Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 164 (2016) 90 – 97

Creative Construction Conference 2016, CCC 2016, 25-28 June 2016

Learning Curve Effect on Project Scheduling

Levente Malyusz*
Budapest University of Technology and Economics, Muegyetem rkp. 3. Budapest 1111, Hungary

Abstract

In this paper a simple learning curve effect on project scheduling is shown with a help of an artificial example. Although learning
is an essential part of our life, traditional scheduling technique can not handle efficiently the learning curve effect. It is assumed
that the duration of upcoming repetitive activities are shorter due to the learning curve effect if the gap between consecutive
activities is small enough. Learning curve theory can be applied to predicting cost and time to complete repetitive activities. Taking
into accounts the effects of learning curve (or experience curve) one can get better future prediction on project duration and can
save money and time. This effect normally result in shorter project duration. Although the effect is “simple” calculations are
cumbersome if the learning effect is applied on traditional project scheduling techniques like Critical Path Method, or Precedence
Diagramming Method as calculation lead to an exponential time algorithm.
© 2016
© 2016 The
TheAuthors.
Authors.Published
Publishedby byElsevier
ElsevierLtd.
Ltd.This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the organizing committee of the Creative Construction Conference 2016.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the Creative Construction Conference 2016
Keywords: learning curve; project scheduling; precedence diagramming method

1. Introduction

In practice project scheduling methods suffer from lack of precision therefore it is a big challenge to create a
realistic and useable project schedule. It is difficult and time consuming to properly estimate time, assign resource,
determine interdependencies between tasks, manage changes… That is why it is important to discover and investigate
the differences between practice and theory of scheduling methods (Francis et al. [1]). In a construction project general
contractor distributes the job among subcontractors. It is an observation that subcontractors rarely start their work at
earliest possible time. It is obvious that they can reduce their cost if they work continuously.
According to the theory of learning when numerous similar or nearly identical tasks are performed, the necessary

* Corresponding author. Tel.: +37 -70-290-5210;

E-mail address: lmalyusz@ekt.bme.hu

1877-7058 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the Creative Construction Conference 2016
doi:10.1016/j.proeng.2016.11.596
 Levente Malyusz / Procedia Engineering 164 (2016) 90 – 97 91

effort is reduced with each successive task (Oglesby et al. [2], Drewin [3], Teplitz [4], Everett and Farghal [5,6], Lutz
et al. [7], Lam et al. [8], Couto and Teixeira [9]). Learning curve theory can be applied to predicting the cost and time,
generally in units of time, to complete repetitive activities (Malyusz and Pém[10]). Based on the theory basically there
are two different methods of calculation the activity time of repetitive activities. Unit time and the cumulative average
time methods. Unit time method means that the time of some doubled unit equals the time of the undoubled unit times
the slope of the learning curve. Cumulative average method means that the cumulative average time of some doubled
unit equals the cumulative average time of the undoubled unit times the slope of the learning curve. This was used in
the original formulation of the learning curve method, referred to as Wright’s model, in Wright’s famous paper on the
subject [11]. A number of researchers have suggested that Wright’s model is the best model available for describing
the future performance of repetitive work (Everett and Farghal [5], Couto and Teixeira [9]). In (Malyusz and Pém
[12]) the exponential average method with Į =0.5 yielded the most accurate predictions.
In construction project management the proper scheduling of project is an essential problem. Estimation of activity’s
time is a crucial part of the schedule. Learning curves impact activity’s time and in recently used management
softwares, which can handle resources, this effect would be handled easily (Hajdu [14], [18]). The objective of this
paper is to show an effect of learning curve that can cause changes in project duration. This occurs when the same
construction team performs similar activities continuously, so after finishing of the predecessor activity successor can
immediately start.
There is little information in the literature about the use of learning curves in scheduling, although it seems that the
principle of learning curves gathers its ground in scheduling of repetitive construction operations (Hinze and Olbina
[15], Fini et al. [17]). In (Zahran et al.[16]) learning curve effect on linear scheduling method is discussed.

1.1. Learning curve

Learning curve theory is applicable to the prediction of the cost or time of future work, assuming repetitive work
cycles with the same or similar working conditions in terms of technology, weather, and workers, without delay
between two consecutive activities. The direct labor required to produce the (x + 1)st unit is assumed to always be less
than the direct labor required for the xth unit. The reduction in time is a monotonically decreasing function, an
exponential curve, as described in Wright’s [11] paper.
Wright's linear log x, log y model is as follows:

݈݊‫ ݕ‬ൌ ݈݊ܽ ൅ ܾ݈݊‫ݔ‬Ǣ‫ ݕ ש‬ൌ ܽ‫ ݔ‬௕ ൌ ܽ‫ ݔ‬௟௢௚మ௥ (1),

where x is the cycle number, y is the time required to complete cycle x in labor hours/square meter, a is the time
required to complete the first cycle, b is a learning coefficient, and r is the rate of learning. For example if r=0.9 (90%),
then b=-0.151 see Figure 1. Wright discovered that when the labor cost decreases at a constant rate, that is, the learning
rate, the production/cycles doubles. So learning rate is the constant rate with which labor time/cost decreases when the
production/cycles doubles in a linear log x, log y model. This feature of the learning rate comes from the logarithms
nature and true only in linear log x, log y model.
92 Levente Malyusz / Procedia Engineering 164 (2016) 90 – 97

ϭ Ϯ ϯ ϰ ϱ ϲ ϳ ϴ ϵ ϭϬ ϭϭ ϭϮ ϭϯ ϭϰ ϭϱ ϭϲ

Figure 1. Learning curve

In the construction industry learning rate is between 85-95%. Let us take an example with 90% of learning rate. In
this case if a job is 10 days a repetition of that is 8 days, if the working conditions are similar...
Learning curve effect is not always apply, of course. It flourishes where certain conditions are present. It is
necessary for the process to be a repetitive one. Also, there needs to be a continuity of workers and they mustn’t be an
abrupt stop during the production process. When the learning curve effect can on occasion come to an abrupt stop,
graphically, the curve jumps up.

1.2. Project scheduling

Least cost scheduling techniques are one of the most sophisticated techniques applied in project management. In
this paper we follow the concept of activity on node network and Precedence Diagramming Method (PDM) is used.
Although there are four relationships can be defined between activities we used only Finish Start relationship and its
variants, namely: FS0, maxFS1, max FS0. The maximal precedence relationship describes the maximum allowable
time between the start/finish point of the preceding and the start/finish point of the succeeding activity. The concept
of non-linear activity-production-time functions has been introduced by Hajdu (Hajdu,[18]).

1.3. An example

The artificial example project can be seen on Fig. 2. Activity B and E are worked by the same group of workers.
Let B and E originally are 10 days job. Each relationship is FS0. According to the results of the time analysis B ends
on day 15 , E starts on day 18 , so there are 2 days between them. In this case activity time of B and E are 10 days. If
there is one day gap between the two jobs let the second job be 9 days.
 Levente Malyusz / Procedia Engineering 164 (2016) 90 – 97 93

B FS0 D
FS0 10 days 6 days
FS0

A F
FS0 5 days
5 days

FS0 FS0
C E
12 days FS0 10 days
Figure 2. Example, project duration is 32 days

In Figure 2. traditional representation of this problem is shown. Minimum project duration is 32 days. Now let us
take into account the effect of learning curve in activity time E.
Activity B and E are similar jobs and they made by the same construction team.
If after finishing B immediately (next day morning) E starts, in this case to complete E needs less time that is 8
days (see Figure 4). In this case the minimum project duration is 30 days. If there is one day between the finishing of
B and starting of E then 9 days necessary to complete E (see Figure 3). In this case minimum project duration is 31
days. Otherwise 10 days the minimum time of activity E.

Table 1. Activity time of E

ƐƚĂƌƚŝŶŐͲĨŝŶŝƐŚŝŶŐ ϮŽƌŵŽƌĞ ϭ Ϭ
ĂĐƚŝǀŝƚLJƚŝŵĞŽĨ ϭϬ ϵ ϴ

Let us modify our example adding a maxFS1 relationship between B and E. After calculating, B ends on day 16
day, E starts on day 18 day, so there are 1 day between them (day 17). See Figure 3. Since max relationship can cause
a cycle in the network it is not sure generally that a feasible solution exists.
94 Levente Malyusz / Procedia Engineering 164 (2016) 90 – 97

Figure 3. Example, project duration is 31 days

And finally let us modify our example changing maxFS1 with maxFS0 relationship between B and E. According
to the results of the time analysis B ends on day 17, E starts on day 18, so there are 0 day between them, E starts
immediately after B ends (see Figure 4.).
In this example activity time of E was reduced as an impact of learning: subcontractor could work continuously in
activities B and E. In addition to activity E has a normal critical type so project duration could be reduced to 31 days.
 Levente Malyusz / Procedia Engineering 164 (2016) 90 – 97 95

Figure 4. Example, project duration is 30 days

In the next example we will see that if B and E are closer to each other and learning curve effect is take into account
project duration is increased.

Learning curve effect on project scheduling with reverse critical activity. According to the results of the time
analysis B ends on day 15, E starts on day 18, so there are 2 days between them.

B FS0 D
FS0 10 days 6 days
FS0

A F
FS0 5 days
5 days

FS0 FS0
C SS12 E
12 days 10 days
FF12 SS17

Figure 5. Example with reverse critical activity

In Figure 5. traditional representation of this problem is shown. Minimum project duration is 41 days.
96 Levente Malyusz / Procedia Engineering 164 (2016) 90 – 97

Now let us take into account the effect of learning curve in activity time E.

