AFWED FOR4 CU HISTORICAL REVIEW PROGRAM
TITLE: Production At An Aircraft Plant
AUTHOR: Randolph Payne
VOLUME: 6 ISSUE: Spring
STUDIES IN
INTELLIGENCE
A collcclion ol articles on tho historical, operational, doctrinal, and iheoreilcat aspects ot intolligonco,
All staiements of fact, opinion or analysis expressed in Studies in Intelligence are ihose of
the authors. They do not necessarily reflect official positions or views of the Central Intelligence Agency or any other US Government entity, past or present. Nothing in the contents should be construed as asserting or implying US Government endorsement of an article's factual statements and inicrprctations.
7
The standard date, required for estimating numbers of air-craft producedoviet plant are the model of the plane and the approximate weight of Its airframe, the floor area of-the plant and the number of shift* It works, and thedate when production of this model began.imple
t calculation from these data rests upon average figure* that have been empirically derived fornd transports respectively expressing pounds of production as a
i function ot plant floor space: one need only multiply tbe floor area by thla factor, adjust for number of shifts worked, divide by the weight of one airframe plus spare parts, and multiply by the number of months since production began to get the total number of units produced.
This method, however.omparatively crude one. in that It rollsingle average some variables ofrequired per pound of production, worker density on the production line, and man-hour* worked per month. In particular, it disregards the ImportantIn rate of production that Is always achievedlant gains experience inew model. The graphicof this phenomenonalled the learning curve. It Is the moat Important, versatile, and widely used of the many statistical tools employed by. aircraft Industry inplanning, and evaluating the production of airframes.
The Learning Curve
The formula for the learning curve expressesa persistent and well-defined relationship between the hours of labor expended directly on airframe building and the number of airframes produced. With each doubling of the cumulative number of airframes built, the man-hour*per pound of airframeercentage decrease
that is constant for any particular model. If the man-hour requirements decrease byercent when twice as many airframes have been built, the rate of learning is reflected in what Is calledpercent learning curve, because the direct labor required to produce the second frame is only SOof that required for the first, that for the fourth only BO percent of that for the second, that forth onlyercent of that forth, and soelationship of this type.^feataring successive powersas theproperty of appearingtraight line when graphed on logarithmic scales. Graphed arithmetically it appearsurve of constantly changing slope as in Figure 1.
This form of the learning curve, which shows the man-hours required to produce any one of the aeries of airframes, isnit curve. There can also be constructed a
latlve total curve lo ihow the cumulative total of man-hours required for all airframes up through any given number in the series. Both are illustrated In logarithmic representation in
The learning curve can be applied In estimatingoviet airframe plant with the assistance of four other standard curves also derived from the Worldnd rxotwar experience of. airframe Industry. These represent:
Man-hours per pound of airframe required for production of the initial unitunction of airframe weightS).
Floor area devoted directly to productionunction of total covered floor area
Square footage per direct worker on the largest shiftunction of airframe weight
Flow span (number of working days and cumulative man-hours expended from start of fabrication to Initial flight) for the initial unit of production
Crate and Coot
The refined methodology based on these curves is ina matter of calculating the man-hours expended per month against the man-hours required for the Initial unit of production and then, by application of the learning curve, for subsequent units. It can be Illustrated by an account of how
' The equation ot the unit learning curve Is
where
redirect man-hour* required to produce airframe unitan-hour* required lo produce unit numberny number ol unite produced, and
ope (tangent) of the learning- com (log of the percent of the learning curve, expressedecimal, divided by
The cumulative total earn la closely aproxtmated by the definite Integral from Ute first unit minus one-half to the last unit plua one-halt The in leers, doe gives lb* equation
otal direct mm hours required to produce an unite through unit X. the other symbol* remaining as above.
It was used to estimate the production of the Soviet transportCrate)Coot) at Moscow Airframe Planthis estimateomplex example in that It covered the phasing of the Crate aircraft out of production and the phasing of the nev and much larger Coot in. Inhen the estimate was made, the followingpertaining to the problem was available.
The total covered floor at Plant No.asillion
square feet. The plant was working two shifts per day. The Crate airframe weighs0 pounds. The first series-produced Crate was probably completed in
Inccording lo the reportisiting foreign delegation, the plant was tooling for series production of the Coot aircraft,rototype of the Coot was being tested.
In7 Crate aircraft continued to be assembled while work on the Coot was in progress. It was reported by other foreign visitors.
In7 the new Coot prototype was seen
The Coot was displayed publicly for the first time
The Coot airframe weighs0 pounds.
The first series-produced Coot was probably completed in either August orneptember,tot successfullyestonservative estimate sets the start of
That average parameters of production might be on the low side when applied to Plant Noas indicated by the fact that it had received many awards for production andeople had worked there for more thanears. On the other hand, negative allowances would have to be made for handicaps to production of the large Coot imposed by constrictions In Plants final assembly building.eetide, andigh, this building was known to be Interrupted by rows of columns, one of them running lengthwise through the middle of the building. The doors were only wide enough to accommodateoot wing span of the Crate; the Cooting spanoreign delegation visiting the plant8 reported that several chords and other portions of the roof trusses had been cut out to allow the tall fins to protrude above the soffit of tbe lower members. Openings had also been cut into one outside wall, with lean-to's built around them, to accommodate one of the wings, the other being allowed to protrude through the row of columns that ran down the middle of the bay. The Coot fuselage was thus rendered Immobile until the aircraft had been assembled completely, and then tbe outboardof the wings and the vertical tall assembly had to be removed in order to get it outside.
