Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

ME 6973 Spring 2017

Project Management: Scheduling Expedited Time-Cost using


Linear Programming
Ian Ibarguen, Rajitha Meka, Syed Hasib Akhter Faruqui
Department of Mechanical Engineering
University of Texas at San Antonio

Instructor: Dr. Krystel Castillo

Abstract
A construction project management case study was taken to discuss the approaches to deal with the scheduling
problems that appear frequently in the real-time projects. The project is to construct a road scheduled in two phases
with a total of 270 days. On the decision of expediting the project, a linear programming model was formulated to
minimize the cost of the project associated with the expedition of activities. The model helped to decide on activities
to expedite at less cost. The model was also compared with CPM, which was most used in construction projects.

Keywords
Optimization, Integer Linear Programming, Project Management, CPM

1. Introduction

Knowledge of process optimization is important for many engineering tasks specially for those managers who may
be responsible for decision making, especially in industrial fields. Since so many engineers seek to continually
improve their processes, it is easy to see why finding an optimal situation would be valuable to find the current
limits of the process. A common tool for process optimization is the use of linear programming (LP), which seeks an
optimal outcome (such as maximizing output or minimizing cost) for a situation restricted by a number of linear
inequalities call constraints. The purpose of this report is to apply the techniques of linear programing to a real-
world scenario to show how an engineer and their company can benefit from it.

Construction is a good field to examine for a number of reasons, one being that it is desirable to minimize the
overall cost of a project while still meeting the demands of the customer. Especially when it comes to completing
the project on time. Deadlines in the construction industry can be affected by several facts such as late material
deliveries, poor weather, or changes in customer requirements. Depending on the cause, a construction company
may be required to accept the extra cost associated with deadline issues, and sometimes the customer may wish to
pay extra to have a job finished earlier. In these cases, it is not uncommon for a construction firm to feel the need to
“Crash” a project schedule. Being able to crash a project is a way of saying that a task may not need to take as much
time as it is scheduled for, but that performing the task in less time will cost more. It is often best used as a last
resort, especially if the project has not already been fast-tracked (a method of finding the critical path to minimize
project time) due to the extra costs involved. But if there is no other choice, it is best to find ways to minimize the
loss in profits associated with crashing a project.

A construction project tasked with building a road was chosen as the focus of this report. The Project in question is
originally scheduled to take 270 days to complete. But due to unknown circumstances, the project manager has
decided that construction needs to be finished in 264 days. The problem is that the project is already fast-tracked so
the schedule needs to be crashed. Using linear programing techniques, a model will be developed with the intention
of minimizing the cost of this crash on the construction company.

1
ME 6973 Spring 2017

2. Literature Review

Construction projects require several resources demanding large amount of capital to be spent. The sequence of
activities to meet the final objective of constructing the road should be scheduled in a realistic way to avoid chances
of scheduling failures. Another important concept of scheduling is being able to schedule all the activities in the
optimal way. The project manager should be responsible to make sure to think about various possibilities the project
can be done in an effective way and should be able to predict the effect of each decision on the project. The
management should develop and evaluate all alternate ways and determine the optimal schedule.

Generally, when the project is to be completed early than the due date, the activities will be expedited by crashing
the time by increasing the cost by hiring more labor or purchasing a machine or using other specialized resources.
This is the tradeoff between the time and cost. Assuming that the cost varies linearly with time of activity
completion, there are several linear programming approaches determining the optimal way to project crashing.
These algorithms help to expedite the project by providing the information about which activities can be crashed and
by what amount. From all the real projects studied by many experts, there is a theory that, there exists a small
number of activities, which are critical when crashed show a large effect (Kelley Jr 1961, Zhou et al. 2013).

CPM is one of the methods widely used in scheduling the construction projects. However, later on for some specific
projects the usage of CPM decreased due to its complexity. Yamin and Harmelink(2001) provided a comparison
between CPM and a linear scheduling model based on their performance in three important stages of project
management. The framework of comparison was based on list of attributes that any ideal scheduling tool should
meet. The linear scheduling model has great visualization features, which can be helpful in the communication for
specific types of projects, but lacks some resource management capabilities. CSM on the other hand is a complete
scheduling tool because of its multiple resource management techniques and statistical analysis developed for it.

