Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Introduction

The effects of the For decades, with more appreciation of


variability of project financial management, there was a surge in
planning on cost interest in cash flow forecasting in both
commitment curves: contractor and client organisations. Research
funding was more forthcoming from
a case study construction clients and hence more effort
was dedicated to the study of cash flow
A.P. Kaka forecasting for clients. This was demonstrated
J. Lewis and by the development of a series of typical value
H. Petros S-curve models. Examples of such
endeavours include studies based on data
The authors collected from health authorities (Drake,
A.P. Kaka is based at Department of Building Engineering 1978; Berny and Howes, 1983; Tucker,
and Surveying, Heriot Watt University, Edinburgh, UK. 1988), commercial, industrial and residential
J. Lewis and H. Petros are both based at the School of projects (Oliver, 1984; Singh and Woon,
Architecture and Building Engineering, University of
1984; Kenley and Wilson, 1986;
Liverpool, Liverpool, UK
Khosrowshahi, 1991), chemical projects
(Miskawi, 1989), educational, offices and
Keywords
apartment projects (Christian and Kallouris,
Planning, Scheduling, Estimating, Construction management, 1990). Most of the models produced by these
Cost control, Regression analysis
studies were obtained by fitting selected
functions to the available data. Unfortunately,
Abstract most these models did not achieve a
Standard value/cost flow models (often referred to as reasonable degree of confidence with respect
S-curves) are widely used in cash flow forecasting, to the accuracy of fit of the S-curves.
particularly at the tender stage. A substantial amount of Kaka and Price (1993) argued that the
research has concentrated on improving the accuracy of
these curves. Categorizing construction projects into
reasons behind the failure of previous models
groups and subgroups has helped, but the fact remains to produce accurate standard value curves
that different construction projects possess different could be attributed to the following factors:
profiles of cost flow. This paper is an attempt at assessing . Construction projects are unique and the
the extent of influence of planning and programming the progress of work varies from one project
work on the cost flow curves. One real project was used
to another.
as a case study and four planners were independently
asked to produce programmes for executing the project.
. In previous studies the choice of project
These programmes were analysed and converted to cost groupings was poor.
flow curves using one database of productivity and unit . Unbalancing (front-end loading) and
cost rates. Results confirmed that the variations in overmeasure distort the shape of the value
programmes produced less variations in cost flow curves curve. This can be shown by comparing
than the errors to be expected from the use of average
curves derived from project groups (mean SDY 2.88 the bills of quantities of several tenders
compared to previous studies of 5.5, 8.5 and 10.67). The for the same contract.
results suggest, taking due account of the limitations of . Errors in estimating affect the shape of
the scope of the study, that further effort at the the value curve, and lead contractors to
categorisation of projects into subgroups will result in a submit different tenders for the same
reasonable improvement in accuracy.
project.

Electronic access Based on the above factors, Kaka and Price


concluded that, due to differences in
The Emerald Research Register for this journal is
available at estimating and unbalancing, for the same
http://www.emeraldinsight.com/researchregister project and the same schedule of work, two
contractors are likely to produce two different
The current issue and full text archive of this journal is
available at value curves. These two factors have no effect
http://www.emeraldinsight.com/0969-9988.htm on the actual commitment curve of a project.
Therefore, it was proposed that the
Engineering, Construction and Architectural Management
development of standard cost commitment
Volume 10 . Number 1 . 2003 . pp. 15-26
# MCB UP Limited . ISSN 0969-9988 S-curves for different groups of contracts may
DOI 10.1108/09699980310466514 yield more accurate results. In addition, more
15
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

effective classification and grouping of the comparison. The methodology is explained


