Anda di halaman 1dari 8

INTERNATIONAL JOURNAL OF

PROJECT
MANAGEMENT
International Journal of Project Management 23 (2005) 215–222
www.elsevier.com/locate/ijproman

A tool for managing projects: an analytic


parameterization of the S-curve
Denis F. Cioffi *

Project Management Program, The George Washington University, Monroe Hall 302, 2115 G Street NW, Washington, DC 20052

Received 27 January 2004; received in revised form 30 March 2004; accepted 5 August 2004

Abstract

The solution to a differential equation used frequently in ecology is found to reproduce the well-known S-curve seen in various
aspects of project management. The solution is modified in a minor way to fit project management boundary conditions. An excel-
lent fit of this theoretical curve to two samples of project cost data shows the utility of the formula. Numerical approximations valid
under typical project conditions are utilized to produce an analytic expression that can easily generate classic project management
evolution curves under a variety of conditions. The curves are normalized to two basic parameters: the total of the relevant quantity
(e.g., project costs) and the duration of the project. The user can choose the steepness of the climb and the point in time at which half
the total has been accumulated.
 2004 Elsevier Ltd and IPMA. All rights reserved.

Keywords: Managing projects; Progress; Cost; Cash flow management; S-curve; Earned Value

1. Introduction: population growth in nature and cost duration [2]. The curve was used to monitor progress in
growth in projects a series of related Argentine projects.
By comparison, the more flexible form presented here
When displayed as a function of time, accumulated should have uses in different contexts and in different
efforts or costs of a project usually take a form described types of projects. The first question that comes to mind
as the S-curve. The guide to the Project Management would be to discover whether project S-curves have
Body of Knowledge (the PMBOK) defines the term common characteristics within a single specific, narrow
only by its appearance: ‘‘Graphic display of cumulative industry (e.g., construction of houses) or type of indus-
costs, labor hours, percentage of work, or other quanti- try (construction as a whole), but differ as the type of
ties, plotted against time. The name derives from the S- industry changes (to information systems, say). The
like shape of curve (flatter at the beginning and end, existence of known common characteristics would
steeper in the middle)...’’ [1]. greatly facilitate future industry-specific simulation and
One earlier work in the project management litera- planning.
ture showed how to construct a curve numerically by As an example within a single project, an S-curve of
building it from a normal distribution and forcing it planned cash flow may assist risk analysis of project fi-
through the fixed point of 5% progress at 10% of project nances by showing altered spending rates needed to
achieve completion by different dates. Actual progress
measured against the planned S-curve can also influence
scheduling adjustments. Also, project resources, both
*
Tel.: +1 202 994 9533; fax: +1 202 994 2736. human and material, show similar evolution, and the
E-mail address: denis.cioffi@gwu.edu. curve can be applied to these data, too. Finally, earned

0263-7863/$30.00  2004 Elsevier Ltd and IPMA. All rights reserved.


doi:10.1016/j.ijproman.2004.08.001
216 D.F. Cioffi / International Journal of Project Management 23 (2005) 215–222

