2nd International Conference on Emerging Trends in Engineering & Technology, April12, 13, 2013

College of Engineering, Teerthanker Mahaveer University.

Applying Learning Curve: Incorporation of Learning

Curves to calculate labour productivity
Krishna Dwivedi 1, Dr D.N. Raut1

This paper discusses the application of learning curves to real world problems. Learning curves have several
alternate names, such as improvement curves, progress curves, startup functions, and efficiency curves, but we
prefers the original historical name “learning curves” to the more recent alternatives. Learning curves are
mathematical models used to estimate efficiencies gained when an activity is repeated, Main motive behind this
paper is to measure the change in labour productivity due to effect of learning curve which will help to forecast
manpower resource planning thereby reducing direct costs to company,In this paper data is collected via time study
method conducted on three different workers (experienced, Intermediate, trainee) so that it can be easy to predict the
increase in productivity with increase in experience, also with the help of learning curve formula for change in
productivity is derived.

Keywords: Labour Productivity, Learning curve, Manpower resource planning.

1. Introduction
How would a company use learning curves?
T.P. Wright first documented the idea of
The most important question in regards to learning
learning curves and its effect on efficiency in the
curves is how would a company use learning curves in
American aircraft industry in 1936. Wright found that
their business? One practical application of learning
there was a pattern in the way that people learned; he
curves is in forecasting budgets if companies can
found that per unit production time reduced at an
accurately estimate the rate of learning they can also
unvarying rate. Researchers took Wrights’ idea and
estimate product cost. Understanding cost is also
expounded upon it during World War II, where the
paramount to pricing decisions, should the company
emphasis was primarily on cost efficiency.
follow a prestige pricing or price skimming strategy.
Since World War II, the term learning curve has
Determining human resource needs is another
been expanded to other business and non-business
application in which learning curves can be used, as
applications such management, distribution, team
companies become more efficient they may eliminate
learning, health care, investing, and even exercise.
jobs or choose to move them to other functional areas.
These more general applications of learning curves are
However as G.J. Steven points out, “It is important to
more accurately described by the term experience curve
appreciate that the learning curve is not a cost-reduction
or practice curve. For the purposes of this paper a
technique since the rate of future time reduction can be
learning curve describes the nonlinear relationship
predicted by the learning curve model. Cost reduction
between labour hours per unit and units of output, we
only occurs if management action is taken, for example,
assume a single driver units of output for our purposes
to increase the rate of time reduction by providing
though more complex formulas exist.
additional training, provision of better tools etc.
1) Prof., Department of Production Engineering, Veermata
Jijabai Technological Institute, Mumbai, India. Email:
dnraut@vjti.org.in

1
2nd International Conference on Emerging Trends in Engineering & Technology, April12, 13, 2013
College of Engineering, Teerthanker Mahaveer University.

2. Basic Learning Curve Calculations 4. Learning curve application in Die-casting

There are different models for learning curves that have Industry
been proposed and used by businesses. None are Since learning curve helps to determine the time
generally accepted as being superior. For explanatory required to produce the nth unit hence we will first find
purposes we will use the Incremental Unit-Time the learning curve of labour in die-casting industry.
Learning Model also called u learning model. This i) Time study is being carried out three
model is based on the equation different labours i.e. Experienced with (5000
hrs of work (experience), Intermediate with
b
Y = aX approx. 1000 hrs of experience, Trainee with
11 hrs of experience.
Y = the time to produce the last unit of output, ii) Mentioned cycle time in table is average time of all
a = the time to produce the first unit of output, recorded data for 3 labours with whole operation is
X= cumulative number of units produced, divided into many work elements to clearly define
and b= the rate of learning. work into variable and constant element (human and
machine).
The rate of learning is calculated by taking the natural iii) Also during learning curve calculation
log of the percentage of learning and dividing by the experience (no of hours) is converted into no of
natural log of 2 (ln(% of learning)/ ln2). Ultimately castings produced for eg.
what this equation is describing is that the incremental Experience of 5000 hrs = 5000*141=705000
unit time decreases at a constant rate each time the castings
quantity of units produced doubles. The lower your Intermediate of 1041 hrs=1041*126=131166
percentage of learning is the faster individual units are castings
produced and thus the faster average time per unit. As a Trainee of 11 hrs =11*31.32=344
practical matter there is no hard and fast rule for castings.
determining the percentage of learning.
iv) Productivity is calculated for the following
3. Steps to forecast future production. labours in (casting produced/hours).
v) Change in Productivity is calculated between
1) Calculate learning curve slope i.e b using past
all of these labour w.r.t experience.
recorded data (time study in my case).
vi) Now learning curve slope is determined.
2) Once the percentage of learning is established the
vii) From learning curve slope, we will be able to
numbers can be easily generated using an excel
forecast the future production rate.
spread sheet. The individual unit times are based on
b
the equation y = a X where a and b will be constant
throughout the range of units. 5. Time study chart:
3) Now putting the current data in developed learning Times mentioned here are average times of 3
curve equation will give us the future (forecasted) readings per labour
production. Time mentioned in seconds
Experienced workmen = 5000 hrs = 625 days
Intermediate workmen = 1041 hrs = 130 days
Trainee workmen = 11 hrs = 1 day
Activity type
C: Constant time (Machine element)
V: Variable time (Human element)

4) We can measure the change in productivity which

can also help us to determine quality of training
provided to the employees.
5) Calculated results will also help us in manpower
resource planning.

