SL3003Ch04Frame Page 69 Tuesday, November 6, 2001 6:10 PM …yahuza.weebly.com/uploads/2/2/2/2/2222048/sl3003ch4.pdf · 2018-09-05 · 4.2 DESIGN FOR SIX SIGMA (DFSS) DFSS methodology

69

4

DFMA/DFSS

John W. Hidahl

Design for manufacture and assembly (DFMA) and design for Six Sigma (DFSS)are complementary approaches to achieving a superior product line that maximizesquality while minimizing cost and cycle time in a manufacturing environment.DFMA is a methodology that stresses evolving a design concept to its absolutesimplest configuration. It embodies ten simple rules, which can have an incredibleimpact on minimizing design complexity and maximizing the use of cost-effectivestandards. DFSS applies a statistical approach to achieving nearly defect-free prod-ucts. It uses a scorecard format to quantify the parts, process, performance, andsoftware (if applicable) capabilities or sigma level. It facilitates the effective designof a product by aiding the selection of (1) suppliers (parts), (2) manufacturing andassembly processes (process), (3) a system architecture and design (performance),and (4) a software process (software) that minimizes defects and thus produces ahigh-quality product in a short cycle time.

4.1 DESIGN FOR MANUFACTURE AND ASSEMBLY (DFMA)

The DFMA methodology consists of six basic considerations and ten related rules,as shown in Table 4.1.

DFMA is intended to increase the awareness of the engineering design staff tothe need for concurrent product and process development. Several studies haveproven that the design process is where approximately 80% of a product’s total costsare determined. Stated differently, the cost of making changes to a product as itprogresses through the product development process increases by orders of magni-tude at various stages. For instance, if the cost of making a change to a productduring its conceptual design phase is $1000, then the cost of making the same changeafter the drawings are released and the initial prototype is fabricated is approximately$10,000. If this same change is not applied until the production run has started, thecost impact will be approximately $100,000. If the need for the design change isnot recognized until after the product has been purchased by the consumer ordelivered to the end user, the total cost for the change will be approximately 1000times as great as if it had been implemented during the conceptual design review.In addition to driving product cost, design is also a major driver of product quality,reliability, and time to market. In today’s marketplace, customers are seeking thebest value for their investment, and the most effective way to incorporate maximumvalue into a product’s design disclosure is through the use of DFMA.

SL3003Ch04Frame Page 69 Tuesday, November 6, 2001 6:10 PM

© 2002 by CRC Press LLC

70

The Manufacturing Handbook of Best Practices

4.1.1 S

IMPLICITY

Simplicity is the first design consideration, and it bridges the first five DFMAcommandments, namely, (1) minimize the number of parts, (2) minimize the use offasteners, (3) minimize reorientations, (4) use multifunctional parts, and (5) usemodular assemblies. There are several approaches that can be used to minimize thepart count in a design, and specific workbook and software techniques have beendeveloped on this, but the driving principles revolve around three questions: (1)Does the part move? (2) Does the part have to be made from a different materialthan the other parts? and (3) Is the part required for assembly or disassembly? Ifthe answer to all three is no, then that part’s function can be combined with anotherexisting part. Using this approach progressively, existing assemblies that were notbased upon DFMA principles can often be redesigned to eliminate 50% or more oftheir existing parts count. Reduced part counts yield (1) higher reliability; (2) lowerconfiguration management, manufacturing, assembly, and inventory costs; (3) feweropportunities for defects; and (4) reduced cycle times. Minimizing the use of fas-teners has several obvious advantages, and yet it is the most frequently disregardedprinciple of DFMA. Excessive fasteners in a design are often the result of engineeringdesign uncertainty, and are often justified as offering flexibility, adjustment, quickcomponent replacement, or modularity. The reality is that excessive fastenersincrease the cost of assembly, increase inventory costs, reduce automation opportu-nities, reduce product reliability, and contribute to employee health risks such as

TABLE 4.1DFMA Considerations and Commandments

Considerations

1. Simplicity2. Standard materials and components3. Standardized design of the product itself4. Specify tolerances based on process capability5. Use of the materials most processed6. Collaboration with manufacturing personnel

The Ten Commandments

1. Minimize the number of parts2. Minimize the use of fasteners3. Minimize reorientations4. Use multifunctional parts5. Use modular subassemblies6. Standardize7. Avoid difficult components8. Use self-locating features9. Avoid special tooling

10. Provide accessibility



DFMA/DFSS

71

carpal tunnel syndrome. Prototype designs may require additional fasteners andinterfaces to test various design or component options, but the production design shouldbe stripped of any excessive fasteners. The five

why

’s approach as used commonly inroot cause analysis is recommended for testing the minimal requirements for fasteners.Unless one of the sequential answers to, “