value calculations, where different S-curves are used to ing rate to decline. Because the tasks depend on each
represent actual costs, planned spending, and the budg- other, most often in a start–finish mode, they cannot
eted cost of work performed, hold particular potential all end simultaneously. Activities dwindle to a small
(see, for example [3]). With this new formalism, produc- number again as deliverables are completed, and eventu-
ing curves to illustrate different possibilities for discus- ally the entire project ends, with no further funds being
sion becomes a simple algebraic exercise, which should allocated. Thus, in projects the S-curve is driven by the
encourage practitioners to take this perspective of pro- multiple interconnected activities that occur in the mid-
jects more frequently. dle of a projectÕs life.
In what follows, I will first use the exact solution to Appendix A sketches the derivation of the logistics
the differential equation that generates an S-curve in differential equation (Eq. (A.1)) in its common ecologi-
population studies to fit project management data. cal context. There three quantities determine the exact
Then, approximations to the exact solution produce an progress of the solution (Eq. (A.3)) that models a popu-
expression that can be used more easily, and this lationÕs growth in a given system: the birth rate, the
approximate solution meets project management needs. death rate, and the so-called carrying capacity of the
Most important, with this approximation the progres- environment. Whether or not useful management ana-
sion of the total cost of any project can be characterized logs of these quantities exist (and they may), the initial
by two numbers: the slope of the rise in the S-curve and boundary condition of the ecology equation (below)
the time at which half the total funds have been spent. does not further the cause of well-planned projects,
Eager readers may bypass the details of the equationÕs and its solution should be modified.
generation by proceeding immediately to the answer, The growth of a living population, Y, from a low-lying
Eq. (7). initial state begins with some pre-existing, ground-state
population, so at time zero (t = 0) the initial boundary
condition of the ecology equation requires a non-zero
2. An equation for project management S-curves quantity, Y = Y0. But in project management, the quan-
tity Y represents cost or effort (task duration multiplied
Before proceeding with the mathematics, let us ask by resource intensity, e.g., [6]). A comprehensive plan
why an S-shaped curve arises, whether in nature or in allows a well-managed projectÕs costs to accrue only after
project costs. The common characteristics of all systems the project begins, and thus subtracting Y0 from the
that demonstrate this behavior include slow growth fol- ecological solution changes the initial condition to a
lowed by rapid growth, which in turn is followed by more appropriate one: Y = 0 at t = 0.
slow growth again to an asymptotic maximum – hence One can write the new solution in terms of the same
an S-shaped curve. Note that the Gompertz curve, parameters found in the ecological solution: the con-
which is used in mortality calculations [4], does not show stants d and Y0, and the maximum cost of the project,
the initial slow growth [5], and so it would not be useful Y1 (to which the solution asymptotes at infinite time).
for modeling this effect. But the ratio of Y1 to Y0, which is generally much
In natural settings S-curves arise in closed systems greater than one, turns out to be a useful parameter.
where the growth of a relatively small population is ini- Thus, we divide Y1 by Y0 and instead of using Y0
tially unlimited. At first the population increases slowly explicitly, define a constant c ” Y1/Y0; one can then
because the growth rate depends on the number of its write a generic solution for the project management
individual elements. As that number increases, the rate S-curve as
increases and the population grows more swiftly. At 1  expðdtÞ
some point, however, limits are imposed, e.g., the food Y ¼ Y1 : ð1Þ
1 þ c expðdtÞ
or energy supply is finite, or in the case of an epidemic,
the number of potential new victims declines (essentially At early times, i.e., at t  0, the growth is linear,
everyone has already been infected). The population Y  Y0dt. At late times, as t gets very large, Y  Y1,
then reaches a maximum. as desired. However, this equation will be more useful
Even the largest projects start with an initially small for planning if the summed costs equal the total project
number of tasks, but projects soon begin to tackle mul- cost at the end of the project (t = t1), and not just at infi-
tiple activities simultaneously. These parallel, often nite time. The proper normalization in the solution that
interconnected activities – a key characteristic of pro- follows (Eq. (7)) will ensure this equality, but for now we
jects that mandates project management – increase the need not worry about it; in most cases it will be approx-
spending greatly compared to the work at the beginning, imately correct as written (of order 10%, which will be
which by comparison is usually more limited and nar- shown later, in Table 2).
rowly focused (e.g., planning). But in almost all projects These behaviors at the temporal extremes of the curve
(with the exception of some defense work!), a cap on the replicate the verbal PMBOK description noted previ-
total cost limits the budget and forces this large spend- ously, but does the equation truly work with project
D.F. Cioffi / International Journal of Project Management 23 (2005) 215–222 217

Table 1
2500
This table shows the statistics of the sample data set taken from PS8
Statistics of sample data
Number of points 207 2000
Minimum 0

