Project Scheduling: Pert/Cpm: Learning Objectives

ste13809_ch08_359-392.
qxd 1/27/09 7:28 AM Page 359
Project Scheduling:
CHAPTER 8
PERT/CPM
LE A R NI NG OB J E C T I VES
After completing this chapter, you should be able to:
1. Describe the role and application of PERT/CPM for 7. Compute the project completion time given a
project scheduling. certain level of probability.
2. Define a project in terms of activities such that a 8. Find the least expensive way to shorten the
network representation can be developed. duration of a project to meet a target completion
3. Develop a complete project schedule. date.
4. Compute the critical path, the project completion 9. Formulate the crashing problem as a linear
time and its variance. programming model.
5. Convert optimistic, most likely, and pessimistic 10. Formulate project scheduling as a linear
time estimates into expected activity time programming model.
estimates. 11. Know some of the specialised software available
6. Compute the probability of the project being in the market for scheduling and tracking project
completed by a specific time. activities.
C HA PT E R OU T LI NE
8.1 Introduction 360 8.8 Using Linear Programming in Project
8.2 A Project Example: Replacement of an Airport Scheduling 378
Gate-Management System 361 8.9 Project Scheduling Software 378
8.3 Project Network Representation 361 Summary 379
8.4 Project Scheduling with Deterministic Activity Glossary 379
Durations 364 Solved Problems 380
8.5 Project Scheduling with Probabilistic Activity Discussion and Review Questions 384
Durations 368
Problems 385
8.6 Uses of Simulation in Project Scheduling 372
End of Chapter Case 390
8.7 Project Crashing 372
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 360
360 Part 2 Deterministic Decision Models
8.1 INTRODUCTION
Project scheduling is at the heart of many decision problems that are, in nature, non-
repetitive, unique, and clearly defined in terms of scope, objectives, and time frame. These
decisions are referred to as a project; an interrelated set of activities directed towards the
accomplishment of a unique, often major outcome and that have a definite starting and end-
ing point. Every organization may be involved at any time in a project of any size, duration,
and complexity level. Examples of typical projects are setting up of a stage for a rock concert,
the construction of a new plant or facility, the design and the marketing of a new product or
service, the redesign of a business process, the construction (or repair) of a bridge, the devel-
opment of a new drug, and the acquisition and installation of an enterprise planning system.
All these projects consist of several activities that have to be completed and some of them are
interdependent, that is, they cannot start before the completion of some other activities. In
many situations, managers face the challenges of planning, coordinating, and monitoring
these activities so that the project of interests is completed on time and within the allocated
budget. Project management provides a number of approaches to cope with these challenges.
This chapter introduces the critical path method (CPM) and the program evaluation
review technique (PERT), two management science techniques developed in the late 1950s
to plan, schedule, and control large, complex projects with many activities. These
approaches differ primarily on how the duration and the cost of activities are processed. In
the case of CPM, it is assumed that details about these inputs are known with certainty,
whereas for PERT, these details are not known with certainty. Both approaches use a net-
work representation (see Chapter 5) to display the relationships between project activities
and to help managers to address questions such as:
1. What is the total time required to complete the project (the expected total time for
PERT)?
2. What are the start and the completion times for individual activities?
3. Which critical activities must be completed as scheduled to meet the estimated project
completion time?
4. How much delay can be tolerated for non-critical activities without incurring a delay
in the estimated project completion time?
5. What is the least expensive way to speed up a project to meet a targeted completion time?
PERT also provides answers to the following additional questions:
6. What is the probability of completing a project within a given time frame?

7. What is the variability in the project completion time?
We will first introduce a project example that will be used in the chapter to illustrate the
project graphical representation and to discuss both approaches, CPM and PERT. The
chapter ends with a discussion on time–cost trade-offs in project acceleration and on the
use of linear programming in project acceleration as well as in project scheduling. It is
worthwhile to note that project scheduling, the focus of this chapter, is only one of the
major phases involved in the management of a project. Project planning and project con-
trol processes, as well as their related challenges, are two other phases, not discussed in this
chapter, that are required to ensure the effective use of resources to deliver the project
objectives on time and within cost constraints. Other management variables that also mat-
ter include the senior managers’ support, a clear definition of roles and responsibilities, the
communication systems, and the human resource management practices.
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 361
Chapter 8 Project Scheduling: PERT/CPM 361
8.2 A PROJECT EXAMPLE: REPLACEMENT OF AN

AIRPORT GATE-MANAGEMENT SYSTEM
AC Inc. is a full-service, major airline company in the international market. It relies on a
user-friendly gate management software system at its hub airport to decide how to assign
aircrafts to gates. Based on a flight schedule and various business criteria, the system serves
for long-, medium-, and short-term planning of gate requirements. It also serves for real-
time allocation of gates during the day of operation. The company faces the challenge of
replacing its current system because its hardware platform is obsolete and no longer sup-
ported. As well, a considerable increase of activities at its hub facility is making the current
environment less suitable. The company is concerned with the overall cost involved if a
manual process would have to take place in case of a breakdown of the current system.
Therefore, a project has been set-up with a mandate of acquiring and implementing a new
gate management system to replace the current system. Soad El-Taji is project manager at
AC Inc., and has many years of experience. Her exemplary performance in similar projects
in the past has earned her the confidence of senior management. It was with unanimity
that she was called once again to manage this acquisition project to ensure that it was suc-
cessfully completed on time. As in the past, she resorted to PERT/CPM approaches. The
first step she completed was to accurately establish a list of activities that needed to be
undertaken, their precedence relationships and the time estimates for each activity. Table 8-1
shows the list of activities. The immediate predecessors of an activity refer to those activi-
ties that must be completed prior to the starting time of a given activity. Similarly,
immediate successors of an activity refer to those that follow the completion of a given
activity. The “—” in this table indicates an activity without a predecessor.
After the completion of the project activities, Soad El-Taji wants to know how to better
visualize these activities and how to develop answers to questions listed at the end of the
previous section.
8.3 PROJECT NETWORK REPRESENTATION

A project network representation is used to depict the project activities and their relation-
ships. As discussed in Chapter 5, a network consists of a set of circles referred to as nodes
and lines connecting nodes together referred to as arcs. The two common approaches of a
project network representation are activity on node (AON) or activity on arc (AOA). In
the first approach, the nodes of the network represent the project activities and the arcs
show their precedence relationships. In the second approach, the project activities are
reported on arcs and nodes represent the starting or the completion of activities. We focus
in this textbook solely on the activity on node representation, given its large adoption in
many software packages. It is common in this approach to add one dummy source activity
node (referred to as Start) and connect it to all activities nodes with no immediate prede-
cessors as well as one dummy destination activity node (referred to as Finish) and connect
it to the project network with arcs from activities with no immediate successors. This
ensures that there is one starting point and one finish point in the project network.
Figure 8-1 shows the network representation of the airport gate management system acqui-
sition project example. Note that a dummy Finish node is added to ensure that the project
network has one starting node and one finish node. Since there is only one node without a
predecessor (A), no dummy Start node is added in this network.
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 362
List of Activities for the Airport Gate Management

TABLE 8-1 System (AGMS) Acquisition Project
Immediate Estimated
Activity Description Predecessors Time (weeks)
A Set up the project acquisition team — 2

