange everything. Which means don't formally change the critical path. The other necessary condition is Ruth's point, we cannot afford to leave the true critical path exposed, which means we must formally change the critical path.
I use the trick Johnny taught me—concentrate on the arrow you dislike the most. Like Ted, I don't like rearranging the project midway; I don't want to formally change the critical path. It ruins the whole elegance of the solution, and will make it impractical. Moreover, we didn't have to do it in the projects we worked on. So what is the assumption there?
In order to not leave the true critical path exposed, we must formally change the critical path. Why? Because if we don't do it, the critical path will not be protected. Is it really the case? Now I see it.
Confidently, I ask the class, "What assumption do we make when we claim that if we don't formally change the critical path it will be exposed?"
They don't answer, so I ask again, "Are the original critical path and the true critical path very different from each other?"
"No," Ruth answers. "As a matter of fact, from the point where the two merge until the end of the project, they are the same. I see. So the only exposure is from step N up to, but not including, the step where they merge. In most cases there will be no additional feeding buffers involved, but what about the resource buffers