prioritizing process improvements

Prioritizing Process Improvements

Govind RamuJDS Uniphase CorporationASQ World Conference on Quality and ImprovementW07

http://www.asq.org/sixsigma/about/govind.html

How effective is your engineering meeting?-

Are you discussing the appropriate topics?

-

Do you have the right audience?-

Do you have actionable discussions?

-

Do you leave the meeting knowing exactly what you need to do?

-

Would you consider your time spent productive and worthwhile?

Group Interaction(3 to 5 minutes)

Meetings -

Are they productive?• Overwhelming information.• Tons of presentations.• Information unrelated to agenda.• Attendees not well prepared.• Delegated attendees.• Too many side conversations.• Blackberry/laptop usage during discussions.• Working on next meeting while listening.• Engaging in arguments on “how to.”• No clear conclusions as to “what to do.”• No clear “actionable”

discussions.

Productivity Survey•

People work an average of 45 hours a week; they consider about 17 of those hours to be unproductive (U.S.: 45 hours a week; 16 hours considered unproductive).

•

People spend 5.6 hours each week in meetings; 69 percent feel meetings aren't productive (U.S.: 5.5 hours; 71 percent feel meetings aren't productive).

•

The most common productivity pitfalls are unclear objectives, lack of team communication and ineffective meetings – chosen by 32 percent of respondents overall (U.S.: procrastination, 42 percent; lack of team communication, 39 percent; ineffective meetings, 34 percent).

Source: The Microsoft Office Personal Productivity Challenge (PPC)Responses from more than 38,000 people in 200 countries

http://www.microsoft.com/presspass/press/2005/mar05/03-15ThreeProductiveDaysPR.mspx

Typical Engineering Meeting

Product by volume

Product volume by year by part #Product yield by part #Product volume and yield

Any actions?

What really matters at the engineering meeting?

Engineers want to know:•

How the process health is doing.

•

What the risks are for internal/external customers.

•

What process parameters to fix.•

What actionable intelligence exists.

The eight-step model for process improvement:

•

Step 1: Identify CTQs and CTPs.•

Step 2: Create a CTQ-CTP relationship matrix.

•

Step 3: Conduct a process FMEA.•

Step 4: Develop a control plan.

•

Step 5: Conduct gage R&R studies.•

Step 6: Set up statistical process control.

•

Step 7: Use process capability & gage R&R to identify improvements.

•

Step 8: Prioritize improvement efforts.

need

VOCI want…

CTQ

CTQ

CTQ

CTQ

CTQ

CTP

CTP

Customer Needs

Business Needs

Product CTQs

Need Drivers CTQs

General Specific

Hard to measure Easy to measure

Customer Drivers

Business

Drivers

CTP

CTP

CTP

Primary Needs

Secondary Needs

Tertiary Needs

CTPs

With right participants, engineering meetings to focus heavily here for improved effectiveness

Currently engineering

meetings may be spending more time

at this level

Step 1: Identify CTQs and CTPs

Some Definitions•

Customer drivers –

quality, cost, delivery, response.

•

Business drivers –

first-pass/rolled throughput yields, work in progress material cost, inventory cost, cycle time, etc.

•

Primary, secondary and tertiary needs –

customer needs from abstract to tactical.

•

CTQ –

critical to quality characteristics of products that customer expects from the product or service.

•

CTP –

critical process parameters that have cause and effect relationship to one more CTQ.

need

VOCI want…

Need Drivers CTQs/ CTCs

General Specific

Hard to measure Easy to measure

CTPs

Product quality

Delivery

Cost

Servicequality

I want tasty pizzaI want hot pizza

I want my pizza to be crispyI want my pizza to have fresh toppings

I want my pizza to be quickerEvery time I get either the wrong pizza or wrong toppings! Not so expensive

Taste

Average order-deliverytime

Selling price

Quantity &right product

Customerrecovery

Complaint handlingtime

Oven process Controltempr. X1 deg +/- 5 deg F

Raw material aging (days)

Vegetable aging (days)

Oven process controltime. X2 min +/- 2 min

Order processing time

Order handling time

Order delivery time (By type & volume)Material cost

Processing cost

Yield %

Margin %

I want to get a replacement for the mistake

Order check

Response time

Replacement time

Step 1: Identify CTQs and CTPs

CTP Vs

CTQ

Screening DOE

EngineeringJudgment

Step 2: Create a CTQ-CTP Relationship Matrix

CTP Vs

CTQExplore InteractionsInterrelationships

Similar idea referenced by Mikel Harry : http://www.isixsigma.com/forum/ask_dr_harry.asp?ToDo=view&questId=82&catId=11

Step 2: Create a CTQ-CTP Relationship Matrix

http://www.isixsigma.com/forum/ask_dr_harry.asp?ToDo=view&questId=82&catId=11

Step 3: Conduct a Process FMEA

Establish severity,

occurrencedetection scales

Identify process steps

to perform FMEA

Identify failure modes, causes,

effects, current controls,

& risks

FMEA

Identify critical process

variables tomonitor, assign

RPN

Developcontrol

plan (CTQ,CTP)

Creating customized severity, occurrence, and detection scales for the nature of your business or industry can make a difference – one size does not fit all!