B Write down the software requirements A 2
C Develop a contractor evaluation grid B 1
that will be used to evalute proposals
D Identify and select potential contractors A 1
E Develop and send out a request for B, D 4
proposal to potential contractors
F Audit candidate contractors, select C, E 2
one contractor, negotiate and sign an
agreement contract with the
selected contractor
G Prepare the definition of functional F 5
specifications
H Develop a software testing plan G 2
I Software customization phase I G 12
J Purchase and install the hardware G 2
K Test the first release H, I, J 1
L Develop a training plan for key users K 1
O Train key users L, N 2
M Software customization phase II K 6
N Test the second release M 1
P Software customization phase III N 3
Q Test the final release P 2
R Software deployment and project Q 4
sign-off
For a small project network, one convenient way to determine its duration and critical
activities (questions 1 and 3 above) is through the enumeration of all the different paths in
the network. A path is a sequence of connected nodes in the network from the start node
to the finish node. The length of the path is given by the sum of the durations of the activ-
ities on the path. For the network shown in Figure 8-1, the corresponding paths are shown
in Table 8-2. In this list, path # 1 is the longest, while paths # 23 and # 24 are the shortest.
Path # 1 is critical, because any delay in the duration of an activity located in this path will
delay the entire project. For instance, increasing the duration of activity A by 2 weeks
will increase the length of path # 1 by 2 weeks for a total duration of 46 weeks. Whereas, for
path # 23 and # 24, the length will only increase to 20 weeks, respectively. Hence, in any
project network, the path with the longest duration is called a critical path and the corre-
sponding activities are called critical activities in that they must be completed as scheduled
to meet the scheduled project completion time. The estimated duration of the project is
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 363
FIGURE 8-1 Network Representation for the AGMS Acquisition Project
C J
M
B F N P Q R
A E G I K
O
L
D Finish
H
TABLE 8-2 List of Paths for the AGMS Acquisition Project
Path # Sequence of Nodes Length (total time in weeks)
1 A-B-E-F-G-I-K-M-N-P-Q-R-Final 44
2 A-B-E-F-G-J-K-M-N-P-Q-R-Final 34
3 A-B-E-F-G-H-K-M-N-P-Q-R-Final 34
4 A-B-C-F-G-I-K-M-N-P-Q-R-Final 41
5 A-B-C-F-G-J-K-M-N-P-Q-R-Final 31
6 A-B-C-F-G-H-K-M-N-P-Q-R-Final 31
7 A-D-E-F-G-I-K-M-N-P-Q-R-Final 43
8 A-D-E-F-G-J-K-M-N-P-Q-R-Final 33
9 A-D-E-F-G-H-K-M-N-P-Q-R-Final 33
10 A-B-E-F-G-I-K-M-N-O-Final 37
11 A-B-E-F-G-J-K-M-N-O-Final 27
12 A-B-E-F-G-H-K-M-N-O-Final 27
13 A-B-C-F-G-I-K-M-N-O-Final 34
14 A-B-C-F-G-J-K-M-N-O-Final 24
15 A-B-C-F-G-H-K-M-N-O-Final 24
16 A-D-E-F-G-I-K-M-N-O-Final 36
17 A-D-E-F-G-J-K-M-N-O-Final 26
18 A-D-E-F-G-H-K-M-N-O-Final 26
19 A-B-E-F-G-I-K-L-O-Final 31
20 A-B-E-F-G-J-K-L-O-Final 21
21 A-B-E-F-G-H-K-L-O-Final 21
22 A-B-C-F-G-I-K-L-O-Final 28
23 A-B-C-F-G-J-K-L-O-Final 18
24 A-B-C-F-G-H-K-L-O-Final 18
25 A-D-E-F-G-I-K-L-O-Final 30
26 A-D-E-F-G-J-K-L-O-Final 20
27 A-D-E-F-G-H-K-L-O-Final 20
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 364
therefore given by the length of the critical path. It is possible to find more than one critical
path in a network project, but all critical paths will have the same length. For the network
shown in Figure 8-1, the critical path is A-B-E-F-G-I-K-M-N-P-Q-R and its corresponding
length is 44 weeks. Therefore, the project’s estimated duration is 44 weeks. All activities in
the critical path are called critical activities, whereas the remaining activities (C, D, H, L,
and O) are non-critical activities.
The procedure described above is prohibitive for large project networks. In addition, it
does not provide answers to the scheduling time of individual activities and the delay that
can be tolerated for non-critical activities without incurring a delay in the scheduled proj-
ect completion time. In the next two sections, we discuss how techniques such as CPM and
PERT help managers to address these questions.
8.4 PROJECT SCHEDULING WITH DETERMINISTIC

ACTIVITY DURATIONS
For larger project networks, CPM provides a most efficient approach for project scheduling
when the duration of activities are known with certainty. The approach consists of find-
ing the earliest and the latest schedules to avoid delays in project completion. Therefore, for
all activities in the network the following information is computed:
1. The earliest start time (EST): the earliest time at which an activity can start if no
delays occur in the project;
2. The earliest finish time (EFT): The earliest time at which an activity can finish if no
delays occur in the project;
3. The latest start time (LST): The latest time at which an activity can start without
delaying the completion of the project;
4. The latest finish time (LFT): The latest time at which an activity can finish without
delaying the completion of the project.
A forward pass (from the starting node to the finish node) is used to compute the EST and
EFT, whereas a backward pass (from the finish node to the starting node) is used for the
LST and LFT. To determine the critical path and the project schedule, the approach consists
of calculating, respectively, the starting time and the completion time for each activity as
well as identifying the corresponding slack.
The basis of the forward pass is the EST rule that states that all immediate predecessors
must be completed before an activity can begin. Let t be the duration (estimated) of an
activity. If no delay occurs anywhere in the project, the earliest finish time for an activity is
EFT EST t (8-1)
It results from the EST rule that the earliest start time for each activity is equal to the largest
earliest finish times of the immediate predecessors.
For example, consider the network shown in Figure 8-1. The project-starting activity A
has no predecessor. Therefore, it can start as soon as the project starts, which we assume to
be time 0. The earliest finish time of activity A can now be computed as 0 2 2. Activi-
ties B and D have activity A as their immediate predecessor. The EST for B is 2 and the EFT
is 2 2 4. The EST for D is 2 and EFT is 2 1 3. Activity B is the only immediate
predecessor of activity C and its EFT is known. Therefore, the EST for C is 4 and the EFT is
4 1 5. Activity E has two predecessor activities, B and D, with known EFT (4 and 3).
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 365
FIGURE 8-2 Nodes Notation
EST EFT
A
2
LST LFT
The EST for D is 4, or the maximum between 4 and 3. The EFT of activity D is 4 4 8.
If we continue this forward pass, the EST and the EFT for all activities can be computed as
summarized in Figure 8-3, in which the node is expanded to include additional informa-
tion (as shown in Figure 8-2). We show in the top left side of the node, the activity name
(e.g., A); in the bottom left side, the activity duration; and in the top right side, the activity’s
EST and EFT. Later in our discussion we will place the activity’s LST and LFT on the bot-
tom right of the node.
The forward pass to obtain the EST and the EFT can be summarized as follows:
1. For each activity with no predecessor, set EST 0.

2. For each activity with known EST, calculate EFT using Equation 8-1, that is,
EFT EST t.
3. For each new activity where all immediate predecessors have known EFT values, apply
the EST rule to obtain the EST and step 2 to calculate EFT.
4. Repeat step 3 until EST and EFT have been obtained for all activities.
The basis of the backward pass is the LFT rule which states that an activity can start at the lat-
est time if and only if all its immediate predecessors are completed. Hence, the latest finish
time for each activity is equal to the smallest latest start times of the immediate successors’
FIGURE 8-3 EST and EFT Computation for the AGMS Acquisition Project
ES, EF
C ES, EF
4, 5 P 35, 38
ES, EF ES, EF
B 1 ES, EF J ES, EF
2, 4 F 8, 10 15, 17 M 3
2 28, 34
2 ES, EF ES, EF
2 6 Q 38, 40 R 40, 44
ES, EF
ES, EF E ES, EF ES, EF 2 4
A 0, 2 4, 8 K N 34, 35
ES, EF 27, 28
4 I
2 ES, EF 15, 27 1 1
G 12
10, 15 ES, EF
ES, EF ES, EF O 35, 37
D 5
2, 3 L
ES, EF 28, 29 2 ES, EF
1 H Finish 44, 44
15, 17 1
2 0
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 366
activities. Therefore, if no delay occurs anywhere in the project, the latest start time for
an activity that will result on not delaying the completion of the project is:
LST LFT t (8-2)
The backward pass starts by setting up the LFT of all activities without successors (includ-
ing the finish node) equal to the maximum EFT and then works backward from the finish
node to the starting node. The procedure can be summarized as follows:
1. For each of the activities without successors (including the finish node), set LFT equal
to EFT of the finish node.
2. For each activity with known LFT value, calculate LST using Equation 8-2, that is,
LST LFT t.
3. For each new activity where immediate successors have known LST values, apply,
respectively, the LFT rule to obtain the corresponding LFT and step 2 to calculate LST.
4. Repeat step 3 until LFT and LST have been obtained for all activities.
For the network shown in Figure 8-4, we set the LFT and the LST of the finish node equal to
its EFT 44 weeks. The immediate successor of activities R and O is the finish node. Hence,
the LFT is 44 weeks and LST 44 4 40 weeks for activity R. For activity O, LFT is 44
weeks and LST 44 2 42 weeks. Activity R is the immediate successor for activities Q.
Hence the corresponding LFT for activity Q is 40 weeks, whereas the LST 40 2 38
weeks. Activity Q is the immediate successor for activity P. The LFT 38 for activity P and
the LST 38 3 35 weeks. Activity P and O are immediate successors for activity N.
Hence, LFT minimum (35, 42) 35 weeks for activity N. If we proceed backward until
the starting node, the LFT and the LST for all activities can be computed as summarized in
Figure 8-4. For example, Activity G has activities J, I, and H as immediate successors. Its cor-
responding LFT min (25, 15, 25) 15 weeks and LST 15 5 10 weeks.
The slack time for an activity refers to the length of time that can be tolerated without
incurring a delay in the scheduled project completion time. The slack time per activity
needs to be calculated first to identify the critical path(s), by considering either the start
times or the finish times. Hence, for each of the activities in the project network, the
slack time can be calculated as follows:
Slack LST EST or LFT EFT (8-3)
FIGURE 8-4 LST and LFT Computation for the AGMS Acquisition Project
ES, EF ES, EF
C 4, 5 P 35, 38
ES, EF
B 2, 4 1 LS, LF ES, EF
ES, EF
ES, EF LS, LF
7, 8 J 15, 17 3
F 8, 10 M 28, 34 35, 38
2 LS, LF LS, LF
2, 4 2 LS, LF 2 LS, LF ES, EF
25, 27 6 Q 38, 40
8, 10 28, 34
ES, EF ES, EF
ES, EF E
4, 8 ES, EF ES, EF 2 LS, LF R 40,44
A 0, 2 ES, EF K 27, 18 N 34, 35 38, 40
4 LS, LF
2 LS, LF 4, 8 ES, EF
I 15, 27 LS, LF LS, LF 4 LS, LF
1 1 34, 35 40, 44
0, 2 G 10, 15 1 2 LS, LF 27, 28
15, 27 ES, EF
ES, EF
D 2, 3 5 LS, LF ES, EF O 35, 37
10, 15 L 28, 29
ES, EF 2 LS, LF
1 LS, LF H 15, 17 1
LS, LF 42, 44 ES, EF
3, 4 41, 42 Finish 44, 44
2 LS, LF LS, LF
25, 27
0 44, 44
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 367
Table 8-3 summarizes the activities’ slack times for the airport gate management system
acquisition project example. Activities A, B, E, F, G, I, K, M, N, P, Q, and R have zero or
no slack, meaning that these activities cannot be delayed without delaying the entire
project. Alternately, a reduction in the duration of one these activities will result in the
reduction of the entire project’s length. They are called critical activities and belong to
the critical path(s). The remaining activities (C, D, H, J, L, and O) are non-critical activ-
ities since they provide managers with some degree of freedom about when to start or
complete them without delaying the entire project. Therefore, Table 8-3 provides the
project manager with answers to some of the questions raised at the end of the intro-
duction section.
For the airport gate management system acquisition project example, the critical
path is A-B-E-F-G-I-K-M-N-P-Q-R-Finish, the estimated project completion time is
44 weeks, and each activity has to be completed according to the schedule shown in
Table 8-3. As stated earlier, note that it is possible for a project to have multiple critical
path(s). The slack times provide valuable information for the overall management of
the project. For example, an activity with smaller slack time (e.g., activity D with
1 week) will need tighter control than an activity with a larger slack time (e.g. activities
H and J with 10 weeks).
TABLE 8-3 Summary of Activities’ Start, Finish, and Slack Times
Activity EST EFT LST LFT Slack Critical Activity
A 0 2 0 2 0 Yes
B 2 4 2 4 0 Yes
C 4 5 7 8 3
D 2 3 3 4 1
E 4 8 4 8 0 Yes
F 8 10 8 10 0 Yes
G 10 15 10 15 0 Yes
H 15 17 25 27 10
I 15 27 15 27 0 Yes
J 15 17 25 27 10
K 27 28 27 28 0 Yes
L 28 29 41 42 13
M 28 34 28 34 0 Yes
N 34 35 34 35 0 Yes
O 35 37 42 44 7
P 35 38 35 38 0 Yes
Q 38 40 38 40 0 Yes
R 40 44 40 44 0 Yes
Finish 44 44 44 44 0 Yes
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 368
8.5 PROJECT SCHEDULING WITH PROBABILISTIC