Why

do we need this fastener?” can betraced directly to a stated operational requirement, the fastener(s) should be elimi-nated from the production design disclosure. With respect to minimizing reorienta-tions during assembly, the guiding principles are to create a design that can be easilyassembled (with a minimum amount of special tooling) and to always use gravityto aid you in assembly. Minimizing the number of fasteners will obviously contributetoward minimizing the number of reorientations necessary. The use of multifunc-tional parts is a primary method of reducing the total parts count, thus enhancingdesign simplicity. Similarly, the use of modular subassemblies is a good designmethod to predesign for continuous product improvement through block upgradesand similar product line enhancements over time. As new technology moves intopractice and becomes cost effective, modular subassemblies can be easily replacedto provide expanded capabilities, higher processing speeds, or more economical(market competitive) modular substitutions. Although modular subassemblies mayincrease the total part count of the original product, the added ease and speed ofimplementing improvements are a positive trade-off for many products or productfamilies.

4.1.2 U

SE

OF

S

TANDARD

M

ATERIALS

C

OMPONENTS

AND

D

ESIGNS

The second and third design considerations, standard material and components andstandardized design of the product, are described by the sixth commandment: stan-dardize. Design reuse is one of the most cost-effective methods used in the designprocess. By defining company- or product family-related standard materials, standardparts, and specific design process standards, the product cost and time to marketwill be reduced, while reliability and customer value will be maximized. The keyelement in standardization is establishing the discipline within the organization tokeep the standards current and readily available to the product development team,and enforcing their effective and consistent use.

4.1.3 S

PECIFY

T

OLERANCES

The fourth design consideration is specifying or establishing design tolerances basedupon process capability rather than the typical design engineer’s affinity for closelytoleranced parts. This approach is embodied in the seventh design commandment:avoid difficult components. The most effective way to apply this consideration isthrough the concurrent product development team environment where the designengineer and the manufacturing (producibility) engineer work collaboratively toensure that the designed parts can be efficiently manufactured without excessivecosts or scrapped material. This imposes the requirement that the manufacturingengineer have full knowledge of the process capabilities of in-house equipment andprocesses, as well as supplier equipment and processes.



72


4.1.4 U

SE

OF

C

OMMON

M

ATERIALS

The fifth design consideration is use of the materials most processed. This simplymeans that materials that are commonly machined or processed in some mannerwithin the company or within the company’s supplier base should be the firstmaterials of choice for the various components. Exotic or state-of-the-art processesand materials should be avoided whenever possible to preclude extended processdevelopment activities associated with low process capability, which typicallyincrease cost and cycle time while reducing quality and reliability.

4.1.5 C

ONCURRENT

E

NGINEERING

C

OLLABORATION

The sixth and final design consideration is collaboration with manufacturing per-sonnel. As identified previously, it is essential that the design team include cross-functional personnel such as manufacturing engineers, quality engineers, and procure-ment specialists to ensure that all the appropriate design trade-offs are properly analyzedand selected throughout the product development process by the experts in the respectivedisciplines involved. The traditional “Throw the design over the wall to manufacturingwhen engineering is done with it” approach is guaranteed to produce product attributesthat contribute to higher production costs and extended time to market.

The other three design commandments that remain to be described are (8) touse self-locating features, (9) to avoid special tooling, and (10) to provide accessi-bility. The use of self-locating features is an assembly aid that can dramaticallyreduce assembly costs and cycle time. Parts that naturally nest together or containself-centering geometries reduce the handling, alignment, reorientation, and inspec-tion costs of assembly. Automated assembly processes in particular benefit tremen-dously from self-locating features to minimize the tooling and fixturing oftenrequired to ensure proper part alignment during assembly. Similarly, the avoidanceof special tooling is a key consideration in complex assembly processes. Specialtooling should be used only when other design elements or part geometries cannotincorporate self-locating features. Special tooling harbors an extensive array ofhidden costs when fully analyzed. In addition to the cost of designing, fabricating,checkout, inventory, maintenance, spares, and planned replacement of special tool-ing, it can also add substantial cycle time to the assembly process. The added cycletime can accrue from issuing it from stores, moving it, installing it, and then verifyingits proper placement, alignment, attachment, and operation over its intended designlife. The final commandment is to provide accessibility, which implies the need formaintenance, inspection, part adjustment, part replacement, or other product accessrequirements over its design life. The key here is to define the requirements foraccessibility based on the customers’ (end-users’) needs and the product develop-ment team’s comprehensive vision of the product’s possible applications, as well asits growth or evolution in the future. This requires a balance between satisfyingcurrent minimum needs and anticipating the most likely future needs, while stillkeeping the design simplicity DFMA consideration in mind.

All the aforementioned DFMA considerations and commandments should beapplied as an integrated and balanced approach in the design process. A well-documented product development process, in combination with clearly defined team



DFMA/DFSS

73

member roles and responsibilities, will greatly improve the application of DFMAin most organizations.

