DMAIC

Define, Measure, Analyze, Improve, Control

The Problem: A company which supplies parts and services has certain requirements from its customers, which include low cost, reliable assemblies, and on-time delivery. While all three factors are important to both the supplier and the customer, the customer has identified delivery time as their primary concern since they do not have the resources to handle the variation. The supplier is late far too often, and if the supplier ships the assemblies too early, it causes problems for the customer.

The company held an internal meeting to see what they can measure at their facility as a predictor of what their facility needs to improve. Delta Days to Required Date was determined to be the most vital measurable factor.

Product Requirements

Customer Expectations

Impo

rtan

ce

Del

ta D

ays

to R

equi

red

Dat

e

$ ch

arge

d pe

r un

it

Mea

n T

ime

to F

ailu

res

Inst

anta

neou

s F

ailu

re R

ate

% m

arku

p

Mea

n T

ime

Bet

wee

n F

ailu

res

Day

s fr

om d

ock

to d

ock

Mat

eria

l $

Tot

al

Voi

ce o

f C

usto

mer

On-Time Deliver 5 H L 50Low Cost 3 H M L M L L L 57Reliable 2 H H M L 44

Total 45 27 27 21 9 9 8 5

In the above, each H(igh) has a value of 9, each M(edium) has a value of 3, each L(ow) has a value of 1To find the total value, multiply the value of each letter by its customer importance for that row

Areas of Application: Regression Statistics – histogram, mean, standard deviation, correlation plot Excel Mini-Tab (optional instead of Excel)

Materials Included:Excel files:

DataFile.xls GageRR.xls MSAResults.xls ICD.xls

Goal: Since the supplier has decided that delta days to required date has the highest priority in keeping their customers satisfied, the supplier would like to explore the various factors that cause the variability in delta days and determine mathematically which, if any, can be controlled. The delivery dates acceptable to the customer are in the time frame of 10 days early and 20 days late 95% of the time.

Questions: Define, Measure, Analyze, Improve, Control

There are a number of questions the supplier would like to explore. Use the DMAIC process to drive out the variability.

Define: What is important to the customer?

Measure: How high is the quality of the data available? Could it be that the problem lies in the

measurement system and not the process itself? The quality of data is and always should be an area of concern. There is always the tendency to have human errors and differences in measuring. Use the Gage R&R (GageRR.xls) tool and MSA Results (MSAResults.xls) to determine if the variation is too large between the two operators. Explain and justify your answer.

Working with the Gage R&R Tool. You must enter data in all of the green fields:A. Enter operator measurement data from MSA Results.xls.

o The spreadsheet then calculates the range for each part, then calculates an average range between the two operators.

B. Enter the identical Part ID next to Part Number C. Enter the number of operators: 2

o Enter the number of parts: 10 The spreadsheet will then determine the d2* value, based on a standard

table of reliabilities that takes into account the number of operators versus the number of parts.

D. Your tolerance for delivery has a range from 10 days early to 20 days late.o Enter your lower and upper limits for delivery time in LSL and USL, respectively.

The spreadsheet will calculate the toleranceE. The spreadsheet will now calculate several more values:

o First, the spreadsheet estimates standard deviation [Sm (Est.)] by dividing the average range by the d2* value.

o Next, the standard deviation estimate is multiplied by 5.15 (because 5.15 standard deviations represent 99% of the expected population of ranges) to determine the Gage Error.

o Finally, % R&R is calculated by dividing the Gage Error by the calculated tolerance. If the % R&R is less than 30%, we can be confident that measurement is not the primary cause of the variability.

How good is the supplier relative to the customer’s requirements (show graphically)? How far off is the company? Express the answer in terms of mean and standard deviation. What should the mean and standard deviation be in order to meet the customer’s requirements?

Analyze: Look at the fishbone diagram and discuss what you think the two major causes for the

customer delivery problem are. Use data analysis to verify these causes.

Looking at the next fishbone diagram, discuss what you think would be the top two causes of Supplier Delivery variation. Use graphical and/or mathematical verification.

Measurements Materials Men

Environment Methods Machines

Data Base Supplier Delivery

Plant Location

Customer Delivery

Near end of quarter Grinder

Near holiday

Internal TAT

Cleaner

Manual Documentation

Supplier

Potential X’s

Measurements Materials Men

Environment Methods Machines

Tracking System Weld Supplier

Supplier Delivery

Strike

Holidays

Schedule variation Type

Potential X’s

What drives schedule variation? Look at the next fishbone diagram and remember to use graphical and/or mathematical verification.

Looking at the internal process below, which of the three manufacturing “Gates” appear to have the greatest influence on TAT? Prove using the Excel file Data File.xls and making regression plots (scatterplots) between the Gates and TAT and looking at the correlation.

Process Map

Measurements Materials Men

Environment Methods Machines

Parts

Planner

Schedule Variation

End of Quarter End of Quarter

Sales input

Potential X’s

Receive Clean Inspect Machine Clean Inspect Ship

Gate 1 Gate 2 Gate 3

Using the “Family Tree” concept below, fill in the X’s based on the information gathered so far.

Customer Delivery

Supplier Delivery TAT

X X

X X X

X X X

Summary of Variation Sources

Improve and Control: Based on your findings, create an Improve plan complete with recommended actions and

a Control plan for each of your corrective actions that will need to be implemented to control and minimize the variability in the customer delivery problem. Support the Improve plan with your data analysis as you will be trying to convince the finance department and other groups to support your actions.

With the newly implemented changes, customer delivery has improved. Use the new data in Improved Customer Delivery.xls to determine if you have reached your statistical goals. Show your results using graphical and statistical methods. Has the customer delivery improved enough that 95% of the deliveries occur within the given tolerance?

