Beruflich Dokumente
Kultur Dokumente
School of Architecture and Building, Deakin University, Geelong, Victoria 3217, Australia
Abstract Delays are an endemic feature of the construc- Thus, using system dynamics modelling, the effects of
tion industry. Typically, when a delay occurs in a project, prolonged overtime work on project cost and quality are
the project manager often expedites progress through examined. To overcome project delays, several options
activity-crashing with respect to available float and time- representing various combinations of prescribing over-
cost relationships. An accelerated schedule is thus ob- time work and injecting additional resources are analysed.
tained either by prescribing overtime working hours or by Utility theory is then applied to determine the most
procuring additional resources or a combination of both. appropriate solution for mitigating project delays. The
However, excessively prolonged overtime work can gen- modelling approach offered in this paper should be partic-
erate quality problems, such as rework, and additional ularly useful for large projects and for projects on
resources. With this in mind, there is a need for a model confined sites where potential cost savings and improved
to assist project managers with understanding the com- quality standards are likely to be the most significant.
plex nature of attaining a trade-off between overtime Keywords cost, quality, rework, systems dynamics,
working and the procurement of additional resources. time, utility theory
decision-making process and, in recent times, has be- tional resources increases the overall project cost as
come a popular technique for modelling change in compared with the overtime working mode, which
project management (Sterman 1992; Rodrigues & results in less rework being perceived. To examine the
Bowers 1996; Shen et al. in press). For a detailed combined effect of prescribed overtime work and addi-
review of the system dynamics methodology, refer to tional resources in activity-crashing, the authors have
Forrester (1961), Mohapatra et al. (1994) and Coyle developed a system dynamics model to simulate the
(1996). relationships among the key variables that are consid-
ered to be important for activity-crashing: additional
cost, cumulative work scope, quality and rework.
Reference mode
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7 | 3 , 211 -220
Effects of overtime work and additional resources on projects 213
additional resources in order to reduce the amount of In the overtime work and additional resources sector,
work remaining. However, sanctioning overtime work the parameter schedule pressure generates the need for
and/or injecting additional resources may cause de- prescribing overtime work and/or injecting additional
clines in quality, which can result in rework. resources. Thus, additional resources increase a
It is recognized that prescribing overtime work and/ project's cost, while prescribing overtime work affects
or assigning additional resources are activity-based (as the cost, schedule, additional recruitment of human
opposed to project-based) decisions. As such, the resources, personnel burnout, motivation and fatigue.
ordering of overtime work (or assigning additional The progress sector simulates desired progress based
resources) typically happens in a particular activity or on the scope of work, actual progress as a result of
some selected activities rather than in the entire pro- normal and overtime working hours and their differ-
ject. The model illustrated in Fig. 2 aims to simulate ence. The cost sector keeps track of the available
the interactions within a particular activity rather than budget from project revenues and the required budget
in the whole project. for resources. Additional cost is defined as the differ-
ence in the required and the available budget. Simi-
M O D E L SECTORS larly, the human resource sector simulates the inflow of
additional human resource requirements caused by
In the scope and rework sector, as shown in Appendix 1, prescribed overtime work and personnel burnout. The
schedule pressure is defined as a non-linear graphical changes in the levels of fatigue and motivation of the
function of schedule discrepancy and budget inade- personnel are modelled in the fatigue and motivation
quacy. Parameter schedule discrepancy is the differ- sector. A generic structure for each sector identified and
ence between desired and actual progresses at the job their respective functional relationships and parame-
site, while budget inadequacy is deemed to be the ratio ters can be seen in Appendices 1 and 2.
of the difference between available and required bud-
gets over the required budget. Work scope accumu-
lated as a result of schedule discrepancy and rework
MODEL V A L I D A T I O N
generated is drained through work outflow. The im-
pact of schedule pressure on quality is modelled in the Model validation is undertaken to ensure the sound-
quality sector. This sector simulates the decline of qual- ness and usefulness of the model (Forrester & Senge
ity when the pressure to expedite the project increases. 1980). Validation requires not only that the model
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7 | 3 , 211 -220
214 Li. H. et al.
complies with known 'physical laws' but also that its The total duration of a typical Harmony Type 1
results comply with the behaviour of the real world. project is 28 months. According to the schedule, the
Thus, the model was tested for both structural and planned progress at the end of the tenth month should
behavioural validity according to the guidelines de- be that 18% of the total work is completed. This
scribed in Forrester & Senge (1980). In addition, it has assumption is incorporated into the model and simula-
been examined for structural validation inasmuch as tion results are shown in Figs 3 and 4. Figure 3 shows
the major factors identified in Love et al. (1999b) and only the effect of overtime work on project cost, qual-
from the literature have been used and the values used ity and rework because no additional resources have
in the model have been derived from completed been prescribed. Similarly, Fig. 4 indicates only the
projects. The model is tested for behaviour prediction effect of additional resources on project cost, quality
so as to assist practitioners with particular scenarios and rework. Comparison of the simulation results of
that they may be faced with. Figs 3 and 4 with the respective reference modes in
To simulate the model for a base run, it was cali- Fig. 1 confirms the behavioural validation of the
brated to replicate the delay process according to data model. Specifically, in Fig. 3, a decline in quality is
collected from 14 projects carried out in Hong Kong. observed after proceeding with overtime work. In-
These projects are all Harmony Type 1, which is a creases in rework are also observed as quality declines.
standardized modular design adopted by the Hong This suggests that a decline in quality is a primary
Kong Housing Authority (1989-90). Of the 14 cause of rework. Similarly, simulation results in Fig. 4
projects, eight contained activity-crashing data as they indicate a considerable increase in cost, but a lesser
were used to generate the non-linear time-cost curves. decline in quality and less rework.