4.2 DESIGN FOR SIX SIGMA (DFSS)

DFSS methodology encompasses all the DFMA principles and adds proven statis-tical techniques to drive the design process, and thus the product, to lower defectcounts. The typical DFSS statistical applications in design include (1) toleranceanalysis, (2) process mapping, (3) use of a product scorecard, (4) design to unitproduction costs, and (5) design of experiments.

4.2.1 S

TATISTICAL

T

OLERANCE

A

NALYSIS

Statistical tolerance analysis employs a root-sum-squared approach to evaluatingtolerancing requirements in lieu of the more traditional “worst-case analysis.” Itsmethodology is based on the statistical fact that the probabilities of encounteringthe worst-case scenario are extremely remote. For instance, if an assembly involvesthe interfacing of four different parts, and each part is known to have a ±3 sigmadimensional capability, then the defect probability can be calculated to be 2.7 in1000, or 0.0027. By applying statistics, the probability of encountering the worst-case situation can be calculated to be 5 in 100 billion or 0.0000000000534. Thisclearly demonstrates the ultraconservatism of this approach and the consequentextremely tightly toleranced part call-outs required to achieve it. Tightly tolerancedparts have inherent hidden manufacturing costs associated with them, because theydictate detailed inspection requirements and often require scrap or rework of asignificant percentage of the manufactured parts. Most of these scrapped or reworkedparts would have, in fact, worked perfectly well, but were rejected due to excessivelydemanding part tolerancing.

A product generally consists of both parts and processes. This relationship meansthat to be successful you should seek to understand both the upstream and downstreamcapabilities of the various processes that will be used to produce the product. A productmust be designed to not only meet the customer’s requirements, but must also comple-ment the process capabilities of the manufacturing company and its supplier base. It isunlikely that a company will ever reach a goal of Six Sigma quality without under-standing the capability of the entire supply (or value) chain. Design teams must under-stand and properly apply the process capabilities of their manufacturing facilities andthose of their suppliers in order to repeatedly produce near zero-defect products. Processcapability data are the enabling links needed to create robust designs. The preferredgraphical method of describing the key process capabilities and how they relate to theoverall product manufacturing activity is through the process map.

4.2.2 P

ROCESS

M

APPING

Six Sigma process-mapping techniques encompass several statistical measures ofprocess performance and capabilities in addition to the typical process flows andrelated process operation information. As you will see, this information is extremelyuseful when a team of individuals has been assigned to improve a process. Let’s



74


start with some of the common vocabulary used in process mapping to becomefamiliar with the terminology (Table 4.2).

Now that the basic terms have been defined, why do you suppose a process mapis important when improving an existing process or implementing a new one? Thereare several visual features that a process map provides to aid a team’s understandingof the operations involved in a given process:

1. A process map allows everyone involved in improving a process to agreeon the steps it takes to produce a good product or service.

2. A map will create a sound starting block for team breakthrough activities.3. It can identify areas where process improvements are needed most, such

as the identification and elimination of non-value-added steps, the poten-tial for combining operations, and the ability to assist with root-causeanalysis of defects.

4. It will identify areas where data collection exists and ascertain its appro-priateness.

5. The map will identify potential X’s and Y’s, leading to determining theextent to which various x’s affect the y’s through the use of designedexperiments.

6. The map serves as a visual living document used to monitor and updatechanges in the process.

7. It acts as the baseline for an XY matrix and a process failure modes andeffects analysis (PFMEA).

A Six Sigma process map for a manufacturing operation is shown in Figure 4.1.The map was created by a focused team working on a product-enabling process. Theteam consisted of operators, maintenance technicians, design engineers, material andprocess engineers, shop floor supervisors, and operations managers. The basic elementsof this process map include (1) the process boundaries, (2) the major operationsinvolved, (3) process inputs, (4) process outputs, and (5) the process metrics. There areseveral steps that must be followed to create a valid process map, as outlined in Table 4.3.

TABLE 4.2Process Mapping Vocabulary

Process map:

a graphical representation of the flow of a process. A detailed process map contains information that is beneficial to improving the process, i.e., cycle times, quality, costs, inputs, and outputs.

Y:

key process output variable; any item or feature on a product that is deemed to be “customer” critical, referred to as “y1, y2, y3.”

X:

key process input variable; any item which has an impact on Y, referred to as “x1, x2, x3.”

Controllable X:

knob variable; an input that can be easily changed to measure the effect on a Y.

Noise X:

inputs that are very difficult to control.

S.O.P. X:

standard operating procedure; clearly defined and implemented work instructions used at each process step.

XY matrix:

a simple spreadsheet used to relate and prioritize X’s and Y’s through numerical ranking.



DFMA/DFSS

75

The key statistical information often described on a Six Sigma process mapincludes the defects per unit (DPU) at each operation step, rolled throughput yield(RTY), and key process capability (CPk) values. The design team needs to analyzethese process parameters and understand their influence on RTY in order to designquality into the product rather than attempting to inspect quality into the product.