Ipsilandis (2007) proposed a linear programming model to address multi objective nature of decisions that
management have to face in repetitive construction projects. Repetitive scheduling method is one more method other
than CPM/PERT used in planning and controlling repetitive construction projects. All these traditional methods
focus on minimization of duration of the project by providing critical path and the optimum time-cost tradeoff by
crashing the critical activities but have some inadequacies, as they cannot guarantee the work continuity of the tasks.
The linear programming model addressed these issues so that the managers can develop several schedules and can
select the best suited meeting the multiple objectives during the project.

Chassiakos and Sakellaropoulos (2005) proposed two methods an exact and approximate method to provide the
optimal solution for the construction projects by considering the project characteristics. The time-cost tradeoff
analysis was performed by taking into account the precedence relationships of the activities, planning and external
time constraints; and late/early project completions of the project. The proposed exact method provides the optimal
solution by considering all the possible time-cost alternatives for each activity thus minimizing the project cost. The
appropriate method cannot guarantee the optimal solution but flexible when compared to exact one. This method
helps to reduce the project duration by assessing the critical activities and choosing the best activity based on time-
cost effectiveness to expedite the project. A highway upgrading project having two phases with 29 activities was
tested using two methods. The approximate method provided the solution that is accurate more than 99% in terms of
cost.

3. Linear Programming

Linear programming, sometimes known as linear optimization, is the problem of maximizing or minimizing a linear
function over a convex polyhedron specified by linear and non-negativity constraints. Simplistically, linear
programming is the optimization of an outcome based on some set of constraints using a linear mathematical model.
Linear programs are problems that can be expressed in canonical form as-

Maximize CTx
Subject to Ax ≤ b
x≥0

2
ME 6973 Spring 2017

Linear models can be solved in a number of ways. But before the start of optimization it is necessary to approach the
problem in a way so that the formulation doesn’t over constrain the overall system. For the problem described in this
paper, data requirements are matched so that it is on per with the regulations. The assumptions made to reach the
best result for the model is given as-
1) Normal Duration of the Project activities and costs associated with the activities.
2) Number of days an activity can be expedited and the time-cost associated with that.
3) Activity dependencies (Either all the activities can be started simultaneously or there remains
interdependencies).
4) The final expedited time (saving 6 days from overall project time).
Terms associated with the model is described below-

Associated Unit Cost for Crashing:


( 𝐶𝑐𝑟𝑎𝑠ℎ𝑒𝑑 – 𝐶𝑁𝑜𝑟𝑚𝑎𝑙 )
𝐶 𝑢𝑛𝑖𝑡 =
𝑇𝑈𝑝𝑑𝑎𝑡𝑒𝑑

Where,
𝐶 𝑢𝑛𝑖𝑡 = Unit Cost or Cost Slop
𝐶𝑐𝑟𝑎𝑠ℎ𝑒𝑑 = Cost associated with the crashing time
𝐶𝑁𝑜𝑟𝑚𝑎𝑙 = Normal Cost associated with the task
𝑇𝑈𝑝𝑑𝑎𝑡𝑒𝑑 = 𝑇𝑁𝑜𝑟𝑚𝑎𝑙 -𝑇𝐶𝑟𝑎𝑠ℎ𝑒𝑑 = Updated Time after Crashing = Allowable crashing time

Variables:

Xi = Number of days to crash for an activity.


Si = Start Time of an Activity. (Not slack)
Where, i= 1,2, …………., n

This is a time constrained problem where days has to be reduced in a whole rather than partial. Thus, integer
problem is used to solve this problem.
Now, to setup the objective function. The objective of this project is to minimize the cost of crashing time by
minimizing the number of day an activity can run. Thus, the optimization function will be the multiplication of Unit
Cost and time amount. Mathematically-

Minimize, Z = ∑𝑛𝑖=1 𝐶𝑢𝑛𝑖𝑡 𝑖 𝑋𝑖


Constraints:

If the work is to complete within the specific time frame, then either the number of workforce has to be increased or
increasing the hours of productivity. Both the task includes cost for hiring the additional workforce or for the
existing workforce. But there remains a threshold time to which a single task can be expedited. Considering that,
from the available data the total amount of time that can be saved from the overall process may be 17 days but
reducing that whole time may cause excessive additional cost to the available work system and may include
excessive workforce than required which is not an efficient approach. Thus, an optimization process is required for
the available system. The constraints required for the system can be given as-

𝑋𝑖 ≤ Allowable crashing time, 𝑇𝑈𝑝𝑑𝑎𝑡𝑒𝑑 , where i = 1,2, …., n


Si = (Start Time + Normal Duration – Crash Duration).