Cost (Dollars) Per Day


Maximum $2456
Sum $92,464
1500
Mean $447
Standard deviation $590

1000

Table 2
This table shows the calculated values of c and y1  1 for five 500
combinations of r0.67 and b1/2. Also shown are the values of the
corresponding b 0 , which are the times at which y = 0.5 exactly, and
parenthetically, the b1/2 parameters that one would need to use to
obtain the actual halfway points represented by the original numbers 0
used for b1/2
r0.67 c b1/2 b0 y1  1 (b1/2) 0 50 100 150 200
Time(Days)
0.5 5.389 0.500 0.454 0.119 0.5607
1 5.389 0.250 0.250 0.002 0.2505 Fig. 1. The data points show the daily costs, in dollars, for a 207
1 52.60 0.500 0.496 0.018 0.5045 calendar-day project. Many days (e.g., weekends) have zero cost. The
1 401.4 0.750 0.720 0.135 0.7889 solid curve is the (arithmetically calculated) derivative of the exact
2 2979. 0.500 0.500 0.0003 0.5000 solution shown in Fig. 2, and so its units are also dollars per day.

data? That is, does the equation fit well the accumulated 1 105
cost data of a project? Without overestimating the value
of a least-squares fit to summed numbers, we should first
ask if the standard (PearsonÕs) coefficient of determina- 8 104
tion, R2, approaches one. For this test, instead of using
Summed Costs (Dollars)

actual project data, which are often proprietary and dif-


ficult to understand or duplicate, I took a set of numbers 6 104
from an example used in Version 8 of ScitorÕs Project
Scheduler software, PS8 [7].
Fig. 1 shows daily cost data taken directly from the 4 104
intranet example in PS8. PS8 calculates the total cost
of each day for every day of the week; sometimes the
dayÕs cost is zero (e.g., weekends). Table 1 gives some 2 10
4

statistics of the raw data (i.e., not the accumulated


costs).
This example was chosen initially because of its avail- 0
ability, but the wide dispersion of the data and the
roughness of the resultant curve of the accumulated 0 50 100 150 200
costs make the test particularly appealing. The jagged Temporal Units (Days)
curve in Fig. 2 shows the accumulated cost data as a
Fig. 2. The exact solution, Y (Eq. (1), dashed line), fit to the sample
function of time (in the unit days, which is irrelevant
data of Fig. 1 (jagged solid line). The four fit parameters here are:
to this discussion). The smooth, dashed curve is a Y1 = 89,437; d = 0.03392; c = 16.88 and R2 = 0.99638.
least-squares fit 1 of the project management S-curve
equation, Eq. (1). An eyeball view of the graph alone
gives a powerful impression about the accuracy of the (1) represents well the S-curve seen frequently when
fit. The value of the coefficient of determination, managing projects.
R2 = 0.996, confirms this impression (a low number
would cause concern). In this one example, then, Eq.
3. A form more useful for project management

1
Via the general curve-fit function of Kaleidagraph software, The above form works well with project data, and
Version 3.52. one can determine numerically the three parameters
218 D.F. Cioffi / International Journal of Project Management 23 (2005) 215–222