4.2.3 S

IX

S

IGMA

P

RODUCT

S

CORECARD

The Six Sigma product scorecard is an excellent method for applying process capabilityinformation to the conceptual phase as well as subsequent phases of the design evolu-tion. The scorecard is derived from the Six Sigma requirements for process definition,measurement, analysis, improvement, and control. By individually analyzing four ele-ments of a design (parts, process, performance, and software), scorecard sigma levelscan be identified. Initial scorecard values can be used to evaluate conceptual designalternatives and to influence the downselect criteria; refined scorecards can be used toaid trade studies to optimize the baseline design configurations. In these design studies,product sigma levels can be evaluated as independent variables that drive cost, schedule,and other critical parameters. Baseline design selection at an overall 3 Sigma level, forinstance, would yield 66,807 parts per million (ppm) defective, whereas achievementof a 6 Sigma design level would yield only 3.4 ppm defective, or a ratio of approximately20,000 to 1 in improved quality!

An example of a Six Sigma product scorecard is shown in Figure 4.2. Thissummary-level scorecard includes the four assembly level evaluation elements: parts,process, performance, and software, with the software element being nonapplicablefor this simple mechanical configuration. Note that for each of the elements, theDPU estimate and the opportunity counts are described for each major subassembly.These are then totaled near the bottom of the table, and first time sigma, DPU/oppor-tunity, sigma/opportunity long term and short term are all calculated through algorithmsbuilt into the Excel spreadsheet. Each element results in a separate short-term sigma

TABLE 4.3Steps to Creating a Process Map

Step 1:

Define the scope of the process you need to work on (actionable level).

Step 2:

Identify all operations needed in the production of a “good” product or service (include cycle time and quality levels at each step).

Step 3:

Identify each operation above as a value-added or non-value-added activity. A value added operation “transforms the product in a way that is meaningful to the customer.”

Step 4:

List both internal and external Y’s at each process step.

Step 5:

List both internal and external X’s at each process step.

Step 6:

Classify all X’s as one or more of the following:

•

Controllable (C)

•

Standard operating procedures

•

Noise

Step 7:

Document any known operating specifications for each input and output.

Step 8:

Clearly identify all process data-collection points.



76


that is used as the design basis for most applications. The minimum sigma value forany of the elements constitutes the design sigma limitation. Unless all the elements arefairly equivalent in value, the overall sigma score will be heavily influenced by thelowest element sigma value. Each of the elements uses a separate worksheet accessiblethrough the Excel worksheet tabs at the bottom of the spreadsheet layout.

The parts worksheet shown in Figure 4.3 is completed by defining all the majorpurchased or manufactured individual parts that will make up the assembly orsubassembly. This is most easily accomplished through the use of a bill of materials,or parts listing. The supplier, part number, part description, quantity, part defect ratein ppm defective, and the total DPU, an alternate description for ppm, are all defined.A separate worksheet is completed for each major subassembly to be built bymanufacturing. The overall intent of this methodology is to drive the previously

FIGURE 4.1

Solid rocket motor strip winding process map. CT = cycle time, DPU = defectsper unit, MBOM = manufacturing bill of materials, NVA = non-value added, RTY = rolledthroughput yield, SOP = standard operating procedures, VA = value addeed, X = inputvariables, Y = output variables.

Receive MaterialDPU=.01

CT=2.0 hrs

X’s Y’s

NVA

Quality of Material

Technician, SOP,Specifications

Material Handler,SOP, 40°F ColdBox, ProperStorage

Material Handler,SOP, Forklift

Technician, SOP,Controllers, BarrelTemp., ScrewTemp.,Head Temp.,Hopper Temp.,Rollaformer Temp.

Technician, SOP,Rollaformer Profile

Technician, SOP,Diode Settings

MaterialConforms toSpec



MaterialReceived atStrip Winder

PreheatedOperatingSystem

Thickness ofStrip meetsRequirements

Width of StripmeetsRequirements

Verify and Test MaterialQuality Properties

DPU=.001CT=40.0 hrs

Transport and Store in40°F Cold Box

DPU=.001CT=4.0 hrs

Issue Material perMBOM to floor

DPU=.001CT=2.0 hrs

Preheat TemperatureControl UnitDPU=.001

CT=1.0 hrs

Set Gap on Upper/LowerRollaformersDPU=0.05

CT=1.0 hrs

Set Diode (width) onthe Controller

DPU=.001CT=.01 hrs

NVA

NVA

NVA

NVA

NVA

NVA

Technician, SOP,Material Condition,Machine Settings




Technician, SOP,MaterialConveyanceSystem

X’s Y’s

Material FeedIntiated

Material pre-conditioned

System atAcceptablePressureRange

Hot StripMolded

Molded Stripready forApplication atWinder

NVA

VA

SCRAPAcceptable

Strip atRollaformers?

