MODULE 9 CLOSURE
Spring 2010
Compiled by Greg Kinney
“MUDDIEST ITEMS”
QUESTION:
Activity on Arrow/ADM/PERT. I am comfortable drawing up a CPM schedule, but when it comes to activities on arrows, I get confused and have to be very careful in what I am doing. I find CPM/AON/PDM much more clear, and much more intuitive. Does anyone actually use an activity-on-arrow setup or network?
ANSWER:
The answer is yes. Some people just prefer the nodes to represent completions rather than activities, and arrows representing activities rather than simply linkages. In reality, I think many more people would prefer AOA except that Microsoft Project is configured around the opposite, which has made everyone used to it. Note that AOA is the proper one to use if there is a time distribution.
QUESTION:
The muddiest part was (and still is for me) how to compensate for bad time estimations. I see how the variances work, but it seems if you have bad estimations, the calculations are gong to be inaccurate as well.
ANSWER:
This is absolutely true. The authors have done a great job talking about the statistical modeling approach, which is important and will become more so. But the foundation is still a quality estimate. I don’t think they have emphasized that enough. The subject is beyond the scope of the book, but it can’t be beyond the scope of the PM or the team.
QUESTION:
The muddiest item concerned “reverse critical” activities that the text refers to. This occurs “when the critical path enters the completion of an activity through a finish constraint, continues backward through the activity, and leaves through a start constraint”. Does this only occur with ADM/PERT or with PDM/CPM as well? What constitutes a finish constraint and start constraint? How can the critical path move backward through a forward-moving activity? This is not clear to me.
ANSWER:
One other student asked about this as well. Think of constraints as linkages. In a finish to start relationship, the start of Task B is constrained to follow after completion of Task A. This is what we’re used to. There are also start to start constraints, so that Task B can’t start until Task A is started. A good example: if Task A is complete mobilization of equipment to a site, you can’t start work at the site under Task B until Task A is at least started. To get to reverse critical: The text provides examples of this using Fig. 8-17. Two of the relationships illustrated are: finish to finish, and start to finish. If the normal critical path would go through the second task (the “finish” on start to finish, such as activities 7 to 8 on Fig. 8-17 and the second finish on finish to finish on tasks 10 to 11), then you can envision how a disruption on the start of 7 could affect the finish of task 8, etc. That’s what they mean. This is rare, of course. As for examples, the 7 and 8 situation given by the text involves priming walls (activity 7) and getting/installing wallpaper (activity 8), where you can’t hang wallpaper in a portion of the house you’re working on until the primer has dried 24 hours there. The Activity 10 to 11 thing involves a cruise example that doesn’t really resonate with me. They actually admit there are almost always better ways to represent the situation.
QUESTION:
It was never entirely clear to me why we continue to use both AON and AOA when it seems, to me anyhow, that it would be easier to drop one system and work with the other. It all seems to come down to preference between the two, but at the same time, it seems unlikely that both systems are still utilized just because of preferences.
ANSWER:
AOA is better for time distributions (e.g., PERT). Aside from that, it’s mostly a matter of taste.
QUESTION:
The muddiest part is about simulation. Does the term “Simulation” mean to merely use triangular distribution method to calculate the probability? I think beta distribution should also be one type of simulation methods.
ANSWER:
This is confusing for many students. The triangular distribution is just one of many that can be used. The beta distribution is another, as are normal distributions. It is just the simplest and most convenient one, and one that’s simple to implement whether on a spreadsheet or MSP.