projects were proposed and tested. Cost data below in more detail.
from more than 70 industrial and commercial
projects were collected. Projects were then Selecting a completed construction
analysed and classified according to certain project as a case study
criteria. Tests were conducted to identify The project was a three-story extension of an
which of these criteria affected the shape of existing building, built in 1990. The
the cost commitment curves. Results showed extension had three floors, each with a clear
that the size of the project and type of height of three metres. The structure of the
procurement were statistically influential and building used composite steel frame and
thus used to group the projects and build the in-situ concrete flat slabs with 450m2 total
standard S-curves. The accuracy of these gross floor area. It was designed to house
curves was assessed and compared with facilities for chemical research. The rationale
previous standard value curves, and although for choosing such a case study was the
found to be superior, they were still short of availability of information needed in terms of
being sufficiently accurate for the the bill of quantities and the detailed
performance control of individual projects. construction drawings. There were no prices
included with the various items in the Bill of
Quantities, though this did not cause
problems since the research was concerned
Research objective and methodology with cost flows as opposed to value flows.
The study by Kaka and Price (1993)
Preparation of the case study
highlighted important issues about the factors
In order to present the case study to the
causing inaccuracy in standard value (or
interviewees (planners) in an efficient manner,
indeed, cost) curves. One additional factor
a prototype programme was developed using
that had not been explored was the
Microsoft project. The prototype programme
contractor's own approach to the way in
was only for presentation purposes and
which construction activities are scheduled
planners were asked not to be influenced by the
and the extent of resources deployed by them.
way in which activities were scheduled. To ask
This research work was conducted to
planners to produce a complete and detailed
investigate whether different contractors plan
project plan from scratch would have been very
their work differently, and if so, to evaluate
difficult and time consuming. Results showed
the effect that difference could have on cost
considerable variation between the project
commitment curves. A significant effect plan, which confirm that the prototype
would lead to the conclusion that standard programme did not influence planners.
S-curves do not exist, and any accurate Initially, the strategy was to include as many
analysis should be based on detailed estimates activities as possible in the programme. This
and specific programmes, rather than would enable the research to assess the effect
standard S-curves. A non-significant effect of different programmes even at a detailed
would lead to the conclusion that more level. In the initial process of cost and time
research (possibly through further calculation, the project was split into over 500
classification of projects) is worthwhile. activities and sub-activities. Although a cost
In order to achieve this, an actual and duration could be calculated for each
completed project was selected and used as a activity, it proved difficult to manipulate 500
case study. Four planners were asked to activities using Microsoft project as the
produce a programme for the work and also planning system. Therefore, it was necessary
to specify the type and level of resources to be to reduce the number of activities to a number
used in each activity. These individual that was both manageable and realistic for a
programmes were then analysed and priced contractor's programme. A pilot study was
based on direct cost only. Indirect cost and carried out with the assistance of an expert
preliminaries were ignored in the analysis as planner, inter alia in this process, the number
they are separately priced, measured and paid of activities was reduced to just under 100.
for. As a result, a cost commitment S-curve This was principally achieved by combining
was produced for each of the four activities, though a few very short duration
programmes followed by analysis and and low cost items were removed.
16
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

The following step was to assign resources needed estimates of the cost of these activities,
to the activities. In practice, the level and which would have been difficult to assign by
application of resources to be assigned is the planners, hence the use of the price books.
significantly influenced by the specified
overall duration of the project. It is also Production of the four programmes
influenced by: the nature and complexity of Using the resource and precedence
the project; labour and plant availability; and information provided by the planners, the
experience of workers, etc. cost and duration of each activity was
For the purpose of this study, it was recalculated and four new programmes were
thought neither feasible, nor desirable, to try produced. From each programme, total
to control the level of resources the planners duration, total cost and monthly cost flow
were to use when programming the work. were predicted.
Therefore, the project's total duration was not The results were sent to the interviewees for
fixed prior to presentation of the case study to validation and revision if necessary. This was
the interviewees. done using a Delphi type of approach, which
For a cost flow calculation, the cost and consisted of two stages: first, with the
duration of each activity in the programme interviewee examining only his/her
needs to be estimated. In order to do this a submission, and second, with each looking at
uniform set of productivity and cost rates was all the submissions. These reviews resulted in
needed. These were obtained from a ``standard'' only very minor changes in each case.
builders' price book for major works. Appendix 1 (Table AI) provides the final
planning data supplied by the four planners.
The interviews Since the size of the labour force and plant
Having produced an initial Gantt chart of the used differed from one planner to another, the
works, the next step was to obtain individual total cost of the works also varied slightly
planner's programmes. Four planners were between planners. The duration, in contrast,
identified from two collaborating varied significantly between planners. The
organisations: two were planning managers, results are illustrated in Table I.
each with over ten years experience in the Although it was considered important to
field, one was a project manager with over 15 evaluate these differences and the reasons
years experience, and the fourth a structural behind them, the primary aim of this research
engineer with considerable site experience. was to compare the shape of the cost flow
The interviews started with a brief curves in their normalised forms (i.e.
presentation of the research, outlining aims percentage of total cost plotted against
and methodology. This was followed by a percentage of total duration). In making such
detailed description of the case study a comparison, neither the small absolute value
including the drawings, characteristics of the differences in costs, nor the larger differences
site and the method to be used in assessing in duration, were considered significant.
productivity and estimating cost. The As far as the durations are concerned, one
prototype programme was then shown to the result showed a large variance from the rest of
interviewees and this acted as an interface for the set. It is interesting to note the extent to
eliciting information and producing each of which planners can influence a project's
the four programmes. duration. It also indicates, given that this
Each of the planners was asked to provide result was generated by one of the two
three types of information: the level and type contractors' planning managers involved, the
of resources to be assigned to each of the potential for variability between a contractor's
activities, the precedence information defining performance and a client's demands. It
the relationships between activities and,
finally, a description and justification of the Table I Total cost and duration
main the decisions taken. The duration of Estimated total Estimated
each activity was calculated by the research Planner cost (£) duration (weeks)
using productivity rates published in Laxton.
This was to minimise the effort needed by the 1 293,990.00 39
collaborators and hence ensure reasonable 2 294,890.00 34
and similar levels of accuracy given these time 3 295,580.00 70
4 294,890.00 30
restraints. In addition, the exercise also
17
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

