P. 1
Project Management

Project Management

|Views: 3|Likes:
Published by hema16

More info:

Published by: hema16 on Dec 18, 2013
Copyright:Attribution Non-commercial


Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less





461 Journal of Asian Architecture and Building Engineering/November 2005/466

Enhanced Resource Leveling Technique for Project Scheduling
Jaejun Kim*
, Kyunghwan Kim
, Namyong Jee
and Yungsang Yoon
Professor, Dept. of Architectural Engineering, Hanyang University, Seoul, Korea
Research Scientist, Construction Research Institute, Hanyang University, Seoul, Korea
Professor, Dept. of Architectural Engineering, Hanyang University, Seoul, Korea
Graduate Student, Dept. of Architectural Engineering, Hanyang University, Seoul, Korea
Since a construction project involves a series of activities that utilize resources to achieve its goal within
a given time period, efficient resource management is a prerequisite for its success. Two major techniques
of resource management are resource constrained scheduling (RCS), which focuses on limited resource
availability, and resource leveling, which focuses on a smooth resource usage pattern for a fixed project
completion time. Although they can successfully generate schedules that accomplish their own objectives,
they have certain limitations. RCS assumes a single fixed resource supply, and resource leveling may not
improve the smoothness of resource usage beyond a certain level due to the fixed project completion time. In
order to overcome these limitations, this study proposes a model to enhance the minimum moment algorithm
of resource leveling for the efficient use of resources within an appropriate project completion time.
Keywords: planning; scheduling; leveling; resource; resource constrained scheduling
A construction project involves a group of activities
to accomplish its goals within a specific amount of
time. Each activity requires certain resources, such
as labor, materials, equipment, etc., to carry out
assigned tasks. However, these resources are not
always provided at construction sites upon demand
because some of them are scarce in the market. If some
resources are not fully available for certain activities,
the work cannot be properly executed as scheduled,
probably resulting in a project completion delay. In
addition, it might not be a good strategy to acquire
the demanded resources without planning for their
effective utilization. Therefore, in order to manage a
construction project successfully, resources should be
allocated carefully throughout the project phase by
considering demand, supply, and effective utilization.
Typically, two techniques—resource constrained
schedul i ng (RCS) and resource l evel i ng—are
employed for allocating resources in a scheduling
process; they are compared in Fig.1. Both techniques
can be applied after the critical path method (CPM).
The RCS technique sets the maximum availability of
resources and attempts to eliminate resource overuse
periods by delaying activities. If an activity is delayed
*Contact Author: Jae-Jun Kim, Professor,
Dept. of Architectural Engineering, Hanyang University,
Haengdang-dong 17, Sungdong-ku,
Seoul 133-791, Korea
Tel: +82-2-2220-0316 Fax: +82-2-2296-1583
e-mail: jjkim@hanyang.ac.kr
(Received May 10, 2005; accepted August 25, 2005)
beyond its total float, the project completion time
should be extended, as shown in Fig.1.(b). On the
other hand, the resource leveling technique attempts to
reduce the sharp variations of resource demands, while
maintaining the original project completion time, as
shown in Fig.1.(c). Although each method successfully
generates a schedule that accomplishes its own
objective, there are several limitations to both methods.
In the RCS technique, the maximum availability of
resources must be fixed as a single number prior to the
scheduling process, as shown in Fig.1.(b). It does not
allow a range of resource availability, e.g., 8~10 laborers,
which occurs in practice in most construction projects.
On the other hand, resource leveling may not improve
some high peaks or valleys, as shown in Fig.1.(c). The
peak resource demand (12) is still present in order to
maintain the initial project completion time. Furthermore,
it cannot resolve the case in which a high peak is beyond
the resource availability. Finally, there is no systematic
way for applying both methods simultaneously.
This study proposes an enhanced resource leveling
technique with an increased capability to overcome
the aforementioned limitations of the current resource
handling techniques.
Heuristic Methods
Various analytical and heuristic methods have been
developed to apply resources during the scheduling
process (Ahuja 1976; David and Patterson 1975;
Hegazy 1999; Kelly 1963; Wiest 1967). Analytical
methods attempt to determine the optimum solution in
terms of minimum project resources and/or duration.
However, they require a large amount of computational
462 JAABE vol.4 no.2 November 2005 Jaejun Kim
time and are therefore impractical for real construction
projects. On the other hand, heuristic approaches
provide reasonable solutions within a practical period
of time (Boctor 1990; Hegazy 1999). The important
indexes of typical heuristic approaches for evaluating
the resource usage efficiency are described as follows.