It is to be mentioned, the data for normal starting time of an activity in normal time is not being provided. So, it is
assumed that each work starts at the end of its predecessor work in as shown in the network activity figure- 1.

3
ME 6973 Spring 2017

Non-negativity constraints-
Xi, Si ≥ 0
Xi , Si Є Z

This formulation is a disaggregated model for ILPs. The project expedition time is compared by solving the problem
again using the CPM method. But, it is to be mentioned that CPM focuses on the activities on the critical path
whereas, using the method of ILP accounts for both the critical and non-critical activities which in term becomes
critical as the project time is expedited. The problem is solved using Excel Solver for LP’s.

4. Data

This project will discuss the implementation of Linear programming to solve an Expedited Time-cost Scheduling
problem. This problem deals with a project where a road infrastructure company has to construct a 463.875-meter-
long road and it consists of two phases (Elmabrouk, 2012).
Phase I: Foul and Storm drainage work
Phase II: Roads and sidewalks
Both phases will need a fixed amount of time to complete and a cost is associated with each step. Table 1 contains
the duration of the activities for both the phases. These activities have a dependency to each other i.e. one task
cannot start before the other ends. As there was no other chart provided from the project management is assumed
that a new activity starts only after the end of its predecessor. The Normal cost and Expedition Cost is also shown in
the table. These dependencies along with the critical path is shown in figure -1.

Table 1: List of Schedule tasks with their Normal and Expedited Cost
Crash
Activity Normal Duration Normal Cost Expedition
ID Activity Description Time
Dependecies (Days) ($) Cost ($)
(Days)
Phase-I
1 Excavation foul - 10 9 $ 4,400.00 $ 4,840.00
2 Foul pipe installation, backfilling & testing 1 10 9 $ 3,035.00 $ 33,385.00
3 Foul manhole construction 1 10 9 $ 4,200.00 $ 4,620.00
4 Foul laterals & hours connections 2,3 10 8 $ 87,600.00 $ 105,120.00
5 Excavation storm 4 10 9 $ 4,400.00 $ 4,840.00
6 Storm pipe installation, backfilling & testing 5 10 9 $ 30,350.00 $ 33,385.00
7 Storm manhole construction 5 10 9 $ 4,200.00 $ 4,620.00
8 Storm pipe junctions backfilling 5 10 9 $ 2,400.00 $ 2,640.00
Phase-II
9 Excavation to formation 6,7,8 30 28 $ 7,650.00 $ 8,160.00
10 Water house connections 9 30 28 $ 10,950.00 $ 11,680.00
11 Cable crossing ducts 9 30 28 $ 12,570.00 $ 13,408.00
12 Gully& gully connection to storm 9 30 28 $ 10,400.00 $ 11,093.00
13 Road endings 10 30 28 $ 29,643.00 $ 31,620.00
14 Compaction to formation 13 30 28 $ 20,161.00 $ 21,505.00
15 Sub base laying & compaction 14 30 28 $ 28,292.00 $ 30,178.00
16 Granular base coarse laying & compaction 15 30 28 $ 33,133.00 $ 35,342.00
17 Asphalt binder to roads 16 10 10 $ 33,416.00 $ 33,416.00
18 House concrete steps and ramps 16 12 10 $ 5,460.00 $ 6,370.00
19 Sidewalk backfill & base coarse 17 10 10 $ 7,400.00 $ 7,400.00
20 Sidewalk wearing 11,12,19 10 10 $ 12,600.00 $ 12,600.00
21 Asphalt wearing to roads 18,20 10 10 $ 38,567.00 $ 38,567.00
22 Finish roads& sidewalks road 21 0 0 $ - $ -

4
ME 6973 Spring 2017

11

2 7

1 4 5 10 6 10 9 30 10 30 13 30 14 30 15 30 16 30 17 10 19 10 20 10 21 10 22

3 8

12 18

Figure 1: Activity Dependency and Critical Path

At the beginning of this construction work it was planned to be completed within 270 days based on a critical path
analysis. Now, due to management decision this whole project has to be finished within 264 days. This expedited
time has a cost associated with it. It has to be planned which activities are to expedite so that the additional cost
associated with it is kept at minimum. Table-2 shows the cost associated with the expedition of activity and the
maximum expedition time.
Table 2: Cost-time associated with the tasks

Time Saved Additional Cost Crash Cost/Day


ID
(Days) for Crashing ($) ($/day)

