WMOD00380

• Title: One Viewpoint Concerning Unit Sizes in the Development of Wind Turbines
• Date: 11/1/1981
• Weather Modification - Doc Type: Report

In support of the Federal program of solar energy
development, MITRE Corporation prepared a series of
studies [15] to develop experience curves and cost trends and
apply them to the problem of predicting the cost of solar-
related equipment, including large wind turbines. Their
studies did not, in general, make allowance for improvement
in product performance that would result from technological
improvements during the evolution of these products. Much
of the data was limited to only a few doublings of ac-
cumulated production, and the cost index used differed from
the value index used by the Boston Consulting Group. Their
curves show general trends common to a large majority of
products undergoing a cost bulge in the early to mid-1950s, an
easing of costs at the end of the 1950s and into the mid-1960s,
and a sharp upward inflection point at 1973.
Looking at both sets of data, two influences can be seen.
One is the experience effect; this is production-related and is
independent of the calendar, with each product having its own
rate of growth. The other is an external effect, calendar-
related, that tends to affect all products in the same way at the
same time. The external effect shows a lesser experience rate
in the 1950s, a greater experience rate in the early and mid-
1960s, and a sharp unfavorable inflection point at 1973. These
trends coincide almost exactly with trends in the productivity
of the private nonfarm business economy as shown in Fig. 5.
The MITRE study mentioned these external factors, but did
not investigate their causes beyond attributing them generally
to surges in the cost of labor and energy. The report men-
tioned a tendency for long-established products to exhibit
flatter experience curves than those of novel products.
However, if the calendar-related trend shown in Fig. 5 is
removed from the experience curves of the Boston Consulting
Group (14) and MITRE (15), little evidence remains of such a
distinction. The portion of the cost trend related to
production, the experience rate proper, continues in effect
even when externalities such as changes in the nationwide
productivity superimpose other cost trends of a less predic-
table nature.
In bringing these general considerations to bear on
estimating the cost of large wind turbines, the mission
analyses by General Electric [2], Kaman [3], and Lockheed [4]
allowed experience effects for relatively short production
runs, up to the thousandth unit (ten-thousandth in the case of
Lockheed). General Electric included estimates based on 85
and 90-percent experience rates, since a wide variety of their
products in mass production have consistently demonstrated
rates close to 85 percent over the long pull. Analyses of ex-
pected experience rate made by Kaman [3] were based on the
assumption that certain major components of wind machines,
such as electrical elements and steel for towers, would be
purchased off-the-shelf from present manufacturers at
present prices and therefore would not contribute to the
experience rate. Kaman arrived thereby at an estimate very
close to 95 percent. This assumption may be justified so long
as wind machines represent only a small fraction of the
market for such off-the-shelf components, but appears
questionable as it becomes economically worthwhile to
redesign each component and produce it specifically for the
given product line, as Ford did [13]. The numbers adduced in
the analysis that follows suggest that this point would be
surpassed except for the very lowest national target, and that
the Kaman estimate of the experience rate may therefore be
unduly low.
In estimating the cost of large wind turbines, the MITRE
study followed the guidance of the General Electric [2] and
Kaman [3] mission analyses. It assumed that the cost of 958
dollars per kW for the first "mass-produced" machine (in
1976 dollars) would decrease by 35 dollars with the second
machine and with each doubling thereafter until the
thousandth had been built, when cost would reach 609 dollars

Journal of Solar Energy Engineering

130
/
/
120 /
;;; Trend ------y/
-'
<t 110 /
u /
(J)
0 100 130
~ /
a: *Cytlically /
0 90 /
Q adjusted ^ 120
t-- / /
<D 60 / "
!!! / .'
X / ! 110
w 70 /; I
0
~ /,/
60 100
1950 1955 1960 1965 1970 1975

-". .

YEARS

* The cyclical adjustment is based on a regression of productivity
on the current and lagged unemployment rate from 1960 to 1968.
Sources: Department of Labor and Council of Economic Advisers.

Fig,5 Productivity in the private nonfarm business economy, from the
Economic Report of the President (1978)

per kW and remain constant thereafter. This substituted an
assumption of fixed-dollar decline for the fixed-proportion
decline of the experience law, and terminated its efficacy at a
level of production two to four orders of magnitude smaller
than that characteristic of the other studies cited. In this form,
the MITRE projections became an input to the SPURR
(System for Projecting Utilization of Renewable Resources)
model used to generate estimates of relative market
penetration rates for solar technologies [16]. Beyond the
references cited, no justification was adduced for assuming a
constant product cost after the thousandth unit, and in their
search of the literature, the authors have found no other.
On the basis of these considerations, it was assumed for the
present calculations that an experience law would continue to
characterize the cost of large wind turbines, at least in relation
to other sources of energy. Experience rates of 95,90, and 85
percent were selected for exploratory calculations. The 95
percent rate has been characterized by Lindley [17] as con-
servative and appears to be so by comparison with other
authorities cited. The 85 percent rate may be described as
middle-of-the-road or optimistic, depending upon what
authority one chooses to accept.

Performing the Calculations. Calculations were made for
four different machine capacities: 0.3, 0.6, 1.2, and 2.4 MW.
The experience curve was expressed in analytic form:

cents/MJ =Anlog2b

where A is a constant dependent on the cost of energy from a
particular machine, n is the ordinal number of that machine in
the production run, and b is the experience rate expressed as
the cost decrease per doubling of production. The average
energy cost was obtained by integrating this expression over
the number of machines of each size required to reach each
capacity goal and dividing the sum by the number of
machines. For example, the number of machines of 0.6 MW
capacity needed to reach the capacity goal of 120 GW is
200,000.
As a starting point for the calculations, it was first assumed
that the envelope of cost minima for each windspeed
represented the cost of wind energy from the thousandth
machine; the first doubling would then be achieved at the two-
thousandth machine. Then the ordinal numbers (the "serial
numbers" in a production run) shown in Table 1 were
assumed for the production run called for by the energy
target, and the experience rate of 95 percent (that used in the
Kaman analysis from which the envelope was obtained) was

NOVEMBER 1981, Vol. 1031309