ACTIVITY DURATIONS
The CPM approach assumes that the duration of activities are known with certainty and
the actual duration will turn out to be exactly as estimated. However, in practice this is not
always possible and many projects involve variability in activity times due to factors such as
lack of prior experience, equipment breakdown, unpredictable weather conditions, late
delivery of supplies, and others. PERT analysis is used when the duration of activities are
not known with certainty. It involves three types of estimates of the duration of an activity
instead of one single value as in the case of CPM:
1. The optimistic duration a the time an activity will take under the most favourable
conditions.
2. The pessimistic duration b the time an activity will take under the most unfavourable
conditions.
3. The most likely duration m the most realistic time an activity will require to be
completed, that is, the time an activity will take under normal conditions.
The duration of an activity is therefore assumed to have a beta probability distribution in

PERT analysis. Following this distribution, the expected activity time, te, and the variance
of the activity completion time, 2 can be obtained as follows:
te a 4m b (8-4)
6
1 b 6 a 2
2
2 (8-5)
Figure 8-5 illustrates the shape of the beta distribution, where at the two extremes we have
the two estimates a and b with a very small probability, whereas the third estimate m pro-
vides the highest point (mode) of the probability distribution.
In PERT analysis, the project completion time is computed in a similar manner as in the
CPM approach, but by substituting the three estimates of the activity duration with
the expected activity time, te , as obtained according to Equation 8-4, and by the variance of
the activity completion time, 2, as obtained according to Equation 8-5. Therefore, the
expected completion time of the project (p) can be derived as well as the variability in
the project completion time (p), as follows:
p max (EFT) max(LFT) (8-6)
sum of the expected duration for the activities in the critical path
P √ (sum of the variances of the duration for the activities in the critical path) (8-7)
Assume that, due to variability in activity times, the three estimates of activities duration for
the airport gate management system acquisition project example are as shown in Table 8-4.
(see columns 2–4). Therefore, by applying Equations 8-4 and 8-5 the mean and the variance
of the activity duration shown in the last two columns of Table 8-4 can be computed.
The estimated start and finish times for all activities according to the forward pass and
the backward pass were summarized in Figure 8-4. The critical path is A-B-E-F-G-I-K-M-
N-P-Q-R-Finish. The expected project completion time (p) is 44 weeks (given by the
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 369
FIGURE 8-5 Illustration of the Shape of the Beta Probability Distribution
x
a m b
Activity Duration
Activities Expected Times and Variances for

TABLE 8-4 the AGMS Acquisition Project
1 b 6 a 2
Mean, Variance,
2
ACTIVITY a m b te a 4m b 2
6
A 1 2 3 2 0.111
B 1 2 3 2 0.111
C 0.5 1 1.5 1 0.028
D 0.5 1 1.5 1 0.028
E 3 4 5 4 0.111
F 1 2 3 2 0.111
G 3 5 7 5 0.444
H 1 2 3 2 0.111
I 10 12 14 12 0.444
J 1 2 3 2 0.111
K 0.5 1 1.5 1 0.028
L 0.5 1 1.5 1 0.028
M 4 6 8 6 0.444
N 0.5 1 1.5 1 0.028
O 1 2 3 2 0.111
P 1.5 3 4.5 3 0.250
Q 1 2 3 2 0.111
R 2 4 6 4 0.444
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 370
maximum EFT or LFT). Given the data in Figure 8-4, the standard deviation for the proj-
ect equals 1.62 weeks, computed as follows:
p 冑 (var(A) var(B) var(E) var(F) var(G) var(I) var(K)

var(M) var(N) var(P) var(Q) var(R)
Using the numerical values in Table 8-4, we get:
p 冑 (0.111) (0.111) (0.111) (0.111) (0.444) (0.444) (0.028)

(0.444) (0.028) (0.250) (0.111) (0.444)
p √(2.64) 1.62
In addition to providing answers to questions about the project’s critical activities, the
start and completion times of activities, the expected completion time of the project, and
the variability in the project completion time, PERT analysis also answers questions such
as the probability of whether or not the project will be completed on time or, conversely,
what the project completion time will be, given a certain probability.
Probability of Whether a Project Can Be Completed

or Not by a Specific Deadline
Let c and d denote the possible project specified deadlines (assuming c d) and X the total
time required to complete the project. Due to the central limit theorem, which indicates
that the sum of independent random variables can be approximately represented by a nor-
mal distribution as the number of random variables becomes larger, the project comple-
tion is approximated by a normal distribution with mean p and standard deviation p.
Without loss of generality, the three possible situations of interests are illustrated in
Figure 8-6 and refer, respectively, to the probability that the project duration (X):
• does not exceed the deadline b, i.e., P(X d);
• does exceed the deadline b, i.e., P(X d);
• falls between c and d, i.e., P(c X d).
The approach consists of converting X into a standard normal distribution and determin-
ing the area under the normal curve using Table A in Appendix B. To that end the z value is
computed as follows,
xp
z (8-8)
p
where
x c or d
For the airport gate management system acquisition project, what is the probability of
completing the project within 46 weeks? The Z value for the normal probability distribu-
tion at x 46 is
z 4644 1.23. Therefore from Table A in Appendix B,

1.62
P(X 1.23) 0.50 0.3907 0.8907 or 89.07%. Hence, there is about 89% probability
that the project will be completed on time. Alternatively the Excel function NORMSDIST(z)
can be used to seek the probability corresponding to the z value.
What is the probability that the project lasts more than 46 weeks?
P(X
46) 1 P(X 46) 1 P(X 46) 1 0.8907 0.1093 or 10.93%.
There is a 11% probability that the project will not be completed on time.
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 371
Illustration of Possible Situations of Interest for Probability

FIGURE 8-6 Computation
P(X ≥ d)
X X
d
P(c ≤ X ≤ d)
X X
c d
Project Duration
P(X ≤ d)
X X
d
Project Duration
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 372
FIGURE 8-7 Project Completion Time Given a Certain Probability
P(X ≤ d)
X X
d
Project Duration
Project Completion Time Given a Certain Probability

In this case one must find the target project completion time (value of d) that corresponds
to a specified probability level, as shown in the shaded area of the standard normal curve in
Figure 8-7.
The approach lies on finding the z value using Table A in Appendix B. Therefore from
Equation 8-8 the following relation can be obtained:
x p z p (8-9)
For the airport gate management system acquisition project, what would be the com-
pletion time under which for the project has 95% probability of completing?
From Table A in Appendix B, z 1.65 for P(X b) 0.95. Alternatively, the Excel
function NORMINV(probability) can be used to seek the corresponding z value. Hence,
using Equation 8-9, x 44 1.65 1.62 46.67 weeks, that is, if the project completion
time is 46.67 weeks, the company will complete it on time with a 95% probability.
8.6 USES OF SIMULATION IN PROJECT SCHEDULING

Resorting to PERT analysis to cope with uncertainty presents a number of difficulties in
practice. The underlying assumption that the activity durations are independents is some-
times difficult to justify. In addition, the activity durations may not follow a beta distribu-
tion. One of the approaches to use when other distribution functions are involved is
simulation, the topic discussed in Chapter 13.
8.7 PROJECT CRASHING

It is common in project management that additional resources are used to either speed up
some activities to get the project back on schedule or to reduce the project completion
time. Late penalty costs, monetary incentives, cost savings, or strategic benefits are some of
the reasons for shortening a project completion time. Crashing an activity refers to the
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 373
speeding up or shortening of the duration of an activity by using additional resources.