No

Yes

Convey Strip toApplication System

Extruder Charged at Steady-State Pressure

DPU=.001CT=.05 hrs

Strip Formed atRollaformersDPU=.001CT=.05 hrs

VA

VA

VA

Final RTY=92.5%

Material Conditioned atExtruder

DPU=.001CT=.10 hrs

Feed Insulation Materialinto Extruder

DPU=.01CT=.050 hrs



DFM

A/D

FSS

77

FIGURE 4.2

Six Sigma product scorecard.

Date AI&T Cost $2,599 DPUPart Number Critical Path Cycle Time 0 DPMOName ACME Raw Process Multiplier 8.29 SigmaPeriod of Data

Part (σ) Process (σ) Performance (σ)

Assembly DPU Opp. Count Parts Cost DPU Opp. Count Labor CostCycle Time

(min.)Total Time

(min.VA Time

(min.)Scan Drive 0.1649 1 $2,500 37.9711 2680 $99 250 290 35 8.29 0.07667732 2191AntennaReceiverElectronicsSystem

Totals 0.1649 1 $2,500 37.9711 2680 99 250 290 35 8.29 0.0767 2191First Time Sigma 1.03 <-6 1.45RTY 84.8% 0.0% 92.6%DPU/Opp 0.1649 0.0142 0.0000Sigma/Opp 1.03 2.20 3.98

4/1/00 -

04/04/00xxxxxxxx

2.42

38.21277843.3

DPU Opp. CountRPM

Date AI&T Cost $2,599 DPUPart Number Critical Path Cycle Time 0 DPMOName ACME Raw Process Multiplier 8.29 SigmaPeriod of Data

Part (σ) Process (σ) Performance (σ)

Assembly DPU Opp. Count Parts Cost DPU Opp. Count Labor CostCycle Time

(min.)Total Time

(min.VA Time

(min.)Scan Drive 0.1649 1 $2,500 37.9711 2680 $99 250 290 35 8.29 0.07667732 2191AntennaReceiverElectronicsSystem

Totals 0.1649 1 $2,500 37.9711 2680 99 250 290 35 8.29 0.0767 2191First Time Sigma 1.03 <-6 1.45RTY 84.8% 0.0% 92.6%DPU/Opp 0.1649 0.0142 0.0000Sigma/Opp 1.03 2.20 3.98

4/1/00 -

04/04/00xxxxxxxx

2.42

38.21277843.3

DPU Opp. CountRPM

SL3003C

h04Frame Page 77 T

uesday, Novem

ber 6, 2001 6:10 PM


78


described DFMA principles of fewer parts and part types into the design and toultimately select quality suppliers and processes to manufacture the individual parts.

The process worksheet portrayed in Figure 4.4 describes the assembly processinformation, much of which is taken directly from the process map previouslyconsidered. Here again, one worksheet per major assembly or subassembly is com-piled for each assembly level built by manufacturing. The process worksheet iden-tifies all the major internal processes used to build the product. The DFMA intenthere is to use high quality processes and simplify the build process to the greatestpractical extent. For each process step, the load center, cycle time, labor hours andcost, process target, specification or tolerance, upper specification limit (USL), lowerspecification limit (LSL), process mean value, standard deviation, process capability(CPk), number of applications, process opportunities, and product opportunities areall defined. From this information the spreadsheet algorithms are used to calculatethe total number of product opportunities, average defects per opportunity, averageyield per opportunity, average process sigma long term (LT), average process sigmashort term (ST), as well as the total defects per unit, the rolled throughput yield,and the sigma (z) score. As evidenced by the amount of statistical process datarequired, this methodology involves extensive process capability data collection andknowledge to be used successfully. It requires taking the operator “black magic”out of the process capability equation, and replacing it with parametrically drivenprocess knowledge and control features, which can be derived from design ofexperiments, and other Six Sigma methodologies.

An example of the performance worksheet is presented in Figure 4.5. It is usedto identify all the customer-focused, top-level system performance parameters, andto quantify the probability that the design configuration will successfully achievethem. Its intent is to quantifiably assess the design’s capability against the definedsystem-level requirements. It also provides insight into the production acceptancetesting requirements and needed measurement system accuracy (MSA). The work-sheet lists the key customer-based performance parameters that can be obtained froma customer’s specification, a technical requirements document, or from a qualityfunction deployment (QFD) process. It defines target values, units, upper specifica-tion limit (USL), lower specification limit (LSL), performance mean value, standarddeviation, z score USL, z score LSL, rolled throughput yield, and DPU.

A software worksheet is presented in Figure 4.6. It identifies the entire softwarebuild process, tracks defects found during each phase of the software development,and calculates the efficiency of each software phase in detecting and eliminatingdefects. It also provides a future extrapolation of overall delivered software quality,based on defect rates demonstrated during the build process.

The top-level product scorecard results are calculated by algorithms internal tothe spreadsheet using all the individual worksheet inputs. As previously identified,Figure 4.2 illustrates the combined results from this Six Sigma tool, and its influenceon designing quality into the product. This methodology provides a powerful methodof positively influencing the design process through the use of data and removes themystery (or mystique) that surrounds many modern-day manufacturing facilitiesabout their ability to produce high-quality products on a consistently repetitive basis.



DFM

A/D

FSS

79

FIGURE 4.3

Six Sigma product scorecard — parts worksheet.

Part Number TDUName Scan Drive YieldPeriod of Data 4/1/00 - SigmaTotal Part Count 1Avg Defects/Part 0.1649 COQAvg Yield/Part 84.8% Part CostAvg. Part Sigma 1.03 Variance

Supplier Part No. Description Feature Qty. LSL USL Mean St.Dev.Units Defects PPM DPU Sigma Unit Cost COQ Total Planned VarianceAce 1349594-1 Printed Wiring Board 1 291 48 164948 0.1649 1.03 $2,000 $500 $2,500 $2,000 ($500)

Cost DataMeasured FeaturePart Description

$2,500($500)

0.164984.8%1.03

$500

Defect Data

SL3003C

h04Frame Page 79 T

uesday, Novem

ber 6, 2001 6:10 PM


80

The M

anu

facturin

g Han

db

oo

k of B

est Practices

FIGURE 4.4

Six Sigma product scorecard — process worksheet.

Part Number TDU Total Unit Cost Cycle Time - Mean (min.)Name Yield Total COQ Cycle Time - Std. Dev. (min.)Period of Data Sigma (Z) Total Cost Total Process Time - Mean (min.)Total # of Product Opps. Total Variance Total Process Time - Std. Dev. (min.)Average Defects/Opp. Value Added Time (min.)Average Yield/Opp. Raw Process MultiplierAvg. Process Sigma

Process Step Ope

rati

on N

umbe

r

Def

ect

Iden

tifi

cati

onM

etho

d

LSL

USL

Mea

n

Std.

Dev

.

Cpk

Num

ber

of U

nits

Num

ber

of D

efec

ts

# of

Tim

es U

sed

Ope

rati

onO

ppor

tuni

ties

Pro

duct

Opp

ortu

niti

es

Def

ects

per

Uni

t

DP

MO

Fir

st T

ime

Sigm

a

Sigm

a/O

ppor

tuni

ty

Num

ber

of T

imes

Pro

cess

Im

plem

ente

d

Std

Hou

rs/U

nit

Uni

t L

abor

Rat

e ($

)

Ext

ende

d C

ost

($)

CO

Q (

$)

Tot

al C

ost

($)

Pla

nned

Cos

t ($

)

Var

ianc

e (P

lan-

Act

ual)

($)

Cri

tica

l Pat

h P

roce

ss

Cyc

le T

ime

- M

ean

(min

utes

)

Cyc

le T

ime

- St

d.D

ev. (

min

utes

)

Val

ue A

dded

Tim

e -

Mea

n (m

inut

es)

Raw

Pro

cess

Mul

tipl

ier

Form & Tin 2306 Insp. 3324 < 0 382 3647 1 908 908 9.547 10514 -3.80 2.31 1 0.3 $37 $11 $2 $13 $200 $187 1 120 30 10 12.00Identification 3044 Insp. 3445 0.56 332 16 1 0 0.048 48193 1.67 1.67 1 0.3 $37 $11 $2 $13 ($13) 0 0 58.00Stencil Print 3196 Insp. 3324 < 0 382 3475 1 886 886 9.097 10267 -3.69 2.32 1 0.1 $37 $4 $1 $4 ($4) 1 100 20 0 infPick & Place 3196 TOTAL 0.67 1 886 886 19.279 21760 < -6 2.02 1 $37 $73 $15 $88 ($88) 1 0 120 1.50

Insp. 3196 < 0 361 405 1.122 1266 -0.45 3.02Insp. 3324 0.68 382 6936 18.157 20493 < -6 2.05

2.20

37.971

0.01417

0.0%

98.59%

< -6

xxxxxxxxScan Drive4/1/00 - 2680

$99$20$119$81

357.14

2503629036

0

2

54

3

SL3003C

h04Frame Page 80 T

uesday, Novem

ber 6, 2001 6:10 PM


DFM

A/D

FSS

81

FIGURE 4.5

Six Sigma product scorecard — performance worksheet.