necessary for a good fit (d, Y1, and c). However, what if ical approximation to this slope that permits an estimate
the detailed planning that makes a fit possible has not of d.
yet occurred? How can we take advantage of this solu- The classic S-curve is described as having three parts:
tion to create a first guess of what the spending profile a gentle rise, a steep slope, and a gradual path to the
might look like? asymptote. To start, we can assume that these three
Again three quantities will parameterize the solu- parts occur over equal intervals of time, each occupying
tion, but using the equation in the absence of detailed t1/3. We can further assume, temporarily, that the steep
data suggests a slightly different set of constants. First slope in the center portion comprises two-thirds of the
one uses the final cost of the project, now written rise from Y = 0 to Y = Y1. The remaining one-third of
explicitly as Y1. The other two parameters are the total the rise occurs throughout the two gently rising end por-
duration of the project, t1, and the time at which the tions. To give the individual project manager the free-
project has used half its total funds, t1=2 . With esti- dom to choose an approximation to this central slope,
mates of these three numbers, at the start of any pro- we introduce r as the factor that describes the magnitude
ject a manager can write an S-curve equation that can of the rise in the middle third, and define r0.67 through
indicate how labor or funds might be specified 2
throughout the project. r ¼ r0:67 : ð4Þ
3
First we calculate this new time, t1=2 . Letting Y = 0.5
Y1 yields When r0.67 = 1, two thirds of the rise occurs in the mid-
     dle interval; r0.67 > 1 steepens the slope.
1 Y1 Y1 With this definition, we find an expression for d in
t1=2 ¼ ln c þ 2  ln 2 1
d Y1 Y1 terms of r and t1: d = 8r0.67/t1. Using this form for d with
1 the previous equation for t1/2 (Eq. (2)) yields the follow-
’ lnðc þ 2Þ: ð2Þ
d ing expression for determining the c parameter:
In the second line of the above equation, we take advan- lnðc þ 2Þ ¼ 8r0:67 b1=2 : ð5Þ
tage of the near equality that occurs in many cases,
These expressions may be inserted back into the original
Y1  Y1. Retaining the distinction here between Y1
equation (Eq. 1) to find:
and Y1 would not justify the resulting algebraic com-
plexity, and the difference introduced can be corrected Y ðbÞ 1  expð8r0:67 bÞ
¼ :
easily with a simple numerical solution that converges Y1 1 þ c expð8r0:67 bÞ
quickly (see Appendix B).
As written, this equation would provide a good fit to
We want to express this time in terms of the total
cumulative cost or effort data under the assumption that
duration of the project, t1, and so now define b as the
Y1 = Y1. When planning, however, one does not want
ratio
to imply that the workersÕ total effort (and hence total
t project cost) will be expended only after an infinite time!
b :
t1 We can fix this problem by dividing both sides of the
We term the specific b where Y . 0.5Y1 as b1/2, subject equation by Y1, defined through
to the condition that 0 < b1/2 < 1. If the project manager 1  expð8r0:67 Þ
uses a so-called front-loaded distribution, where the Y1  Y1 ; ð6Þ
1 þ c expð8r0:67 Þ
funds are disbursed earlier rather than later, b1/2 will
sit closer to the beginning of the project. Other projects and writing the final form of the equation in terms of the
will have this time more towards the middle (b1/2  0.5) ratio y = Y/Y1:
or perhaps nearer the end of the scheduled duration 1  expð8r0:67 bÞ
(say, b1/2 J 0.7). An approximation to the slope of yðbÞ  y 1 ; ð7Þ
1 þ c expð8r0:67 bÞ
the curve at b = b1/2 will allow us to take full advantage
of this new formalism. From the original differential where y1 = Y1/Y1 > 1. By definition, y = 1 at the end
equation (Eq. (A.2)) and its solution (Eq. (A.3)), where of the project, and y ! y1 > 1 beyond that time. At
Y1 = d/b, we find b = b1/2, y = y1/2.
Because this equation forces y to equal 1 at b = 1, it
dY dY 1
’ ; ð3Þ has less freedom to fit the data, but it still does an excel-
dt 1=2 4
lent job. When the normalized data are fit with Eq. (7),
which can only be an approximation because of the the R2 value drops insignificantly, from 0.996 to 0.993,
changed boundary conditions; the ‘‘1/2’’ on the left- and with the exception of the end point (y = 1) the
hand side of the equation reminds us that this quantity graph is nearly identical to the display of Fig. 2 (and
is calculated at b = b1/2. Based on the PMBOKÕs verbal so it is not reproduced); c = 12.1 and r0.67 = 0.746. To
description of the curve, we will next introduce a numer- ensure that this one example is not a special case,
D.F. Cioffi / International Journal of Project Management 23 (2005) 215–222 219

Appendix C shows the good fit to another randomly


chosen data set. 1
Finally, how is it all used? For any project, the man-
ager can choose values for b1/2 and r0.67, calculate c from
Eq. (5), find the normalization factor (y1) from Eq. (6), 0.8
and plot the desired evolution curve with Eq. (7). If the β = 0.25 0.50 0.75
1/2
manager wants to insist that y = 1/2 at b = 1/2, perform-
ing the b 0 calculation (see Appendix B) will produce a 0.6
new b1/2 (and a corresponding new c) that will fulfill

y
the desired condition. In the fit to the PS8 data, for
0.4
example, b1/2 = 0.435, but y = 1/2 at b = b 0 = 0.433.
The conversion to actual durations and costs (i.e., with
dimensions of time and money) comes through the ac-
0.2
tual Y1 and t1.