The resource moment of variance (Mv) and the
resource moment about the x axis or time axis (Mx)
represent the level of resource fluctuations throughout
the project duration. Smaller values of these indexes
indicate a more uniform or effective resource allocation.
The resource moment about the y axis or the project
start time (My) indicates the uncertainty of the demand
and supply of future resources. A higher value of My
indicates greater uncertainty. The rate of resource usage
(RR) shows a wait or idle status of resources. It indicates
the rate of total resource usage in comparison with the
maximum amount of resources utilized.
This study considers these four indexes to evaluate
the effectiveness of a schedule generated by the
proposed technique. They are independent of each
other, so that a survey was conducted to determine
their relative importance in the scheduling process. The
survey adopted the analytic hierarchy process (AHP)
(Satty 1980).
Analytic Hierarchy Process
The AHP method is appropriate for considering
objective or subjective evaluation factors. One of its
advantages is its relatively simple logic based on a
respondent's intuition rather than mathematical results.
Some of the features of the AHP are described as follows.
First, it has clear structures and factors that can be
understood easily. Second, it separates complicated and
uncertain problems into several parts and analyzes their
importance by performing a one-to-one comparison.
Third, it makes experiential decisions based on
subjective judgments and their collation, followed by
conclusions. Fourth, it decides between participants
by drawing a conclusion based on a comparison of
each person's opinion. Due to these advantages, AHP
was adopted to gather the opinions of construction
Fig.1. Comparison between Resource Profiles of RCS and
Resource Leveling Techniques
(a) Initial Schedule
(b) RCS Schedule
(c) Resource Leveling Schedule
Table 1. Survey Result of a Professional
Table 3. Average Weight of All Surveys
Table 2. Result of AHP Analysis of Table 1.
Mx My Mv RR GM Weight
Mx 1 3 5 3 2.59 0.507
My 1/3 1 3 3 1.316 0.258
Mv 1/5 1/3 1 1/5 0.34 0.067
RR 1/3 1/3 5 1 0.863 0.169
Sum 5.109