Part Number xxxxxxName Scan Drive

Period of Data 4/04/00 - # of Parameters 7

Avg. Defects/Parameter 0.0110Avg. Yield/Parameter 98.91%

Avg. Parameter Sigma 2.29

PerformanceParameter

Process Step atWhich Measurement

is Made OperationNumber Target Units Failures LSL USL µ, mean Std. Dev. Z, LSL Z, USL Cpk Calc yield

ActualYield Calc. DPU

ActualDPU

UUT CRNT 313 0 0.95 1.5 1.13855017 0.03 7.0 13.4 2.3 100.0% 100.0% 0.0000 0.0000V1 313 2 5.38 5.565 5.44587197 0.19 0.4 0.6 0.1 53.6% 99.4% 0.6230 0.0064V12 313 0 11.5 11.96 11.8114671 0.03 9.9 4.7 1.6 100.0% 100.0% 0.0000 0.0000ACT DELAY 313 1 575 745 666.189273 17.60 5.2 4.5 1.5 100.0% 99.7% 0.0000 0.0032ACT AMP F1 313 0 1.4 3.25 2.10889273 0.20 3.6 5.8 1.2 100.0% 100.0% 0.0001 0.0000ACT AMP 196 313 0 2.11858131 0.22 100.0% 0.0000ACT AMP F3 313 0 1.85948097 0.20 100.0% 0.0000

0.931.45

Units TestedUnits Failed

TDUYield

Sigma (Z)

31324

0.0767

SL3003C

h04Frame Page 81 T

uesday, Novem

ber 6, 2001 6:10 PM


82


4.2.4 D

ESIGN

TO

U

NIT

P

RODUCTION

C

OST

(DTUPC)

The design to unit production cost (DTUPC) methodology is yet another opportunityto apply statistical methods to a design optimization process. In this case, the criticaldependent variable is cost, and the design must be evolved to meet this drivingrequirement. DTUPC offers a method of determining how much it costs to build aproduct, what each DPU costs the company, how much work-in-progress (WIP) thefactory or shop has, and how much WIP your suppliers are holding that you areultimately paying for. Many companies do not find out until the end of their account-ing cycle, whether annually, monthly, or weekly, what profit they have made. DTUPCoffers the opportunity to know the true cost of every unit produced. The cost ofdefects is typically ignored in most factory operations, but in reality, the additionallabor, inventory, overhead, inspection, and other hidden costs, including warrantycoverage can completely undermine the product profit margin. Excessive WIP,whether in your factory or at the supplier’s, is yet another indication of carryingcosts that limit profitability and cash flow. Six Sigma DTUPC includes seven basicmanufacturing cost elements: (1) setup and assembly labor costs, (2) applicableoverhead and general and administrative costs (G&A), (3) bill of material (BOM)cost of parts, (4) inspection costs, (5) DPU, (6) rework cost to correct defects, and(7) warranty costs for escaping defects. Most organizations have cost estimating orcollection methods for determining the contributions of cost elements (1), (2), (3),and (4), but the “hidden-factory” or Cost-of-Poor-Quality elements (5), (6), and (7)are often overlooked or ignored, and yet can contribute substantially to the cost ofthe product. For instance, if supplier A prices a part at $35/unit that has a DPU of1.0, and your labor (hidden) to repair the part is

¼

hour

×

$60/hour = $15

then the total cost is $50/unit. If supplier B offers the same part for $42/unit, buthas a DPU of 0.05, and your hidden repair costs are, therefore, reduced to

5 defects/100 units

×

¼

hour

×

$60/hr = $0.75/unit average

then the total cost is $42.75/unit, or a savings of $7.25/unit (roughly 17% of supplierB’s total cost). This simple illustration points out the importance of knowing yoursupplier’s part defect rates and avoiding merely selecting the apparent low-costsupplier in the source selection process. Detailed statistical analysis of DTUPC canbe applied as an extension of the product scorecard to ascertain true unit productioncosts using various suppliers, in-house processes, and materials. This type of SixSigma analysis facilitates cost trades and the ultimate approach to achieving theminimum production cost of any given product.

4.2.5 D

ESIGNED

E

XPERIMENTS

FOR

D

ESIGN

O

PTIMIZATION

The use of design of experiments (DOE) to solve design problems is yet anothermethod of applying Six Sigma principles to the engineering design process. Similar



DFM

A/D

FSS

83

FIGURE 4.6

Six Sigma product scorecard — software worksheet.

