Making a Multiplant Supplier Rating System Produce

9
MAKING A MJLTIPLANT SUPPLER RATING SYSTEM PRODUCE Paul Jay Goldin Radio Corporation of America Camden, New Jersey Defense Electronic Products, a division of the Radio Corporation of America, instituted a supplier rating system at its Camden, Moorestown, and Los Angeles plants two years ago. This rating system was implemented with the following broad objectives in mind: 1. Concrete statistics would become available on the incoming quality levels of each type part (resistors, castings, etc.). '2. Sharp discrimination would be -possible among preferred, average, and unsatisfactory suppliers. DEP purchasing could channel purchase orders to preferred suppliers while withholding them from unsatisfactory suppliers. 3. Test and inspection costs could be reduced and realigned in purchased materials inspection (PM:[) through knowledge of trouble and nontrouble areas, and the elimination of substandard suppliers. 4. Since many suppliers service more than one DEP plant, comparison of a sup- plier's rating at each plant would be possible. 5. Quality control field representatives located throughout the country could make their plant visitations on the basis of factual inspection data plus delivery schedules, rather than just on delivery schedules. 6. Supplier test and reliability data could be used as a basis for PMI acceptance. This report will indicate the progress made in attaining these objectives, as well as the concrete improvements, innovations, and cost reductions made pos- sible by the rating system. The mathematical theory and machine mechanics of the system will not be discussed, since they are similar to those used throughout the industry. INCOMING QUALITY LEVELS Three measures of incoming quality are used by DEP plants. The first is the over-all process average for each class of parts (found by dividing the total number defective by the total number inspected). Detailed in Table I are the Camden plant incoming process averages for six electrical and four mechanical part classifications for two time periods. The first period represents the qual- ity levels for the six months immediately following the implementation of this system, while the second period represents the last six months up to September 30, 1958. These averages are calculated every six months and are used to indi- cate over-all incoming quality performance. The second measure. of inlcoming qualit;y is a further refinement of our first measurement tool. Within each parf. classification, the process average of each

Transcript of Making a Multiplant Supplier Rating System Produce

Page 1: Making a Multiplant Supplier Rating System Produce

MAKING A MJLTIPLANT SUPPLER RATING SYSTEM PRODUCE

Paul Jay GoldinRadio Corporation of America

Camden, New Jersey

Defense Electronic Products, a division of the Radio Corporation of America,instituted a supplier rating system at its Camden, Moorestown, and Los Angelesplants two years ago. This rating system was implemented with the followingbroad objectives in mind:

1. Concrete statistics would become available on the incoming quality levelsof each type part (resistors, castings, etc.).

'2. Sharp discrimination would be -possible among preferred, average, andunsatisfactory suppliers. DEP purchasing could channel purchase ordersto preferred suppliers while withholding them from unsatisfactorysuppliers.

3. Test and inspection costs could be reduced and realigned in purchasedmaterials inspection (PM:[) through knowledge of trouble and nontroubleareas, and the elimination of substandard suppliers.

4. Since many suppliers service more than one DEP plant, comparison of a sup-plier's rating at each plant would be possible.

5. Quality control field representatives located throughout the country couldmake their plant visitations on the basis of factual inspection data plusdelivery schedules, rather than just on delivery schedules.

6. Supplier test and reliability data could be used as a basis for PMIacceptance.

This report will indicate the progress made in attaining these objectives,as well as the concrete improvements, innovations, and cost reductions made pos-sible by the rating system. The mathematical theory and machine mechanics of thesystem will not be discussed, since they are similar to those used throughout theindustry.

INCOMING QUALITY LEVELS

Three measures of incoming quality are used by DEP plants. The first is theover-all process average for each class of parts (found by dividing the totalnumber defective by the total number inspected). Detailed in Table I are theCamden plant incoming process averages for six electrical and four mechanicalpart classifications for two time periods. The first period represents the qual-ity levels for the six months immediately following the implementation of thissystem, while the second period represents the last six months up to September30, 1958. These averages are calculated every six months and are used to indi-cate over-all incoming quality performance.