Phase-I
1 1 $ 440.00 $ 440.00
2 1 $ 30,350.00 $ 30,350.00
3 1 $ 420.00 $ 420.00
4 2 $ 17,520.00 $ 8,760.00
5 1 $ 440.00 $ 440.00
6 1 $ 3,035.00 $ 3,035.00
7 1 $ 420.00 $ 420.00
8 1 $ 240.00 $ 240.00
Phase-II
9 2 $ 510.00 $ 255.00
10 2 $ 730.00 $ 365.00
11 2 $ 838.00 $ 419.00
12 2 $ 693.00 $ 346.50
13 2 $ 1,977.00 $ 988.50
14 2 $ 1,344.00 $ 672.00
15 2 $ 1,886.00 $ 943.00
16 2 $ 2,209.00 $ 1,104.50
17 0 $ - -
18 2 $ 910.00 $ 455.00
19 0 $ - -
20 0 $ - -
21 0 $ - -
22 0 $ - -

5. Model Formulation

Based on discussion on section 3 the model for analysis can be formulated from the data. A sample formulation for
the system is shown below-

Activity 4 can be crashed for 2 days. The Normal Cost for the Activity is $87,600.00 and the Expedition Cost
associated with it is $105,120.00. Thus,

( 𝐶𝑐𝑟𝑎𝑠ℎ𝑒𝑑 – 𝐶𝑁𝑜𝑟𝑚𝑎𝑙 ) = $ (105,120.00-87,600.00)


= $ 17,520

5
ME 6973 Spring 2017

𝑇𝑈𝑝𝑑𝑎𝑡𝑒𝑑 = 𝑇𝑁𝑜𝑟𝑚𝑎𝑙 -𝑇𝐶𝑟𝑎𝑠ℎ𝑒𝑑


= 10 – 8 = 2

$ 17,520
𝐶 𝑢𝑛𝑖𝑡 =
2

= $ 8,760.00
formulation the same way for all the activities the objective function becomes-

Minimize, Z= 440x1 + 30350x2 + 420x3 + 8760x4 + 440x5 + 3035x6 + 420x7 + 240x8 + 255x9
+ 365x10 + 419x11 + 346x12 + 988.50x13+ 672x14 + 943x15 +1104.50x16 + 455x18

Constraints:
Crashing time Constraints,
X4 ≤ 2

Start time Constraints,


Si = Si-1 + Normal Duration – Crash Duration
Si - Si-1 + Crash Duration = Normal Duration
Thus, S4 – S3 + x3 ≥ 10;

Project Duration Constraint,


S22 ≤ (270-6)
≤ 264
Xi ,Si Є Z

6. Analysis and Results

Considering the problem, it is a straight forward LP problem. Thus, reducing a task activity time may affect the
whole calculation and minimize the cost associated with it. But to meet the scheduling constraint for the whole
project the analysis of Critical Path is necessary as some tasks must be finished before the other. This activity
dependency and critical path is shown in figure 1. These activity dependency is important scheduling and optimizing
the project. A task could theoretically cost only $2 to crash from 10 days to 9 days, but if the task isn’t part of the
critical path it won’t do the schedule any good. Based on the critical path, the overall project should take 270 days to
complete. The original paper (Elmabrouk, 2012) mentioned the project taking less time since it is implied that one
task does not have to wait for the preceding task(s) to finish before starting. However, the information on how long
each task must wait before starting is not provided in the paper. So, it is assumed that a task must be finished before
the next in sequence to begin. It is also assumed that, the manager has to expedite the time as a whole day. The time
crashed cannot be of any fraction of the day. Thus, making it a Integer Linear Programming (ILP) and the second
assumption made was that a task could not be “partially crashed”. If a task saves a set amount of time from crashing,
it cannot save half of that time for half the cost.

Microsoft Excel with the solver add in was used to solve the IP problem using the simplex method. The problem is
solved in two steps. The first involved using the data on production times and requirements to create a diagram
using the Critical Path Method (CPM). And in second step is LP optimization. Using the CPM diagram it can be
seen that tasks 17, 19, 20, 21, and 22 are not part of the critical path and can be ignored. In addition, task 2 and 3, as
well as 6, 7, and 8, must be completed simultaneously in order to minimize the overall time. This means these could
be considered as one task for the linear problem, with their costs added together. Table 3 shows the modified data
variable systems for the analysis. It would not make sense to crash the time for one of these tasks when the
subsequent task depends on the completion of another un-crashed task. Using this information, the number of
variables that need to be examined is only 11 and can be solved with 1 main constraint, along with a non-negativity
and binary constraint. The spreadsheet and solver inputs are shown below. The boxes in blue contain the formulas
used for the solver. The two formulas are ∑𝑛𝑖=1 𝑋𝑖 𝐶𝑖 for the objective function, where C is the cost associated with
crashing an activity, and ∑𝑛𝑖=1 𝑋𝑖 ≥ 𝑑𝑒𝑠𝑖𝑟𝑒𝑑 𝑡𝑖𝑚𝑒 𝑠𝑎𝑣𝑒𝑑 for the constraint, where T is the time saved by crashing.