Identify the critical process flow of the product lineand hold a team brainstorming session to identify allprobable failure modes, causes, and interim and endeffects.

Document current controls as you would in standardFMEA practice.

Further develop the FMEA to include the severity,occurrence, and detection ratings from your customizedscales.

Calculate risk priority numbers (RPNs) by multiplyingthe severity, occurrence, and detection ratings.Prioritize risks based on RPN values.

(See more detailed flow next slide)

Reference: http://www.qualitytrainingportal.com/resources/fmea/index.htm

Brainstorming all potential causes for failure modes.

Inputs: Process flow charts, manufacturing work instructions,historical process defect Pareto, lessons learned, etc.

Reference: March 2009 QP article, “FMEA Minus the Headache.”

Reference: March 2009 QP article, “FMEA Minus the Headache.”

Populating the FMEA table with discussion outputs.

Step 4: Develop a Control Plan

Inputs from Step 1 (Identify CTQs and CTPs) and FMEA recommended actions corresponding to CTQ, CTP merge here.

Measurement system that may require R&R –

ensure if adequate before

proceeding with SPC.

Back to Basics•

Let us refresh our memory on some basic definitions:–

Repeatability: Variation in measurements obtained with one measuring instrument when used several times by an appraiser (operator) while measuring the identical characteristic on the same part.

–

Reproducibility: Variation in the average of the measurements made by different appraisers (operators) using the same gage when measuring a characteristic on one part.

–

Process capability compares the output of an in-control process to the specification limits by using capability indices (Cp, Cpk). The comparison is made by forming the ratio of the spread between the process specifications (the specification "width") to the spread

of the process values, as measured by 6 process standard deviation units (the process "width").

–

Process performance indices (Pp, Ppk) basically try to verify if the sample generated from the process is capable to meet customer CTQs (requirements). Process performance differs from process capability and is only used when process control cannot be evaluated.

Step 5: Conduct Gage R&R Studies

SGage

5.15 SGage

Gage repeatability

0.5%0.5%

Standard deviation of gage (one appraiser)

= =

∑i 1

n

Sn - 1

f i −( )X Xi

2

Definition of standard deviation:

Width

99% of measurements fall in the gage repeatability

range

X

Appraiser 1Appraiser 3

Appraiser 2

Standard deviation of gage (more than one appraiser)

(Back to Basics)

Measurement Systems Analysis (MSA)


Sources of Variation

PartOperatorBy2

Operator2

ityRepeatabil2

Product2

Total2 σσσσσ +++=

100.00%

Overall Part to Part Repeatability Operator Operator by Part

Reproducibility (Back to Basics)

Reference Minitab Help: GR & R Study (Crossed)- ANOVA Method


How can a measurement system contribute in accepting BAD product and rejecting GOOD product ?

LSL USL

True measurement of the product

Accepting BAD product Rejecting GOOD product

Operator variation(E.g., reading from analog panel)+/-

10 Deg F

Oven instrument variation+/-

5 deg F

True measurement of the product

(Back to Basics)


Effects of Sources of Variation

Product

Overall production variation

Target

What wasproduced

What wasobserved

Process adjustment

Operator variation—reading from analog panel

Oven instrument variation

(Back to Basics)


Step 6: Statistical Process Control (SPC)

Assign Unique ID XXX Man

Machine

Material

Method

Environment

+

+Extended free text about

Special cause

OCAPData base

Inputs from FMEABrainstorming

Short-term Vs Long-term Capability

LSL USLTarget

Time 1

Time 2

Time 3

Time 4

Over long term conditions, a “typical” process will shift and drift by approximately 1.5 standard deviations*.

“Short-term capability” (Cp, Cpk)

“Long-term performance” that includes changes to material, multiple shifts, Operators, environmental changes (Pp, Ppk)

(Back to Basics)

Step 6: Statistical Process Control (SPC) – Calculate Cp, Cpk if the process is stable.