The second measure. of inlcoming qualit;y is a further refinement of our firstmeasurement tool. Within each parf. classification, the process average of each

Page 2: Making a Multiplant Supplier Rating System Produce

TABLE I

Incoming Process Averages by Part Classification

Per CentNo. Inspected No. Defective Process Average Dnprove-

g Jan.- Apr.- Jn -Apr.- ment orPart. June Sept. June Sept. June Sept. Deteri-

Classification 1957 1958 1957 1958 1957 1958 oration

Resistors 33,006 34,728 649 362 2.0 1.0 +50.0Capacitors 29,075 26,551 674 435 2.3 1.6 +30.4Diodes 15,786 17,789 525 204 3.3 1*2 +63.6Switches 4,967 2,403 187 105 3.8 4.4 -15.9Relays 22,611 17,574 934 612 4.1 3.5 +14.7Transformers 6,903 5,432 655 348 9.5 6.4 +32.6

Castings 6,145 5,105 148 110 2.4 2.2 + 8.3Gears. 1,437 1,292 39 20 2.7 1.6 +40.7Machined metalparts 6,714 13,790 315 341 4.7 2.5 +46.8

Sheet metalparts 6,351 12,537 457 634 7.2 5.1 +29.2

supplier is calculated every three months, published, and circulated throughoutDEP. We know, for example, that our largest supplier of resistors contributed55% of those inspected and only 15% of those defective during the April-Septemberperiod. On the other hand, one of our smallest resistor suppliers contributedless than 1% of those inspected arnd more than 33% of those defective during thissame period. It is concrete facts like these which provide our PurchasingDepartment with the ammunition to demand definite corrective action fromsuppliers.

The third measure of incoming quality, and one on which definite dollar val-ues can be placed, is the number of lots returned to suppliers, and the number oflots reworked at RCA. Several cost studies have indicated that the average costof returning a lot to a supplier on a material discrepancy report (MDR) is $25.This figure takes into account handling, paperwork, investigation and evaluationtime, and reinspection. The average cost of reworking a defective lot is $11 andis based on a labor cost of $5 per hour. Since the inception of this system, thereturn-to-supplier rate at the Camden plant has dropped from an average of 10.8%of all incoming lots to an average of 7.7%. Since Camden regularly receivesupwards of 40.,000 lots a year, this represents a reduction in returns of 1,240lots, or a saving of $31,000. At our Moorestown plant, the return-to-supplier*rate has dropped from 6.0% to 3.9%. Based on receipts of about 25,000 lots ayear, this represents a reduction in returns of 525 lots or a saving of $13,125.This reduction in returns has been accomplished while the number of lots reworkedhas remained constant.

QUALITY DISCRIMINATION AMONG SUPPLIERS

Every three months, each DEP supplier is given a concrete rating of between

0 to 100. Seventyr points of this rating are dependenlt on the supplier's process8

Page 3: Making a Multiplant Supplier Rating System Produce

average, while the remaking 30 points are allotted for delivery performance.Each supplier is classified according to the following schedule:

Class Rating Meaning

A 90-100 PreferredB 50. 89 AverageC 0- 49 Unsatisfactory

The suppliers are then placed in their proper classification by part typeand are listed in order of their numerical rating. At the present time, we have25 classes of parts ranging from motors to hardware. These listings are placedin book form and issued quarterly to purchasing, engineering, and quality controlpersonnel. Those suppliers who are rated as unsatisfactory for any three monthperiod may be called into the appropriate DEP plant for a purchasing conference.This conference serves to notify the supplier of his unsatisfactory rating andthe need for corrective action. It is our feeling that the calling in of unsat-isfactory suppliers is a more useful technique than sending a rating to each sup-plier. This is especially true with regard to our small suppliers, who greatlyoutnumber our large ones. In these cases, usually the president or vice-presidentattends the meeting and finds out firsthand why his company has been singled out.Since RCA's business withl these companies constitutes a considerable part oftheir total sales and means money in the pocket to the chief executive, theseconferences have resulted in the executive bringing pressure to bear on the manu-facturing activity to correct the discrepancies on future shipments.

