235

Research

Learning to Apply Interactive Graphics in Computer-Aided Design - Some UK. User Experience ’

Erik Arnold 1. Introduction Srwncr Policy Reseurcit Untt, Mutttell Butlding, Unicrrsr~v oj

S~c.~se.u. Fdntrr, Brtghtun BN 1. UK New CAD users, like new users of other tech- nologies, normally fail to take adequate account of the learning processes involved in technical change. In particular, the importance of learning by management is underestimated. Training needs and payback periods can be underestimated as ti resuh. This paper analyses data obtained during struc- tured interviews with 34 CAD user firms in the UK engineering industry. Supplementary informa- tion was obtained through discussions with con- sultants, bureaux, CAD suppliers, and trade un- ionists.

Rapid diffusion of interactive graphics became possible in the UK in about 1978, when American CAD technology was marketed in ‘turnkey’ form. Interviews with CAD suppliers and 34 CAD users in the UK engineering industry yielded information about the learning processes which new users have to undergo. The dlfrerent learning processes of operators and of managers are discussed, and conclusions are drawn about way.< of easing the processes needed for the adoption of CAD.

Kavw~r~~: Computer Aided Design, CAD. CAD/CAM. Inter- active graphics, Learning. Two levels of learning are distinguished here:

operator learning; and management learning. Three distinct stages in management learning are set out. These chart new CAD users9 progress

I This rcsearcb was sPo;isored by the Engineering Industry Training Board. iiowever. the views expressed here are those of the author, and do not necessarily reflect the views of the Board. The support of the Board, the helpful co-operation of the people and organisations involved in the study and helpful comments by Professor E.H. Siblcy and anonymous referees on an earlier version of this paper are gralefully

acknowledged.

North-Holland Information & Management 6 (1983) 235-240

Erik Arnold holds a BA in English Literature and an MSc in History and Social Studies of Science, both from Sussex University in the UK. He has worked in market research and as a computer programmer with AGB Re- search. Since 1980, he has been a Fel- low of the Science Policy Research Unit, Sussex University. rIis principal research interest is the relationship be- tween technological change, corn’ petitiveness, labour force skills and employment in the engineering in-

dustry. Mr. Arnold is a committee member of the British Computer Society’s specialist group on computers and employ- ment.

()378_7206/83/$3.00 ! 1983. Elsevier Science Publishers B.V. (North-Holland)

towards best practice. Ccnclusions are drawn about wavs of easing the iearning processes involved in unroducmg CAD.

2. The Importance of ‘Turnkey’ Systems

Interactive graphics CAD programs were devel- ripcd at the Massachusetts Institute of Technology and at General Motors as long ago as 1963. How- ,*ver. these programs - and CAD graphics tech- nt,Iogy developed in the UK at the end of the I%% -- relied on expensive mainframe computers. fhree important innovations occurring around I97tj paved the way for CAD systems at prices tc-hrch medium and small sized firms could afford. fhe\c innovations were:

the ‘storage* tube, providing a cheap flicker-free drsplay with modest computational require- ments IQ-hit minicomputers. offering relatively power- ful computing at a modest and declining price v rrtual memory and structured programming techniques, allowing large programs and data sets to be written for, and handled by, small computers.

Wtth these innovations. new companies were able tt~ produce and market CAD hardware in the US tn a packaged form and at an acceptable price.

in the 1960s. CAD users were obliged to build much. if not all. of their own system. The packaged ‘turnkey’ systems provided CAD to users who did ngbt know - and did not necessarily want to know

how to buitd a CAD system. Their need was for d C’AD system they could buy, install, and then u~c. In the second half of the 197Os, leading US turnkcv suppliers were growing at spectacular rates 4 up to 100% per year).

OutGde the microelectronics industry. CAD had BrttIr impact in the UK before 1978. when Ameri- LXX suppliers began marketing to the UK. Most 1 K u.ms experience. therefore. relates to the .ra?;iqxion and use of tuakey products - primarily * J I3 origin -- in the period since 1978, although .kn rmportant minority of systems now stem from I& % :K itself.

