DESIGN FOR MANUFACTURING

for Productivity•...............ons

Reducing the number of aproduct's components andits manufacturing stepsstar ts in the designphase.

RUSS GAGERTechnical Editor

Design ing for product ivity is a pro­cess which combines both design andmanufacturing disciplines. A sere,...'may be a quick and easy solut ion tohold two assemblies togethe r for adesigner , but for the manufact ur ingdepart ment , dr iving that screw forhundreds of t housands of parts in­stead of mak ing the asse mbly snap-fitcan cost far more than the design timesaved.

Simplifying an assembly by reduc­ing t he number of parts can also pro­vide ast ronomical savings for a com­pany mass producing prod ucts . Manytimes , even th e costs of additionaldesign t ime, ret ooling, specialty fas­teners. and new materials do not out­weigh the savings generate d by themore efficient design.

Whet her manual or automatic as­semblv is used is many t imes lessimportant t han t hat th~ analysis tosimplify the assembly was done. Somecompanies have simplified an assem­blv, done a cost analvsis on automat­ing it, and decided to"stay with manu ­al assembly. But t hose cost sa vingsfrum simplificat ion of t he assemblyremain, and sometimes are even larg­er because there is no addi tional capi­tal expense from t he automatedequi pment to payoff.

The basic princi ples of Design forProductivity were outl ined by Geof­frey Boothroyd and Pete r Dewhurstof the University of Rhode Island (for-

merl y of t he University of Massachu­setts). Booth royd and Dewhurst havedeveloped a soft ware system for IBMPC, XT, AT and compatible personalcomputers which encapsulates t heirDesign for Assembly Handbook. A-I5-minute instruct ional videotape isalso available.

The handbook and software detail asvstern which uses numbers to eva lu­ate whether a product is being assem­bled most efficiently . The seven soft­ware prog rams operate by ente r ingt he names of a product 's parts on t hescree n and answering basic quest ionsabout each part.

As each quest ion is answered , aworksheet is automatically created ont he computer scree n. Complete analy­ses for t he basic assembly systems arepri nted out after t he quest ions areanswe red .

Boot hrovd and Dewhu rst are alsowork ing on a Design for Elect ronicAssembly handbook to analyze t hemost efficient meth ods of assemb lingcomponents on a circuit board. Anoth ­er way of speeding t he desig n forassembly ana lysis is also being inves­t igated. This met hod would includethe soft ware in CAD/CA~I systems sothat ult imately t he manufactu r ingcost of a product 01' assembly could bedeterm ined by pressing a button witha minimum of input s.

DFA analysi s used extensivelyMany companies are using design forassembly analysis regular ly. Hitachihas been teaching a method for years.Xero x and IBM teach its employeesthe entire Boot hroyd/Dewhurst sys­tem .

IBM offers a two-dav se minar unDesig n for Assembly methodology atregular intervals at its IBM Corpo­rate Quality Insti tute.

"\Ve st ress the idea t hat success fulassembly automation hinges uponpurposeful design for automati c as­sernblv." says Morris Krakinowski, asenior"enginee r at IBM.

A major project at IBM involveddesign for asse mbly evaluat ion of1,300 drawings t hat make up 50 sub­assemblies. A funct ional prototypewas deve loped and then analyzed formanufacturability,

U\Ve run the desig n analysis soft­ware on an IBM PC, which requiresno programmi ng s k ills, " saysKra kinows ki. "But if users are notcarefully introduced to t he system,the result s can he somewhat mislead­ing.

"The te rminology is simple , but themeaning of t he te rms are r igorouslydefined . For example, the mean ing ofa te nn such as 'par t is easy to handle'may appea r self-evide nt , but it has avery specific meanin g in t he context ofth e analysis system. That is why weuse th e t wo-day workshop to intro­duce users to t he subject and givethem enough of a working knowledgeof t he system."

General Electric has been conduct­ing Design for Assembly workshopsthroughout t he company. They haveresulted in reduct ion of part s require­ments by an average of20 percent andcut asse mbly labor by an average of ~Opercent. Handling and inventory costshave also bee n redu ced due to theimproved designs.

The 2~-da\' sessions are led bv twoDFA consult~nts from GE , The" firstday explains DFA evaluat ion andbreaks t he group of 20 attendees fromdesign engineer ing, manufacturingand quality personnel into small teamswhich each analyze two practice as­semblies.

On the second da v, th e small teamsanalyze their own produ cts, Then onthe third dav, t he teams brainstormideas to improve each product' s as­semblv,

At Black & Decker, the BoothroydDewhurst design fur assembly tech­nique has been used ext ensively. Oneproduct which has used it is a speciallydesign ed , high-st rength flashlight forfirefighters called t he Sun Lance.

Designing automated manu facturing for the Type 91 relayat Hamilton Standard Controls in Logansport , Ind . resu ltedin a Iw bstantiul reduction of direct labor. h i the photo at theupper right is sluJlcn robotic assembly ofthe magnetic subas­sembly and term ina! block subassembly. hi the photo at

bottom right is robotic assembly of the niocab!e contact» andactua tor to the terminal block w~sembly . Shown in the largephoto at left is final assembly rireting (~f the term inal blocka lUl magnetic subassembly. Behind the feeder bouls , term i­IInls (I re solde red.

Sequential stacking motions are usedfor this n"hirl/xH)1 mClium cleonamotor assembly (the enclosllre of whichis sluncn being inserted on the line). Theassembly cOl/si."lts of the electronic con­trole. f onr screws, a mounting bracket ,the seal ami then the motor

The number of parts on the flash­light were reduced from 30 to 10 andt he number of manufacturing pro­cesses dropped from 16 to five. As­se mbly t ime was reduced by 60 per­cent. Among t he techniques usedbesides design for assembly tec h­niques were eliminating adhesivebonding and solder ing.

Hamilton Standard Cont rols, a unitof United Technologies, has also de­signed many of its products for pro­d ucti vit y usi ng t he Boot hr oydDewhu rs t app roach. Besides t heirelectronic cont rols, the company hasalso used desig n for product ivity int he manu facture of two relays.

Direct labor on t he Type 91 relay

was substantially red uced by design­ing it for automation. There was nonet reduction in parts or assemblyopera t ions, but some of t he metal fab­rication processes were changed sothey were more consistent. The use offeeder bowls, pick and place ma­chines , and automated tes ting re­sulted in t he labor redu ction. Within

Xerox, IBM, GE, Whirl­pool and Black & Deckerhave used design for pro­ductivity extensively.

four months, the line was running at70 to 80 perce nt of its rated out put.

The T~'I'" 1:14 re lay was designedusing the knowledge the company hadgained by redesigning th e TYI'" 91product ion line. The covel' and base ofthe 1:14 relay snaps toge ther insteadof using screws, and fly w indingequipment , which is new to the com­pany , will be used to wind wire on therela v's bobbin.

The st andar d wind ing met hodwhich had been in use involved put­ting the re lay's bobbin on an arbor androtat ing it at high speed. But since the

184 re lay has terminals which are longrelative to most re lay bobbins, t herewas the possibility that t he cent rifu­gal force of rotating the bobbin wouldthrow the terminals out of place. Butwith the fly winding equipm ent , t hismishap won't be poss ible because thebobbin will be stationa rv and the flverwill rotate the wire around it. •

The company also has a proprietaryway of maintaining the critical dimen­sion bet ween the magn et core and thepivot point of the armature , A patentis being applied for it.

Designing vacuum cleanersWh en desig ning its new vacuumcleaner line , Whirlpool used seve raldesig n for productivity met hods in­cludin g th e Boothroyd Dew hurstmeth od. One meth od used was classi­fying parts by their material proper­t ies and de si gni ng t he m to bemulti functional.

An ex ample of this is the wrapa­round bumpe r on the canister vacuum

cleaner. It serves not only as a bump­er , but as a seal for the dust bagcove r.

The motor mount, which is made ofan injection moldable thermoplast icrubber , also se rves as a vibrationaldampe ning mount and as a se al toseparate the two vacuum chambers ofthe machine.

Another produ ctivi ty meth od wasto class ify part s by production toolingprocesses by incorporating more fea­tures into one mold. The number ofparts and molds were reduced in sev­eral locations of the canister vacuumcleaner.

A new sw ive l caste r design reducedt he number of parts from 20 to foul'.The base of the caniste r includes th esupport for the real' wheels and isretained by snap catche s. The fronthandle on the base also se rves as abumper support, and the base in­cludes the cord reel support.

The net benefit of these sophist i­cated too ling ideas was that the molds

Difficult to assemble Preferred

Thi s slot would be hard to detect

The parts in the 100rer IV/(' were rede­signed fur ,'iy m met r y .'iO the poes lblenumber (~" o,.ielltat iol1s was kept loU'.

Sett-tix turina (~" IXl ,.ts iritl. notches orgu ide pillS mn ('(Ise assembly .

This part could be orient ed in any direction.

These parts can be oriented only one way .

Hole to accept notched partHole to acceptswaged part

~~~~v~"D'tsbaped hole/

Chamfer to help orient s lot

External teatures C(l1I be added 10 ori­ent !HII'fN.

If par t» C(J n't be des igned s yUlllletrica lly , their as ym met ry should be inc reased IW

they CUl l be orien ted ollly one way .

The redesign un the bottom /('(lSone of three done by ,NCR Corp. for (l paper guideassembly in a teller machine. It "educes the total tlumber OflXlr txfor the (lssemblyfrom the ./.J shown on top to sereno

The est imated asse mbly time is atheoretical minimum assembly timeplus t ime penalti es for problems inha ndling and inserti ng as semblyparts. If a part is eliminated , theasse mbly t ime for that par t is zero.

A genera l rule of designing for as­sembly is to avoid faste ners, especial­ly non-standard ones. But when theyare necessary , the same type of fas­te ner should be used. If washers arenecessary, they should be attached toa bolt or screw ("captured") ratherthan se parate , Also, tapped holesshould be avoided and self-threadingfasteners subst ituted.

In manufacturing, "pancake" as­sembly, in which parts are layered ontop of each other along the Z axis, canbe used not only to simplify assemblymotion, but also to reduce the numberof fasteners. To provide compliance or"give" in part assembly, so accuracyer rors can be forgiven, chamfered orbeveled part edges are helpful.

Simple mechanical processes shouldbe used in manufacturing when possi­ble. Difficult ones such as welding orbrazing , soldering, and certain adhe­sives should be avoided.

Nesting a part in an indentation inthe work surface which matches itscontour can eliminate clamping thepart down during assembly. Cableswhich must be connected, as on aloudspeaker, should have connectorsattached to them so the cable end canbe located. Electronic componentsshould not be connecte d to a circuitboard wit h ca bles, but rathe rmounted on a slave circuit board andthe slave board inserted into the maincircuit board.

Other design for assembly ideas areshown in the accompanying diagrams.

oare pieces of fixed automation-theyredu ce the number of assembly opera­tions.

The third productivi ty method usedwas to design the product for assem­bly efficiency. Lineal' motion is usedextensively in the asse mbly of thevacuum cleaner. For example, themotor was designed to use st raightline assembly motions.

Included in the motor asse mbly isone of the electronic controls, foul'screws , a mounting bracket, the sealand then the motor. These compo­nents are all a.ssembled in a sequentialstacking operat ion where the net ben­efit results from a simplified and moreefficient assembly method .

DFA rulesThe basic pr inciples of Design forAssembly relate to two areas : theease of component handling and as-

sernbly of a product or subassembly;and whether the minimum numbe r ofparts have been used for a product orsubassemblv.

Th e criteria Boothroyd andDewhurs t established for the mini­mum number of parts test was:• Does the part move relative to itsmating part during operation or serv­ice?• Do the parts have to be of differentmaterials?• Do the parts have to be disassem­bled during asse mbly or service?

The assembly efficiency is then cal­culated by dividing the theoreticalminimum assembly time by the esti­mated assembly time . The theoreticalminimum assembly time is assignedby the Boothroyd Dewhurst met hodaccording to the product. The timevalue is for ideal parts and is standardfor all parts.

Reprinted from APPLIANCE MANUFACTURER January. 1986© 1986 by CAHNERS PUBLISHING COMPANY