In addition to the quarterly rating reports, a set of three books has beenpermanently distributed to key purchasing and engineering personnel and are keptup-to-date by quality control. The three books have blue, red, and black coversand contain what we call our blue-line or preferred, red-line or borderline, andblack-line or most unsatisfactory suppliers. Any supplier who has been rated forat least four of the latest five quarters, and whose ratings place him in one ofthese three categories is listed. Our Purchasing Department has been instructedby DEP procedure to take the following action on suppliers listed in each ofthese books:

1. Blue-line suppliers Should receive preference over any other suppliersin placing orders.

2. Red-line suppliers -- Must be called into the applicable plant for confer-ence immediately. New ordersmay be placed only as long as correctiveaction is indicated.

3. Black-line suppliers -- No new orders may be placed with these suppliersexcept in urgent situations. Orders may be resumed only when correctiveaction has been taken and verified by quality control field representatives.

Two of the most interesting cases which clearly demonstrate the vali.e offace-to-face conferences have just recently been conc'Ludedl. The first caseinvolved a sheet metal company supplying cabinets and related parts to our CamndenRand Moorestown plants. This supplier's quality was rated unsatisfactory at bothplants for a number of quarterly periods. When no correctivre action was takenfollowing a conference in Camaden, the supplier was notified that no more orders

9

Page 4: Making a Multiplant Supplier Rating System Produce

would be placed with him. We recently received a letter from the president ofthis company outlining an entire new quality control organization which he had toset up for the purpose of improving his product. Where previously the functionof the inspection department had been one of finding and repairing defects, thisreorganization made it possible to put a defect prevention program into effect.We have asked our- field representative to visit the company and report on theadequacy of its corrective action. If we are satisfied that this reorganizationwill produce the desired results, orders will once again be channeled to thissupplier.

The second case involves a large relay manufacturer with one central officeand several plants. We discovered that our corrective action requests and MDR'swere being buried in the central office and never forwarded to the manufacturingfacilities. The vice-president of the company took the required action to makehis plants aware of our findings, and this company is now one of our best relaysources.

REDUCTION AND REALIGNMENT OF INCOMING INSPECTION COSTS

Individual instruction cards for each part number is the key to our programfor reducing PMI costs (Figs. 1 and 2). Each card details the characteristics

NUWBER OF 03HARACTERISTICS CLASSIFICATION OF SAMPLING PLANCLASSITFIED AS MAJOR, MINOR CHARACTERISTICS TO BE USEDAND INC MME=L//

HAOS 0 ~~~~~~~~~~PAW?NO. SUL

|Relav, aratuare / |CA +21P-

_____ _____ _____ _____ _____ ____ ____Ref: TP-67-25loo.-5o3

C DCENTA1 PP6859R sub. 7

DATERAE. DAS

1 G. S. 1-20-55 Adew hi-Dot test.2 G. S. 2-22-56 evise item k to agree with vibration spec. PP6859P note C.3 O. S. L-25-56 Standardize characteristic numbering.l G. So 5-43-58 Reviewed to sub 15. General revision changing all items to A plan.

/_______REVISIONS DUE TO FACTORY CR PHI FAIIJrRESs GOOD) WRFORI4NCECRPLNGE OF SUPPLIER; E SUIB; ETC.

Fig. 1 - Purchased materias inspection (Phi) instruction card (aide 1).

10

Page 5: Making a Multiplant Supplier Rating System Produce

SPECIFIC CHARACTERISTICS THATAPPLY TO THIS PART NUMBER

/ ~~~INSPECTION INSTRUCTION 629482-3NO. CLS CHARACTERISTIC$ TOL - !2WMET1 AtJI Prkingsj including vendor's C stamnp Vlisual4 A Terminal finish and condition Visi-tal9 A Drowing di-mensions and mounting stud threads SMI, thread gage11 A Vibrate: 30cps .06" TE for i hour. Check for loosened TX 8332-503

or broken parts Holding Fixture17 A Hi-pot: Apply 350V R! 6Ocps between paralleled TX 8232-501

terminals and case for 5 seconds29 A Operate, rom temp. asing 15V DC energized coil volt- TX 8?32-501

age (REF TP-67-25100)27 A Operate, oold temp. using 18V DC energized coil volt- LT 8232-501

age (REF TP). Stabilize at -55 deg. C bhen remove Cold boc -55 deg. C28 A Operate, high temp. using 18V DC energized coil volt- TX 8232-501

age (REF TP) Stabilize at 85 dego C then test a few Oven 85 deg. Cat a time in bench fixture

13 A Hermetic seal Hot water bath 85 deg. C8 A. Inspect for header damage or loosened parts Visual/rmanual

*PS4 SPECIFIC INSPECTION TEST EQUIPMENTINSTRUCTIONS REQUIRED

Fig. 2 - Purchased materials inspection (PMI) instraction card (side 2).

to be checked, the sample size to be used on each characteristic, the test equip-ment required, and the requirements to be met. This setup lends itself very eas-ily to computations, as the characteristics within any class of parts have stand-ard numbers. Characteristic 13 in the transformer class, for example, is ahermetic seal test regardless of whether a pulse transformer or a power trans-former is being inspected. If no hermetic seal test is to be performed, charac-teristic 13 is omitted from the instruction card.

The PMI data card is the complement to the instruction card (Fig. 3). Adata card is completed for each lot received by PMI. Note in the right-handthird of the card the blocks numbered 1 to 32 which correspond to the character-istic numbers on the instruction card. After the inspection is completed, thenumber defective for each characteristic step is noted in the corresponding blockon the data card.

Besides using the data cards as the means of calculating each suppliertsthiee month process average, print-outs are made in part number order. Theseprint-outs inxdicate the total numrber of failures occurrinlg for each characteris-tic for which a particular part is being tested. The PMI planning activity reg-ularl;y revriews the instruction cards and revises the characteristics to be

11

Page 6: Making a Multiplant Supplier Rating System Produce

checked when the print-outs indicate a change is warranted. Changes can take theform of deletion of characteristics due to a proven record of successes; reduc-tion or increase (up to 100% screening) in the frequency of inspection; and addi-tion of characteristics to complement existing ones. Where it is necessary toincrease our testing, we will immediately notify our supplier so that he may takethe needed corrective action at his plant.

At the present time, we are in the process of determining what percentage ofthe total of each type part submitted is being inspected, and what is the averagecost of inspection for each type part. On the two part types studied to date,relays and transformer-coils, we have found that our average cost to inspect arelay is $0.67, and for a transformer-coil $1.82. Furthermore, we are inspecting52% of all relays, 35% of all transformers, and 13% of all coils submitted. When

UNIQUE VENDOR NUMBER FAILURES BY CHARAMCERISTIC

CHARACTERISTIC STEP

6294L8223RELA50s12 19802-0331 1 81522 29DRAWING NUMBER PART NAME VEND6R PURCHASE ORDER NUMBER _ _z

2 ~9 16 23 30 ze0

861225J1 |' 3 I1l2I 30723 A 10-7 3 t10 17 .24 31NUMBER NUMBER LOT SAMPLE DATE 1

LOT SiZE NSPECTEO DEFECTIVE IiPsosirN CONTROL NUMBER MDR. NUMBER PLAN USED INSPECTEf -4 1118 2 32

_~~~_.

0

5 1.2 19 26

INSPECTION STAMP INSPECTION TAMP SUB MEG DATE E EGNEERING RATING___EXPLANATION: 6 13 20 27

7 . . .14 21 28

mImo 074S5

DESIGNATES ACCEPTANCE RETUWNED To SUPPLE ON THISOR REJECTION OF LOT MTTERUL DISC.REPANCY REPORT

Fig. 3 - Purchased materials inspection (PMI) data card.

we have completed this study covering all classes of parts, we feel we will be inan excellent position to judge whether we are spending too much or too little tojustifly the quality levels being attained in assembly and test.

Recently, another step toward reducing incoming inspection costs has beentaken by placing a special code on the purchase orders sent to all preferredhardware suppliers; i.e., those listed in the blue book. This code tells ourreceiving department to send these shipments directly to stock without any incom-ing inspection whatsoever. We feel that our experience over the past two yearshas given us adequate assurance that these lots will be good, and indicationo todate have provred this assumption to be correct. Approximatel 1,200 lots a yeaof jut hardware ites will be sent directly to stock, providing a saving of atleast $10,000 in hadling, paperwork, and inspection time.

12

Page 7: Making a Multiplant Supplier Rating System Produce

Although we have emphasized the reduction of PMI costs, the realignment ofinspection dollars to provide imcreased reliability is equally important. Ourprogram is designed to eliminate inspection where it provides no additional bene-fits, and to add inspection and special testing where it will provide additionalreliability. For example, in a recent pilot production of an airborne receiver-transmitter, our incoming inspection was concentrated on relays, quartz czrstals,and switches, with many special reliability tests being run on samples by ourquality control engineering laboratory. These three part types were chosen afterreviews with the engineering group that designed the equipment and our standardsengineering activity., and a review of the inspection history of equipments simi-lar to the one in question.

As it turned out, these three part types did provide a majority of the prob-lems. Two of these problems which were solved before the parts were ever assem-bled into units were the direct result of increased testing. The first involveda relay used over 100 times in the equipment. Special accelerated life testsshowed that the welds from the buffer arm to armature and from the coil lead toheader were opening up. Our supplier was notified. By increasing the wire sizeused, the junction characteristics were improved and a more uniform weld resulted.The second problem involved a switch used in a critical application. Tests attemperature extremes showed that this switch was very unstable. Engineeringworked with the supplier, and by changing the switch material from a paper basephenolic to glass silicon, the problem was eliminated.

As with all problems of this type, DEP will eliminate its special testing assoon as the supplier has taken acceptable corrective action. We firmly believethat the burden of assuring a good part must be placed on the suppliers' shoul-ders, and our incoming inspection assures us that he has done an acceptable job.Once this assurance is forthcoming, acceptance will be made on a limited inspec-tion basis.

INTERPIANT COMPARISONS

By maintaining separate purchasing departments at each of its three manufac-turing facilities, DEP must assure itself that its policies regarding suppliersservicing more than one plant are uniform. That is, one plant should not tell aparticular supplier that he is doing an excellent job, while another plant hasthat supplier listed as highly unsatisfactory. As a means of solving this prob-lem, all suppliers who service more than one plant are listed every three monthswith their ratings at each of the plants. Due to differences in personnel,inspection techniques, etc., we expect rating variations to occur. A one classvariation from an A to a B, or a B to a C, is not regarded as -significant, butdifferences of two classes (A to a) are regarded as significant and requiringinvestigation.

Upon investigation, 60% of the differences were explainable by the fact thatthe part numbers bought by each plant were totally different. In one case, Cam-den was buying small brackets while Moorestown was buying 6 foot cabinets fromthe same supplier. Camden rated this supplier as A while Moorestown rated him aC. Since the large dollar value of our business with this supplier emanated fromthe Moorestown plant, the supplier was treated as unsatisfactory,.

Most of the remaining differences are caused by poorer quality units being-shipped to one of our plants. In one particular case, a large manufacturer Of

143

Page 8: Making a Multiplant Supplier Rating System Produce

terminal boards was consistently being rated A by Moorestown and C by Camden.Camden, in fact, was encountering problems with finish, cracked boards, rollover,etc., which made it difficult to believe that Moorestown was receiving goodunits. It was finally uncovered that Moorestown had had earlier problems withthis supplier and had been returning all discrepant lots. Camnden, on the otherhand, was notifying the supplier of discrepant lots, but repairing them, due tomaterial shortage. Thus, this supplier had tightened up on all material going toMoorestown, but was still shipping his usual quality to Caamden.

QUAITL!fY CJON2tROL FIELO COTERAGE

DEP maintains a number of field quality assurance representatives locatedthroughout the United States, as a service to all DEP plants. It had been ourpractice to have these representatives monitor only those suppliers who had beensingled out by the individual plant puirchasing and quality control groups. Thismonitoring took the form of initlal visits to assure adequate provisions forquality manufacture, followed by periodic visits based on delivery schedules toassure conformance of the product and processes to requirements.

Last May DEP instituted the procedure of issuing monthly print-outs of eachsupplier's incoming quality performance. These print-outs denote the supplier'sname and address in the upper left-hand corner of an 8! inch by 11 inch page, andthen list in part number order all lots submitted for a particular month andtheir eventual disposition. Only one supplier is listed on a page, which allowsa sorting operation by area to be readily performed. The listings are then for-warded to our representatives.

Each field representative is now responsible for all suppliers in his areawhetlher he visits them on a sceduled basis or not. It is his responsibility toleaf through the set of pages he,receives and pull out those suppliers who areperforming below par as evidenced by the lot disposition codes. He will thenvisit these suppliers to explain our rating system, review their processes andprocedures, provide information as to our standards and requirements, and mostimportant of all, see that adequate corrective action is taken. Although thiseffort does reduce the time that can be spent with scheduled suppliers, compensa-tion has been forthcoming in the form of lower rejection rates.

SUPPLIER TEST AND RELIABILITY DATA

Based on supplier process averages, DEP has undertaken the program ofrequesting certified test data and tagged samples from certain suppliers. Thisprogram has been implemented with suppliers of relays, transistors, and diodes,with other electrical parts to follow shortly. When lots are received with cer-tified test data and tagged samples, PMI makes variable measurements and performscorrelation and limit tests. Where these tests have proven satisfactory on agiven number of consecutive lots, PMI will accept future lots on the basis ofsupplier data.

The taking of variable measurements and the subsequent correlation with sup-plier data has proved very useful in uncovering differences between DEP and sup-plier methods Of test and measurement. One particular case, for example,involved the testing of a diode for Ir at 15O0°C Our readings differed signif-icantly fromn our supplier's, although the same test equipment was used at both

14

Page 9: Making a Multiplant Supplier Rating System Produce

places. After investigation, the supplier discovered that he had been checkingthe units at 129'C rather than 1500C.

Another case which gave us quite a bit more trouble involved the measurementof frequency vs. temperature for quartz crystals. Our data for this test, whichwas conducted in accordance with MIL-C-3098B, repeatedly showed out-of-limitreadings, while the supplier's data showed satisfactory values. It was finallydiscovered that although the supplier's method of test swept the temperaturerange within tne time limits of the specification, his recording device was notcapable of tracking this rapidly, and rejects were being accepted. This was cor-rected by establishing a slower rate of temperature change to be used at the sup-plier's test facility. Many other instances, some in which our methods requiredcorrection, others in which our suppliers required correction, have provred to usthe benefits of performing these tests.

In addition to requesting test data, DEP has taken another step to assurethe reliability of incoming parts. We are now placing probability of survivalrequirements on electrical part prints in many of our contracts. The values forprobability survival are derived from comprehensive analyses perfornied on equip-ments prior to the release of any purchase orders. The techniques involved inthis an alysis are thoroughly described in RCA's TR-llOO entitled tRelriabilityStress Analysis for Electronic Equipment." These analyses involve the additionor subtraction of failure bits to base failure rates dependent upon the mechan-ical and. electrical stresses present in the circuit and surrounding environment.

The probability of survival is made a supplier requirement by purchaseorder. DEP fully realizes that the number of unrits that would be needed to provethis probability of survival for any one part number would be extremely high. Itis felt, however, that suppliers have sufficient experience and data on the partsthey manufacture to know whether the requirement can be met. On many occasions,suppliers have informed us that our survival requirements were unreall stic andshould be changed. In those cases where the supplier could furn-ish sufficientdata to support his claims, the requirement was changed. Conversely, many timeswe proved to suppliers that they were capable of meeting our requirements.

We feel that the placing of survival requirements on prints does have thefollowing advantages:

1. It makes the supplier perform at least a prelimfinary review of his testrecords to determine if the rate is feasible.

2. It makes the supplier aware of the importanlce of reliability to us.

3. It brings unrealistic rates to our attention.

CONCLUSION

The implementation of a supplier rating system has proven extremely bene-ficial in reducing rejection rates of incoming material, aligning our inspectiondollar where it will do the most good, and providing a guide for the schedulingof field visits. Its greatest benefit, however, has been in the enlightenment ofDEP mnanagement as to where our inspection dollars go and how they may be usedmost judiciously.

15