B FS0 D
FS0 10 days 6 days
FS0 FS0

A F
FS0 5 days
5 days
maxFS1

FS0 SS12 FS0

C E
12 days 9 days
FF12 SS17

Figure 6. Example with reverse critical activity

Activity B and E are similar jobs and they made by the same construction team.
If there is one day between the finishing of B and starting of E then 9 days necessary to complete E. In this case
minimum project duration is 42 days (see Figure 6). If after finishing B immediately (next day morning) E starts, in
this case to complete E needs less time that is 8 days. In this case the minimum project duration is 43 days. Otherwise
10 days the minimum time of activity E with project duration 41 days.

1.4. Algorithm

This problem leads to a mixed integer programming problem with some combinatorics. Correct solution is to
calculate project duration of 3 networks.
1. Relationship between B and E is FS0.
2. Relationships between B and E is FS0 and max FS1.
3. Relationships between B and E is FS0 max FS0.
If there are more (let us say n) similar relations between 2n activities - where effect of learning curve take into
account - in the network, then number of networks what we have to calculate is 3n , so it is an exponential time
algorithm.

2. Conclusions and further investigations

In this paper a learning curve effect in project scheduling is shown. This effect can change project duration but an
exponential time algorithm is necessary to solve it. Further investigation is necessary to solve this problem in
polynomial time and to make other models for learning curve effect in projects where more activities are repetitive or
the structure or substructure of the different projects are the same.
 Levente Malyusz / Procedia Engineering 164 (2016) 90 – 97 97

References

[1] Francis, A., Bibai, J., Miresco, E. T. (2013). Simulation of scheduling logic using dynamic functions. Management, Procurement and Law,
166(3), 145-158.
[2] Oglesby, C. H., Parker, H. W., and Howell, G. A., Productivity improvement in construction, McGraw-Hill, New York, 1989.
[3] Drewin, F. J., Construction productivity: Measurement and improvement through work study, Elsevier Science, New York, 1982.
[4] Teplitz, C. J., The learning curve deskbook: A reference guide to theory, calculations, and applications, Quorum Books, Westport, Conn.,1991.
[5] Everett, G. and Farghal, H., Learning Curve Predictors for Construction Field Operations. Journal of Construction Engineering and Management
vol 120. (1994), pp603-616.
[6] Everett, G. and Farghal, H., Data representation for predicting performance with learning curves. Journal of Construction Engineering and
Management vol 123. (1997), pp46-52.
[7] Lutz, J. D., Halpin, D. W., and Wilson, J. R., Simulation of learning development in repetitive construction. J. Constr. Eng. Manage., 120. 4.,
(1994) 753–773.
[8] Lam, K. C., Lee, D., and Hu, T., Understanding the effect of the learning-forgetting phenomenon to duration of project construction. Int. J. Proj.
Manage., 19.7., (2001), 411–420.
[9] Couto, J. P., and Texiera, J. C., Using linear model for learning curve effect on highrise floor construction. Constuction. Management Economics,
23.4, (2005), 355–364.
[10] Mályusz, L. and Pém, A., Predicting future performance by learning curves, Procedia-Social and Behavioral Sciences, vol.119. pages 368-
376, doi:10.1016/j.sbspro.2014.03.042, 2014.
[11] Wright, T.P., Factors affecting the cost of airplanes. Journal of Auronautical Science, (Feb.),(1936), pp. 124-125.
[12] Mályusz, L. and Pém, A., Prediction of the Learning curve in Roof Insulation, Automatization in Construction, (2012), DOI: 10.1016/j.autcon.
2013.04.004
[13] Hajdu M (1993), An algorithm for solving the cost optimization problem in precedence diagramming method. Periodica Polytechnica ser.
Civil Engineering 37, 231-247.10.1139/l94-100
[14] Hajdu M, Klafszky E. (1993), An algorithm to solve the cost optimization problem through an activity on arrow type network (CPM/cost
problem), Periodica Polytechnica ser. Architecture 37, 1-4. ,
[15] Hinze, J. and Olbina, S., Empirical analysis of the learning curve principle in prestressed concrete piles. Journal of Construction Engineering
and Management vol 135. (2008) pp425-431.
[16] K. Zahran , M. Nour , O. Hosny, The Effect of Learning on Line of Balance Scheduling: Obstacles and Potentials, International Journal of
Engineering Science, Volume 6 Issue No. 4, DOI 10.4010/2016.889 ISSN 2321 3361, 2016.
[17] A. F. Fini, A., Rashidi, T., Akbarnezhad, A., and Travis Waller, S. (2015). "Incorporating Multiskilling and Learning in the Optimization of
Crew Composition." J. Constr. Eng. Manage., 10.1061/(ASCE)CO.1943-7862.0001085, 04015106, 2015.
[18] M. Hajdu (2015) Continuous Precedence Relations for Better Modelling Overlapping Activities, Procedia Engineering Volume 123, pp 216-
223, doi:10.1016/j.proeng.2015.10.080.