The PERT method

Eric S. TRAORE

1. Generalities
At the end of the 1950s, the US Navy devised a new technique
scheduling that was expected to lead to significant time savings in the accomplishment of its
nuclear warhead missiles Polaris: this is the PERT technique (Program Evaluation and Review Technique)
Technique - scheduling and program control technique). This technique enabled
to coordinate the work of nearly 6000 builders within the imposed deadlines by the
American government.
The POLARIS project represented among other things:
250 suppliers
9000 subcontractors,
-7 years of achievement.
The use of PERT has reduced the overall project duration from 7 to 4 years.
This method then spread to the American industry and later to the Western industry.
PERT is "a method consisting of organizing several tasks in the form of a network
tasks that through their dependency and chronology all contribute to the achievement of a
finished product". It provides a set of project planning techniques.
Unlike the GANTT chart, the PERT method focuses mainly on highlighting
the connections that exist between the different tasks of a project and to define the so-called 'critical' path.
The PERT method is most often synonymous with managing important and long-term projects.
That is why a number of actions are necessary to successfully implement it.
1. Clearly identify the tasks and the steps.
2. Determine the appropriate succession (ordering) of tasks.
3. Build the network graph.
[Link] the time required for each task
[Link] the critical path.
[Link] out periodic checks and update the PERT chart during execution
of the project.
The main advantages of the PERT method:
provide the following information: the expected project completion date, the probability
completion before a given date, the tasks on the critical path that directly affect the
completion deadline of the project, the tasks that have a margin in the execution time and that
can allocate resources to the critical path tasks, the start and end dates of the tasks.
use of the method to optimize resource allocation and costs.
The limitations of the method:
the duration estimation is somewhat subjective
the completion date is supposed to be that of the critical path tasks, while others
Critical tasks may arise during the project and lead to an underestimation of this date.
completion.

1
 2. Basic Concepts
A PERT network consists of STEPS and TASKS. A network is a graph.
involving subnodes connected by arcs. There are two modes of representation of a
PERT network:
the Potential-Step mode, where the steps are the nodes, and the tasks are the directed arcs. It is the mode
representation originally used in PERT, and this is the one we will adopt.

or

- the Potential-Task mode where tasks are the nodes, and where the directed arcs express the
predecessor relationships between tasks. This mode is rather suitable for representing the method
MPM (Method of Matra Potentials). However, it is the mode of representation adopted by the
most software.

2
2.1. Definitions
Task: A well-defined activity that fits into the realization of the project
♦ It consumes resources (labor, equipment, ...).
♦ is graphically represented by an arrow whose length has no
temporal significance in the Potential-Stage.
♦ is identified by a code.
A4

Step: Start (starting point) or end of a task

♦ has no duration
♦ is graphically represented either by a numbered circle, or by a rectangle, or
by a square
1 1 1

hence the aspect of a partial graph:

2
A4 B2 3 C7
1

2.2. Configuration types of steps and tasks in a network

Soient les tâches A = mise en place de la plomberie ( durée d’exécution : 4 j )
B = electrical installation (execution duration: 3 days)
C = painting (execution duration: 2 days)
Sequential tasks: they occur one after the other, separated by steps.

A4 B3 C2
1 3 4
2

Simultaneous tasks: they can start at the same time from the same step.

2
A4
1
B3
C2
3 4
Between steps 2 and 3, there is a fictitious task with no duration or cost, which expresses a constraint of
link between these two stages ( C can only start if A and B are completed ).
Converging tasks: they lead to the same step.
1 A4
C2
B3 3
2
Task A starting from step 1 is time-consuming for step 3 which cannot occur before
time 4 while B is finished at time 3.

