Dr. Gary Richardson
Program Director
Technology Project Management Program
University of Houston

I suppose what you suggest re milestone could be done. It would not be part of Full Monte though; it would I suppose be a button on the add-in ribbon to do that between two selected tasks?

I did wonder if it might be possible to support the introduction of a dummy milestone between the end of a predecessor and the end of a successor, which is how I work around MSP’s limitation.

I presume that you overlay the native MSP data when you imtroduce a branching point.

Re. point 1, There is not much we can do because Project blocks this on input. Our own code makes no assumptions like that. Unlike the way i understand other systems work we do not rely on Project’s time analysis algorithms or even their date calculations. (This is why we are so fast.) So Full Monte could replace their regular time analysis — basically just doing one iteration with fixed durations — but again how do we get the data in? I suspect it would get blocked even if you tried to create those relationships programmatically though I have not tried.

Point 2 is a good idea, though regrettably we have not included it in 1.0. (I suspect there would be some pretty hairy math involved for the beta distribution.) We do document that for the normal, lognormal, and beta distributions, we interpret the input values as 3-sigma limits, which from memory I think means about a 99.5% confidence interval. Even then it’s approximate in the case of the beta. We also truncate the distribution at these points to avoid the odd outliers which might confuse people. (For the uniform and triangular distributions of course we treat the input as absolute.)

Two queries:

1 Having augmented the network logic capabilities of MSP to include branching, is there any chance of overcoming its ban on having SS and FF links between the same two tasks? This is very useful when modelling high level representations of projects with overlapping activities such as Design, Procurement, Construction. Primavera Risk Analysis allows it but one has to use dummy activities to have the same effect in @Risk for MSP.

2 Are you making provision to define distributions using percentile points instead of extreme values. That is, if I elicit information from a client in the form of forecast P10, Likely and P90 outcomes, which has all sorts of benefits for the process, can I use those directly in your tool to define the distribution or would I need to extrapolate to the extremes and use those values?

Steve