463 JAABE vol.4 no.2 November 2005 Jaejun Kim
professionals on the relative importance of each
heuristic index.
A survey of 49 professionals was conducted in
October 2004. Most of them had worked as scheduling
engineers for 10 years or more at major construction
corporations in Korea. After a detailed explanation
wi t h a si mpl e schedul i ng exampl e, t hey were
interviewed individually for their opinions on the
four indexes. Table 1. shows the survey result of an
interviewed professional, and Table 2. shows the AHP
analysis of Table 1. This professional believes that Mx
is the most important and Mv is the least important as
compared with the others. The average weights of all
the 49 responses are listed in Table 3.
Enhanced Resource Leveling Technique
The conventional resource leveling technique
attempts to minimize the peaks and valleys of a
resource usage pattern for improving the resource
usage efficiency throughout the project period.
However, since it assumes a fixed project completion
time, it may not improve some high peaks or valleys
and cannot resolve a high peak that extends beyond
the resource availability, as shown in Fig.2.(a). The
enhanced resource leveling technique overcomes this
problem. It enhances the features of the minimum
moment algorithm (MMA) (Harris 1978) by extending
the project completion time, as shown in Fig.2.(b). The
four heuristic indexes introduced earlier are applied in
order to determine the best project schedule in terms of
resource utilization.
The enhanced resource leveling technique consists
of four steps. Table 4. shows an example of a schedule
used to explain each step. In this table, each activity is
assigned an ID, a duration, resource requirement, and
Step 1: CPM
In step 1, the CPM forward and backward passes are
performed. After the CPM, the project duration and
early/late time and float of each activity are computed,
as shown in Fig.3.
Step 2: MMA
Step 2 applies the MMA to determine the initial
resource leveled schedule. The detailed process of
the MMA is not described because it is beyond the
scope of this study. Instead, the CPM Level Program
developed by Martinez (1992) is applied for depicting
the scheduling outputs. Fig.4.(a) shows the initial CPM
schedule and its resource profile. Fig.4.(b) shows the
MMA result in which resources are leveled smoothly
as compared to the CPM schedule.
Step 3: completion extension
The project completion time is extended day by day
to find the most suitable schedule. In order to extend
the completion time without affecting the current
condition of activities, a dummy activity that does
not require any resources and whose duration is the
sum of the initial project duration and the amount of
extension is applied. For example, when the initial
project duration is 17 days, a dummy activity with a
duration of (17 + n) days is used to extend the project
completion time by n days. Fig.5. shows the result of
extending the completion time by one day. Activity z
that denotes the topmost bar in the chart is the dummy
activity with a duration of 18 days.
For each time extension, the values of the four
indexes are computed, as shown in Table 5. The
extension continues until there is no change in the
index values. In the example schedule, there is no
change in the index values on the 7th day of extension,
as shown in Table 5.
Step 4: weighted indexes
The four indexes computed in Step 3 are independent
of each other. In order to integrate them, each value
(a) Result of Conventional Resource Leveling
(b) Result of Enhanced Resource Leveling
Fig.2. Comparison between Traditional and Enhanced Resource
Leveling Techniques
Table 4. Example of Schedule Data
ID Dur. Res. Pred.
A 1 1 -
B 2 6 A
C 2 4 A
D 1 4 A
E 4 3 B
F 1 4 E
G 3 5 C,D,E
H 2 4 E
I 2 1 G
J 2 3 I
K 2 4 F,J
L 1 2 K,H
464 JAABE vol.4 no.2 November 2005 Jaejun Kim
is translated to a percentage value. For example, the
best and the worst Mx values are converted to 100%
and 0%, respectively. Similarly, the other values are
also converted to relevant percentage values. Each
percentage value of Mx, My, Mv, and RR is listed
in Table 6. Therefore, the total value of each time
extension is computed by adding the weighted index
values calculated for each day. For example, the total
value of the time extension of one day is computed as
follows: (21.429 × 0.245) + (93.496 × 0.248) + (92.000
× 0.235) + (75.539 × 0.272) = 70.584. Based on the
combination of all weighted index values, the time
extension of one day generates the best schedule for
the example considered above.
It should be noted that this time extension of one
day is not always the best schedule. It is decided by
the weight of each surveyed index. In other words,
if a project manager believes that Mx and Mv are
more important than the others, he/she may select the
Fig.3. CPM Network of Schedule of Table 4.
(a) Resource Profile with CPM Schedule
(b) Resource Profile with MMA
Fig.4. Scheduling Output of CPM Level Program
Fig.5. Resource Leveling Output after Extension of One Day
Table 5. Results of Extension of Project Completion
Extension Mx My Mv RR (%)
0 448 698 76 68.91
1 424 714 58 65.08
2 416 740 74 61.65
3 408 761 74 58.57
4 384 818 86 55.78
5 360 883 62 53.25
6 336 944 54 59.42
7 336 944 54 59.42
Table 6. Converted Indexes and Combined Weighted Values
Ext. Mx My Mv RR Total
0 0.000 (7)
1 0 0 . 0 0 0
56.000 (7)
1 0 0 . 0 0 0
64.588 (2)
1 21.429 (6) 93.496 (2) 92.000 (3) 75.539 (2) 70.584 (1)
2 28.571 (5) 82.927 (3) 60.000 (5) 53.666 (3) 55.965 (5)
3 35.714 (4) 74.390 (4) 60.000 (5) 33.981 (6) 50.355 (6)