0
4. Adjusting the curve by changing the rise or the halfway
point 0 0.2 0.4 0.6 0.8 1
β
Textbooks (e.g., [8]) present typical S-curves as sym-
Fig. 3. The approximate solution, y = Y/Y1 (Eq. (7)), shown at
metric about both the abscissa and the ordinate, that r0.67 = 1 for three values of b1/2.
is, symmetric both in time and in cost. But this solution
(Eq. (7)) allows a project manager to change either or
both of the two determining parameters, r0.067 and b1/2,
either to match the curves that result from the actual
1
scheduling or to influence schedule planning by playing
with alternative cash outflows while still in a projectÕs
r =2 1 0.5
development stage. In this section, we see how the shape 0.8
0.67

of the curve changes as b1/2 moves away from 1/2 and as


r0.67 moves away from 1. At r0.67 = 1, the curve produces
the strong rise in the middle noted in the derivation of 0.6
Eq. (4). Increases in r0.67 steepen the rise.
y

If a manager wants spending symmetrical in time, b1/2


should be set to about 1/2. When r0.67 = 1, c = 54.6, 0.4

which is reassuringly much greater than 1, and


y1  1 = 0.018. One front-loads projects by moving
0.2
b1/2 closer to the beginning of the project. Similarly, a
project with most of its spending much later than the
midpoint would use a larger b1/2, closer to the end of 0
the project.
Table 2 shows the values of the different parameters 0 0.2 0.4 0.6 0.8 1
at three choices for the parameters r0.67 and b1/2. Note β
that when r0.67 is small (0.5) or when b1/2 is large
Fig. 4. The approximate solution, y = Y/Y1 (Eq. (7)), shown at
(0.75), the curve does not have time to approach its b1/2 = 0.5 for three values of r0.67.
asymptote sufficiently quickly, and y1  1 is then of or-
der 10–15%.
Figs. 3 and 4 show two sets of three curves corre- the common b1/2 parameter, the curves do not intersect
sponding to the values in Table 2. The curves behave at y = 0.5 because the actual halfway points (the b 0 s) are
as expected, giving confidence in the approximations. 0.454, 0.496 and 0.500, respectively. The large rise in the
Fig. 3 groups curves with constant rise, r0.67 = 1, at three slope for the last one produces a large value of c that re-
values of b1/2: 1/4, 1/2, and 3/4. The middle one is the sults in b1/2 . b 0 .
familiar textbook curve. The first and third ones corre- The curve with the lowest rise, r0.67 = 0.5, does not
spond to families of front-loaded and rear-loaded show much of the familiar S-curve shape, but doubling
projects. the standard rise produces a curve with little time
Fig. 4 shows three curves at a common b1/2 = 0.5, allowed from minimal spending to full cost. Thus,
with the rise r0.67 doubling from 0.5 to 1 to 2. Despite again we have some confidence that a reasonable
220 D.F. Cioffi / International Journal of Project Management 23 (2005) 215–222

approximation was used originally, when r0.67 = 1, and given time, i.e., the spending distribution. (Fig. 1 shows
in the current absence of further examined data, we this curve for the sample data.) For a symmetric accu-
may expect most real projects to show r0.67  1. The mulated cost (b1/2 = 0.5), the plotted derivative looks
range of the values about unity awaits analysis. similar to a Gaussian (normal) distribution, but it does
differ slightly.
As one might expect, an asymmetrical spending rate
5. The derivative of the accumulated cost: the cost generates an asymmetrical accumulated cost. Many ref-
distribution erences (e.g., [8]) show a symmetrical accumulated cost
but then one or more asymmetrical spending rates with-
The derivative of the accumulated cost as a function out pointing out that these graph sets are internally
of time yields the instantaneous cost rate, dY/dt, at any inconsistent.
The three curves of Fig. 5 derive from the curves of
2.5 Fig. 3, where r0.67 is held constant but b1/2 takes three
different values. Similarly, the three curves of Fig. 6 hold
b1/2 at 0.5 while taking the three values of r0.67 that gen-
2
erate the curves of Fig. 4.