These include overtime, hiring temporary staff, renting more efficient equipment, and
other measures. Project crashing refers to the process of shortening the duration of the
project by crashing the duration of a number of activities. Since it generally results in an
increase of the overall project costs, the challenge faced by the project manager is to iden-
tify the activities to crash and the duration reduction for each activity such that as the proj-
ect crashing is done in the least expensive manner possible. This section discusses a
procedure that can be used for small-sized projects. We also discuss how linear program-
ming (Chapter 3) can be used to investigate project crashing decisions, especially for larger
size projects. Figure 8-8 illustrates the activity costs and activity duration relationships,
where the normal time refers to the estimated activity duration used with CPM or PERT in
the computation of earliest (latest) start or finish times. The normal cost refers to the activ-
ity cost under the normal activity time. The crash time refers to the shortest possible time
to complete an activity with additional resources. The crashing cost refers to the activity
cost under the crashing activity time.
This relationship is assumed to be linear. Hence, for each activity a crash cost per
period (e.g., per week) can be derived as follows:
(Crash cost Normal Cost)
crash cost per period (8-10)
(Normal time Crash time)
FIGURE 8-8 Activity Cost and Activity Time Relationship
Crash
Crash cost
Activity Cost
Normal cost Normal
Crash time Normal time

Activity Duration
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 374
The general procedure for project crashing involves the following four steps:
1. Compute the crash cost per period for all activities using Equation 8-10.
2. Find critical path(s) in the project network using the normal times and identify critical
activities.
3. Select a critical activity with the smallest crash cost per week that can still be crashed,
in the case that there exists only one critical path. Otherwise, select one activity from
each critical path that can be still crashed and yield the smallest total crash cost per
period (including a common activity among critical paths). Crash the selected activity
or activities by one period. Update the length of paths.
4. Stop the procedure if the completion deadline is reached. Otherwise, check to ensure
current critical path(s) are still critical and find the new ones, if any. Return to Step 3.
To illustrate the procedure above, consider the network shown in Figure 8-9, which shows
the activities for a new product development project and their precedence relationships.
Start is a dummy activity with zero duration added to ensure that all activities have one
starting node.
Table 8-5 provides the information about the activities normal times and costs, crash
times and costs, the maximum crashing reduction in time as obtained by the difference
between the normal time and the crash time, and the crash cost per week as obtained using
Equation 8-10. For example, the normal time for activity C is 10 weeks and its cost is $45 000.
It can be shortened by up to 4 weeks at an additional cost of $36 000 or $9000 per week of
reduction.
The project critical path obtained by using normal times is Start-A-C-E-H. The esti-
mated project completion time is 28 weeks. Suppose that management wants to shorten
the project to 24 weeks to beat competition. Which activities should be crashed, and for
each crashed activity provide the total number of weeks crashed and the total cost. What is
the overall project crashing cost?
To reduce the project completion time from 28 weeks to 27 weeks, one of the activities
on the critical path needs to be reduced. Activity A has the lowest crash cost per week
among all critical activities ($3000). It is therefore selected and crashed by one week, that is,
the duration for activity is 4 weeks instead of 5. This shortens the project completion time
Project Network for the New Product Development

FIGURE 8-9 Example
F
A
H
C E
Start
D
G
B
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 375
Time and Cost Information for the New Product

TABLE 8-5 Development Project
Maximum
Time (weeks) Cost ($) Reduction in Crash Cost
Activity Normal Crash Normal Crash Time (weeks) per Week
A 5 2 25 000 34 000 3 3000
B 4 2 30 000 40 000 2 5000
C 10 6 45 000 81 000 4 9000
D 5 3 30 000 38 000 2 4000
E 7 6 30 000 37 000 1 7000
F 5 3 20 000 26 000 2 3000
G 4 2 35 000 44 000 2 4500
H 6 3 35 000 65 000 3 10 000
to 27 weeks and Start-A-C-E-H is still the single critical path. To shorten the project com-
pletion to 26 weeks, activity A is once again selected, given that it can still be crashed and it
has the lowest crash cost per week among all critical activities ($3000). The project comple-
tion time is now 26 weeks. Start-A-C-E-H is still a critical path but another path, Start-B-
C-E-H, has also become critical. To reduce the entire project to 25 weeks, one activity from
each critical path needs to be crashed. One option is to choose activity A and B, respec-
tively, but the total cost would be $8000 ($3000 $5000). Activity E is common to both
critical paths, has a crash cost per week of $7000, which is lower than $8000 (if A and B are
crashed), and will reduce both path simultaneously if selected. We crash activity E by one
week to reduce the project completion time to 25 weeks.
At this stage, we have 3 critical paths: Start-A-C-E-H, Start-B-C-E-H, and Start-B-C-
F-H. Activity E cannot be crashed further. To reduce the project completion time to
24 weeks, Activities A and B will be crashed for a total cost of $8000 ($3000 $5000), any
other combination being more costly. Note that the option of shortening A and F is
cheaper, but it does not reduce the length of the critical path Start-B-C-E-H.
In conclusion, to shorten the completion project duration to 24 weeks, activity
A should be crashed by 3 weeks at a cost of $9000 (3 $3000), activity E by 1 week at a cost
of $7000, and activity B by one week at a cost of $5000. The total project crashing cost is
$21 000 ($9000 $7000 $5000). Table 8-6 provides a summary of the procedure after
4 weeks of project crashing, where * indicates a critical path.
Using Linear Programming to Make Crashing Decisions

The manual procedure described above is suitable for projects involving a small number of
activities. For large-scale projects, linear programming can be used to make crashing deci-
sions. The problem consists of minimizing the total cost of the project (including the extra
cost of crashing activities), subject to the constraint that project duration must be less than
or equal to the desired deadline. The decision variables are the start time of each activity,
the reduction in the duration of each activity due to crashing, and the finish time of the
project. The constraints are that the maximum reduction in time for each activity cannot
be exceeded, the project finish time must be less or equal to the desired finish time; and the
precedence relationships of all activities must be respected.
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 376
TABLE 8-6 Summary Output from the Crashing Procedure
Length After Crashing x Weeks

Paths 0 1 2 3 4
Start-A-F-H 15 14 13 13 12
Start-A-C-E-H 28* 27* 26* 25* 24
Start-A-C-F-H 21 20 19 19 18
Start-B-C-F-H 25 25 25 25* 24
Start-B-C-E-H 26 26 26* 25* 24
Start-B-D-E-H 22 22 22 21 20
Start-B-D-G-H 19 19 19 19 18
Activity crashed A A E A,B
Crashing cost $3000 $3000 $7000 $8000
For the new product development project, the linear programming will be formulated
as follows, where X j starting time for activity j and Y j number of weeks by which
activity i is crashed.
The cost for completing the project using normal times is fixed. Hence, the objective
function is formulated as minimizing the project crashing:
minimize Z 3000YA 5000YB 9000YC 4000YD 7000YE

3000YF 4500YG 10 000YH
Precedence Relationship Constraints To express the precedence relationships between

activities, the following three relations are used for each activity: (1) Earliest start
time Finish time of preceding activity(ies); (2) Finish time Earliest start time activity
duration; and (3) Activity duration Normal activity time number of weeks by which an
activity is crashed. Therefore, the following constraints are developed:
XA 0 (Earliest starting time for activity A)

XB 0 (Earliest starting time for activity B)
XC XA 5 YA (C cannot start earlier than the completion of activity A)
XC XB 4 YB (C cannot start earlier than the completion of activity B)
XD XB 4 YB (D cannot start earlier than the completion of activity B)
XE XC 10 YC (E cannot start earlier than the completion of activity C)
XE XD 5 YD (E cannot start earlier than the completion of activity D)
XF XA 5 YA (F cannot start earlier than the completion of activity A)
XF XC 10 YC (F cannot start earlier than the completion of activity C)
XG XD 5 YD (G cannot start earlier than the completion of activity D)
XH XE 7 YE (H cannot start earlier than the completion of activity E)
XH X F 5 Y F (H cannot start earlier than the completion of activity F)
XH X G 4 Y G (H cannot start earlier than the completion of activity G)
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 377
Maximum Reduction Constraints This set of constraints refers to the limits by which
each activity can be crashed.
YA 3 (activity A can be crashed up to 3 weeks)
YB 2 (activity B can be crashed up to 2 weeks)
YC 4 (activity C can be crashed up to 4 weeks)
YD 2 (activity D can be crashed up to 2 weeks)
YE 1 (activity E can be crashed up to 1 week)
YF 2 (activity F can be crashed up to 2 weeks)
YG 2 (activity G can be crashed up to 2 weeks)
YH 3 (activity H can be crashed up to 3 weeks)
Project Completion Deadline Since H is the finish activity of the project, its finishing
time must be no longer than the project deadline, that is, XH 6 YH 24 weeks.
Non-negativity Constraints This set of constraints stipulates that all variables must be
non-negative, that is, all Xj and Yj 0, j A, B, C, D, E, F, G, H.
Exhibit 8-1 shows the Excel Solver layout for the new product development crashing proj-
ect. The Excel’s function SUMPRODUCT is used in the LHS column as well as in the Z value
cell. The solution is shown in the bottom section of the Exhibit. It shows that activity A should
be crashed by 2.5 weeks and activity C should be crashed by 1.5 week for a total crashing cost
of $21 000. Note that the Solver’s solution has the same objective function as the manual pro-
cedure, but with a different solution. This indicates that the problem has multiple solutions.
EXHIBIT 8-1 Excel Solver Output for the Crashing Example

ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 378
8.8 USING LINEAR PROGRAMMING IN PROJECT

SCHEDULING
The project scheduling problem can also be formulated as a linear programming problem
that seeks to determine the project completion time subject to meeting the precedence rela-
tionships between activities. To determine the EST and the EFT for activities the following
model can be developed. Let X j earliest start time for activity j, t j the duration for
activity j, and P(j) set of immediate predecessors of activity j. Therefore, the linear pro-
gramming model consists of:
minimize Z a X j (8-11)
j
Subject to
Xj Xi ti, for all j, for all i 僆 P(j) (8-12)
Xj 0, for all j (8-13)
The objective function minimizes the sum of earliest start times of activities. Its value is
only to ensure that each activity starts at the earliest time. Constraint 8-12 states that an
activity cannot start unless all its immediate predecessors are completed. Constraint
8-13 is the non-negativity constraint.
To determine the LST and LFT for activities, the linear programming model can be
written as follows, where W j latest start time for activity j:
maximize Z a W j (8-14)
j
Subject to
Wj Wi ti, for all j, for all i 僆 P(j) (8-15)
Wj tj EFTj , j finish activity (8-16)
Wj 0, for all j (8-17)
The objective function to maximize the sum of all activity start times to ensure the latest
activity start times. Constraint 8-15 defines the activity precedence relationships. Con-
straint 8-16 states the completion time of the finish activity in the project network.
8.9 PROJECT SCHEDULING SOFTWARE

For efficient management of large scale projects, many specialized project management
software packages have been developed for scheduling and tracking project activities.
Examples include Microsoft Project (www.microsoft.com/office/project), Primavera
(http://www.primavera.com/products/p6/index.asp), and Artemis Project (http://www
.aisc.com/). Microsoft Project, a project management software program developed and
sold by Microsoft, is one of the most-used products of this type in the market. It allows
users to draw the project network, develop the project schedule, assign resources, track the
project’s progress, manage the budget, and analyze the project workload. As the focus of
this chapter is how to develop a basic project schedule, the capabilities of these software
packages will not be discussed further.
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 379
Summary
This chapter introduced the critical path method (CPM) the probability to complete a project according to a
and the program evaluation review technique (PERT), given time frame and the variability in the project com-
two management science techniques developed in the late pletion time. The chapter also showed how the network
1950s to plan, schedule, and control large, complex proj- representation can be used to depict the project activities
ects with many activities. We showed when and how these and their relationships, how to handle crashing decisions
approaches can be used to help managers address ques- heuristically or using a linear programming model, and
tions such as what is the expected total time required to how the project scheduling problem can be formulated as
complete a project; what are the start and the completion a linear programming problem that seeks to determine
times for individual activities; which critical activities the project completion time subject to meeting the prece-
must be completed as scheduled; how much delay can be dence relationships between activities. Finally, a brief list
tolerated for non-critical activities; what is the least was provided of specialized project management software
expensive way to speed up a project; and in the case where packages that have been developed for scheduling and
the durations of activities are not known with certainty, tracking project activities.
Glossary
Activity A task that needs to be completed within a project and consumes both time and resources.
Activity on arc (AOA) A project network representation in which the project activities are reported on arcs and the
nodes represent the starting or the completion of activities.
Activity on node (AON) A project network representation in which the project activities are reported on nodes
and the arcs represent their precedence relationships.
Backward pass The process used to determine the latest start time and the latest finish time of an activity, which
consists of moving backward through the project network from the finish node to the starting node.
Beta probability distribution A form of distribution used to represent the duration of an activity in PERT analysis.
Critical path method (CPM) An analysis approach used in project scheduling when the project inputs (e.g., activi-
ties duration and costs) are assumed to be known with certainty.
Crashing Speeding up or shortening the duration of an activity by using additional resources.
Crash time The shortest possible time to complete an activity with additional resources.
Crash cost The cost to crash an activity.
Critical path(s) The path(s) with the longest length in the project network.
Critical activities Activities in the critical path. These activities must be completed as scheduled to prevent delay-
ing the project completion.
Earliest start time (EST) The earliest time at which an activity can start if no delays occur in the project.
Earliest finish time (EFT) The earliest time at which an activity can finish if no delays occur in the project.
Forward pass A process used to determine the earliest start time and the earliest finish time of an activity, which
consists of moving forward through the project network from the starting node to the finish node.
Immediate predecessors Activities that must be completed prior to the starting time of a given activity.
Immediate successors Activities that follow the completion of a given activity.
Latest start time (LST) The latest time at which an activity can start without delaying the completion of the project.
Latest finish time (LFT) The latest time at which an activity can finish without delaying the completion of the project.
Length of path The sum of the durations of the activities on the path.
Normal time The estimated activity duration used with CPM or PERT in the computation of earliest (latest) start
or finish times.
Normal cost The activity cost under the normal activity time.
Path A sequence of connected nodes in the network from the start node to the finish node.
Program evaluation review technique (PERT) An analysis approach used in project scheduling when some of the
project inputs (e.g., activities duration and costs) are not known with certainty
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 380
Project A set of activities interrelated by their precedence relationships that need to be undertaken in the accom-
plishment of a unique, often major outcome.
Project crashing The process of shortening the duration of a project by crashing the duration of a number of activities.
Project network A representation used to depict the project activities.
Slack time The length of time that can be tolerated for an activity without incurring a delay in the estimated proj-
ect completion time.
Solved Problems
Problem 1 deviation for each path are shown in Table 8-7. Path S-1-5-
Expected durations (in weeks) and variances for the 8-End is the critical path. Therefore, the project expected
major activities of an R&D project are depicted in the completion time is 51 weeks and the corresponding vari-
precedence network diagram chart. Determine the proba- ance is 1.488 weeks.
bility that project completion time in Figure 8-10 will be a. The z value for the normal distribution at x 50 is
a. Less than 50 weeks. z 50tS1 .82. Hence, P(x 50) .2061.
b. More than 50 weeks. 1.22
b. P(x
50) 1 P(x 50) .7939.
Solution
Because S and End have zero durations, we can ignore
them in the following calculations. The mean and standard
FIGURE 8-10 Precedence Network Diagram Chart for Problem 1
Variance
Mean
.69 11, .69 24, .11
16,
1 5 8
18, .25 26, .11

S 4 7 End
5, 0
2 10, .25 14, .36 12, .11
3 6 9
Table 8-7 Probabilities for Problem 1

Path Expected Time (weeks) Standard Deviation (weeks)
S-1-5-8-End 16 11 24 51 √.69 .69 .11 1.22
S-2-4-7-End 5 18 26 49 √.00 .25 .11 .60
S-2-3-6-9-End 510141241 √.00 .25 .36 .11 .85

ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 381
Problem 2 Solution
Table 8-8 shows the information related to a project that
a. The project network is shown in Figure 8-11.
involves the merger of two marketing firms (in days).
b. The project schedule is as follows:
Table 8-8 Data for Solved Problem 2 Earliest Latest Earliest Latest
Activity Start Start Finish Finish Slack
Immediate Estimated
Activity predecessor(s) duration(days) Start 0 0 0 0 0
A 0 13 10 23 13
A — 10 B 0 0 15 15 0
B — 15 C 10 23 15 28 13
C A 5 D 15 15 27 27 0
D B 12 E 27 28 41 42 1
E C, D 14 F 15 19 23 27 4
F B 8 G 27 27 42 42 0
G D, F 15 H 41 47 51 57 6
H E 10 I 42 42 48 48 0
I E, G 6 J 48 48 57 57 0
J F, I 9 Finish 57 57 57 57 0
a. Draw the project network. c. The critical activities are: B-D-G-I-J

b. Develop the project schedule (EST, EFT, LST, LFT). d. The project completion duration is 57 days.
c. What are the critical activities? e. Activity A or C since they have the largest slack times.
d. What is the project completion duration? f. The LP formulation of the model is as follows, where
e. If there is an option to delay one activity without X j earliest start time for activity j A, B, C, D, E, F,
delaying the entire merge project, which activity G, H, I, and J.
would you delay and why?
f. Formulate the LP model that would determine the minimize Z XA XB XC XD XE XF
earliest start times of activities. XG XH XI XJ
FIGURE 8-11 Network for Solved Problem 2
A C H
E
Start
I
B D
Finish
G
F J
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 382
subject to b. The expected project completion time is 57 days and