should be noted that in practice this the use of polynomial regression analysis for
variability is normally limited by the curve fitting is ``logit transformation''. This
specification of a contract period (or a was initially proposed and used by Kenley and
completion date) by the client. Wilson (1986) and later adopted by Kaka and
Price (1993). The fitted curve that results
from this transformation can be uniquely
The S-curves described by two shape parameters ± alpha
and beta ( and ). It was decided to apply
Having generated each of the programmes for the same methodology in this research for the
the building, the monthly cost commitment following reasons: the accuracy of the
data were determined for each and four individual fits produced in the past value
different cumulative cost flow curves were models demonstrated its efficiency; it is
generated (Appendix 2 (Table AII) lists the reliable, easy to understand and flexible;
actual cumulative monthly values for each accuracy of fits in previous work was
project). The curves are illustrated in Figure 1. available for comparison. Therefore, in
Although each of the four cost commitment order to assess the effect of the four
curves has an approximately S-shaped profile, different programmes on the shape of the
they are by no means close to each other. In S-curve profiles, the total durations and costs
fact, a quick visual inspection of the curves of the projects were normalised. In order to
suggests that there are significant differences, do this, all cost and time values were
not only between the absolute curves, but also converted to percentages of their
between the shapes themselves. However, in corresponding totals. The following step was
order to conduct a meaningful examination the to fit curves to these profiles using the logit
results have to be looked at from modelling and transformation. The process of fitting these
forecasting perspectives. A standard S-curve is curves is as follows:
usually expressed in terms of percentages of (1) The linear equation is found by a logit
total cost and duration and is only converted to transformation of both the independent
absolute values when applied to a particular and the dependent variables (in this case,
project. Also, the shape of a standard S-curve is time and cost respectively):
usually expressed in terms of a mathematical Z
model, that is they are ``fitted'' to a Logit ˆ Ln …1†
1 Z
mathematical expression whose parameters
can be adjusted to suit particular projects. where Z is the variable to be transformed
Most of the previous S-curve models have and Logit is the transformation.
been developed using polynomial regression (2) Once the two variables are transformed, a
for the curve fitting and for representing the linear equation can be fitted into the
standard curves themselves. An alternative to transformed data and thus, and can

Figure 1 Cumulative cost commitment data

18
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

be found. This linear equation is . Curve 3 is very linear, showing a constant


expressed as follows: rate of cost build-up. Compared to the
Y ˆ ‡ X other curves, curve 3 shows consistently
higher rates of progress in the early stages
where: of the work.
c t . Curve 2 is concave in shape and shows the
Y ˆ Ln and X ˆ Ln …2†
100 e 100 t slowest rate of progress, of all the curves,
during the middle phase of the work.
(3) The S-curve is then generated using the
following equation:
 t  Test 1. SDY values between the
100  F
cˆ where F ˆ e …3† individual fitted S-curves
1‡F 100 t
The first test was conducted to determine the
variability between the shapes of the four
Please refer to Kenley and Wilson (1986) for
fitted cost curves. SDY values were calculated
more details.
for pairs of curves, and these values,
presented as percentages of cost, are
Analysis of results illustrated in Table II.