3
 3. Construction of a PERT network
The following operations must be performed chronologically:
1. Establishment of a task list
Provide the exhaustive list of tasks to be performed; assess the duration of the tasks and determine
the necessary resources to accomplish them; codifying tasks to facilitate the
construction of the network (A, B, C, D,…)
2. Determination of prior tasks and immediately prior tasks
Which task(s) must be completed immediately before another one starts?
What task should follow a determined task?
3. Construction of partial graphs
Graphically represent the relationships of immediate precedence, knowing that tasks
successive ones are connected by a step. Multiple previous tasks without another
information is temporarily considered as immediately preceding and
converging.
4. Grouping of partial graphs
Resolve any contradictions between partial graphs arising from the decision
provisionally considering the multiple previous tasks as convergent.
5. Determination of the start and end tasks of the work
Beginning tasks have no predecessors (no preceding tasks). End tasks
have no successor (are prior to no task).
6. Network construction
It is sometimes necessary to introduce a dummy task to represent certain simultaneity.
or convergences (for example when two simultaneous tasks are also convergent).
3.1. List of tasks in table form
Table 1
Task Code Duration (days)
Study, implementation, and acceptance of the plans A 4
B
Ground preparation 2
C
Material order (wood, bricks, cement, sheet metal for the roof) 1
D
Excavation of foundations 1
E
Doors, windows orders 2

Delivery of materials F 2
G
Pouring of foundations 2
H
Delivery of doors, windows 10

Construction of the walls, the roof I 4

J
Installation of doors and windows 1

3.2. Task Priorities

When a task has only one preceding task, then this task is necessarily
immediately preceding. When there are several preceding tasks, it is not obvious to
determine those that are immediately prior. On the other hand, a task that is only prior to
Only one task is immediately prior to this task.
Note: this last rule should be used with caution as the assessment of previous tasks
is generally not done exhaustively.

4
3.3. Construction of partial graphs (see Table 2)
Representation rules:
A network always has a starting point and an ending point. A network is read from left to right.
to the right. The arrows are pointing in this direction. There is never a return.
A task can only be represented by a single arrow. Every task has a single starting point.
and a single final step. A subsequent task can only start if the previous task is
finished.
Table number 2
Tasks Immediate tasks Marker
Task Partial graphs
previous previous line
A - - 1
B - - 2
A C
C A A 3
A D
D A, B - 4
B
A E
E A A 5
C F
F C C 6
D G
G D, F - 7
F
E H
H E E 8
G I
I G G 9
H J
J H, I - 10
I

3.4. Grouping of partial graphs

Before proceeding with the grouping, here are some situations that require the creation of a
fictional task:

A and B simultaneously A and B converge to C; B immediately A and B both converge to C and D;

and convergent prior to D and A not prior to D E immediately before D, not C
X A Y A C A C
B B
B D E D

11 = 3 + 5 + 6 + 8
The partial graphs 3 and 5 imply that tasks E and C are simultaneous because they have the same
immediately preceding task, namely task A, which gives us partial graph 11 of the
figure 1-a.
H

E
A
C
F

Figure 1-a
5
12 = 11 + 7 + 9
The partial graphs 11, 7, and 9 fit together without any contradiction to give the partial graph.
12 of figure 1-b.
H
E
A
C I
F G

Figure 1-b

13 = 12 + 4 + 10

A joint according to figure 1-C is not correct because B is neither anterior to C nor to E.

H
J
E
A C
D F I
G
B

Figure 1-c

On the other hand, A is immediately prior to E and C and converges with B towards D. It is then
it is necessary to introduce a fictitious task as shown in figure 1-d.

H
J
E
A
C I
F G

B
Figure 1-d
Notion of idleness task
Fictitious tasks, which have zero duration, express immediate precedence constraints.
When it comes to expressing timeframe constraints, we create idle tasks that play the same role.
role, but have durations equal to the imposed deadlines. For example, if G must start as early as 3
Days after the end of E, we create a leisure task of duration 3 immediately preceding G.
simultaneous with H (immediately following E).

