Zelda Turner
EM 530 Applications in Constraints Management
Summer 2000
Multi-Tasking In A Multiple Project Environment
The project used for this presentation is the same used in class, the bead project. To meet the requirement of
a minimum of three projects an additional bead project was added, the "green project." The first file
shows the adaptation of the bead project to the software. Tasks that would have been measured in seconds were
given completion times of fifty hours as a default. This assumes that the fifty hours corresponds to an eighty-five
percent probability of completion. The linking of the tasks is as described in the bead project. All beads must
be sorted before they can be checked. All checking operations can be done simultaneously, but must be completed
before the beads can be turned, and so forth.
The second file shows the "multi-tasking" environment. Each task is broken into many segments to show
the effects of multi-tasking. The project was then subjected to the same probability distribution given in the
lecture materials by rolling the die and mapping to a number of days' worth of work as shown:
| Pips on Die |
No. of Days of Work |
| 1 |
1 |
| 2 |
3 |
| 3 |
5 |
| 4 |
6 |
| 5 |
7 |
| 6 |
8 |
The results of the run are tabulated as shown in the following table:
| Week |
Roll |
Mapped to |
Red Days Left |
Blue Days Left |
Green Days Left |
| 1 |
6 |
8 |
22 |
21.5 |
23 |
| 2 |
3 |
5 |
20 |
20.5 |
21 |
| 3 |
3 |
5 |
19 |
18.2 |
19 |
| 4 |
6 |
8 |
16 |
15.5 |
17 |
| 5 |
3 |
5 |
14 |
14.5 |
15 |
| 6 |
1 |
1 |
14 |
13.5 |
15 |
| 7 |
6 |
8 |
11 |
11.5 |
12 |
| 8 |
2 |
3 |
10 |
10.5 |
11 |
| 9 |
2 |
3 |
9 |
9.5 |
10 |
| 10 |
2 |
3 |
8 |
8.5 |
9 |
| 11 |
1 |
1 |
8 |
7.5 |
9 |
| 12 |
5 |
7 |
6 |
5.5 |
6 |
| 13 |
2 |
3 |
5 |
4.5 |
5 |
| 14 |
6 |
8 |
2 |
2.5 |
2 |
| 15 |
1 |
1 |
1 |
2.5 |
2 |
| 16 |
3 |
5 |
0 |
.5 |
0 |
| 17 |
1 |
1 |
0 |
0 |
0 |
Note that the projects were scheduled to be completed in about seventy-five days and still came in nearly two weeks
late, taking a total of seventeen weeks or eighty-five days once the effect Murphy was taken into account.
The third file is set up according to CCPM, with each task scheduled for only twenty-five hours. This assumes
a schedule time with only a fifty percent probability of completion, and is half of the time previously scheduled
with a buffer added of about half the hours removed from the total project estimated. No resource buffer was included
in this analysis, nor was the load completely leveled. There is a week where resource one is over-allocated.
These would have been easy enough to add, but their absence does not invalidate the impact of the demonstration
in this simple scenario.
The simulation was then run with the same mapping. The results of that run are tabulated as follows:
| Week |
Roll |
Mapped to |
Red Days Left |
Blue Days Left |
Green Days Left |
| 1 |
5 |
7 |
113.5 |
19 |
19.63 |
| 2 |
6 |
8 |
5.5 |
11 |
17 |
| 3 |
2 |
3 |
2.5 |
8 |
14 |
| 4 |
6 |
8 |
0 |
0 |
6 |
| 5 |
6 |
8 |
0 |
0 |
0 |
This included running out the project buffers, but completes the entire set of projects in 25 days or five weeks,
less than a third of the time used in the multi-tasking environment. This corresponds, roughly, to the times shown
in class for the bead projects. It is not clear to me whether or not probability or Murphy has been taken into
account twice, once in setting the probabilities of completion and again in rolling the die to determine the amount
of work completed in any given time frame. Perhaps skewing the roll of the die in a different way would model
the effects of Murphy better, once the fifty percent schedule probability has been taken into account. But it
is clear that even if the individual projects were given back the full time taken out in CCPM and the time was
used as additional buffer, this plan would still be completed much sooner than the multi-tasking plan for completing
a multiple series of projects.
Zelda.Turner@PSS.Boeing.com