Crate Production
The application of the established parameter for man-hours required per pound of production according to total airframe weight (see Figureave an estimate that the first Crate airframe could have been produced Toran-hours per pound. But Plant No.as here following In the footsteps of Tashkent Airframe Planthe first to produce the Crate, and-USWaU indicate that the second builder*of the same aircraft requires for his Initial unit of production only
ercent of the man-hours needed by the first. Thus Plant No. SO was estimated to have produced Its initial airframe atan-hours perotalan-hours forpound Crate. It remained to estimate the monthly rate at which these man-hours were expended.
The relationship between total covered floor area and that devoted directly to production (seepplied torulllon Muare feet ofroductive Boor1 areaquare feet. The relationship between square
^fitter
Aircraft Production
footage per direct worker and airframe weight (see Figurehowed for theequirementquare feet per worker on the largest shift, which could thusInquare feet. On the assumption that the second shift wasercent as large as the first, the total number of direct workers was estimated to.
These workers were each assumed toours fire days per weekours on Saturday.nnualthere wouldour andour days,otal
r
i
ours, per year. an's average working time In one month would thenours, which, however, wasbyercent factor for direct workersirect work, on sick leave, on vacation, and on rest periods, toours per month per worker. Multiplication byenan-hours per month expended on the Crate.ercent effort devoted to the production of the nuuvhours per.mon.th remaining fcjr^urdtair-uctlon figured. Dividing into this monthly expenditurean-hours calculated above to have been required for the Initial unitatio of effort expended to effort requiredressed In percentage. That Is. the Initial unitroduced Int the rate of aboutnits per month, and the rate for succeeding units would Increase according to the learning curve. The average learning curve for transport aircraft in the vs. aircraft Industryut In estimating Plants production of the5 percent curve was used because of the manyworkers at the plant and because of IU past production record,
A schedule of Crate production could now have beenby accumulating the monthly expenditureof first-unit requirements and converting this Into units of production In accordance with5 percent learning curve. But6 production of the Coot had also beenand the full manpower In the plant was not being used on the Crate. An estimate bad first to be made of tbe manpower requirements for production of the Coot In order toigure for the remaining manpower available for work on tbe Crate.
Coot Production
The initial unit of series production forpound Coot airframe would, according to the curve of Figureave requiredan-hours perotal ofan-hours. This figure was first used to determine when work on the Coot began through the correlation between man-hours required for Initial unit production and working days from start of fabrication to initial flight shown in Figure 6.
i
Flow Spin for Manufacture of an Aircraft from the Start of Fabrication to the Initial Flight, for the Initial Unit of Production |
an-hours expended on the firstaircraft would, on the basis of experience in theIndustry, meanorking days werea working dayours to allow for the shortandours per month calculated above forthe total elapsed time for the first Coot wasbeonths; but It was decided to increase this tobecause of ^eAanqUcar^<underf| which-theIt was established in
our data that the first unit was completed inut the start of fabrication sometime in
Production of the Coot would have utilised
j square feet per worker on the largest shift, according to the worker density curve of Figureith the productive floorquare feet, previously calculated for Plant No.nd again assuming the second shift to be atercent the strength of the first, the total number of direct workers
i was estimated to be
540
This figure was rounded offirect workers andas beforeours per month tooan-hours for the monthly expenditure of effort The allowance of
ercent for production of spare parts leftroductive man-hours per month expended on Coot airframe
1 units.
The monthly expenditure of man-hoursercentage of
I
| man-hours required for the Initial unit then equaled
ate of production for the first unitnits per month.
The average learning curve5 percent, which had been sharpened for the Crate, was slacked off toercent InCoot production to compensate for the handicaps under which the aircraft was being assembled. Using this curve to convert accumulated monthly expendituresercent of first-unit man-hours Into units produced, the schedule of production shown inas compiled.
The Phasing Out
The device used to detennlne the number of man-hours that were devoted each month beginning with February lBSfl to the Coot and therefore the number of workers out ofotal remaining available for Crate production was the Coot work-in-process curve shown In Figureere the expended cumulative percentages of first-unit man-hours represented by deliveries were plotted by month from7 delivery date of the first completed unit. These form the straight curve "Completion ofhen each unit of production was plotted horizontally to the leftistance representing the lapsed time from start of fabricationonths for the first unit and reducedearning curve for succeeding units. These form the "Leading Line." Perpendiculars were drawn to this curve from the
Compleuoo of Production line,hird cum ww perpendiculars and arbitrarily
faired from the origin to the first perpendicular. This is theroceas curve.
This curve determined by month the cumulative percentage of nrst-unit man-hours devoted to Coot production, and the difference between successive percentages wu the percentage expended each month. The latter were multiplied by the number of man-months required for the Initial) to give the number of workers absorbed each month by the Coot. These were then subtracted each month7 if8l^otT^'^-lcuutlon
ii
SHSiHSiHI
8
I
lilli*
=
<'
fig
r
Production
mmmm mm mmmm mm
sa
-
8 =IBim
yi|Siill!
Validation
information received since these estimates were madethat atrate aircraft were produced at Planthis figure compares well with the totaln Table n. The estimate of Coot production, however, was less accurate; It was learned later that the second work shift at the plant was probably discontinued inateoots p- r. nvinth. not in8 as estimated, but only In October As8 the total produced Is believed to have beennstead of theiven In Table I. and production remained constant at the rateer month
These results show that successful application of ais dependent on how well the seasoned Judgment of the analyst can cope with the Imponderables and on the accuracy of the Intelligence information used in the calculations.under any methodology, although useful in the absence of more direct information, never preclude the need for concrete and reliable Intelligence on the production of Soviet aircraft.
Original document.
Comment about this article, ask questions, or add new information about this topic: