Solution to Building a House1

Consider the (simplified) list of activities and predecessors that are involved in
building a house, as listed in Table 1.

Activity Description Predecessors Duration (days)

Activity A Build foundation — 5
Activity B Build walls and ceilings A 8
Activity C Build roof B 10
Activity D Do electrical wiring B 5
Activity E Put in windows B 4
Activity F Put on siding E 6
Activity G Paint house C, F 3
Table 1: House-Building Activities

a. Draw a project network and use LP to find the critical path and the
minimum number of days needed to build the house.
Here is one possible network diagram (an activity-on-arc diagram). The arrows
(arcs) represent activities; the balls (nodes) have been numbered for ease of
discussion:
F
1 3 4
6
A G
5 E 3
4
0 C 5
B
Start 8 10 End

D
5
2

1
Based on 5-45 (p. 259) in Practical Management Science (2nd ed., Winston and Albright, 2001
Duxbury Press). Solution by David Juran, 2002.
Managerial Formulation
Decision Variables
We are trying to decide when to begin and end each of the activities.
Objective
Minimize the total time to complete the project.
Constraints
Each activity has a fixed duration.
There are precedence relationships among the activities.
We cannot go backwards in time.

Mathematical Formulation
This is an example of a type of project-scheduling problem that can be solved
with the Critical Path Method (CPM). These problems can be solved by hand, or
with any of a large number of project management software packages. They can
also be solved using linear programming, as is demonstrated here.
Decision Variables
Define the nodes to be discrete events. In other words, they occur at one exact
point in time. Our decision variables will be these points in time.
Define ti to be the time at which node i occurs, and at which time all activities
preceding node i have been completed.
Define t0 to be zero.
Objective
Minimize t5.
Constraints
There is really one basic type of constraint. For each activity x, let the time of its
starting node be represented by tjx and the time of its ending node be represented
by tkx. Let the duration of activity x be represented as dx.
For every activity x, t kx  t jx  d x
For every node i, t i  0

B60.2350 2 Prof. Juran

Solution Methodology
Here’s the spreadsheet model:
A B C D E F G H I J K L M N O P
F
1
1 3 4
2 6
3 1
4 A G
5 t0 t1 t2 t3 t4 t5
6
5
0 1 1 1 1 1
E 3
7 4
8 0 B C 5
9
Start
10 8 10 End
11 t0 t1 t2 t3 t4 t5
12 A -1 1 0 0 0 0 1 >= 5
13
14
B
C
0
0
-1
0
1
-1
0
0
0
1
0
0
0
0
>=
>=
8
10
D
15 D 0 0 -1 0 0 1 0 >= 5 5
16 E 0 0 -1 1 0 0 0 >= 4
17 F 0 0 0 -1 1 2
0 0 >= 6
18 G 0 0 0 0 -1 1 0 >= 3
The matrix of zeros, ones, and negative ones (B12:G18) is a means for setting up
the constraints. The sumproduct functions in H12:H18 calculate the elapsed time
between relevant pairs of nodes, corresponding to the various activities. The
duration times of the activities are in J12:J18.

B60.2350 3 Prof. Juran

Here is the optimized spreadsheet:
A B C D E F G H I J K L M N O P
F
1
1 3 4
2 6
3 26
4 A G
5 t0 t1 t2 t3 t4 t5
6 0
5
5 13 17 23 26
E 3
7 4
8 0 B C 5
9
Start
10 8 10 End
11 t0 t1 t2 t3 t4 t5
12 A -1 1 0 0 0 0 5 >= 5
13
14
B
C
0
0
-1
0
1
-1
0
0
0
1
0
0
8
10
>=
>=
8
10
D
15 D 0 0 -1 0 0 1 13 >= 5 5
16 E 0 0 -1 1 0 0 4 >= 4
17 F 0 0 0 -1 1 2
0 6 >= 6
18 G 0 0 0 0 -1 1 3 >= 3
Any activity for which t kx  t jx  d x is said to have slack time, the amount of time
by which that activity could be delayed without affecting the overall completion
time of the whole project. In this example, only activity D has any slack time (13
– 5 = 8 units of slack time).
Any activity x for which t kx  t jx  d x is defined to be a “critical” activity. Every
network of this type has at least one critical path, consisting of a set of critical
activities. In this example, there are two critical paths: A-B-C-G and A-B-E-F-G.

B60.2350 4 Prof. Juran

We have used Excel’s conditional formatting feature here to identify the activities
on the critical path.

It is also possible to identify the critical path by looking at the Solver answer
report:
Constraints
Cell Name Cell Value Formula Status Slack
$H$12 A 5 $H$12>=$J$12 Binding 0
$H$13 B 8 $H$13>=$J$13 Binding 0
$H$14 C 10 $H$14>=$J$14 Binding 0
$H$15 D 13 $H$15>=$J$15 Not Binding 8
$H$16 E 4 $H$16>=$J$16 Binding 0
$H$17 F 6 $H$17>=$J$17 Binding 0
$H$18 G 3 $H$18>=$J$18 Binding 0

Conclusions
The project will take 26 days to complete.
The only activity that is not critical is the electrical wiring.

B60.2350 5 Prof. Juran

 b. Suppose that by hiring additional workers, the duration of each
activity can be reduced. The costs per day of reducing the duration of
the activities are given in Table 2. Use LP to find the strategy that
minimizes the cost of completing the project within 20 days.

Cost per Day to Maximum Possible

Activity Reduce Duration Reduction (Days)
Build foundation $30 2
Walls and ceilings $15 3
Build roof $20 1
Electrical wiring $40 2
Put in windows $20 2
Put on siding $30 3
Paint house $40 1
Table 2: Reducing House-building Activity Times

Managerial Formulation
Decision Variables
Now the problem is not only when to schedule the activities, but also which
activities to accelerate. (In CPM jargon, accelerating an activity at an additional
cost is called “crashing”.)
Objective
Minimize the total cost of crashing.
Constraints
The project must be finished in 20 days.
Each activity has a maximum amount of crash time.
Each activity has a “basic” duration. (These durations were considered to have
been fixed in Part a; now they can be reduced.)
There are precedence relationships among the activities.
We cannot go backwards in time.

Mathematical Formulation
Decision Variables
Define the number of days that activity x is crashed to be Rx. For each activity
there is a maximum number of crash days Rmax, x
Define the crash cost per day for activity x to be Cx

B60.2350 6 Prof. Juran

Objective
7
Minimize Z = R
x 1
x Cx

Constraints
For every activity x, t kx  t jx  d x  R x
For every activity x, R x  R max, x
For every node i, t i  0

Solution Methodology
Here’s the spreadsheet model:
A B C D E F G H I J K L M N O
F
1
1 3 4
2 6
3 30
4 A G
5 t0 t1 t2 t3 t4 t5
6 0 1 51 1 1 1 E 3
7 4
8 0 B
20 <-- Max Completion Time C 5
9
10Start 8 10 End
11 t0 t1 t2 t3 t4 t5 Improved Duration Basic Duration Crash Time Max Crash Cost/Time
12 A -1 1 0 0 0 0 1 >= 4 5 1 2 $ 30
13
14
B
C
0
0
-1
0
1
-1
0
0
0
1
0
0
0
0
>=
>=
8
10
D 8
10
0
0
3
1
$
$
15
20
15 D 0 0 -1 0 0 1 0 >= 5 5 5 0 2 $ 40
16 E 0 0 -1 1 0 0 0 >= 4 4 0 2 $ 20
17 F 0 0 0 -1 1 0 2
0 >= 6 6 0 3 $ 30
18 G 0 0 0 0 -1 1 0 >= 3 3 0 1 $ 40

G3 now contains a formula to calculate the total crash cost.

The new decision variables (how long to crash each activity x, represented by Rx)
are in M12:M18.
G8 contains the required completion time, and we will constrain the value in G6
to be less than or equal to G8.
The range J12:J18 calculates the revised duration of each activity, taking into
account how much time is saved by crashing.

B60.2350 7 Prof. Juran

Here is the optimized spreadsheet:
A B C D E F G H I J K L M N O
F
1
1 3 4
2 6
3 145
4 A G
5 t0 t1 t2 t3 t4 t5
6 5
0 3 8 11 17 20 E 3
7 4
8 0 20 <-- Max Completion Time C 5
9 B
10Start 8 10 End
11 t0 t1 t2 t3 t4 t5 Improved Duration Basic Duration Crash Time Max Crash Cost/Time
12 A -1 1 0 0 0 0 3 >= 3 5 2 2 $ 30
13
14
B
C
0
0
-1
0
1
-1
0
0
0
1
0
0
5
9
>=
>=
5
9
D 8
10
3
1
3
1
$
$
15
20
15 D 0 0 -1 0 0 1 12 >= 5 5 5 0 2 $ 40
16 E 0 0 -1 1 0 0 3 >= 3 4 1 2 $ 20
17 F 0 0 0 -1 1 0 2
6 >= 6 6 0 3 $ 30
18 G 0 0 0 0 -1 1 3 >= 3 3 0 1 $ 40

B60.2350 8 Prof. Juran

Conclusions
It is feasible to complete the project in 20 days at a total crash cost of $145, as
shown in Table 3.

Basic Duration Crash Time Improved Duration Cost/Time Cost

A 5 2 3 $30 $60
B 8 3 5 $15 $45
C 10 1 9 $20 $20
D 5 0 5 $40 $-
E 4 1 3 $20 $20
F 6 0 6 $30 $-
G 3 0 3 $40 $-
$145
Table 3

1 3 F 4
6
A G
3 E 3
3
0 C 5
Start B End
5 9

D
5
2

B60.2350 9 Prof. Juran