β = 0.25
1/2
6. Summary
1.5

This paper shows the derivation of a modified logis-


dy/dβ

0.5 tics equation that project managers can use, in any stage
1 of a project, when working with costs or other limited
quantities. When plotted versus time, such quantities
typically take a form commonly known in the project
0.5
0.75
management community as the S-curve, and Eq. (7)
reproduces that shape. The modifications used to pro-
duce the equation required only a few minor mathemat-
0 ical assumptions, and their validity is bourne out in the
0 0.2 0.4 0.6 0.8 1
results.
β With this representation of the curve, a manager has
complete flexibility to generate any desired smooth pro-
Fig. 5. Derivatives of the three curves shown in Fig. 3. file by selecting the strength of the rise of the curve and
the point at which half the total (e.g., of costs) has been
expended. Future studies anticipate examining real data
to find the parameters b1/2 and r0.67 and to determine,
r
0.67
=2 for example, if they vary in some regular fashion with
4
projects in different industries.

3 Acknowledgements

The original solution of the Prigogine differential


dy/dβ

1 equation is courtesy of Dave van Buren. Nabil Bedewi


2 and Homayoun Khamooshi gave helpful comments that
improved the paper. Two anonymous referees also
0.5 zprovided constructive comments that led to important
fine-tuning, and one of them pointed out the epidemic
1
example that is noted.

0 Appendix A. The differential equation, via Priogine [9]


0 0.2 0.4 0.6 0.8 1
β
In the chapter called ‘‘Self-Organization,’’ in a section
entitled ‘‘Ecology,’’ Nobel-Prize winner Ilya Prigogine
Fig. 6. Derivatives of the three curves shown in Fig. 4. discusses ‘‘structural stability’’ and writes about a popu-
D.F. Cioffi / International Journal of Project Management 23 (2005) 215–222 221

lation X in a given medium. Prigogine relates that one We find the new b1/2 through:
often describes the growth of the population through
b1=2 ¼ b0 þ Db ðB:1Þ
the following differential equation,
dX where
 ;
¼ bX ðN  X Þ  dX ðA:1Þ   
dt 1 y1  1 c þ 1
Db ¼ : ðB:2Þ
where ‘‘b is related to the rate of birth, d is related to 4r0:67 y1 cþ2
mortality, and N is a measure of the milieuÕs capacity As an aside, I note that in deriving the above expres-
to support the population.’’ Prigogine does not provide sion, one sees the consistency with the original slope cal-
an analytic solution to the equation, but instead refers culation. The slope used above equals that from Eq. (3)
the reader to a figure that any project management stu- multiplied by (c + 2)/(c + 1), which approximately
dent or practitioner would recognize immediately as a equals 1 for any large c.
prototypical S-curve. Most project managers will probably not be con-
This curve, often known now as the ‘‘logistics cerned about the small difference Db, but for those who
curve,’’ was used, for example, by Hubbert, who made would like to specify the exact point at which y = 0.5, a
the first calculations of the finite duration of fossil fuels quick iteration will generate the b1/2 that corresponds
(e.g., [10,11]). The curve also models the growth rate of with the desired b 0 , as follows. First, set b 0 to the desired
a population of interstellar clouds created in galactic time. The given r0.67 and an initial guess at b1/2 (one can
simulations [12]. Those simulations determined the use the b 0 itself) determine the first c. Calculate Db. Use
cloud birth rate and mortality intrinsically (i.e., the this Db with b 0 to obtain the first new b1/2. The new b1/2
rates were not input), so the variables were redefined yields a new c, from which one obtains a new b1/2, and so
(and X changed to Y) to produce a form more appro- forth. In usually fewer than ten loops, the equations
priate both to the interstellar medium and project converge to a fixed value of b1/2 that corresponds to
management: the desired b 0 . Thus, the b1/2 term is now only a param-
dY eter that is near the actual b 0 at which y = 1/2.
¼ dY  bY 2 ; ðA:2Þ
dt
where in terms of PrigogineÕs original variables,
 Appendix C. Fitting the curve to a second data set
d ¼ bN  d.
The general solution to this equation is
To demonstrate that the example in the main text
 1
b does not represent any specially chosen case, the
Y ðtÞ ¼ þ cedt ðA:3Þ approximate solution, Eq. (7), was fit to an example ta-
d
ken randomly from a textbook [13] (in the book, it is not
as can be demonstrated by differentiating it (i.e., Eq. used to demonstrate an S-curve). The data in Fig. 7 are
(A.3)).
Imposing the boundary conditions appropriate to
project management changes its form slightly, to Eq.
(1), as explained in the main text. The transformation re- 1000
quires subtracting the initial value, Y0, from the general
solution and defining Yss = d/b = Y1 + Y0 to remove Yss
in favor of Y1. 800
Cumulative Expense ($1000)