its variance is .44 .44 .88.
XC XA 10
XD XB 15 c. The probability that the project will be completed
XE XC 5 in 60 days or more: P(X 60) P(X z), where
XE XD 12
XF XB 15 z 6057 3.198. Therefore from Table A in
XG XD 12
√0.88
XG XF 8 Appendix B, P(X 3.198) 0.5 0.4990 0.001
XH XE 14 or 0.1 percent.
XI XE 14 The probability that the project will be completed
XI XG 15 in 55 days or less: P(X 50) P(X z), where
XJ XF 8
XJ XI 6 5557
z 2.13. Therefore from Table A in
XA XB 0 √0.88
Appendix B, P(X 2.13) 0.5 .4788 .0212
All variables 0
or 2.12 percent.
Problem 3 d. z0.96 1.76. Using Equation 8-9, the project comple-
Consider the network information shown in the previous tion time X 57 1.76 .938 58.65 days. There-
problem (Problem 2) and assume that the duration of fore to meet this completion time, the latest start and
some activities is not known with certainty. The estimates finish times for each activity should be as follows:
of these activities are shown below, assuming that the
duration for the other activities remains unchanged. Activity Latest Start Latest Finish
Start 1.65 1.65
Activity Optimistic Most Likely Pessimistic A 14.65 24.65
A 8 10 12 B 1.65 16.65
C 3 5 7 C 24.65 29.65
D 10 12 14 D 16.65 28.65
G 13 15 17 E 29.65 43.65
H 8 10 12 F 20.65 28.65
G 28.65 43.65
a. What is the critical path? H 48.65 58.65
b. What is the project’s expected completion time and I 43.65 49.65
its variance? J 49.65 58.65
c. What is the probability that the project will be com- Finish 58.65 58.65
pleted in 60 days or more? In no more than 55 days? Problem 4
d. If the company wants a 96% probability of complet-
ing the project on time, state the latest time each Indirect cost for a project is $12 000 per week for as long
activity should have started and completed. as the project lasts. The project manager has supplied the
cost and time information and precedence network dia-
Solution gram shown in Figure 8-12.
a. The mean and variance for activities for which the Crashing
duration is not known with certainty: Potential Cost per Week
Activity (weeks) to Crash
1 ba6 2
Mean, Variance,
2 A 3 $11 000
te a4mb 2 B 3 3000 first week, $4000 after that
Activity a m b 6
C 2 6000
A 8 10 12 10 0.44 D 1 1000
C 3 5 7 5 0.44 E 3 6000
D 10 12 14 12 0.44 F 1 2000
G 13 15 17 15 0.44
H 8 10 12 10 0.44 Use the information to:
a. Determine an optimum crashing plan.
The critical path is Start B D G I J Finish. b. Graph the total costs for the plan.
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 383
FIGURE 8-12 Network Diagram for Solved Problem 4
10 weeks 14
A B
13 6
S C D End
15 8
E F
Solution 4. Check to see which path(s) might be critical:

S-A-B-End and S-E-F-End would be 22 weeks in
a. 1. Calculate path lengths and identify the critical
length, and S-C-D-End would still be 19 weeks.
path:
5. Rank activities on the critical paths:
Path Duration (weeks)
Path Activity Cost per Week to Crash
S-A-B-End 24 (critical path)
A-B . . . . . B $4000
S-C-D-End 19
......... A 11 000
S-E-F-End 23
E-F . . . . . E 6000
2. Rank critical activities according to crash costs: ......... F (no further crashing possible)
Activity Cost per Week to Crash

Crash B on path S-A-B-End and E on S-E-F-End
B . . . . . . . . . . . . . $3000 first week, $4000 after that
for a cost of $4000 $6000 $10 000, for a net
A . . . . . . . . . . . . . 11 000
savings of $12 000 $10 000 $2000.
Activity B should be shortened one week since it has 6. At this point, no further reduction is cost-
the lower crashing cost per week. This would reduce effective: paths S-A-B-End and S-E-F-End would
indirect costs by $12 000 at a cost of $3000, for a net be 21 weeks in length, and one activity from each
savings of $9000. At this point, paths S-A-B-End and path would have to be shortened. This would
S-E-F-End would both have a length of 23 weeks, so mean activity A at $11 000 and E at $6000 for a
both would be critical. total of $17 000, which exceeds the $12 000
3. Rank activities by crashing cost on the two critical potential savings in indirect costs. Note that no
paths: further crashing for activity B is possible.
b. The following table summarizes the results, showing
Cost per Week the length of the project after crashing n weeks:
Path Activity to Crash
S-A-B-End . . . . . . . B $4000 Path n0 1 2 3
................. A 11 000 S-A-B-End . . . . . . . . . . . . . . . . . . 24 23 22 21
S-E-F-End . . . . . . . F 2000 S-C-D-End . . . . . . . . . . . . . . . . . . 19 19 19 19
................. E 6000 S-E-F-End. . . . . . . . . . . . . . . . . . . 23 23 22 21
Activity crashed . . . . . . . . . . . . B B,F B,E
Choose one activity (the least costly) on each path to Crashing costs ($000) . . . . . . . 3 6 10
crash: B on S-A-B-End and F on S-E-F-End, for a total
cost of $4000 $2000 $6000 and a net savings of A summary of costs for the preceding schedule would
$12 000 $6000 $6000. look like Table 8-9.
Note: There is no activity common to the two critical
The graph of total cost is shown in Figure 8-13.
paths.
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 384
Table 8-9 Summary of Costs for Solved Problem 4

Cumulative Cumulative
Project Weeks Crashing Indirect Total
Length Shortened Costs ($000) Costs($000) Costs ($000)
24 0 0 24(12) 288 288
23 1 3 23(12) 276 279
22 2 369 22(12) 264 273
21 3 9 10 19 21(12) 252 271
20 4 19 17 36 20(12) 240 276
FIGURE 8-13 Total Cost for Solved Problem 4
290 288
279
280 276
273
271
Total Cost
270
Crash
260
0
20 21 22 23 24
Project Duration (weeks)
Discussion and Review Questions

1. Explain differences and similarities between CPM and 8. Why must all immediate predecessor activities be
PERT. considered when determining the activity’s earliest
2. What are the different types of information that need start time?
to be gathered to depict a project network? 9. Why must all immediate successor activities be con-
3. Explain briefly two approaches that can be used to sidered when finding the activity’s latest finish time?
determine the project critical path(s) and when each 10. What is project crashing and how can it be done
is the most suitable. manually?
4. What is the meaning of activity slack times and how 11. Provide some of the reasons for a company to con-
are they computed? sider project crashing.
5. Explain why the project manager should monitor the 12. What constraints are involved in formulating the
progress of activities with small slack times. project crashing problem as a linear programming
6. Why is it important in project management to iden- model?
tify and monitor the critical activities? 13. What constraints are involved in formulating the
7. When is simulation approach the most appropriate project scheduling problem as a linear programming
in project scheduling? model?
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 385
Problems
1. For the precedence network diagram in Figure 8-14, 4. ET is an entertainment group that specializes in
determine both the critical path and the project the management of band tours around the world.
duration by determining the length of each path. The The company is considering placing a bid for the
numbers above the nodes represent activity duration management of the next 2-day concert of a very pop-
in days. ular rock band in the city of Toronto. As part of the
2. Using the data in Table 8-10 construct a precedence planning process, the company has determined that
network diagram. The project is completed when the list of activities in Table 8-12 would need to be
activities H, I, and J are all finished. performed to carry out the project.
3. Using the data in Table 8-11 construct a precedence Construct the precedence network diagram that can
network diagram. The project is completed when be used in the scheduling of these activities.
activities D, F, J, and K are all finished.
FIGURE 8-14 Network Diagram for Problem 1
10 14 13 7
1 2 3 4
15 6 4
14
6 8 10 End
5
S
11 13 3 10
7 9 11 17
3 8 4 7 6
12 13 14 15 16
Table 8-10 Data for Problem 2

Activity A B C D E F G H I J
Immediate — — — A B B C F F, G D, E
Predecessors

Activity A B C D E F G H I J K L
Immediate — — A C A E — B G H, I J, L —
Predecessors

Activity A B C D E F G H I
Immediate Predecessors — A A B B C, E D, E F, G H
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 386
5. Assume that the activities in Problem 4 have the 8. Reconsider the list of activities in Problem 6 and their
duration (in days) as shown in Table 8-13. corresponding durations in Problem 7. Suppose that
a. What is the maximum number of working days after 12 days, activities A, B, and I have been finished,
the company would have to write in their bid to activity E is 75 percent finished, and activity J is half
complete all activities of this project if no delays finished. How many days after the original start time
occur? would the project finish?
b. What are the critical activities? 9. The following table contains information related to
c. When would each activity have to start and the major activities of a research project. Use the
finish at the earliest to complete this project on information to do the following:
time? a. Draw a precedence network diagram.
d. When would each activity have to start and fin- b. Find the critical path by identifying all the start
ish at the latest to complete this project on time? to end paths and calculating their lengths.
6. Softbank is a consulting company that specializes in c. What is the expected duration of the project?
the customization of software in the banking industry.
Immediate Expected
The company has been hired to serve on a project that
Activity Predecessor(s) Duration (days)
involves the replacement of the user interface currently
in place for online banking in the website of a large A................ — 5
C................ A 8
Canadian bank. It has been determined that the list of
D................ C 2
activities in Table 8-14 would need to be performed to
B................ A 7
carry out the project. E ................— 3
Construct the precedence network diagram that can F ................ E 6
be used in the scheduling of these activities. I . . . . . . . . . . . . . . . . . B, D 10
7. Assume that activities in Problem 6 have the dura- M ............... F 8
tions (in days) as shown in Table 8-15. G................ — 1
a. Determine the following values for each activity: H................ G 2
the earliest start time, the earliest finish time, the K................ H 17
latest start time, the latest finish time, and
10. Chris received a new word-processing software pro-
the activity slack time.
gram for her birthday. She also received a cheque,
b. Identify the critical activities, and determine
with which she intends to purchase a new computer.
the expected duration of the project. What is the
Chris’s university instructor assigned a paper due next
maximum number of working days the com-
week. Chris decided that she will prepare the paper on
pany would have to write in their bid to
the new computer. She made a list of the activities and
complete all project activities if no delays occur?