2
2nd International Conference on Emerging Trends in Engineering & Technology, April12, 13, 2013
College of Engineering, Teerthanker Mahaveer University.

Time study chart:

Observed time
Sr Activity
Activity Experienced Intermediate Trainee
no type
(5000 hrs) (1041 hrs) (11 hrs)
01 Die Closing V 1.87 1.84 1.92
02 Picking up molten metal V 0.47 1.6 0.80
03 Pouring metal into sleeve V 1.59 2.07 2.52
04 Injection C 2.90 2.90 2.90
05 Solidification time C 7.20 7.20 7.20
06 Die opening C 0 0 0
07 Ejection C 2.00 2.00 2.00
08 Pickup plier V 0 0 0
09 Pick up brush V 0 0 0
10 Casting taken out V 1.74 1.78 3.54
11 Apply Diecoat on die surface V 0 0 0
12 Put down plier and brush V 0.98 1.38 1.18
13 Pick up Spray gun V 0.45 0.64 2.87
14 Spraying water on die V 3.98 4.23 5.59
15 Put down Spray gun V 0.33 0.40 0.79
Total Activity time 23.52 26.04 31.32
Productivity units/hr 153 138 115
Productivity units/shift 1225 1106 920
change in productivity 10% 25%

6. Calculating learning curve parameter (b): k=log of learning rate = 3.2415*(log of 2)

According to wright unit model equation = 0.9758
b Hence learning rate is 97.58%
Y=aX
Finding learning parameter from trainee labour to Above solution is programmed into excel to forecast the
experience labour, pattern of cycle time with respect to experience,
Here Y is no of castings produced i.e Screenshot of excel template:
Y= experience in hours* no of units produced/hour,
1256 = a * (31.30) 1+b …………….eqn 1
765415 = a * (23.52) 1+b ……….…….eqn 2

Converting these to the log forms we have:

log 1256 = log a + (1 + b) (log 31.30)
log 765415 = log a + (1 + b) (log 23.52)

solving the equation we get,

b=3.2415
i.e. converting learning parameter(k) in learning curve
percent,
k = log of learning rate / log of 2

3
 2nd International Conference on Emerging Trends in Engineering & Technology, April12, 13, 2013
College of Engineering, Teerthanker Mahaveer University.

Also shown below is learning curve graph plotted 8. Conclusion:

between cycle time and no of units. This paper discussed about the application of learning
curve in Die-casting industry which is 96-98% also
developed formula for change in productivity which will
help to determine the company’s progress and employees
future performance w.r.t current performance, also
learning curve as we saw help to determine the future
production rate of employee ,hence it will help in
accurately forecasting manpower needed to accomplish
the future demand, thereby reducing the unnecessary
manpower cost to company (Direct cost of product).

9. Acknowledgement:
I have great pleasure in expressing my deep sense of
gratitude and indebtedness to Dr. D N Raut Department
of Production Engineering, VJTI, Mumbai.
From the above graph plotted it is easily predicted the
I would also like to thank Matchwell diecasters for
reduction in cycle time after n units,
allowing me to do a valuable project in their esteemed
Also seen above that after some units the curve is almost
organisation, also I hope my research paper would very
straight and very small increment in learning takes place,
helpful to various organisations in the field of die-casting.

7. Relationship between two intermediate units is 10. References:

developed from U learning model. 1) Learning Curves Tutorial by: Eric Gooden
Tn=(Cb/nb)*Tc Operations Management 345 Dr. Tom Foster.
Where, 2) All about Learning Curves, by Evin Stump
Tn= Cycle time of nth unit, P.E.Senior Consultant, Galorath Incorporated.
C = current unit no, 3) Incorporation of Learning Curves in Economic
n = unit whose cycle time is to be determined, Order Quantity (EOQ) and Economic
Tc=Cycle time of current unit Production Quantity (EPQ) Models by Krishan
Productivity at nth unit can be found by: Rana and Ephrem Eyob.
Productivity=Total working time /Cycle time, 4) NBER working paper series, learning curve and
Pc=28800(in case of 8 hr shift)/ Tc ……(i) optimal production under uncertainity,by
Pn=28800/Tn at nth unit ….…(ii) Saman Majd,Robert S Pindyck,working paper
no:2423,National beaureu of economic research.
Change in productivity, 5) en.wikipedia.org/wiki/Learning_curve
% change = (Pc-(28800/(Cb/nb)*Tc)/Pc * 100 6) www.businessmate.org/Article.php?ArtikelId=2
7) www.protech-ie.com/curv-v2.pdf
Change in productivity is calculate and mentioned above 8) Spence, A. Michael, "The Learning Curve and
i.e., Competition," Bell Journal of Economics,
Spring 1981, 12, 49-70.
Change in productivity from 1 day to 625 day = 25%
Learning rate of 97.35% is found to be very small 9) See Merton (1977), and the discussion on
because of very small human element i.e pages 7-8 in Majd and Myers (1986)
C: Constant time (machine element) = 12.10 sec 10) Organisational learning curve: Interplant
V: Variable time (Human element) = 11.42 sec transfer of knowledge acquired through
learning curve by DENNIS EPPLE, LINDA
Which works out to be 48% of total cycle time is
variable time. ARGOTE AND RUKMINI DEVADAS.