6
ME 6973 Spring 2017

Table-3: Modified Data Variables


Task Cost/Day Time Saved Variable
1 440 1 x1
2,3 30770 1 x2
4 17520 2 x3
5 440 1 x4
6,7,8 3695 1 x5
9 255 2 x6
10 365 2 x7
13 988.5 2 x8
14 672 2 x9
15 943 2 x10
16 1104.5 2 x11

Green boxes contain the variable solutions. The red box would be edited by the user based on the number of days
needed to be saved by crashing, up to a max of 17 days (Based on CPM and expedition time it is impossible to save
more time than this) but the associated cost will certainly rise. As the initial plan is to expedite only 6 days. The
considered solution is based on that very fact.

Figure-2: SIMPLEX method in Excel

The problem has 44 variables, X1-X22 for crash time and S1-S22 for task starting time (S in this case is not slack
variables). Based on the CPM it has in total 50 constraints and integer for the S, plus non-negativity constraints. The
use of linear program considers activity in both critical and non-critical path. This is helpful in case the manager
wants to expedite the work even more than 6 days while minimizing the cost associated with it.

Both methods give the same minimum cost for this crash time, $2,120, with activities 1, 5, 9, and 10 being crashed.
The cost associated respectively $440, $440, $510 & $730 and crashed times are 1, 1, 2 & 2 days. The total crash
time needed can be adjusted in both models. The minimum time that can be saved is 1 day for a minimum cost of
$440. Up 17 days can be saved for a cost of $30,751. Table-4 shows different additional cost associated with the
crashing times to give an idea of crash-cost to make a proper schedule management decision. This additional cost
will be added on top of the normal cost $390,827.00 for the whole construction project. The results prove that the
use of linear programing is a valid method for solving this type of problem, but that combining it with other tools
such as CPM can make the process much simpler (5 dual Simplex Iterations for LP and 2 Simplex Iteration for
method including CPM).

7
ME 6973 Spring 2017

Table-4: Crash costs associated with Number of Days Project needs to be Expedited
Crash Time 6 8 10 12 14 17
Additional Cost $2,120.00 $ 3,464.00 $ 5,350.00 $7,327.00 $9,536.00 $30,751.00

7. Conclusion
The formulation presented here support the evaluation of project expedition as set by management. Based on Linear
Programming model which incorporates the provided data and constraints to calculate the task crashing time and
works as a decision support system for the management in terms of optimizing the additional cost. It automates and
improves management’s decision making process by providing better information in a timely manner. The project
requirement for the current project expedition is 6 days. Based on Budget allocation the manager can always
investigate the effect of expediting additional days to finish the project to its earliest possible time.

References
Chassiakos, A.P. and Sakellaropoulos, S.P., 2005. Time-cost optimization of construction projects with generalized
activity constraints. Journal of Construction Engineering and Management, 131(10), pp.1115-1124.
Elmabrouk, O.M., 2012, July. Scheduling Project Crashing Time using Linear Programming Technique.
In Proceedings of the 2012 International Conference on Industrial Engineering and Operations
Management (pp. 3-6).
Ipsilandis, P.G., 2007. Multiobjective linear programming model for scheduling linear repetitive projects. Journal of
Construction Engineering and Management, 133(6), pp.417-424.
Kelley Jr, J.E., 1961. Critical-path planning and scheduling: Mathematical basis. Operations research, 9(3), pp.296-
320.
Yamin, R.A. and Harmelink, D.J., 2001. Comparison of linear scheduling model (LSM) and critical path method
(CPM). Journal of Construction Engineering and Management, 127(5), pp.374-381.
Zhou, J., Love, P.E., Wang, X., Teo, K.L. and Irani, Z., 2013. A review of methods and algorithms for optimizing
construction scheduling. Journal of the Operational Research Society, 64(8), pp.1091-1105.

You might also like