Baseline Capability

Oven temperature

Oven time

Raw material aging Vegetable aging

Order processing time

Order handling time

Order replacement time

Order handling time

Order response time

Pp= 0.8 Ppk= 0.6 Pp= 1 Ppk= 0.93 Pp= 1 Ppk= 1

Pp= 0.8 Ppk= 0.7 Pp= 0.8 Ppk= 0.8 Pp= 0.6 Ppk= 0.6

*Pp= 0.9 Ppk= 0.7 *Pp= 1.2 Ppk= 1.1 *Pp= 1.3 Ppk= 1.2* Data transformed

Baseline GR&R

Step 7: Using Process Capability and GR&R to Identify Improvements

Capability & GR&R Grid

% G

R&

R*

Cpk/Ppk*

Low

High

Low High

>24%

<24%

<1.1 >1.1* GR&R 24%, Cp, Cpk 1.1 are an example. Decide what is acceptable for your organization.


Scenario — High GR&R + Low Cp/Pp & Cpk/Ppk

LSL USL

True value of the part


Instrument variationrepeatability

Operator variationreproducibility




Processshift

% G

R&

R

Cpk/Ppk

Low

High

Low High

% G

R&

R

Cpk/Ppk

Low

High

Low High

(Back to Basics)

Scenario — Low GR&R + Low Cp/Pp & Cpk/Ppk

LSL USL








ProcessShift

% G

R&

R

Cpk/Ppk

Low

High

Low High

% G

R&

R

Cpk/Ppk

Low

High

Low High

(Back to Basics)

Scenario — High GR&R + High Cp/High Cpk

LSL USL








% G

R&

R

Cpk/Ppk

Low

High

Low High

% G

R&

R

Cpk/Ppk

Low

High

Low High

(Back to Basics)

Scenario — Low GR&R + High Cp/Pp & Cpk/Ppk

LSL USL








% G

R&

R

Cpk/Ppk

Low

High

Low High

% G

R&

R

Cpk/Ppk

Low

High

Low High

(Back to Basics)

Capability & GR&R Grid

% G

R&

R

Cpk/Ppk

Low

High

Low High

CTQ1CTP3

CTQ2CTP2

CTQ3

CTP1

Now that we know the baseline data of performance indices/capability & GR&R of our pizza-making CTQ & CTP, let us place them in the appropriate quadrants.


Step 8 - Prioritizing Improvement Efforts (Process Health Card)

•If new test station/equipment added, operator changed, equipment

overhauled, new GR&R study is required. If no changes, 6 months

frequency ofGR&R monitoring is a good practice.** If there has been sudden change in process variation (for good or bad), extended period of lack of stability, a new study has

to be conducted and control limits recalculated. If no changes, 6 months frequency of review of control limits is a good practice.

CTQ: critical to quality characteristics. CTP: critical to process parameters.

Relationship between CTP and CTQ to be established up front.

Pro

duct

Pro

cess

CTQ1

CTQ2

CTQ3

CTP1

CTP2

CTP3

LCL UCL Stability** Cp/Pp Cpk/Ppk

DateCL

ESTB.GRR%DateGRR*

Alpha/Beta Risk%

GRR/ CL Next due

Date

Significant Moderate Weak

CTQ 1 2 3

1

2

3

CTP

Relationship

01/07

01/07

01/07

01/07

01/07

01/07

07/07

07/07

07/07

07/07

07/07

07/07

37%

8%

25%

7%

12%

25%

01/07

01/07

01/07

01/07

01/07

01/07

20

1.30

15

200

1.5

1.7

24

1.80

18

208

1.7

2.0

NO

YES

NO

YES

YES

NO

0.8

0.9

1.2

1.3

1.00

0.95

0.6

0.88

1.15

1.25

0.92

0.82

Alpha/Beta errors can be obtained from statistical software misclassification feature, or by using simulation software.

Data query/retrieve (real time –

where possible)

Validate data sequence

Control chart

Identify special causes (If not stable)

Summarize the Cp, Cpk, Pp, Ppk, GR&R%,

false acceptance, false reject

Establish severity,

occurrencedetection scales

Identify process steps

to perform FMEA

Identify failure modes, causes,

effects, current controls

& risks

Identify equipment

Plan GR&R experiment

Measure capability Cp/Cpk

(if stable)

Cp, Cpkacceptable?

By process date?By lot sequence?By measure date?

SPCFMEA GR&R

Yes

Analyze data

Measure GR&R

Identify critical process

variables tomonitor, assign

RPN

Developcontrol

plan (CTQ,CTP)

Estimate process performance indices

Healthcard

Prioritize improvement efforts

CTQ vs CTPmatrix

Continue monitoringstability & process capability

No

Triggering Actionable Discussions

•

Out of all CTQ and CTP from a given product line, prioritize a vital few for improvement actions. In this example, CTQ1 and CTP3 are prioritized.

•

By improving CTQ1 and CTP3, we can reduce the producer/consumer risks to a set goal acceptable by customers and manufacturing.

•

Improve gage R&R to <10%. Improve process capability indices >1.5.

•

Move items from red, yellow, and blue zones to green based on prioritization.

One might ask…•

Why go through these process steps? Why not focus on low process capability to start with?

•

Answer: This process helps …–

Understand whether the CTQs and CTPs that are measured are traceable to customer needs.

–

Prioritize improvements using the CTQ-CTP relationship matrix.

–

Review priority for improvement in relationship to measurement capability. (As a containment, organizations would rather risk losing yield than sending nonconforming products to customers. 1% of incorrectly accepted products is 10,000 PPM.)

About Engineering Meetings•

Meeting discipline issues narrated in this presentation are common to any organization in general and not targeted on any specific organization.

•

There is more to engineering meetings than Cpk and Gage R&R: e.g., engineering changes, machine maintenance issues, budget control, etc.

•

This presentation is targeted to help quality professionals and engineering professionals involved in quality improvement and does not suggest replacing the entire engineering meeting.

Acronyms & Definitions•

CTQ: critical to quality (characteristics)

•

CTC: critical to cost/customer (characteristics)•

CTP: critical to process (parameter)

•

GR&R: gage repeatability & reproducibility•

LSL: lower specification limit

•

USL: upper specification limit•

FMEA: failure mode effects analysis

•

RPN: risk priority number

•

Alpha risk: probability of rejecting good products•

Beta risk: probability of accepting bad products

Acknowledgements, References, & Bibliography•

Acknowledgements:–

Ms. Noel Wilson, ASQ -

Review, feedback, and support.–

Mr. Steven Hunt-

@ Risk Misclassification & Simulation.–

Ms. Cathy Akritas, Minitab Inc-

Help with Misclassification macro.–

Mr. Ed Russell, Mr. John Noguera, Mr. Andrew Sleeper -

Suggestions and guidance for Misclassification Simulation.

•

References:–

Concepts for R&R Studies, ASQ Press, by Larry B. Barrentine.–

FMEA Minus the Headache,”

QP, April 2009, by Govind Ramu.–

Measurement Systems Analysis Manual,AIAG.–

MINITAB 15 Help Menu. •

Bibliography:–

http://www.isixsigma.com/forum/ask_dr_harry.asp?ToDo=view&questId=82&catId=11–

AIAG Statistical Process Control –

SPC–

http://www.onesixsigma.com/crystalball/Misclassification-Rates-in-Measurement-Systems-

Analysis-Gauge-RR-01011970

–

http://www.qualitytrainingportal.com/resources/fmea/index.htm

55

http://www.isixsigma.com/forum/ask_dr_harry.asp?ToDo=view&questId=82&catId=11

http://www.onesixsigma.com/crystalball/Misclassification-Rates-in-Measurement-Systems-Analysis-Gauge-RR-01011970

http://www.onesixsigma.com/crystalball/Misclassification-Rates-in-Measurement-Systems-Analysis-Gauge-RR-01011970

http://www.qualitytrainingportal.com/resources/fmea/index.htm

Questions and AnswersThanks:Govind Ramu

http://www.asq.org/sixsigma/about/govind.html

Probability of rejecting good products and accepting bad products for Ppk=0.43 and GR&R 30%: E.g., CTQ1 -

Red Quadrant

Incorrectly accepted = 12.02% (Beta Risk) Incorrectly rejected = 1.80% (Alpha Risk)

Reference AIAG Manual Measurement System Analysis 3rd Edition- Pages 16-22

45

Note:

The exact percentage of errors was calculated simulating the distribution with 100,000 random data points using @ Risk software

and MINITAB Macro.

Probability of incorrectly rejecting good parts

Probability of incorrectlyAccepting bad parts

Probability of rejecting bad parts8.99% both tails

Probability of accepting Good Parts 88.08%

Product Distribution

GRR Error Distribution

Probability of rejecting good products and accepting bad products for Ppk=0.43 and GR&R 10%: E.g., CTQ2 -

Yellow Quadrant

Reference AIAG Manual Measurement System Analysis 3rd Edition- Pages 16-22

Incorrectly accepted = 5.22% (Beta Risk) Incorrectly rejected = 0.68% (Alpha Risk)

Probability of rejecting bad parts9.77% both tails

Probability of accepting Good Parts 89.28%

Probability of incorrectly rejecting good parts

Probability of incorrectlyAccepting bad parts

Product Distribution

GRR Error Distribution

Note:

The exact percentage of errors was calculated simulating the distribution with 100,000 random data points using @ Risk software

and MINITAB Macro.