Detected At

Intro. At

Syst

em D

esig

n

Ana

lysi

s

Pre

limin

ary

Des

ign

Det

aile

d D

esig

n

Cod

ing

& U

nit

Tes

t

Inte

grat

ion

&

Tes

t

For

mal

Tes

t

Syst

em

Inte

grat

ion

&

Tes

t

Flig

ht T

est/

Pos

t re

leas

e

Gra

nd T

otal

Lea

kage

Lea

ked

Opp

ortu

niti

es

DP

O

Yie

ld

PP

M

Pro

cess

Sig

ma

System Design 5 4 29 19 4 24 85 100% 85 320 0.266 77% 233273 0.7 Analysis 2 16 128 18 6 70 240 100% 240 1230 0.195 82% 177266 0.9

Preliminary Design 6 31 23 3 7 0 70 100% 70 4330 0.016 98% 16036 2.1Detailed Design 489 182 86 32 31 0 820 40% 331 8660 0.038 96% 37500 1.8

Coding &Unit Test 1921 490 107 28 25 2571 25% 650 109000 0.006 99% 5946 2.5 Integration & Test 177 5 3 0 185 4% 8 285 0.028 97% 27680 1.9

Formal Test 36 10 0 46 22% 10 302 0.033 97% 32570 1.8System Integration & Test 2 0 2 0% 0 433 0.000 100% 0 Infinite

Grand Total 0 0 0 502 2154 933 220 91 119 4019 35% 1394 124560 0.011 99% 11129 2.3

Product Development Sigma* 0.032 0.968 31751 1.9

Delivered Product Sigma** 0.001 0.999 1091 3.1

SL3003C

h04Frame Page 83 T

uesday, Novem

ber 6, 2001 6:10 PM


84


in context to a manufacturing DOE, engineering DOEs can be used to aid indownselecting design concepts and in defining the sensitivity of a design alternativeto various parameters or environmental exposures. As an example, suppose a mate-rials engineer recommends the use of an adhesive to bond two dissimilar materialstogether, which see a shear load and a temperature gradient during system start-upin a reusable application. We want to verify that the adhesive will meet the designrequirements and identify which recommended application process produces thebest bonds when exposed to the operating environmental loads, duty-cycle duration,and repeated cycling. We start by fully defining the engineering requirements. Let’sassume that the shear load is 250 lb and that the temperature of the bond changesduring the start-up transient from 70 to 180°F at a rate of 5°/second. By preparinga process map and a process FMEA, the critical few variables that are influencingthe bond strength can be isolated. Let’s assume that the five variables suspected ofinfluencing the bond strength of the adhesive are (1) material surface preparation,(2) adhesive cure temperature, (3) adhesive pot life, (4) curing pressure, and (5)application area humidity. By running a 2

5-1

order factorial experiment wherein eachof the five variables has two values at which several test coupons were prepared andevaluated, the sensitivity of each of the tested variables can be ascertained, the bondstrength requirement can be verified, and a margin of safety calculated. A designDOE of this type was run, which produced the results shown in Table 4.4.

From these DOE results we can conclude that (1) surface preparation makes asmall difference in the bond strength, but both the low and high test point produceacceptable results; (2) cure temperature likewise has a small effect on the bondstrength, but both the low and high test points produce acceptable results; (3) theadhesive pot life had almost no discernable effect on the bond strength over therange of values tested, and therefore if a pot life of one-shift (or 8 hours) is optimumfrom an operations standpoint, then an 8-hour pot-life test should be evaluated todetermine its effect on bond strength; (4) curing pressure, like pot life, had almostno discernable effect on bond strength over the range of values tested; but (5) thelocal humidity had a great influence on the bond strength over the ranges tested. At20% humidity, the bond strength is acceptable with about a 28% margin, but at 95%humidity, approximately 99% of the bonded parts failed at a shear load of 250 lb.This example demonstrates the importance and value of conducting design-basedDOEs during the design process. By completing this DOE, the design engineer was

TABLE 4.4

Variable Low Value Sheer Strength High Value Sheer Strength

Surface preparation Isopropyl alcohol wipe 319.8 ± 2.5 lb Grit blast 325.2 ± 3.6 lbCure temperature 50°F 314.3 ± 2.9 lb 100°F 320.2 ± 3.0 lbAdhesive pot life 1 hour 321.8 ± 2.4 lb 4 hours 320.6 ± 2.7 lbCuring pressure 0.1 psi 320.9 ± 2.5 lb 1.0 psi 321.5 ± 2.9 lbLocal humidity 20% 325.6 ± 3.5 lb 95% 230.6 ± 20.2 lb



DFMA/DFSS

85

able to specify the desired adhesive for bonding the two parts. He could allow awide range of process variables (as defined by the DOE), as long as the local humiditywas maintained consistent with a humidity-controlled (air conditioned) environmentas is found in most laboratories and clean rooms.

DFMA and DFSS are both effective methods for aiding the design engineer inconceptualizing and detailing the design disclosure package for a wide variety ofparts, components, assemblies, subsystems, and systems. Proper application of thevarious tools described within this chapter will yield tremendous dividends to thecompany or organization that fosters a “near-zero” defect mindset into its designfunctions.



SL3003Ch04Frame Page 69 Tuesday, November 6, 2001 6:10 PM …yahuza.weebly.com/uploads/2/2/2/2/2222048/sl3003ch4.pdf · 2018-09-05 · 4.2 DESIGN FOR SIX SIGMA (DFSS) DFSS methodology

Documents