You are on page 1of 4

The Revay Report

Volume 18 Published by Construction Consultants


Number 3 Revay and Associates and
October 1999 Limited Claims Specialists

By Stephen G. Revay
LEARNING CURVE IN CONSTRUCTION
The article by Zey Emir
in this issue describes by Zey Emir, P. Eng, MBA total cost method because these meth-
the theoretical and the Revay and Associates Limited, Montreal ods do not allocate the responsibility for
practical applications the loss. In other words, the quantity of
of learning curves for the loss is, in effect, offered as proof of
PRODUCTIVITY the loss for which the Owner is responsi-
both managing
construction projects In its broadest sense, productivity is ble. Courts tend to hold contractors to
and analyzing loss of the measure of output (the work pro- more rigorous standards when damage
productivity claims. The theory is well duced) per unit of input (various cost claims are based on hypothetical base-
recognized in an assembly-line-type of items that are incurred). Although pro- line productivities. Such claims are
ductivity is often thought of as relating to viewed as self serving and simply sub-
environment, but its usefulness has, at
labor only, in practice it relates to any jective. Instead, courts seem to favor
times, been questioned in construction. claims that are as closely linked to reali-
resources used to produce a result.
After all, as the non-believers argue, ty as possible.
tradesmen are already well experienced in Lost productivity is not necessarily
their craft when they arrive on a project. measured based on the optimized uti- The method of calculation most favored
Unfortunately, this argument is seldom lization of resources to maximize output. by courts uses some cause and effect
sustainable, particularly during a buoyant Contractors frequently decide for analysis to prove damages. The most
economy. On any new project, workers practical reasons that maximizing out- convincing evidence of loss of efficiency
put is not in their best interest, or they results from comparing the work done
always require time to learn to perform
fail to maximize output due to their own during an impacted period with work
their specific tasks well. It has also been shortcomings. However, no matter how done during a period of unimpaired pro-
said that most construction activities are of inefficiently a contractor has performed ductivity. The best measure of the base-
a non-repetitive nature. It is true that any the contract from the outset, it may still line against which the loss of
two projects are seldom identical; but be exposed to reductions in the level of productivity is established is that pro-
within the same project many tasks are productivity it anticipated due to disrup- ductivity which has been actually
repetitive, at least to some degree. Anyone tion of the work routine by others and achieved performing the same task on
who has done the same or a similar task that loss of productivity may be com- the same project. For example, if a par-
twice recognizes that, barring disruptions, pensable. ticular task is impacted by a change
it takes less time to perform it the second introduced after 30% of that task has
been completed, the productivity real-
time. Nobody could disagree that forming HOW DO WE MEASURE THE ized prior to the occurrence of the
the tenth slab in a high-rise building takes IMPACT OF THE DISRUPTION? change is the best possible baseline for
less time than the fi rst one even though the proving and calculating the loss of pro-
slabs may not be identical. In 1965, the Clearly, one should not compare the pro-
ductivity during the period of disruption ductivity caused by the change.
United Nations issued a report describing
to a theoretical value reflecting optimum This approach is known as the mea-
the results of extensive research on the
or near optimum efficiency as the Con- sured mile method. The measured
beneficial effects of repetition in tractor may not have been working at
construction operations. The report clearly mile method is used to calculate pro-
that level of efficiency prior to the disrup- ductivity losses by comparing productiv-
indicated that more complicated tasks (e.g. tion. Similarly, however, the Contractor ity rates achieved in unimpacted work
erecting formwork) displayed greater should not be deprived of recovery periods or normal periods to produc-
improvement in productivity as a result of because the tradesmen were not efficient tivity rates experienced during periods
repetition than simpler tasks (e.g. stripping prior to the disruption. The loss, there- of claimed or alleged impact.
formwork). These results were subsequently fore, should be the difference between
confirmed by a number of independent the actual manhours expended to pro- The measured mile approach is also
researchers whose findings were published duce the output, and the would have more readily accepted by Owners
in the journals of various learned societies been manhours required to produce the because it inherently adjusts baseline
output but for the disrupting event. productivity to account for inefficiencies
as well as by our own experience,
caused by the Contractor. It takes into
monitoring productivity on numerous
MEASURED MILE consideration the productivity losses
construction projects. A word of warning, which are caused both externally and
however, repetition will not automatically In recent years, techniques least favored internally. For this reason, the mea-
improve productivity. Such improvement by courts and arbitration boards for sured mile method has been recog-
requires a concerted effort by management. measuring damages have been the nized as a superior method of measuring
total cost method and the modified inefficiencies. This approach also repre-
sents the most objective method of Table A curve may start over and the cost for the
claim presentation because, instead of units would be higher than the original
using industry averages for comparison, Unit Cost-Quantity Relationships for cost.
control figures come from the project the Learning Curve Factor of 80%
itself, such that the quantification of loss Cumulative If the extra work assigned to the Con-
Unit Cost tractor is of short duration and relatively
of productivity represents the particular Production
project and the particular circumstances non-repetitive, then the additional work
1 100 is performed at a higher cost. Similarly,
of delay actually experienced. This 2 80
method also eliminates disputes over when a Contractor is working on a large
4 64 number of repetitive units, the earlier
the validity of cost estimates, or factors 8 51
which may have impacted productivity units will have a higher average unit
16 41 cost. If the later units are deleted from
due to no fault of the Owner. As a result,
the measured mile approach is valid the scope of work and the lower cost
even where the Contractors bid esti- struction industry include estimating units are not produced, then the Con-
mates are unrealistic. overall average productivity in order to tractor will not be able to recover its ten-
arrive at a competitive bid price or estab- dered cost estimated using average
The underlying assumption in the mea- lishing a project schedule based on rates.
sured mile method is that the produc- anticipated rates of progress. Costs as
tivity rate achieved during the unim- well as durations of the project can
pacted period would have continued in therefore be quantified with greater
HOW IS THE IMPROVEMENT IN
the impacted period (and even improved accuracy by applying learning curves. PRODUCTIVITY CALCULATED?
with the learning curve) but for the
Another application of learning curves is The learning curve is a mathematical,
Owner caused events that impacted pro-
that of construction claims and disputes. graphical or tabular representation of
duction. The claimed amount is then cal-
Learning curves can be used in deter- how resources such as manhours are
culated based on the difference in the
mining the cost of acceleration, where reduced as the production of a product
productivity rates.
the Contractor is required to work over- or a service is repeated. The learning
time or alternatively use additional curve can be expressed mathematically
LEARNING CURVE crews to perform the same work. An as:
As crews or individuals produce more of obvious result of this acceleration is the y = ax n
a product, the unit cost of production increase in labor costs if the crews work
typically decreases at a decreasing rate. overtime for which they are paid a pre- where,
This effect is commonly called the mium. An equally important but not y = manhours to produce the xth unit
learning curve. The learning curve immediately apparent result however, is a = manhours to produce the first unit
demonstrates that there is a steadily the loss in the improvement in produc- x = number of the unit
decreasing number of manhours tivity which would be expected as the
n = exponent that relates the learning
required for a given operation when that result of repetitive work. In other words,
curve factors to the learning curve
operation is repeated. With a straight if the same amount of work is being
slopes.
line learning curve, the rate of decrease done by two crews rather than one, then
is assumed to be constant each time the each crew will have half the number of This theory states that the unit cost
number of repetitions doubles. The effect repetitions. Simply stated, the crews will decreases by a fixed percentage of the
is that the absolute amount of decrease not achieve the same level of proficiency previous unit cost each time the number
is less with each successive unit pro- as originally anticipated the routine- of units produced doubles. This fixed
duced. After a considerable number of acquiring effect is cut short, thereby rais- percentage is identified as the learning
repetitions, the learning curve approach- ing the average time required to produce curve factor. On operations that are
es a plateau that reflects the minimum a single unit of work. machine paced and for which there is no
time required to perform a task. improvement in output, the learning
When a Contractor is required to stop curve factor would be 100 percent. For
The learning curve theory is often work temporarily on a certain operation, complicated tasks without external
applied in the construction industry the crew may be assigned to other work restraints, the rate of improvement
where the work is repetitive and continu- or may be laid off (depending on the might be as much as 70 percent. In other
ous. For example, when the design calls expected duration of the delay). When, words, the second unit would take only
for repeated utilization of the same con- the cause of the delay is removed and 70 percent as long to produce as the
crete forms in high-rise or multiple build- the Contractor is required to resume the first, and the 32nd only 70 percent as
ing construction, the productivity may be interrupted operation, it may be neces- long as the 16th, and so on.
susceptible to learning curve analysis. sary to restart the learning process.
The same is true for rebar work, concrete Typical learning curve factors for con-
placement, or the rough-in and finish When a Contractor is required to per- struction operations fall in the 70 to 90
work of the trades, as long as there is a form extra work, the cost of the work is percent ranges. Figure 1 shows units and
relatively constant percentage reduction usually reimbursed on either a cost-plus the corresponding manhours for one
in the input required to produce an item or a negotiated price basis. If the extra particular learning curve factor, i.e., 85
as the production quantities double. For work assigned to the Contractor repre- percent.
example, an 80% learning curve repre- sents a large number of repetitive units,
sents a 20% reduction in the time it takes the result may be that the additional The effect of variable learning curve fac-
to produce units one and two, an addi- units are performed at a lower cost tors on the manhours per unit is demon-
tional 20% reduction in the time it takes because the Contractor will move further strated in Figure 2. As seen, a higher
to produce units two and four, units four down the learning curve in performing learning curve factor (i.e., 90%) amounts
and eight, units eight and sixteen, etc. as these units. In this situation, the average to less improvement than an 80% learn-
shown in Table A. bid price may be too high. This would ing curve factor as the cumulative pro-
only apply if both the original and addi- duction quantity increases. It follows,
The most obvious application of learning tional units are produced by the same then, that the learning curve effect is
curves is forecasting manpower require- crews. If the additional units are pro- more evident when operations and even
ments. Specific examples in the con- duced by a different crew, the learning entire projects are of a unique nature.
Figure 1 Figure 2

LEARNING CURVES LEARNING CURVE The Contractor maintained very sophisti-


IN CLAIMS ANALYSIS APPLICATIONS IN CLAIMS cated cost records, such that the
timesheets clearly differentiated between
Learning curve analysis can be very use- ANALYSIS tradesmen placing rebar, pouring con-
ful in measuring loss of efficiency, espe- Case 1 crete or working on the formwork. Over-
cially when detailed productivity records time hours and regular hours were also
are maintained throughout the duration One case where the learning curve identified for each different trade. A dif-
of the project. The learning curve effect analysis was used involved the construc- ferent crew was assigned to each casting
should be used when preparing a mea- tion of a precast segmental bridge. Indi- bed and the Contractor had detailed
sured mile analysis. The damage quan- vidual precast units or segments were records showing which segment was fab-
tification will yield inadequate results if manufactured in a precast yard located ricated in which casting bed.
the unimpacted productivity rate close to the site, which were then erect-
achieved while performing the earlier ed and secured together by longitudinal The first step was to determine the week-
units is compared with the production post-tensioning to form each span. The ly typical segment production based on
during the impacted later period. Instead two lanes of the bridge were composed the as-built information available. The
the unimpacted rate of productivity or of 50 spans each; furthermore, each span typical segment productivity was plotted
the normal period should be adjusted consisted of two pier segments placed per each group of 15 units cast as shown
for what productivity would have been on the piers and ten typical segments. in Figure 3. The reason for sorting the
without the impact i.e., incorporating units fabricated in groups of 15 was
the learning curve effects before com- The precast segments were fabricated in based on having three casting beds oper-
paring it with the productivity during the a special mold or casting bed where a ational five days a week.
impacted period. new segment was cast against its older
neighbor to achieve a perfectly aligned or The second step was to establish a nor-
This is particularly true when one con- match cast joint. There were three cast- mal period. A detailed analysis of the
siders that given an 85% learning curve, ing beds for the typical segments and typical segment productivity chart
the time required to produce the 8th unit one casting bed to produce the pier seg- demonstrates that the average unit cost
will be only 61.5% of the time it took to ments. Each segment was marked by a started to even out and reach a plateau at
produce the 1st unit. Thus, if one were to specific identification number which rep- segment number 285, which represents
use data taken from the first or second resented its proper place on the bridge. approximately 30 percent completion.
time a repetitive task was performed, Once the normal period is identified on
and use that data to prove how long it The obvious profit potential in such a the productivity chart, the analyst must
should have taken the 8th time the same construction is in the repetition of seg- verify that a sufficient amount of work
task was performed (without taking into ments in the casting yard. A long, was performed to demonstrate the Con-
account the learning curve), one would straight bridge or elevated highway is a tractors ability to perform at that
significantly overstate the amount that precasters dream as the Contractor can observed rate over an extended period of
task should have cost. As a result, the achieve a level of expertise relatively time. Another criterion to be met is to
loss of productivity measure would fall quickly. The Contractor on this project, ascertain that the work performed during
short. however, encountered considerable the normal period is representative
delays in the casting yard. One of the and similar in all aspects to work per-
It is important to note that a learning
reasons for the workers slowness in formed during the impacted period. As a
curve is simply a representation of the
mastering the casting of the bridge seg- final test, to avoid mistakes due to biased
expected decrease in input per unit as
ments was the complexity of the units. sampling, asserted production during the
the cumulative production increases. In
The construction drawings revealed that normal period should be checked
order for the calculation to be useful and
these segments had significantly more against the bid estimate and manloaded
accurate, one must have reliable record-
internal tendons than segments in most schedule to see whether production is
ing of the actually achieved productivity.
span-by-span construction. This resulted unrealistically high.
If the data is sufficient in quality and
in an increase in manhours required to
quantity, then it can be used to predict
produce each segment in addition to The productivity during the normal
the manhours required during the
increased overtime in order to accom- period is then designated as the upper
impacted period.
modate the milestones on the span erec- limit of proficiency. This upper limit or
tion schedule. The claim analysis focused normal productivity serves as the basis
on determining the additional unit costs of a scale to determine the level of input
incurred in the casting yard. (i.e. unit cost), for the earlier units based
Figure 3 Figure 4

on learning curve. In other words, the measurement, productivity was ex- the measured mile adjusted for the
project learning curve is calculated using pressed as the ratio of actual input of learning curve effect where the upper
the normal period and is extrapolated labor (i.e. manhours), to the ratio of limit of proficiency was set at 33 percent
over the entire project time line, which progress (i.e., labor content of percent- progress. This percentage complete was
then represents the baseline productivi- age complete) of contract. selected to represent a point at which the
ty. The learning curve is plotted against subcontractor had performed a sufficient
the actual productivity as seen in Figure The next step was to determine the amount of work to reach a plateau on the
3. Any actual hours above the baseline unimpacted or the least impacted peri- learning curve.
are deemed to represent loss of produc- od. The project history revealed that dur-
tivity over and above those inefficiencies ing the first six months (first quarter) of The total loss of productivity was then
which are inherent to or the responsibili- the project, the work force was constant- calculated as the difference between
ty of the Contractor. ly relocated due to lack of access to avail- actual manhours and cumulative base-
able areas. The re-design of mechanical line hours.
The cumulative loss of productivity for and electrical systems was done during
the project represents the difference this time. Once the re-design of the sys- CONCLUSION
between the actual total cost and tems was completed, the manufacturing
the should have cost as illustrated in of additional or modified components The use of learning curves as a tool to
Figure 4. had a further delay on the installations determining equitable adjustments has
and extended well into the second quar- been recognized by courts and arbitra-
Case 2 ter of the contract. The third quarter rep- tion boards. The general theory of learn-
Another example of the application of resented the normal period with the ing curves is that with repetitive tasks
the learning curve involved a drywall least interference impacting the drywall involving a considerable amount of
subcontractor, who incurred delays and system installation. The normal produc- labor, the speed or efficiency with which
additional costs due to interferences and tivity or the measured mile was con- the task is performed increases as the
difficulties encountered in relation to the sidered to be the level of productivity number of units of work increases. In
mechanical and electrical design. which the subcontractor, under normal general, there is leveling off in any
circumstances, could have maintained increase of productivity when the nor-
The first step was to establish the sub- for the duration of the project. mal productivity is said to reach the
contractors actual productivity for stud- upper limit of proficiency.
ding and drywall installation for the The baseline hours to complete the
entire duration of the project. For this entire project were calculated based on

The Revay Report is published by Revay and OTTAWA If undeliverable to addressee return to sender
Associates Limited, a national firm of Construc- 39 Robertson Road, Suite 301 Return postage guaranteed by:
tion Consultants and Claims Specialists, assisting NEPEAN, Ontario K2H 8R2 REVAY AND ASSOCIATES LIMITED
Tel.: (613) 721-6801 4333 St. Catherine St. West, Suite 500
owners and contractors in achieving profitable Fax: (613) 596-8172
and trouble-free construction projects. MONTRAL, Qubec
CALGARY H3Z 1P9
Contents may be reproduced, with a credit as Glenmore Landing
to source appreciated. Your comments and #D262, 1600 - 90th Avenue SW
suggestions for future articles are most wel- CALGARY, Alberta T2V 5A8
Tel.: (403) 777-4900
come. Fax: (403) 777-4903
dition franaise disponible sur demande. VANCOUVER
201 - 1985 W. Broadway
RAL offices: Vancouver, B.C. V6J 4Y3
CONTACT INFORMATION Tel.: (604) 737-2005
MONTREAL
4333 Ste. Catherine St. West, Fax: (604) 737-2008
Please
Suite 500 visit www.revay.com for current office locations.
MONTRAL,
If you would Qubec H3Zsubscribe
like to 1P9 for The Revay Report, click here.
Tel.: (514) 932-2188
Fax: (514) 939-0776 revay@dsuper.net
TORONTO http://www.revay.com
505 Consumers Road, Suite 306
TORONTO, Ontario M2J 4V8
Tel.: (416) 498-1303
Fax: (416) 491-0578

You might also like