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7 | 3 , 211-220
Effects of overtime work and additional resources on projects 215
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7|3, 211-220
216 Li, H. et al.
develop the utility functions using either a straight- The utility functions are created by using either a
line or an exponential function and solve for the straight-line or an exponential relationship. The gener-
constants of each equation. alized equations of straight-line and exponential utility
functions are as follows:
The range of interest identifies the upper and lower
limits (yU,yL) for the options of each criterion and
Straight-line equation: uj(yj) = Ajyj + Bj (1)
formulates the boundaries for numeric inputs. The
threshold point (yT) of each criterion represents the
Exponential equation: uj(yj) = AjeBjyj + Cj (2)
point of neutral desirability. The most preferred point
(yM) represents the best possible option for the partic-
where uj(yj) = utility of criterion j and Aj, Bj and
ular criterion. The corporate policy and decision-mak-
Cj = constants for criterion j .
er's knowledge and experience of similar problems are
Based on the real world pattern, the authors have
usually used to define these points. Values for the
selected an exponential equation for developing utility
threshold points used in this paper are identified in
functions for each criterion. The constants of the
Table 3.
exponential equation are solved using two relative
Fixing the utility values with specific options for
points of the criteria for which the utility is known and
each criterion derives the scale for each utility func-
the hit and trial method to fulfil the need of a third
tion. These options are referred to as relative points
equation.
and a minimum of two is required, depending on the
Utility functions for each criterion are shown in Fig.
method used for developing the utility values. For two
4. The judgement of the decision-maker is required to
relative points, the yT and the yM are used. The utility
determine the most preferred number for each crite-
of the yT is set to 0 and the utility of the yM of each
rion (yM)j. The most preferred number represents the
criterion (j) is set to 1:
relative contribution of the criterion to the achieve-
u(yT)j = 0 and u(yM)j = 1 ment of project objectives. The numbers assigned are
Decision criteria
Rework generation
Alternative decisions Additional cost (% increase) Quality decline (% decline) (% scope)
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7|3, 211 -220
Effects of overtime work and additional resources on projects 217
Cost 250 17.47 0.1736 43.40 26.10 -0.4933 -123.32 25.00 -0.3954 -98.85
Quality 200 28.00 -1.6308 -326.16 25.33 -1.2130 -242.60 14.93 0.0064 1.27
Rework 175 5.01 0.5813 101.73 4.73 0.6086 106.50 1.93 0.8550 149.62
Expected utility -181.02 -259.42 52.05
Cost 250 28.00 -0.6725 -168.13 23.50 -0.2684 -67.11 22.56 -0.1926 -48.14
Quality 200 16.20 -0.1135 -22.70 14.40 0.0544 10.88 13.99 0.0907 18.15
Rework 175 2.43 0.8143 142.50 2.49 0.8093 141.63 2.51 0.807 141.34
Expected utility -48.33 85.40 111.35
Alternative 7 Alternative 8 Alternative 9
Cost 250 20.20 -0.0138 -3.45 20.28 -0.0196 -4.90 22.57 -0.1933 -48.34
Quality 200 13.77 0.1100 21.99 18.72 -0.3729 1 -74.58 14.01 0.0890 17.80
Rework 175 2.50 0.8085 141.48 3.40 0.7314 127.99 2.06 0.8445 14779
Expectant utility 160.03 48.51 117.25
scaling factors. The scaling factor for yj is designated also generate rework. This paper examines the com-
Wj,which is calculated as the multiplication of all of plex nature of attaining a trade-off between overtime
the most preferred numbers of other criteria (exclud- working and the procurement of additional re-
ing the most preferred number of criterion j). Using sources. The system dynamics model developed in
the scaling factor, each preliminary utility function this paper provides a useful way to capture the inter-
described in Fig. 4 can be transformed to an equiva- actions of process variables that exhibit the effect of
lent utility (Uj) value measured on a common scale. prolonged overtime work and additional resources on
The addition of all transformed utilities for all project cost and quality. From the system dynamics
criteria determines the expected utility value (Eu) of model, a number of options is obtained, which rep-
an alternative decision. An alternative with a higher resent different combinations of prescribing overtime
value of Eu represents a better decision. The En of work and injecting additional resources. Utility the-
all alternative decisions are presented in Table 4. ory is then applied to determine the most appropri-
According to the Eu, Alternative 7 is the best solu- ate solution for mitigating project delays.
tion among the alternatives for mitigating project de-
The method presented in this paper represents the
lays. Specificall, assigning 50% overtime work and
first attempt to analytically determine the effects of
30% additional resources is the best way to resolve
delays in the project. prescribing overtime work and injecting additional
resources on project cost and quality. We anticipate
that the method will be particularly useful in mitigat-
CONCLUSIONS ing delays of large projects and projects with
An accelerated schedule is obtained by either pre- confined sites. As in large projects, cost savings as a
scribing overtime working hours, by procuring addi- result of selecting the most appropriate combination
tional resources or a by combination of both. While of prescribing overtime work and injecting additional
excessively prolonged overtime work can generate resources can be significant. For projects with
quality problems, such as rework, and additional re- confined sites, the injection of additional resources
sources, prolonged overtime working may cause de- may result in overcrowded working conditions, thus
clines in productivity and performance, which may causing severe quality problems.
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7 | 3 , 211-220
218 Li, H. et al.
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7 | 3 , 211 -220
Effects of overtime work and additional resources on projects 219
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7 |3, 211 -220
220 Li, H. et al.
2000 Blackwell Science Ltd, Engineering, Construction and Architectural Management 7|3, 211 220