A common way of analysing the variability


Test 2. SDY values for individual plotted
between curves is through the use of the
curves and the original curves
standard deviation of the Y estimate (SDY
Any fitting process will result in some degree
measure). SDY is the standard deviation of the
of error between the fitted curve, represented
difference in Y-axis values (i.e. in this case,
by a mathematical expression, and the
percentage of total cost) between two curves,
original curve from which it was derived, i.e.
at corresponding points (i.e. points of equal
there is rarely a perfect fit. Therefore, a test
percentage of total time). This measure has
was conducted to measure SDY values
successfully been applied to measure accuracy
between individual fits and their original
of fit of S-curves models as well as variability
curves, derived directly from the programmes.
between S-curves (Kenley and Wilson, 1986;
The aim of the test was to determine whether
Kaka and Price, 1993). It was decided to adopt
the error in fitting curves was higher or lower
SDY measure in this study also.
than the variability between individual
Four different tests were performed during
planner's curves. The values of and were
the analysis stage:
also calculated for individual fitted curves.
(1) SDY values between the individual fitted
SDY values are presented in Table III.
S-curves.
The four individual fitted curves were
(2) SDY values for individual fitted S-curves
plotted together with the originals using the
and the original S-curves.
same time/cost scale. The curves are
(3) SDY values for the average fitted S-curve
illustrated in Figure 3.
and the individual fitted S-curves.
The and values are listed in Table IV.
(4) SDY values for the average fitted S-curve
and the original S-curves
Test 3. SDY values for the average fitted
Figure 2 shows the group of fitted S-curves curve and the individual fitted S-curves
after the logit transformation has been The objective of this test was to evaluate the
applied. Visual examination of the curves range of variability of SDY values between the
yields the following: average fitted S-curve and individual fitted
. Curves 1 and 4 are very similar, denoting curves. Figure 4 represents the average fitted
that the plans produced by planner 1 and S-curve determined from the four cost curves.
planner 4 had the same effect on the cost SDY results are calculated and illustrated in
build-up of the project when using a Table V. The values of and for the
unified time scale. They have the classical average fitted curve are also calculated and
``S'' shape, with slow acceleration at the presented in Table VI.
beginning, a steady but faster rate in the
middle period, and gradual tailing off Test 4. SDY values for the average fitted
towards the end. Compared to the other curve and the original curves
curves, they show higher rates of progress The objective of this last test was to evaluate
in the second half of the project. the errors incurred between the average fitted
19
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

Figure 2 Group of fitted cost curves

Table II SDY values measuring the difference between between 1.413 and 8.664 with an average
the fitted S-curves SDY value of 5.24 per cent. We can conclude
from this that fitting errors are small for this
Fitted 1 Fitted 2 Fitted 3 Fitted 4
particular case, and that for all practical
Fitted 1 0.000 ± ± ± purposes, the actual and fitted curves are
Fitted 2 5.528 0.000 ± ± synonymous. Moreover, since no single curve
Fitted 3 3.849 5.181 0.000 ± suffers from markedly different fitting error
Fitted 4 1.333 6.668 5.043 0.000 from the rest of the set, we may also conclude
that, in this case, test 1 provides a good
Table III SDY values measuring the difference between
measure of the variation between cost curves
the individual fitted and the original curves
resulting only from variations in planning.
SDY values The SDY values in test 1 range from 1.3 per
Curve (%) cent to 6.7 per cent, with an average of 4.6 per
Individual fitted 1 and original 1 4.308 cent. There are no previous data from an
Individual fitted 2 and original 2 7.618 analysis of a single project with which to
Individual fitted 3 and original 3 5.958 compare these results directly. However,
Individual fitted 4 and original 4 4.439 sources suggest that they are, as is to be
expected, at the very bottom end of the range
curve and original individual curves. SDY of SDY values obtained by comparing average
values are listed in Table VII. curves for different types of projects (e.g.
Kaka and Price (1993) quote values between
4.9 per cent and 38.8 per cent for
Discussion comparisons between projects). The
significance of the results of test 1 lies in the
Test 2 measures the error in fitting curves to maximum SDY value obtained, since we take
the actual cost/time data. It shows a mean this to be an approximate measure of the
SDY value of 5.58 per cent in all cases. This is likely range of the differences that may arise
higher than previous results. For example, from two different planners' programmes. It
Kaka and Price (1993) report fitting errors should be noted that this maximum value (6.7
causing SDY values in the range 0.475 per per cent) is within the error range to be
cent to 8.339 per cent, with an average of 3 expected when fitting S-curves into individual
per cent, while Kenley and Wilson (1986) projects cost flows.
obtained results ranging from 1.03 per cent to Test 3 measures the error in using an
7.33 per cent. In a more recent study, Evans average curve, derived from all four planners'
and Kaka (1998) obtained results ranging programmes, instead of the individual fitted
20
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

