8.3. Probabilistic Activity Times
* Previously all activity time estimates were single
values
* By using a single activity time estimate, we
assume that activity times are known with
certainty, i.e. does not vary
* In reality, activity estimates can rarely be made
with certainty, thus “probable timespans” are
used in calcuations
‘To this point we have demonstrated the use of CPM and PERT network analysis for deter-
mining project time schedules. This in itself is valuable to a manager planning a project.
However, in addition to scheduling projects, a project manager is frequently confronted
with the problem of having to reduce the scheduled completion time of a project to meet
a deadline. In other words, the manager must finish the project sooner than indicated by
the CPM or PERT network analysis.
Project duration can be reduced by assigning more labor to project activities, often in
the form of overtime, and by assigning more resources (material, equipment, etc.).
However, additional labor and resources cost money and hence increase the overall project
cost. Thus, the decision to reduce the project duration must be based on an analysis of the
trade-off between time and cost.
Project crashing shorensthe Project crashing is a method for shortening project duration by reducing the time of
project time by reducing critical one or more of the critical project activities to a time that is less than the normal activity
acivitytimesat aoe. tine, This reduction in the normal activity times is referred to as crashing. Crashing is
achieved by devoting more resources, measured in terms of dollars, to the activities to be
crashed.
‘To demonstrate how project crashing works, we will employ the original network for
constructing a house shown in Figure 8.8. This network is repeated in Figure 8.19, except
that the activity times previously shown as months have been converted to weeks. Although
this example network encompasses only single-activity time estimates, the project crashing
procedure can be applied in the same manner to PERT networks with probabilistic activity
time estimates.Figure 819
The project network for
building a house
(Cash cost and cash time have a
‘near relationship.
Figure 8.20
Time-cost relationship for
crashing activity 1
In Figure 8.19, we will assume that the times (in weeks) shown on the network activi-
ties are the normal activity times, For example, normally 12 weeks are required to complete
activity 1. Furthermore, we will assume that the cost required to complete this activity in
the time indicated is $3,000. This cost is referred to as the normal activity cost. Nex, we will
‘assume that the building contractor has estimated that activity 1 can be completed in
7 weeks, but it will cost $5,000 to complete the activity instead of $3,000. This new esti-
‘mated activity time is known as the crash time, and the revised cost is referred to as the
crash cost
Activity 1 can be crashed a total of 5 weeks (normal time ~ crash time = 12 ~ 7 = 5
weeks), at a total crash cost of $2,000 (crash cost — normal cost = $5,000 — 3,000 =
'$2,000). Dividing the total crash cost by the total allowable crash time yields the crash cost
er week:
total crash cost __ $2,000
Toalenchtime = 5 = $400 per week r
If we assume that the relationship between crash cost and crash time is linear, then
activity 1 can be crashed by any amount of time (not exceeding the maximum allowable
crash time) at arate of $400 per week. For example, if the contractor decided to crash activ.
ity 1 by only 2 weeks (for an activity time of 10 weeks), the crash cost would be $800 ($400
per week X 2 weeks), The linear relationships between crash cost and crash time and
between normal cost and normal time are illustrated in Figure 8.20.
‘ust be based on.
02 4 6 8 10 12 14 WeeksAs etvities are crashed, thecriti-
cal path may change, and several
‘paths may become critical
Table 84
Normal activity and crash data
for the networkin igure 8.19,
°° Figure 8.21
Network wth normal activity
times and weekly acy
crashing costs
‘The normal times and costs, the crash times and costs, the total allowable crash times,
and the crash cost per week for each activity in the network in Figure 8.19 are summarized
in Table 8.4.
Recall that the critical path for the house-building network encompassed activities
1-+2—+4—+7, and the project duration was 9 months, or 36 weeks. Suppose that the
home builder needed the house in 30 weeks and wanted to know how much extra cost
‘would be incurred to complete the house in this time, To analyze this situation, the con-
tractor would crash the project network to 30 weeks, using the information in Table 8.4.
‘The objective of project crashing is to reduce the project duration while minimizing the
cost of crashing. Because the project completion time can Be shortened only by crashing
‘Normal Crash. ‘Total Allowable Crash
Time ‘Time —»«Normal «= Crash = Crash Time Cost per
‘Activity (weeks) (weeks) __Cost Cost, (weeks) Week
T 2 7 ‘$3000 5,000 5 $00
2 8 5 2000 3,500 3 00
3 4 3 ‘4000 7.000 1 3,000
4 2 9 0,000 ‘74000 3 7900
5 4 1 500 1,100 3 200
‘ 4 1 500 1100 3 200
7 4 3 sooo _22.000 1 7000
$7500 $110,700
activities on the critical path, it may turn out that not all activities have to be crashed.
However, as activities are crashed, the critical path may change, requiring crashing of pre-
viously noncritical activities to further reduce the project completion time.
‘We start the crashing process by looking at the critical path and seeing which activity
hhas the minimum crash cost per week. Observing Table 8.4 and Figure 8.21, we ee that on
the critical path activity 1 has the minimum crash cost of $400. Thus, activity 1 will be
reduced as much as possible, Table 8.4 shows that the maximum allowable reduction for
activity 1is 5 weeks, but we can reduce activity 1 only to the point where another path becomes
critical When two paths simultaneously become critical, activities on both must be reduced
by the same amount. (If we reduce the activity time beyond the point where another path
becomes critical, we may incur an unnecessary cost.) This last stipulation means that we
‘must keep up with al the network paths as we reduce individual activities, a condition that
‘makes manual crashing very cumbersome. Later we will demonstrate an alternative
‘method for project crashing, using linear programming; however, for the moment we will
pursue this example in order to demonstrate the logic of project crashing.
It turns out that activity 1 can be crashed by the total amount of 5 weeks without another
path’s becoming critical because activity 1 is included in all four paths in the network.
. $500Figure 822
Revised network with activity 1
crashed
Crashing this activity results in a revised project duration of 31 weeks, ata crash cost of
$2,000. The revised network is shown in Figure 8.22.
i
Ry,
$3,000
‘This process must now be repeated. The critical path in Figure 8.22 remains the same,
and the new minimum activity crash cost on the critical path is $500 for activity 2. Activity
2 can be crashed a total of 3 weeks, but because the contractor desires to crash the network
to only 30 weeks, we need to crash activity 2 by only 1 week. Crashing activity 2 by 1 week
does not result in any other path’s becoming critical, 30 we can safely make this reduction.
Crashing activity 2 to 7 weeks (i.e. a L-week reduction) costs $500 and reduces the project
duration to 30 weeks.
‘The extra cost of crashing the project to 30 weeks is $2,500. Thus, the contractor could
inform the customer that an additional cost of only $2,500 would be incurred to finish the
house in 30 weeks.
‘As indicated earlier, the manual procedure for crashing a network is very cumbersome
and generally unacceptable for project crashing, Itis basically a trial-and-error approach
that is useful for demonstrating the logic of crashing; however it quickly becomes unman-
ageable for larger networks. This approach would have become difficult if we had pursued
even the house-building example to a crash time more than 30 weeks.