Activity A B C D E F G H I
Duration (in days) 5 18 13 3 10 11 4 9 2

Activity A B C D E F G H I J K L M
Immediate — A B C A E F D, G — I J K L
Predecessors

Activity A B C D E F G H I J K L M
Duration 4 9 5 2 8 7 2 3 10 6 4 5 6
(in days)
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 387
their estimated durations. Chris’s friend has offered Their first project consists of activities listed in the
to shop for, select and purchase a computer, and following table.
install the software. a. Draw the precedence network diagram.
a. Arrange the activities into two logical sequences. b. What is the probability that the project can
b. Construct a precedence network diagram. be completed in 24 days or less? In 21 days or
c. Determine the critical path and its expected less?
duration. c. Suppose it is now the end of the seventh day and
d. What are some possible reasons for the project that activities A and B have been completed while
to take longer than the expected duration? activity D is 50 percent completed. Optimistic,
most likely, and pessimistic estimates for the
Estimated completion of activity D are now 5, 6, and 7 days.
Time (hours) Activity (abbreviation) Activities C and H are ready to begin. Determine
.8 Install software (Install) the probability of finishing the project by day 24
.4 Outline the paper (Outline) and the probability of finishing by day 21.
.2 Submit paper to instructor (Submit)
.6 Choose a topic (Choose) DURATION IN DAYS
.5 Use grammar-checking routine and
Immediate Most
make corrections (Check)
Activity Predecessor(s) Optimistic Likely Pessimistic
3.0 Write the paper using the word-
processing software (Write) A.......... — 5 6 7
2.0 Shop for a new computer (Shop) B.......... — 8 8 11
1.0 Select and purchase computer C .......... A 6 8 11
(Select) D.......... — 9 12 15
2.0 Library research on chosen topic E.......... C 5 6 9
(Library) F .......... D 5 6 7
G.......... F 2 3 7
11. The information in the following table pertains to a H.......... B 4 4 5
project that is about to commence. I .......... H 5 7 8
Immediate Estimated 13. The new director of special events at a large univer-
Activity Predecessor(s) Duration (days) sity has decided to completely revamp graduation
ceremonies. Toward that end, a precedence network
A.................— 15
B................. A 12 diagram of the major activities has been developed.
C................. B 6 The chart has five paths with expected durations and
D................. B 5 variances as shown in the following table. Gradua-
E ................. C 3 tion day is 16 full weeks from now. Assuming that the
F .................— 8 project begins now, what is the probability that the
G................. F 8 project will be completed before:
H................. F 9 a. Graduation time?
I.................. G 7 b. The end of week 15?
J ................. H 14 c. The end of week 13?
K................. J 6
a. As the project manager, which activities would Path Expected Duration (weeks) Variance
you be concerned with in terms of timely project A.................... 10 1.21
completion? Explain. B.................... 8 2.00
b. Determine the following values for each activity: C.................... 12 1.00
the earliest start time, the earliest finish time, the D.................... 15 2.89
E.................... 14 1.44
latest start time, the latest finish time, and
the activity slack time. 14. What is the probability that the following project
12. Three recent university graduates have formed a will take more than 10 weeks to complete if the
partnership and have opened an advertising firm. precedence network diagram, activity means, and
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 388
standard deviations (both in weeks) are as shown 16. The precedence network diagram in Figure 8-15
below? reflects optimistic, most likely, and pessimistic esti-
mates for each activity. Determine:
Activity Mean Standard Deviation
A 5 1.3
a. The expected completion time for each path and
B 4 1.0 its variance.
C 8 1.6 b. The probability that the project will require more
than 49 weeks.
A B c. The probability that the project can be completed
in 46 weeks or fewer.
S End 17. A project manager has compiled a list of major activi-
ties that will be required to install a computer infor-
C mation system in her company. The list includes three
estimates of durations (optimistic, most likely, pes-
15. The project described in the following table and prece- simistic) for activities and precedence relationships.
dence network diagram has just begun. It is scheduled
3-Point
to be completed in 11 weeks.
Activity Precedes Estimates (weeks)
a. If you were the manager of this project, would
A — 2-4-6
you be concerned? Explain.
D A 6-8-10
b. If there is a penalty of $5000 a week for each
E D 7-9-12
week the project is late, what is the probability of H E 2-3-5
incurring a penalty of at least $5000? F A 3-4-8
G F 5-7-9
Expected Standard
B — 2-2-3
Activity Duration (weeks) Deviation (weeks)
I B 2-3-6
A 4 .70 J I 3-4-5
B 6 .90 K J 4-5-8
C 3 .62 C — 5-8-12
D 9 1.90 M C 1-1-1
N M 6-7-11
A B O N 8-9-13
S End If the project is finished within 26 weeks of its start, the

project manager will receive a bonus of $1000; and if
the project is finished within 27 weeks of its start, the
C D
bonus will be $500. Find the probability of each bonus.
FIGURE 8-15 Precedence Network Diagram for Problem 16
11-12-13 9-10-12 5-7-10

2 3 4
14-18-26 8-10-14
5-6-7
6 7 End
8-8-8 5
1 13-13-1
S 3 10.5-13-15.5 -6
6-6
8 9 13
10-12-14 7-10-12 10-11-12

10 11 12
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 389
18. The project manager of a task force planning the con- Crashing costs for each activity are:
struction of a domed stadium had hoped to be able to
complete construction prior to the start of the next CRASHING COSTS ($000)
season. After reviewing construction duration esti- Activity First Week Second Week Third Week
mates, it now appears that a certain amount of crash-
1 . . . . . . . . . $18 $22 $—
ing will be needed to ensure project completion 2 . . . . . . . . . 24 25 25
before the season opener. Given the following infor- 3 . . . . . . . . . 30 30 35
mation, determine a minimum-cost crashing sched- 4 . . . . . . . . . 15 20 —
ule that will shave five weeks off the project length. 9 . . . . . . . . . 30 33 36
5 . . . . . . . . . 12 24 26
6......... — — —
CRASHING COSTS
Normal 8 . . . . . . . . . 40 40 40
Duration First Second 7......... 3 10 12
Activity Precedes (weeks) Week Week 10 . . . . . . . . . 2 7 10
A......... — 12 $15 000 $20 000 15 . . . . . . . . . 26 — —
B......... A 14 10 000 10 000 11 . . . . . . . . . 10 15 25
C......... — 10 5000 5000 12 . . . . . . . . . 8 13 —
D......... C 17 20 000 21 000 13 . . . . . . . . . 5 12 —
E......... C 18 16 000 18 000 14 . . . . . . . . . 14 15 —
F......... C 12 12 000 15 000
G......... D 15 24 000 24 000
a.
Determine a minimum-cost crashing plan
H......... E 8 — — that will take off six weeks from the project
I.......... F 7 30 000 — duration.
J......... I 12 25 000 b. Plot the total-cost curve from part a against
25 000
K......... B 9 10 000 project duration. What is the optimum number
10 000
M......... G 3 — — of weeks to crash?
N......... H 11 40 000 —
20. Chuck’s Custom Boats (CCB) builds luxury yachts to
P......... H, J 8 20 000 20 000
customer order. CCB has landed a contract with a
Vancouver businessman (Mr. P). Relevant data are
19. A construction project has indirect costs totalling shown below. The complication is that Mr. P wants
$40 000 per week. Major activities in the project and delivery in 32 weeks or he will impose a penalty of
their expected durations are shown in the precedence $375 for each week his yacht is late.
network diagram in Figure 8-16:
FIGURE 8-16 Precedence Network Diagram for Problem 19
5 8 7 4
1 2 3 4
11
9
12 5
4
6 8 End
5
S
6 9 1
7 10 15
3 12 9
8
11 12 13 14
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 390
Develop a minimum cost crashing schedule.

CRASHING COSTS 21. Table 8-16 is a list of activities and their expected dura-
Normal
Immediate Duration 1st 2nd tion, used by a component supplier to automobile
Activity Predecessor(s) (weeks) Week Week manufacturers, to plan for QS9000 (the auto industry
K — 9 $410 $415 version of ISO9000) certification (registration).
L K 7 125 — a. Draw the precedence network diagram.
N K 5 45 45 b. Determine the earliest and latest times, and
M L 4 300 350 identify the critical activities and the project
J N 6 50 — duration.
Q M, J 5 200 225
P Q 8 — —
Y Q 7 85 90
Z P 6 90 —

A List of Activities in a QS-9000 Registration Project
Immediate Estimated
Activity Description Predecessor(s) time*
A Appointment of QS-9000 taskforce none 1 week
B Preparation of a feasible plan A 1 week
C Delegation of authority responsibilities B 1 week
D Searching for a QS-9000 registrar C 1 week
E Preparation of three levels of documentation C 12 weeks
F QS-9000 awareness training C 6 weeks
G QS-9000 training of auditors and quality personnel F 6 weeks
H Preparing the plant for QS-9000 registrar C 24 weeks
I Conference with lead auditor D 1 week
J Examination of documentation E, I 3 weeks
K Internal audit of plant sections G, H, J 12 weeks
L Corrective actions of plant sections K 12 weeks
M Lead auditor and audit team audit plant L 1 week
N Audit conference and corrective action plan M 2 weeks
O Implementation of corrective action plans N 12 weeks
P Lead auditor re-audit corrective actions O 2 weeks
Q Lead auditor’s recommendation P 1 week
*These are estimates of time and may vary based upon situation, company, and registrar.
Case 1: Fantasy Products

Company Background foods quickly much as a microwave oven heats them
The Fantasy Products Company (disguised name) is a quickly, although the technology involved is quite differ-
manufacturer of high-quality small appliances intended ent. Tentatively named The Big Chill, the product will ini-
for home use. Their current product line includes irons, a tially carry a price tag of around $125, and the target
small hand-held vacuum, and a number of kitchen appli- market consists of upper-income consumers. At this
ances such as toasters, blenders, waffle irons, and cof- price, it is expected to be a very profitable item. R&D
feemakers. Fantasy Products has a strong R&D department engineers now have a working prototype and are satisfied
that continually searches for ways to improve existing that, with the cooperation from the production and mar-
products as well as developing new products. keting people, the product can be ready in time for the
Currently, the R&D department is working on the all-important Christmas buying season. A target date has
development of a new kitchen appliance that will chill been set for product introduction that is 24 weeks away.
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 391
Current Problem that sales during the first six weeks will be about 400 per
Fantasy Products’ Marketing Vice-President Vera Sloan week, there is an opportunity for $25 per unit profit if the
has recently learned from reliable sources that a competi- early introduction can be accomplished. In addition, there
tor is also in the process of developing a similar product, is a certain amount of prestige involved in being first to
which it intends to bring out at almost the same time. In the market. This should help enhance The Big Chill’s
addition, her source indicated that the competitor plans image during the anticipated battle for market share.
to sell its product, which will be smaller than The Big
Chill, for $99 in the hope of appealing to more customers.
Vera, with the help of several of her key people who are to Data Collection
be involved in marketing The Big Chill, has decided that Since Fantasy Products has been through the product-
to compete, the selling price for The Big Chill will have to introduction process a number of times, Vera has developed
be lowered to within $10 of the competitor’s price. At this a list of the tasks that must be accomplished and the order
price level it will still be profitable, although not nearly as in which they must be completed. Although the duration
profitable as originally anticipated. and costs vary depending on the particular product, the
However, Vera is wondering whether it would be possi- basic process does not. The list of activities involved and
ble to expedite the usual product introduction process to their precedence relationships are presented in Table 8-17.
beat the competition to the market. If possible, she would Duration and cost estimates for the introduction of The Big
like to get a six-week jump on the competition; this Chill are presented in Table 8-18. Note that some of the
would put the product introduction date only 18 weeks activities can be completed on a crash basis, with an associ-
away. During this initial period, Fantasy Products could ated increase in cost. For example, activity B can be crashed
sell The Big Chill for $125, reducing the selling price to from 8 weeks to 6 weeks at an additional cost of $3000 (i.e.,
$109 when the competitor’s product actually enters the $12 000–$9000). Assume that if B is crashed to 7 weeks, the
market. Since forecasts based on market research show additional cost will be $1500 (i.e., $3000/2).
TABLE 8-17 List of Activities and Precedence Relationships
Activity Description Immediate Predecessor(s)
A Select and order equipment —

B Receive equipment from supplier A
C Install and set up equipment B
D Finalize bill of materials —
E Order component parts D
F Receive component parts E
G First production run C, F
H Finalize marketing plan —
I Produce magazine ads H
J Script for TV ads H
K Produce TV ads J
L Begin ad campaign I, K
M Ship product to consumers G, L
ste13809_ch08_359-392.qxd 1/27/09 7:28 AM Page 392
TABLE 8-18 Duration and Cost Estimates
Normal Crash Normal &

Activity Duration (weeks) Normal Cost Duration (weeks) Crash Cost
A 3 $2000 2 $4400
B 8 9000 6 12 000
C 4 2000 2 7000
D 2 1000 1 2000
E 2 2000 1 3000
F 5 0 5 0
G 6 12 000 3 24 000
H 4 3500 2 8000
I 4 5000 3 8000
J 3 8000 2 15000
K 4 50 000 3 70 000
L 6 10 000 6 10 000
M 1 5000 1 5000
2. Is it possible to complete the project in 18 weeks?

Managerial Report What would the associated additional cost be? Which
activities would need to be completed on a crash
Fantasy Products needs to decide whether to bring The
basis?
Big Chill to market 18 weeks from now as Vera Sloan is
3. Is there some time frame shorter than the 18 weeks
recommending. As the project management specialist in
Vera has recommended that would make more sense
the R&D department, you have been asked to answer the
in terms of profits?
following questions:
1. When would the project be completed using normal Source: Adapted from an original case by Robert J. Thieraus, Margaret
durations? Cunningham, and Melanie Blackwell, Xavier University, Cincinnati, Ohio.

Project Scheduling: Pert/Cpm: Learning Objectives

Uploaded by

Copyright:

Available Formats

Project Scheduling: Pert/Cpm: Learning Objectives

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Project Scheduling: Pert/Cpm: Learning Objectives

Uploaded by

Copyright:

Available Formats

ste13809_ch08_359-392.

qxd 1/27/09 7:28 AM Page 359

360 Part 2 Deterministic Decision Models

PERT also provides answers to the following additional questions:

6. What is the probability of completing a project within a given time frame?

Chapter 8 Project Scheduling: PERT/CPM 361

8.2 A PROJECT EXAMPLE: REPLACEMENT OF AN

8.3 PROJECT NETWORK REPRESENTATION

362 Part 2 Deterministic Decision Models

List of Activities for the Airport Gate Management

A Set up the project acquisition team — 2

Chapter 8 Project Scheduling: PERT/CPM 363

FIGURE 8-1 Network Representation for the AGMS Acquisition Project

TABLE 8-2 List of Paths for the AGMS Acquisition Project

Path # Sequence of Nodes Length (total time in weeks)

364 Part 2 Deterministic Decision Models

8.4 PROJECT SCHEDULING WITH DETERMINISTIC

Chapter 8 Project Scheduling: PERT/CPM 365

FIGURE 8-2 Nodes Notation

1. For each activity with no predecessor, set EST 0.

366 Part 2 Deterministic Decision Models

Chapter 8 Project Scheduling: PERT/CPM 367

TABLE 8-3 Summary of Activities’ Start, Finish, and Slack Times

Activity EST EFT LST LFT Slack Critical Activity

368 Part 2 Deterministic Decision Models

8.5 PROJECT SCHEDULING WITH PROBABILISTIC

The duration of an activity is therefore assumed to have a beta probability distribution in

Chapter 8 Project Scheduling: PERT/CPM 369

FIGURE 8-5 Illustration of the Shape of the Beta Probability Distribution

Activities Expected Times and Variances for

370 Part 2 Deterministic Decision Models

p 冑 (var(A) var(B) var(E) var(F) var(G) var(I) var(K)

Using the numerical values in Table 8-4, we get:

p 冑 (0.111) (0.111) (0.111) (0.111) (0.444) (0.444) (0.028)

Probability of Whether a Project Can Be Completed

z 4644 1.23. Therefore from Table A in Appendix B,

Chapter 8 Project Scheduling: PERT/CPM 371

Illustration of Possible Situations of Interest for Probability

372 Part 2 Deterministic Decision Models

FIGURE 8-7 Project Completion Time Given a Certain Probability

Project Completion Time Given a Certain Probability

8.6 USES OF SIMULATION IN PROJECT SCHEDULING

8.7 PROJECT CRASHING

Chapter 8 Project Scheduling: PERT/CPM 373

speeding up or shortening of the duration of an activity by using additional resources.

FIGURE 8-8 Activity Cost and Activity Time Relationship

Normal cost Normal

Crash time Normal time

374 Part 2 Deterministic Decision Models

Project Network for the New Product Development

Chapter 8 Project Scheduling: PERT/CPM 375

Time and Cost Information for the New Product

Using Linear Programming to Make Crashing Decisions

376 Part 2 Deterministic Decision Models

TABLE 8-6 Summary Output from the Crashing Procedure

Length After Crashing x Weeks

minimize Z 3000YA 5000YB 9000YC 4000YD 7000YE

Precedence Relationship Constraints To express the precedence relationships between

p 冑 (var(A) var(B) var(E) var(F) var(G) var(I) var(K)

p 冑 (0.111) (0.111) (0.111) (0.111) (0.444) (0.444) (0.028)