Figure 3 Individual S-curve fittings

Table IV and values for individual fits Table V Individual and average fitted errors (SDY)
Curve SDY
Individual fitted 1 ±0.1 1.268 Curve values (%)
Individual fitted 2 ±0.379 1.135 Individual fitted 1 and average fitted 1.671
Individual fitted 3 ±0.098 1.005 Individual fitted 2 and average fitted 4.059
Individual fitted 4 ±0.062 1.358 Individual fitted 3 and average fitted 2.789
Individual fitted 4 and average fitted 2.988
Figure 4 The average fitted curve

Table VI and values for the average fitted S-curve


Curve
Average fitted ±0.162 1.177

Table VII SDY values between original S-curve and


average fitted S-curve
SDY values
Curve (%)
Original 1 and average fitted 4.862
Original 2 and average fitted 6.675
Original 3 and average fitted 5.527
Original 4 and average fitted 6.410

curves. The SDY values range from 1.7 per Price (1993) report SDY values in the range
cent to 4.1 per cent with a mean of 2.9 per 0.232 per cent to 20.691 per cent (average
cent. Again, there is no directly comparable 5.5) for the differences between average and
data from previous research, but Kaka and individual curves for specific project types. In
21
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

similar previous studies, Kenley and Wilson construction change. Nevertheless, the
(1986) obtained 1.43 to 1.43 (average 10.67) programmes produced by four planners
while Evans and Kaka (1998) obtained 2.123 showed substantial differences and the
to 21.87 (average 8.505). These again breadth of this difference could only have
confirm the relatively small impact the risen with the inclusion of further planners.
planners had on the shape of the S-curve. On the other hand, the case study deliberately
Test 4 measures the error in using the had no constrains with respect to project
average curve in place of curves derived duration and resources to be allocated, the
directly from the specific programmes. The fact that offered the planners opportunities for
SDY values range from 4.9 per cent to 6.7 per alternative sequencing durations of
cent, with an average of 5.9% per cent. Both operations. In real life these constraints are
tests provide additional measures of the effects more visible and thus may yield less
of variation between planners, which compares programming differences and hence even less
quite well with the error to be obtained as a impact on the profile of the project cost flow.
result of using standard/average S-curves.

References
Conclusions
Berny, J. and Howes, R. (1983), ``Project management and
A number of primary conclusions may be control using real time budgeting and forecasting
drawn from the above analysis and the work models'', CIOB, Vol. 2, pp. 19-40.
Christian, J. and Kallouris, G. (1990), ``Predictive cost-time
as a whole. First, different planners'
models for construction activities'', Building
programmes may vary significantly, however, Economics & Construction Management, Vol. 4,
their impact on the shape of the cost flow is CIB 90.
not considerable. Second, variations between Drake, B.E. (1978), A Mathematical Model for Expenditure
the project duration produced by different Forecasting Post Contract, Technician Israel Institute
planners may also vary greatly, when of Technology, Haifa.
Evans, R. and Kaka, A.P. (1998), ``Analysis of the accuracy
unconstrained by specific client's
of standard/average value curves using food retail
requirements. Third, and perhaps most building projects as case studies'', Engineering,
significantly, the difference in cost curves Construction and Architectural Management, Vol. 5
arising from differences in planning are less No. 1, pp. 58-67.
than the errors arising from the use of average Kaka, A.P. and Price, A.D.F. (1993), ``Modelling standard
curves to represent groups of projects. The cost commitment curves for contractors' cash flow
latter is an important result since it suggests forecasting'', Journal of Construction Management
and Economics, Vol. 11, pp. 271-83.
that attempts to further refine project Kenley, R. and Wilson, W. (1986), ``A construction project
classifications, prior to computing average cash flow model ± an idiographic approach'',
curves, will be rewarded by significant Journal of Construction Management and
improvements in accuracy. That is, cost Economics, Vol. 4, pp. 213-32.
curves for projects are different because of Khosrowshahi, F. (1991), ``Simulation of expenditure
differences in the projects' characteristics and patterns of construction projects'', Journal of
Construction Management and Economics, Vol. 9,
not because different people undertake them!
pp. 113-32.
The conclusions, of course, must be treated Miskawi, Z. (1989), ``An S-curve equation for project
with some caution. The results are drawn control'', Journal of Construction Management and
from a single case study of a relatively small Economics, Vol. 7, pp. 115-24.
project, with a few planners. More case Oliver, J. (1984), ``Modelling cash flow projections using a
studies would be needed to substantiate these standard micro-computer spreadsheet program'',
MSc Project in Construction Management,
conclusions. Project size is typically one of the
Loughborough University of Technology,
factors by which projects have been classified Loughborough.
for the purposes of computing average cost Singh, S. and Woon, P. (1984), ``Cash flow trends for high
curves (i.e. it is an important variable in rise building projects'', Proceedings of the 4th
determining curve shape). This fact alone International Symposium on Organisation and
suggests that further work needs to be done Management of Construction, University of
Waterloo, Ontario.
by extending the sample size, before definitive
Tucker, S.N. (1988), ``A single alternative formula for
conclusions can be drawn. It is also noted that Department of Health and Social Security S-curves'',
cost curve shape may also be time dependent, Journal of Construction Management and
if the relative costs of different elements of Economics, Vol. 6, pp. 13-23.
22
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