6
3.5. Determination of start and end tasks
Start tasks: A and B because they have no previous tasks; the partial graph 4 implies that
A and B are simultaneous and convergent, which is impossible in PERT. As they have the same
departure step (step 1) they must have different arrival steps connected by a task
fictive.
Final task: J because it is not prior to any other.
3.6. Construction of the PERT network
Here is the final diagram with the numbered steps:
5

E H

2
C 4
F G I J
6 7 8 9
A
1 D
B
Figure 2
3

4. The determination of time

4.1. Definitions
Date of a milestone: the time taken to reach this milestone by performing the tasks since the beginning.
It is the sum of the durations of the tasks that have occurred up to this stage. But this time varies.
following the path taken in the network to reach the step. For example, for step 3 according to the
the path taken is 4 (task A) or 3 (dummy task + task B), because the dummy task has no duration.
Earliest date of a milestone: it is the maximum time to reach this milestone since the beginning,
that is to say the maximum duration of the task sequences that lead to this step. So the date
the earliest of the first step is 0 and that of the last step is the completion time of
the work.
Deadline for a step: it is the duration of the work minus the minimum delay since this
step to reach the end of the work, that is to say the total duration of the work minus the duration
minimal succession of tasks starting from this step to the end of the work.
Margin of a step: latest date - earliest date.
Free Slack of a task: it is the maximum delay that can be taken at the start of the task
without any effect on the dates of the subsequent stages of the network (therefore without modifying the margins
of these steps).
Total slack of a task: it is the maximum delay that can be taken at the start of the task.
without questioning the completion date (earliest date of completion) of the project.
Critical task: it is a task whose total margin is zero.
Critical path: it is a sequence of steps with zero margins from the beginning to the end of the project, connected.
by critical tasks. The length of a critical path is equal to the duration of the project. Any delay
being involved in the execution of one of its tasks compromises respect for deadlines.
4.2. Calculation of earliest dates
The calculation is done from the beginning of the network to the end.
The earliest date for the initial stage 1 TO1=0
Let us calculate the earliest date TOi of step i:
if step i is the result of a single task Ti-k,i coming from step i-k then TOi = TOi-k +
duration of Ti-k,i

- if several tasks converge to step i, then TOi = max (TOi-k + duration of Ti-k,i)
k
The results of the calculations are recorded above the steps on the diagram (in blue in figure 4)
7
Examples:
Earliest date for stage 2: only task A leads to stage 2; therefore TO2 = TO1 + duration of A,
so 0 + 4 = 4.
Similarly, TO4 = TO2 + duration of C = 4 + 1 = 5
Earliest date for step 3: task B and a dummy task (duration = 0) converge in step 3;
So TO3 = max ( TO1 + duration of B, TO2 + duration of dummy task) = max (0+2, 4+0) = 4.
Similarly, TO6 = max(TO4 + duration of F, TO3 + duration of D) = max(5+2, 4+1) = 7
4.3. Calculation of deadlines
The calculation is done by going from the end of the network to the beginning.
The latest date for the final step is the completion time of the work, that of the
the first step is 0.
The latest date for the last step (no. N) is TAN= TONthe earliest date of the last
step) .
Let us calculate the latest date TAj of step j:
if step j is the start of a single task Tj,j+s resulting in step j+s then TAj = TAj+s - duration
of Tj,j+s
if several simultaneous tasks start from step j, then TAj = min (TOj+s - duration of Tj,j+s)
s
The results of the calculations are written below the steps on the diagram (in red on the
figure 4).

Examples:
Deadline for step 8: only task J starts from step 8; therefore TA8 = TA9 - duration of J =
17-1 = 16. Likewise TA5 = TA8 – duration of H = 16 – 10 = 6. Similarly, the calculations give TA7 =
12 ; TA6 = 10 ; TA4 = 8 ; TA3 = 9.
Deadline for step 2: three tasks go from step 2: E to step 5, C to step 4 and
a fictitious task towards step 3; thus TA2 = min (TA5 - duration of E, TA4 - duration of C, TA3 - 0) =
min(6-2, 8-1, 9-0) = 4.

Figure 4

8
4.4. Calculation of free margins
The free margin of a task X (FMXis equal to the earliest date of the end step (TOFXthe date
as early as the start stage (TOD)Xthe duration of X..
4.5. Calculation of total margins
The total margin of a task X (TMX) is equal to the deadline date of the end step (TAFX) - the
earliest date of the start stage (TOD)XThe duration of X. So we have MTX= MLX+ the margin of
the end step of X and consequently MTX≥ MLX.
Table number 3 presents the margin calculation elements and the results.