!&u-r) UK firms. like their US counterparts, are k f cdtcd h, top managements as series of ‘profir

centres’. Turnkey CAD systems, which are marketed to design and drawing offices, provide peculiarly appropriate investments for such firms, where the investing profit centre (the drawing office) has also to provide investment returns. Senker has explored in some detail the implica- tions of investment appraisal methods for the diffusion of CAD/CAM technology [l]. Clearly, however, the need of the investing profit centre itself to provide returns militates against Computer Aided Design/Computer Aided Manufacture (CAD/CAM) technologies which sacrifice returns in the drawing office to gains at other profit centres in the firm. This, in turn, is not conducive to the increasing interconnection of computers be- tween design and production and the correspond- ing organisational changes which, from an en- gineering point of view, would be desirable.

3. Operator Learning

The rate and extent of learning among opera- tors in our sample varied considerably. Some CAD systems appear to be slower to learn than others, according to the ergonomic qualities of the man/machine interface (both with respect to hardware and software). One supplier’s equipment was particularly hard to learn.

Individual aptitude was also important. Very experienced users said that CAD tended to widen the productivity gap between drawing office per- sonnel. Using manual techniques, one draughts- man might be two and a half times as produr ive as another, but CAD doubled the disparity. Neat- ness became irrelevant, making some untidy indi- viduals more useful. But certain users felt that CAD, like computer programming, requires a certain indefinable aptitude, correlating neither with age nor qualifications. Willingness to experi- ment and inquisitiveness were important aspects of this.

Most users found that younger draughtsmen more quickly learned to operate a CAD system. Some said that older draughtsmen later caught up with an surpassed the younger ones, presumably because of their superior engineering skills. Others found some older draughtsmen unable to adjust to

E. Arnold / Learning to Apply Graphics 237

using the machine as a system, instead simply using it as an electronic drawing board. In certain cases, this extended even to the use of projection lines and reference points, which should have been superseded by use of the machine.

Operator training invariably consisted of one or two weeks in retraining engineering personnel, normally draughtsmen, often followed, after some time, by specialised training. Two-thirds of the interviewed users trained new operators in-house, after some initial training of key personnel by CAD suppliers.

In general, operators took between one and three months to equal their manual productivity. Productivity of 3 : 1 or above tended to take 9-12 months, although these figures were subject to considerable variation. Fig. 1 shows an idealised version of an individual’s learning curve, based on information from one of the vendors. It accords well with the experience of the users we inter- viewed, although it is stressed that the numbers attached to the Figure are purely illustrative.

The first phase of operator learning involves the basic ‘driving’ skills. The second entails complex and grouped commands which instruct the mac- hine to perform a sequence of operations. At this point, a change in the operator’s conceptualisation of drawing work is needed: there is no longer a simple correspondence to the way things are done manually. The third phase requires an apprecia- tion of the machine as a computer system: to use the database efficiently and to cope with some

Productivity mtio

WI@ Of

3:l

a:1

1:l

lh8iC oommradr

3 month8 6 monthm 9 moattlr 12 montha

Fig. 1. General Shape of Operator Learning Curve.

computer programming concepts. While moving from one phase to the next, there is a loss of productivity as the operator explores and experi- ments with new facilities.

Management learning, however, was an im- portant influence and constraint. High operator productivity depended upon appropriate dal.a- bases and standards being established. One firm found at first that it took two to three months to bring a newly trained operator up to manual productivity. Two years later, how -ver, through improved specification of procedures and stan- dards and v ith the help of ‘menus’ of specialised instructions, the time has been reduced by about two-thirds - to three to four weeks. At other establishments, failure by managers to learn this had restrictured the rate at which operator produc- tivi ty rose.

4. Management Learning

Technical change in the drawing office has tended to be slow and incremental. CAD fre- quently represents managers’ first experience at coping with radical technical change and com- puters. Capital invested per worker has been low.

Important decisions and judgements have to be made prior to gaining experience with CAD, and these can prejudice system performance. Since most managers only gain CAD experience after buying and installing a system, management learning hap- pens in the traditional manner (the hard way).

Thus, in the absence of CAD experience, managers are unable to make authoritative esti- mates of the productivity increases which can be achieved. This undermined the bases of many cost justifications put to top management. Training and learning costs were rarely considered. 21 of the 34 engineering users had justified CAD on the basis of anticipated savings in draughtsmen’s labour; in two cases cost-justification had been done purely for form’s sake. Several respondents doubted their own cost arguments! It was not clear whether this was because of mistrust between en- gineering and financial personnel (accountants’ demands could not be met with hard evidence prior to gaining experience), or a sense that the

WI heneflts were unquantifiable. Five firms had llltt Cohi-Jusilfied CAD at all. because it was seen .f. m~pr,tcllcable to design products in their busi- W+ ulthout C‘AD (see Table 1).

$\trm selection was also difficult because mrtnagsrs without knowledge of CAD are not yuahfted to select the best system. However, they r;.mmtl gain CAD experience without a system. I&u\. many selected the market leader’s product ctn the assumpiion that the confidence of the rar+rrlt> ctf other purchasers implied that it was thr best, and hec;,use of highly t:redible after-sales W_WWC’ and *hand-holding’. Data processing staff c~rulti rarei*, offer much help in the selection pro-

ct’~~ hec,tu~c of their lack of engineering knowl-

r&u 44’ rhe firm One manager found a 2% yloc~t~n yuchllonn.Gre in a magazine, which he ii~d iti ’ puke’ vend - ors. He said that he did not aurdsr~tand some of : he vendors’ replies, and that

hlr crftcn did not understand the questions either.

fndu+lnal relations can be an important stum-

krlsng bloch rn (‘AD tmoduction. Gven the slow- IW+Y ctf technical change 1‘1 drawing offices to date, 1~9th managers and union t,ranch members found

tttc rndustrtal relatlc:ns of new technology novel.

i hc ensuing lcarnmg process was sometimes pain- luI (4% would be expected. early consultation often, hug nrlt alu.a\s, made rntroduction easier. Both 4_++ nctrmall~ \aw (‘AD as a good thing. and

UMI~\ rrr.trr+er\ urre surprised at how easily &try

f-A ‘g&t1 C*Al> pasr (he unions’. In two firms.

qurprnent had been ‘blacked’ for a time. In retro- *r)r~*i. managers sard that they had found (his

hrnri~r~l fi)r learning and experimentation. The %ktuJt1on had removed the pressure to cope simul-

TdliCi”U4~. u~th lcarnrng about the system and its

tf~.. 14~ da\ management and the pressure from

Wn:a v mm+er\ for return on the CAD invest-

IsWnl

During discussions with unions, managers had to decide whether to attempt to get agreement on multi-shift operation. Twenty-four of the users operated a single shift system, and a further three ran an extended shift with elements of flexitime and staggered equipment use. Only seven used the system on a two-shift basis. Three of the single-shift users intended to move to two shifts, and one of the two-shift users (the subsidiary of a US com- pany) intended to move to three shifts. Shift oper- ation reduces the payback period of the equip- ment. Payback periods in the UK are typically longer than those in the US, where two-shift oper- ation is the norm.

The question of access to the Equipment was important. In most cases, the initial CAD installa- Lion had very few terminals compared with the number of potential users. Because proficient op- eration required a large amount of continuous access, there was a limit to the number of people who could efficiently use a system. As a result, CAD training for more than a few operators was tokenistic. Operators working full-time at a termi- nal tended to be more productive than those mak- ing occasional use, but some users said that the engineering skills of full-time operators tended to atrophy, regarding this as a waste of scarce and valuable engineering manpower.

Often, CAD provided an excuse for a wider reorganisation of drawing and other activities. New drawing numbering procedures, parts listing and processing, and other procedures relating to archi- ving resulted. Reorganisation could clearly be more important still where attempts were being made to link CAD with computer-aided manufacture of mechanical parts by doing some NC programming using the system. Here it was seen that the tradi- tional division between design and production en- gineering needed to be eroded. CAD/CAM could be a catalyst to the breaking down of this. How- ever, only one of the firms interviewed was COII-

ternplating changes in lines of managerial respon- sibility to facilitate CAD/CAM.

5. Stages in Management Learning

CAD users typically passed through identifiable stages of development in learning to use CAD

E. Arnold / Leaning to Apply Gruphrcs 239

Table 2

Total Sample: Present Stage of CAD Use. by Date of First CAD lnstallatlon

Date ol First Present Stage oC CAD Use Claimed CAD lnstallat~on

Stage I stage 2 ‘Stage 3 Total Productivity

Ratio-Mean

Before 1970 Nil Nil 3 3 4:l 1976 Nil Nil 2 2 4:l 1977 Nil Nil I I 3:) I978 2 2 3 7 2.m : I 1979 2 3 2 7 2.5 : I 1980 4 4 2 IO 2.1 : I 1981 4 Nil Nil 4

TOTAL I2 9 I3 34

* Rased on the 24 lirms able to provide meaningful estimates of operator productivity compared with manual draughting.

cquipmem. These stages are defined in Fig. 2. Table 2 shows the stage of the development re- ached by establishment sampled, classified by the year in which c’AD was first installed. As would be expected, the mean productivity ratio was greater in the older annual tranches, symptomising learning in the sample as a whole. Table 3 cat- egorises the reason for delay B’y individual firms in reaching Stage 3 which it was possible to identify in discussions with them. We split the sample into

three broad sectors: vehicles and aerospace; mechanical engineering; electrical and electronic engineering.

In vehicles and aerospace, industrial relations were an important source of delay. This tended to be the case in large firms with a poor industrial relations history. Paradoxically, some of these firms were declining and had most urgent need of the competitive advantages to be derived from CAD. In-house technical and organisational competence

Table 3

T:>I;II Sample: Rcnhons for any Delays in Reaching Stage 3. by Sector

Reasons for Delay Sector

Vehicles MLwhanieal

and Aerospace t“‘ibil .ccrmg

Electrical

and Electronic

Total

Induslrirrl

Rclatinrrs

Problems 9 2 Nil 7

Supplier Failure

(in particular

to dclivcr

dtwiw I 3 I 5

Manugcrial

lnrlliciency NII 6 NII 6

No ldcntifiuble

D&y 4 4 s 13

svs~cm Dchvercd

Recently Nil 3 2 5

TOTAL IO IX * 8 36 *

* Adds to more than number of firms in sample because both industrial relations problems and managerial inefficiency were

significant in two establishments.

Stage 1 CAD system delivered, installed, and used for experimental purposes. Operator productivity may be quite high, (e.g. as much as 3: l), but not on ‘live’ production drawings of components or products.

stage 2. CAD system used for production drawlrigs, but not (yet) very efficiently. Productivity increases typrcallyl:l to2:l.

Stage 3 Use of CAD at typical state of the art, e.g. 3: 1, 4 1 productivity improvements. May be some in- terest in CAM, but not yet much activity.

~a.4 normally high in this sector. and little delay therefore re:u!ted from technical or organisational *eraknesses. Some of the oldest installations were In this sector.

Managerial inefficiency was more important in mechanical engineering, where limited technical competence in relation to CAD had also allowed users to become vulnerable to supplier failures. Some small suppliers provided incomplete soft- uare systems, and users found themselves waiting qmtr some time after hardware had been installed. Industrial relations was a less important delaying factor than in vehicles and aerospace, but did prolc something of a problem in three larger firms.

E.lectrical and electrictronics establishments, \srth one exception. experienced no undue delays III rcachmg Stage 3.

6. Chclusions

I IX rate and qtktlity of learning is an important &_*tcrmmant of the usefulness of a new technology

such as interactive graphics CAD. The time taken to achieve best practice performance levels can be considerable; although it is to be expected that this will fall as knowledge about. and experience of, CAD becomes more widespread. Best practice performance levels may also rise. Most users did not take sufficient account of learning time in their cost-justification of CAD, and many appear to have underestimated their training needs.

Prior to installing a system, managers need access to expertise about CAD - both technical expertise for evaluating rival suppliers’ equipment, and expertise relating to the organisational impli- cations. Consultants could play a more important role. As the cases which had industrial relations problems show, CAD introduction is more than a technical iss:le. Managers need considerable scnsi- tivity in introducing a new productivity-enhancing technology to workers who may fear that their jobs are threatened.

As CAD becomes increasingly linked to CAM in future, managers will need to cope with organi- sational change linking design and production more closely. This appears likely to extend further their need to acquire a broader competence to cope with technical progress, and top management will need to learn to make decisions about new

forms of organisation within the firm as technol-

ogy changes.

Refercnec

1 l] I’. Senkcr, “some Pmhlcnrs in Justif@ta CAD/CAM”. in proccodings of AUTOMAN ‘83. 2nd Ruropcan Automuted Menufucturin~ Conference, 6th Annual Rritish Robot As- srktion Cnnfercncc. I& I9 Mary 1983. Rirnringhum. UK.