600
Appendix B. Obtaining the exact value of the one-half
point
400

By definition, the value of b at which y . 1/2 is b1/2.


Let b 0 equal the value of b such that y = 1/2 exactly. This
200
section explains how to calculate b 0 .
From the solution represented by Eq. (7), we can cal-
culate the slope at b1/2 and join, with only a slight error, 0
b1/2 and b 0 on a straight line of that slope. With this
approximation, we can find the desired b1/2 parameter 0 10 20 30 40 50
such that y = 1/2 at a chosen b 0 , as shown in Table 2. Weeks

(There the b1/2s corresponding to the ‘‘chosen’’ bs are Fig. 7. A second example of data fit by the approximate solution, Eq.
listed in the column under the parenthetical b1/2.) 7; r0.67 = 0.881 and c = 22.4.
222 D.F. Cioffi / International Journal of Project Management 23 (2005) 215–222

the early-start-time weekly expenses in the ‘‘LOGON’’ [3] Cioffi DF. Completing projects according to plans: An earned
project, Table 9.5, in NicholasÕs book. These data vary value improvement index. J Oper Res Soc 2004 [submitted].
[4] Weisstein EW. Gompertz curve, from MathWorld–A Wolfram
from the classically shown S-curve in two respects: (1) Web Resource. http://mathworld.wolfram.com/GompertzCurve.
The three points sitting at about 20 weeks have identical html, accessed 3 June 2004.
values, and (2) The summed expenses continue to climb [5] Bourke P. Gompertz function. http://astronomy.swin.edu.au/
at the end of the project. (If most projects showed this ~pbourke/analysis/gompertz, accessed 3 June 2004.
type of accumulated cost growth, the phrase ‘‘S curve’’ [6] Cioffi DF. Managing project integration. Vienna, Virginia: Man-
agement Concepts, Inc.; 2002.
might not have arisen.) [7] The Sciforma Corporation, Project Scheduler software, PS8.
Nevertheless, the solution fits sufficiently well. The (http://www.sciforma.com/products/ps_suite/ps8_overview.htm)
difference between the fit and the final data point is (2001).
only about 6 percent: the curve would serve as a [8] Meredith JR, Mantel SJ. Project management: A managerial
working management model for the projectÕs cost approach. 5th ed.. New York: Wiley; 2003.
[9] Prigogine I. From being to becoming: Time and complexity in the
evolution. If we add data points at the same or physical sciences. San Francisco: W.H. Freeman and Company;
slightly greater magnitude at the end to flatten the 1980.
curve and produce a more standard profile, the fit [10] Hubbert MK. Energy from fossil fuels. Science
improves. 1949;109(2823):103–9.
[11] Hubbert MK. Exponential growth as a transient phenomenon in
human history. In: Presented before the World Wildlife Fund,
fourth international congress, the fragile earth: toward strategies
References for survival, San Francisco, 1976. Copyright 1976 by M. King
Hubbert, 1976.
[1] Project Management Institute Standards Committee, A guide to [12] Cioffi DF, Shull JM. Simulations of supernovae-dominated
the project management body of knowledge – 2000 ed. (PMBOK). interstellar media in disk galaxies. Astrophys J 1991;367:96.
Newtown Square, PA: Project Management Institute, Inc., 2000. [13] Nicholas JM. Project management for business and technology:
[2] Murmis GM. ÔSÕ curves for monitoring project progress. Project Principles and practice. 2nd ed.. Englewood Cliff, NJ: Prentice-
Manage J 1997:29–35. Hall; 2000.

Anda mungkin juga menyukai