Table 3: free margins and total margins

Task A B C D E F G H I J
duration 4 2 1 1 2 2 2 10 4 1
TOD 0 0 4 4 4 5 7 6 9 16
TOF 4 4 5 7 6 7 9 16 16 17
TAF 4 9 8 10 6 10 12 16 16 17
ML 0 2 0 2 0 0 0 0 3 0
MT 0 7 3 5 0 3 3 0 3 0

4.6. Critical Path Chart

According to table no. 3, the critical tasks are A, E, H, and J. They are represented by a double
arrow on the diagram in figure 5. The critical path is unique here and appears as the
succession of these tasks.

Figure 5
The critical path is the path whose sequence of tasks has the longest duration.
execution and provides the shortest completion time.
The knowledge of margins is used to optimize the execution of a project's worksite by:
the reduction of time
the optimization of resource allocation
cost control.

These aspects will be the subject of the next module.

9
 5. Resolution of execution constraints
The execution of a project can be subject to various constraints:
limited availability of resources globally or at certain times
necessity to reduce the duration of certain tasks or the entire project, or to extend certain deadlines
These constraints can be known and expressed from the outset, or arise during execution. It
We must then resolve the constraints by readjusting the PERT network.
5.1. Network representation over a time base
A representation of the network through time zones (figure 6) shows the margins
available free resources on which we will be able to play to reduce times.

Dates0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

5 H
E
8
I
2
7 J
C G
A 4 F 6 9

1
B
Tasks B, D, and I have margins.
free so that they can be delayed or
D lengthen (to free up resources) without
no changes in the rest of the network.
3

Staff

Figure 6
The examination of margins also allows for "adjusting the course of the project according to various criteria.
different from the main criterion which is the duration:
decrease in costs because there is often an inverse relationship between the duration of completing a task and
its cost. Indeed, the increasing marginal cost is due to steps in fixed costs (it takes two
machines instead of just one) and at variable costs that are more than proportional (night work,
work on public holidays, temporary labor force;
— better balance of the company's workload, particularly regarding the distribution of
resources in personnel. Indeed, for this last one it is desirable, both in terms of costs and
From a social perspective, to seek a constant workforce.
A judicious scheduling of tasks (shifts or modification of the execution duration of
non-critical tasks allow for equalization of labor needs in many cases;
suppression of a bottleneck caused by the simultaneous execution of several
operations using the same factors of production.1

5.2. Distribution of resources

A task to be executed in a given time requires resources in a given amount.
A common constraint is that the number of human resources is fixed and must be managed.
better to avoid technical unemployment during slow periods, and additional recruitment
during peak activity periods.

1
The PERT method[Link]

10
 Une fois connues les effectifs nécessaires à chaque tâche, le problème est d’occuper au mieux
the overall workforce during each unit of time while remaining within the limits of the workforce
available.
The approach is as follows:
1°) build the network on a time basis and indicate under each task the necessary workforce for
the execution of the task.
2°) create a table of staff by task and by unit of time (UT)
3°) analyze the UT network by UT from the perspective of staff numbers, and adapt the network to the constraints
set by the company by exploiting the margins.
Example:
Here is the table presenting the human resources required in terms of the number of workers for the
tasks of the network in figure 6 :
Table 4: durations and numbers of tasks
Task A B C D E F G H I J
Duration 4 2 1 1 2 2 2 10 4 1
Effective 4 2 3 2 1 4 3 1 2 2

Knowing that we only have 5 workers who can change positions, rearrange them.
affectations and the network in order to make the best use of human resources and reduce if possible
the duration of the project.
1°):

Dates0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

5 H
E
1
1 8
I
2
7 J
C G 2
2
3
A 4 F 6 3 9
4 4

1
B
2
D
2
3

Workforce 6 6 4 4 6 5 5 4 4 3 3 3 3 1 1 1 2
Figure 7
2°) table No. 5: number of staff by task and by UT
UT
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Tasks
A 4 4 4 4
E 1 1
H 1 1 1 1 1 1 1 1 1 1
J 2
B 2 2
C 3
D 2
F 4 4
G 3 3
I 2 2 2 2
Total
6 6 4 4 6 5 5 4 4 3 3 3 3 1 1 1 2
staff

11
 3°) 1st operation: Transfer a worker from task B to task H
Consequences: the duration of B is multiplied by 2 and reaches 4 UT and its free margin drops to 0;
staff during UT 1 to 4 is reduced to 5; the duration of H is halved and goes from 10 UT to 5
UT; step 8 moves to date 13 (movement blocked by task I) and step 9 to date 14;
Task H has a free margin of 2 UT and is therefore no longer critical, nor is task E;
the project duration is reduced by 3 UT, down to 14 UT.
The modified network is presented in figure 8. A new critical path appears, which is defined by
steps 1-2-4-6-7-8-9, namely tasks A,C,F,G,I,J.
Units 5 and 7 still have an overly high number of staff, by 6.
The results are presented in table no. 6 and figure 8.

Table 6:

UT
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Tasks
A 4 4 4 4
E 1 1
H 2 2 2 2 2
J 2
B 1 1 1 1
C 3
D 2
F 4 4
G 3 3
I 2 2 2 2
Total
5 5 5 5 6 5 6 5 5 4 4 2 2 2
staff

Dates0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

5 H
E 2
8
1
I
2 J
C 7 2 2
G
3
A 4 F 6 3 9
4 4

1
B
1
D
2
3

Staff 5 5 5 5 5 5 5 4 4 2 2 2 0 0 0
6 6

Figure 8

2emeoperation: transfer a worker from task D to task E and then delay the start of D by one time unit
Consequences: the duration of D multiplied by 2 becomes 2 UT and its free margin to 1 UT; the duration of E
divided by 2 goes to 1 UT with a margin of 1 UT; the number of UT 5 goes to 5 and that of the UT
7 to 7 (figure 9).

12
 Dates0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
H
5
2
E 8
I
2
2 J
7
C G 2 2
3
A 4 F 6 3 9
4 4

1 D
B 1
1

Workforce 5 5 5 5 5 5 57 5 5 4 4 422 2 0 0 0

Figure 9

3thoperation: delay of the start of H of a unit (large dashed red arrow)

Consequences: the staff of UT 7 is reduced by 2 to reach 5, and that of UT 12 increases.
move from 2 to 4.
In the end, the duration of the project was reduced from 17 to 14, and the workforce is being utilized quite fully.
5 for the UT 1 to 9
- 4 for UT 10, 11, and 12
- 2 for the UT 13 and 14

Tableau 7: initial distribution diagram of staff

UT
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Effective
6
5
4
3
2
1

Table 8: final distribution diagram of personnel

UT
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Effective
5
4
3
2
1

The reduction in times achieved was simply the result of a redeployment of resources without
additional costs. In general, a reduction in project duration is achieved by reducing
durations of critical tasks by allocating additional resources, thus with an extra cost.

13
5.3. Reduction of times
The project completion time is the length of the critical path. To reduce this duration, it is necessary to
reduce the duration of critical tasks by assigning them additional resources. The cost
The unit of these resources varies according to the task and the quantity.
The problem is to satisfy the deadline constraint at minimum cost.
To do this, the method is as follows:
1°) Draw the PERT network based on time
2°) Identify the critical path
Determine the price of the network at normal pace
4°) Establish a table indicating for each task:
the optimal duration of the task (the one initially established)
the minimum duration of the task
possible reductions on the task (optimal duration - minimum duration)
the additional costs due to successive reductions of the task
5°) Choose from the table what costs the least in task acceleration and see what
one can reduce.

Example:
Let the network represented in Figure 10. Its price is 1,500,000 F. Possible reductions and
their costs are provided in table no. 9.
Reduce the duration of the project to a minimum, for a minimum additional cost.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
G
2 5 E
R D
S

1 4 7

B
F A
C

3
H
6

Figure 10
Table No. 9
Durations Reductions Additional costs due to time savings (francs)
Tasks
Optimum Minimum possibles 1EraUT 2èmeUT 3thUT
B 4 1 3 10,000 15,000 20,000
F 3 3 0 impossible
A 5 3 2 15,000 18,000 impossible
C 4 2 2 5,000 9,000 impossible
R 3 3 0 impossible
G 3 3 0 impossible
D 2 2 0 impossible
H 5 2 3 10,000 12,000 15,000
E 3 3 0 impossible
S 2 2 0 impossible

Regarding the reduction of the total duration of the project, only reductions made on tasks of
The critical path has the expected effect. It is noted that reducing B by 3 UT costs 10000 + 15000 + 20000.
= 45000 F while reducing B, A, and C by 1 UT each costs 10000 + 15000 + 5000 = 35000 F for
the same reduction of the overall project duration.

14
To reduce the duration at minimum cost, it is therefore necessary to make successive reductions step by step.
one at a time, by choosing each time the critical path task that can be reduced
for the lowest cost.
Applying this process to the present case, we have:

1erareduction: 1 UT less on C so the end of the project changes from UT16 to UT15
Consequences: additional cost = 5000 F; C goes to 3 UT with 1 UT of
possible reductions; the free margin of E rises to 3 UT (figure 11).

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
G
2 5 E
R D
S

1 4 7

B A
F C

3
H
6

Figure 11
2emeReduction: the second UT of reduction on C costs the least. So we choose it.
and the end of the project changes from UT15 to UT14.
Consequences: additional cost = 5000 + 9000 = 14000 F; C rises to 2 UT without
possible reduction; E's free margin goes to 2 UT.

3threduction: 1 UT less on B and the end of the project goes from UT14 to UT13.
Conséquences: coût supplémentaire = 14000 + 10000 = 24000 F ; B passe à 3 UT
with 2 potential units of reductions; steps 6, 4, and 3 go to respectively
UT11, UT6 and UT3; step 5 moves to UT8; the free margin of D goes to 1 UT, that
from G to 2 UT (figure 12).

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
G
2 5

R D E
S

1 4
7
B F A
C

3 H 6

Figure 12
4threduction: 1 UT less on A or B and the end of the project moves from UT13 to UT12
Let's choose A.
Consequences: additional cost = 24000 + 15000 = 39000F; A changes to 4 UT
with 1 UT of possible reductions; step 6 goes to UT10; the free margins of E and
H passes respectively at 1 UT and 2 UT.
15
5threduction: 1 UT less on B and the end of the project changes from UT12 to UT11.
Consequences: additional cost = 39000 + 15000 = 54000 F; B goes to 2 UT
with 1 UT of possible reductions; steps 6, 5, 4 and 3 go to respectively
UT9, UT7, UT5 and UT2; the free margin of G goes to 1 UT; D no longer has margin
free; the latest date for stage 2 becomes 5-2=3 = earliest date; D becomes
Critique because without free margin and connecting two steps without margin, R as well. We have to
from now on two critical paths (figure 13).

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
G 5
2

R D E
S

1 4 7

B F
A
C

3 H
6

Figure 13
th
6 reduction: 1 UT less on A and the end of the project moves from UT11 to UT10.
Consequences: additional cost = 54000 + 18000 = 72000 F; A goes to 3 UT
without the possibility of reduction; step 6 moves to UT8 E no longer has free margin and that
de H passe à 1 UT ; S et E deviennent critiques ; on a désormais quatre chemins
critiques (figure 14).

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
G 5
2
R D E
S

1 4
7

B F
A C

3 H
6

Figure 14
At this stage and according to table no. 9, the remaining reduction possibilities are 1 Ut for B and 3.
UT for H.
Reducing H costs money and does not shorten the project's duration because H is not on a path.
critique.
If we reduce B by one unit, the total duration decreases by one unit and the common length of all
critical paths must decrease, especially the R-D-A-C path, which is not possible because A
and C have exhausted their possibilities for reduction while R and D have never been reducible.
The process thus stops, with a reduction in time from 16 UT to 6 UT and a price of 1,500,000 F to
1572000 F.

16