4 57.143 (3) 51.220 (5) 36.000 (8) 16.171 (7) 39.363 (8)
5 78.571 (2) 24.797 (6) 84.000 (4) 0.000 (8) 45.909 (7)
1 0 0 . 0 0 0
0.000 (7)
1 0 0 . 0 0 0
39.402 (4) 60.017 (3)
1 0 0 . 0 0 0
0.000 (7)
1 0 0 . 0 0 0
39.402 (4) 60.017 (3)
weight 0.245 0.248 0.235 0.272
( ) rank (%)
total = 0.245 × Mx + 0.248 × My + 0.235 × Mv + 0.272 × RR
465 JAABE vol.4 no.2 November 2005 Jaejun Kim
schedule with a time extension of six days. If early
project completion is the most important priority, then
a schedule can be selected by the conventional resource
leveling technique. Thus, the enhanced resource
leveling technique provides various alternatives for
project managers to select the most suitable schedule
depending on project conditions.
Comparison with Other Scheduling Methods
The best schedule generated by the proposed
technique is compared with other related schedules of the
CPM, conventional resource leveling, and RCS. Fig.6.
shows the resource profile of each scheduling output.
Conventional resource leveling removes many peaks
and valleys of the CPM output, as shown in Fig.6.(a)
and 6.(b). In the best schedule generated by the proposed
technique, the peak on day 13 of the conventional
method is reduced with a project completion delay of
one day, as shown in Fig.6.(b) and 6(c).
The resource profile statuses of Fig.6.(d), 6(e),
and 6(f) are obtained from the RCS of the Primavera
Project Planner™ with maximum units of 7, 6, and 5,
respectively. A resource supply reduction from seven to
six results in a completion delay of five days, as shown
in Fig.6.(d) and 6(e). With a maximum of five resource
units, the profile of Fig.6.(f) is the same as that of
Fig.6.(e). However, the resource requirements of days
2 and 3 exceed the availability because Activity B
requires six units of resources, as shown in Fig.3.
Fig.7. compares the index values of each RCS result
with those of the enhanced resource leveling technique
from Table 6. Based on the combined weighted indexes
shown in Fig.7.(e), it is observed that the best schedule
generated by the proposed technique (18-day completion
time) is superior to all the RCS results for the example
schedule. Since each value indicates relative importance,
a higher value represents a better result.
Summary and Conclusion
Resources should be considered during the scheduling
process in order to manage the construction project
successfully. Resource constrained scheduling (RCS)
and resource leveling are two conventional techniques
for allocating resources. Although they can successfully
generate schedules that accomplish their own objectives,
they have certain limitations. RCS assumes a single
fixed resource supply, and resource leveling may not
improve the smoothness of resource usage beyond a
certain level due to the fixed project completion time.
This study proposes the enhanced resource leveling
technique that overcomes these limitations. It considers
four indexes to determine the most suitable schedule for
effective resource usage. In addition, it generates various
alternative schedules with resource demand status that
provide project managers with greater flexibility to select
an appropriate schedule based on project conditions.
This study could serve as a basis to generate a
cost-effective schedule by considering resource
(a) CPM Result
(b) Output of Conventional Resource Leveling
(c) Best Chedule Generated by Enhanced Resource Leveling
(d) RCS Schedule with a Max of Seven Resource Units
(e) RCS Schedule with a Max of Six Resource Units
(f) RCS Schedule with a Max of Five Resource Units
Fig.6. Resource Profile of each Scheduling Technique
466 JAABE vol.4 no.2 November 2005 Jaejun Kim
effect i veness. Furt her research i s requi red for
devel opi ng a cost i nt egrat ed resource l evel i ng
technique that includes indirect cost, additional cost
to supply extra resources, effect of a congested work
environment on productivity, etc. In addition, computer
software should be developed to simulate the entire
process of the enhanced resource leveling.
1) Ahuja, H. N. (1976) Construction Performance Control by
Networks, Wiley, New York
2) Boctor, F. F. (1990) “Some Efficient Multi-heuristic Procedures
for Resource-constrained Project Scheduling”, European Journal
of Operational Research, 49, pp.3–13
3) David, E. W. and Patterson, J. H. (1975) “A Comparison of
Heuristic and Optimum Solutions in Resource-constrained Project
Scheduling”, Management Science, 21(8), pp.944–955
4) Harri s, R. B. (1978) Precedence and Arrow Net worki ng
Techniques for Construction, John Wiley & Sons, New York.
5) Hegazy, T. (1999) “Optimization of Resource Allocation and
Leveling Using Genetic Algorithms”, Journal of Construction
Engineering and Management, ASCE, 125(3), pp.167–175
6) Kelley, J. E. (1963) “The Critical-Path Method: Resource Planning
and Scheduling”, Industrial Scheduling, Prentice Hall, Englewood
Cliffs, N. J., pp.347–365
7) Martinez, J. and Loannou, P. (1992) CPMLEVEL, Dep. of Civil
Eng., The Univ. of Michigan, Ann Arbor, MI
8) Satty, T. L. (1980) The Analytic Hierarchy Process: Planning
Setting Priorities, Resource Allocation, McGraw-Hill
9) Wiest, J. D. (1967) “A Heuristic Model for Scheduling Large
Projects with Limited Resource”, Management Science, 13(6),
Fig.7. Comparison between Output of RCS and Enhanced Resource Leveling
(a) Changes in Mx Values
(c) Changes in Mv Values
(e) Changes in Combined Index Values
(b) Changes in My Values
(d) Changes in RR Values

You're Reading a Free Preview

/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->