Appendix 1
Table AI The four plans for the same project
Planner 1 Planner 2 Planner 3 Planner 4
ID Name Duration Predecessors Duration Predecessors Duration Predecessors Duration Predecessors
1 Breaking up 1.16d 1.16d 1.6d 1.6d
2 Excavation and earlthwork support 3.99d 1 5.32d 1 3.99d 1 5.32d 1
3 Steelwork 7.07d 35 8.84d 35 17.67d 35 8.84d 35
4 Filling and treatment 3.68d 50 3.68d 50,20 3.68d 50,66 3.68d 50,66
5 FORMWORK Od Od Od Od
6 Foundation, trenches and basement 0.94d 2 13.25d 2 9.94d 2 13.25d 2
7 Columns (foundation ± first floor) 6d 21 6d 3 12d 3,21 6d 3
8 Staircase (ground floor) 1.21d 41 1.61d 37 2.42d 48,23 1.61d 23
9 Beams (first floor) 5.99d 25 7.18d 25 8.98d 36,25 7.18d 3
10 Slab/holodeck (first floor) 3.47d 39 4.16d 39 5.2d 39 4.16d 9
11 Staircase (first floor) 1.21d 45,38 1.61d 41 2.42d 38,27 1.61d 27
12 Columns (first floor ± second floor) 5.13d 41,24 5.13d 24 20.26d 41,24 5.13d 3
13 Beams (second floor) 8.76d 29 7.01d 29 8.76d 40,29 7.01d 3
14 Slab/holodeck (second floor) 5.2d 43 4.16d 43 5.2d 43 4.16d 13
15 Staircase (second floor) 1.21d 48,42 1.61d 45 2.42d 42,31 1.61d 31
16 Columns (second floor ± roof) 5.13d 45,40,28 5.13d 28 10.26d 45,28 5.13d 3
17 Beams (roof) 8.5d 32 6.8d 32 8.5d 44,32 6.8d 3
18 Slab/holodeck (roof) 5.2d 47 4.16d 47 5.2d 47 4.16d 17
19 REINFORCEMENT Od Od Od Od
20 Ground beams 0.57h 2 0.75h 2 1.13h 2 0.75h 2
21 Columns (foundation ± first floor) 0.5h 3 0.67d 3 1d 3 0.67d 3
22 Slab (ground floor) 3h 4 3h 4 6h 4 3h 4
23 Staircase (ground floor) 1.7h 8 1.7h 37 3.4h 48 1.7h 37
24 Columns (first floor ± second floor) 0.42d 25 0.56d 41 0.84d 41 0.56d 3
25 Beams (first floor) 0.94d 22 0.94d 21 1.88d 36 0.94d 3
26 Slab (first floor) 5d 10 5d 10,22 10d 39 5d 10,25
27 Staircase (first floor) 1.7h 11 2.27h 41,23 3.4h 38 2.27h 41
28 Columns (second floor ± roof) 0.42d 29 0.56d 45 0.84d 45 0.56d 3
29 Beams (second floor) 1d 26 1d 24 2d 40 1d 3,28
30 Slab (second floor) 5d 14 5d 14,26 10d 43 5d 14,29
31 Staircase (second floor) 1.7h 15 2.27h 45,27 3.4h 42 2.27h 45
32 Beams (roof) 0.93d 30 0.93d 28 1.86d 44 0.93d 3
33 Slab (roof) 5d 18 5d 18,31,30 10d 47 5d 18,32,30
34 CONCRETE CASTING Od Od Od Od
35 Foundation, trenches and basement 5.54d 6,20 5.54d 6 5.54d 6,20 5.54d 6,20
36 Columns (foundation ± first floor) 0.93d 7 0.93d 7 1.24d 7 0.93d 9,21,7
37 Slab (ground floor) 2.43d 22 3.64d 50 2.43d 4,22 3.64d 9,36,22
38 Staircase (ground floor) 0.42d 23 0.56d 8 0.42d 8,37 0.56d 8
39 Beams (first floor) 1.35d 9 1.08d 9 1.35d 9 1.08d 45
40 Columns (first floor ± second floor) 0.75d 12 0.75d 12 1d 12 0.75d 39,24,12
41 Slab (first floor) 2.2d 26 1.65d 10,38,36 2.2d 26,10 1.65d 45,40
42 Staircase (first floor) 0.42d 27 0.56d 11 0.42d 11 0.56d 11,38,26
43 Beams (second floor) 1.33d 13 1.06d 13,40 1.33d 13 1.06d 48
44 Columns (second floor ± roof) 0.75d 16 0.75d 16 1d 16 0.75d 43,16
45 Slab (second floor) 2.2d 30 1.65d 14,42 2.2d 30,14 1.65d 48,44
46 Staircase (second floor) 0.42d 31 0.56d 15 0.42d 15 0.56d 15,42
47 Beams (roof) 1.25d 17,44 1d 17,44 1.25d 17 1d 37
48 Slab (roof) 2.2d 33 1.65d 18,46,33 2.2d 33,18 1.65d 37,47,33
(continued)

23
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

Table AI
Planner 1 Planner 2 Planner 3 Planner 4
ID Name Duration Predecessors Duration Predecessors Duration Predecessors Duration Predecessors
49 MASONRY Od Od Od Od
50 Substructure 6.46d 36 6.46d 7 5.36d 7 6.46d
51 External (first floor) 7.1d 39,37 5.33d 37 5.33d 54 5.33d 37,46
52 External (second floor) 7.1d 43,51 5.33d 41,51 5.33d 55 5.33d 41,51
53 External (roof) 7.1d 47,52 5.33d 45,52 5.33d 56 5.33d 45,52
54 Internal (first floor) 7.54d 41,66 5.66d 37 5.66d 48,46 5.66d 37
55 Internal (second floor) 7.54d 45 5.66d 41 5.66d 51 5.66d 41,54
56 Internal (roof) 7.54d 48,53,46 5.66d 45 5.66d 52 5.66d 45,55
57 WINDOWS AND DOORS Od Od Od Od
58 Ground floor 2.12d 67 2.12d 56 4.24d 67 2.12d 56
59 First floor 1.56d 72 1.55d 58 3.12d 72 1.55d 58
60 Second floor 1.56d 77 1.55d 59 3.12d 77 1.55d 59
61 LIFT INSULLATION Od Od Od Od
62 First phase installation 6d 52,53 6d 82,53 6d 82,53 6d 52,53
63 Second phase installation 4d 65,81 4d 62,81 4d 62,81 4d 52,81
64 SURFACE FINISHES Od Od Od Od
65 Channels 6.56d 82,58 6.56d 69 9.84d 82,58 6.56d 69
66 Waterproofing (basement) 3.24d 35 1.62d 35 3.24d 35 1.62d 35
67 Screed (ground floor) 1.2d 54 1.2d 54,66 1.2d 54 1.2d 37
68 Plaster/walls and staircase (ground
floor) 14.88d 69,83 14.88d 67 14.88d 69 14.88 67
69 Resing coating/floors (ground floor) 6.51d 65 6.51d 68 16.27d 65 6.51d 68
70 Ceramic tiles (ground floor) 13.75d 68 8.25d 69 20.63d 68 8.25d 69
71 Painting (ground floor) 14.02d 58 9.35d 70 14.02d 58 9.35d 70
72 Screed (first floor) 1.2d 55 1.2d 71 1.2d 55,71 1.2d 41,71
73 Plaster/walls and staircase (first floor) 14.88d 74 14.88d 72 14.88d 74 14.88d 72
74 Resin coating/floors (first floor) 6.51d 59 6.51d 73 16.27d 59 6.51d 73
75 Ceramic tiles (first floor) 3.56d 73 2.14d 74 5.35d 73 2.14d 74
76 Painting (first floor) 6.65d 59,71 4.43d 75 6.65d 59 4.43d 75
77 Screed (second floor) 1.2d 56 1.2d 76 1.2d 56,76 1.2d 45,76
78 Plaster/walls and staircase (second
floor) 14.88d 79 14.88d 77 14.88d 79 14.88d 77
79 Resin coating/floors (second floor) 6.51d 60 6.51d 78 16.27d 60 6.51d 78
80 Ceramic tiles (second floor) 3.56d 78 2.14d 79 5.35d 78 2.14d 79
81 Painting (second floor) 6.65d 60,76 4.43d 80 6.65d 60 4.43d 80,60
82 Roof finishes 8.75d 48 6.56d 48,90 13.13d 48 6.56d 48
83 General fixtures/showers (ground
floor) 3.68d 65 1.84d 69 3.68d 65 1.84d 69
84 Disposal system (below ground) 7.45d 35 7.45d 37 7.45d 35,83 7.45d 37
85 Building fabric sundries (ground
floor) 7.66d 54 7.66d 54,83 7.66d 54 7.66d 37
86 Disposal system (ground floor) 1.43d 54,84 2.86d 69,84,65 2.86d 54,84,85 2.86d 69,85,84,65
87 Building fabric sundries (first floor) 7.66d 55,85 7.66d 55 7.66d 55 7.66d 41
88 Disposal system (first floor) 2.86d 55,86 2.86d 74,86 2.86d 55,86,87 2.86d 74,87
89 Building fabric sundries (second floor) 7.66d 56,87 7.66d 56,60 7.66d 56 7.66d 45
90 Disposal system (second floor) 2.86d 56,88 2.86d 79,88 2.86d 56,88,89 2.86d 79,89
91 Piped supply (ground floor) 0.26d 54 0.26d 54,85 0.26d 54,90 0.26d 86,93
92 Piped supply (first floor) 0.26d 55,91 0.26d 55,91,87 0.26d 55 0.26d 88.91
93 Piped supply (second floor) 0.26d 56,92,90,89 0.26d 56,92,89 0.26d 56,92 0.26d 90,92
94 Paving/planting/fencing (outside) 1.16d 93,81,60 0.58d 93,63 1.16d 93,63 0.58d 93,63

24
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

Appendix 2

Table AII Actual cumulative monthly values for the four projects
Plan 1 Plan 2 Plan 3 Plan 4 Month
0 0 0 0 0
2,166 1,905 2,166 5,355 1
6,157 4,646 6,157 9,102 2
10,162 7,649 10,162 12,106 3
14,337 10,758 14,378 15,216 4
30,924 15,715 20,708 20,174 5
41,334 27,348 26,908 32,505 6
42,726 37,198 33,109 44,272 7
46,167 39,646 37,392 50,797 8
48,025 46,171 37,922 59,355 9
52,894 53,925 38,454 77,306 10
56,873 77,902 42,207 96,684 11
67,243 98,471 45,243 110,870 12
74,469 108,266 50,657 122,251 13
80,685 121,933 56,958 136,137 14
89,176 143,196 59,877 154,406 15
99,280 156,680 62,284 170,012 16
105,691 169,935 62,813 182,825 17
112,018 180,700 63,787 207,779 18
119,265 182,725 64,808 229,328 19
133,358 186,081 66,030 234,002 20
145,014 190,755 71,726 238,230 21
154,381 195,429 75,450 242,874 22
164,537 199,659 78,612 245,752 23
201,352 204,299 79,141 249,550 24
220,209 207,146 79,921 254,224 25
229,301 210,940 80,989 258,898 26
238,649 215,614 82,114 262,996 27
247,677 220,288 86,781 268,053 28
254,259 229,444 91,149 273,817 29
257,032 233,630 94,963 294,595 30
259,805 251,567 97,040 31
262,682 281,567 100,054 32
265,383 295,482 105,618 33
266,668 114,768 34
267,876 120,851 35
269,084 130,795 36
293,332 141,459 37
293,984 152,832 38
179,036 39
202,437 40
207,111 41
210,729 42
212,578 43
214,426 44
216,275 45
218,212 46
219,855 47
221,205 48
222,555 49
(continued)

25
Variability of project planning on cost commitment curves Engineering, Construction and Architectural Management
A.P. Kaka, J. Lewis and H. Petros Volume 10 . Number 1 . 2003 . 15-26

Table AII
Plan 1 Plan 2 Plan 3 Plan 4 Month
224,205 50
225,767 51
227,328 52
229,423 53
234,097 54
238,771 55
242,885 56
244,761 57
246,310 58
247,718 59
249,328 60
250,889 61
252,449 62
256,126 63
260,800 64
265,474 65
268,135 66
270,044 67
271,252 68
295,608 69
296,075 70

26

You might also like