Solutions: Define, Measure, Analyze, Improve, Control

There are a number of questions the supplier would like to explore. Use the DMAIC process to drive out the variability.

Define: What is important to the customer?

Possible Solution: The company needs to deliver goods and services on time and in a consistent manner. Ten days early and twenty days late is the time frame that is acceptable to the customer.

Measure: How high is the quality of the data available? Could it be that the problem lies in the

measurement system and not the process itself? The quality of data is and always should be an area of concern. There is always the tendency to have human errors and differences in measuring. Use the Gage R&R (GageRR.xls) tool and MSA Results (MSAResults.xls) to determine if the variation is too large between the two operators. Explain and justify your answer.

Possible solution: The % R&R is calculated to be 7.84%. This is less than 30%, which indicates that are no significant differences in measurement between the two operators and that measurement is not the primary cause of the variability, so we proceed to look for the true cause of the variability in the process.

How good is the supplier relative to the customer’s requirements (show graphically)? How far off is the company? Express the answer in terms of mean and standard deviation. What should the mean and standard deviation be in order to meet the customer’s requirements?

Possible solution: Develop a histogram in Excel using the file data.xls.

Reality Check

0

20

40

60

80

100

120

140

160

180

-15 -10 -5 0 5 10 15 20 25 30 35 40 45 50 55

-20 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 50

Customer Delivery

Fre

qu

en

cy

55.3%

You can easily see that the company is not meeting the customer’s time frames, as only 55.3% of the deliveries occur within the acceptable timeframe. The mean is 17.96 and the standard deviation is 12.42.

In order to maximize the number of deliveries that occur within the acceptable timeframe, the mean value for the ideal situation would be halfway between the minimum and maximum limits for delivery, 10 days early and 20 days late. Therefore, the mean should be 5 days late.

We know that, based on the empirical rule of a normal distribution, 95% of the data is included within 2 standard deviations of the mean. Therefore, 95% of the data must lie between -10 and 20, or 15 days in either direction of the mean. In order to ensure that 95% of the deliveries occur within the acceptable timeframe, the standard deviation should be 7.5.

Analyze: Look at the fishbone diagram and discuss what you think the two major causes for the

customer delivery problem are. Use data analysis to verify these causes.

Possible solution: Looking at the fishbone diagram, it is logical that Supplier Delivery (how well the suppliers are performing) and Internal TAT (the company’s manufacturing Turn-Around-Time) are the key drivers influencing the customer delivery problem.

To investigate the correlation between Supplier Delivery and Customer Delivery, a scatter plot was generated in Excel, graphing Supplier Delivery (x) vs. Customer Delivery (y). Then the data was regressed (a linear trend line was added) and the correlation coefficient, R, was calculated. The R value was calculated to be .781, which indicates a strong positive relationship between Supplier Delivery and Customer delivery, implying a correlation between the variation in the delivery times of each.

The investigatory process for determining the correlation between Turn Around Time (TAT) and Customer Delivery is similar. The R value was calculated to be .477, indicating only a mild positive relationship between TAT and Customer Delivery.

(See scatter plots on following page)

Measurements Materials Men

Environment Methods Machines

Data Base Supplier Delivery

Plant Location

Customer Delivery

Near end of quarter Grinder

Near holiday

Internal TAT

Cleaner

Manual Documentation

Supplier

Potential X’s

Supplier Delivery vs. Customer Delivery

R = 0.781-30.0

-20.0

-10.0

0.0

10.0

20.0

30.0

40.0

50.0

60.0

-60.0 -40.0 -20.0 0.0 20.0 40.0 60.0 80.0 100.0

Supplier

Cu

sto

mer

TAT vs. Customer Delivery

R = 0.477-30.0

-20.0

-10.0

0.0

10.0

20.0

30.0

40.0

50.0

60.0

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0

TAT

Cu

sto

me

r D

eliv

ery

Looking at the next fishbone diagram, discuss what you think would be the top two causes of Supplier Delivery variation. Use graphical and/or mathematical verification.

Possible solution: Schedule Variation and Supplier are the top two causes of Supplier Delivery variation. Most of the other factors affecting Supplier Delivery cannot be controlled.

Measurements Materials Men

Environment Methods Machines

Tracking System Weld Supplier

Supplier Delivery

Strike

Holidays

Schedule variation Type

Potential X’s

A scatter plot was developed in Excel to compare Schedule Variation vs. Supplier Delivery. The R value was calculated to be 0.728, indicating a fairly strong positive relationship between Schedule Variation and Supplier Delivery.

Schedule Variation vs. Supplier Delivery

R = 0.728

-60.0

-40.0

-20.0

0.0

20.0

40.0

60.0

80.0

100.0

-60.0 -40.0 -20.0 0.0 20.0 40.0 60.0 80.0

Schedule Variation

Su

pp

lier

De

live

ry

To demonstrate that the different suppliers affect supplier delivery variation, the mean and standard deviation were calculated for the delivery times for each supplier, ACME, Best, Kansans, and New Buy. The standard deviations are relatively close for each company: every company delivers 95% of its shipments within about a 72 day time span (2 standard deviations). The differences in variation come into play because of where these times spans are centered, or their mean delivery times. For instance, ACME is on average 6 days early, while New Buy is 35 days late. With a range of 72 days, ACME delivers 95% of its products between 42 days early and 30 days late, while New Buy delivers 95% of its products between 1 day early and 71 days late. Each company is consistent within a 72 day range, but the overlap is minimal, causing a huge variation in overall supplier delivery.

  ACME     Best     Kansans     NewBuyMean -6.3   Mean 24.3   Mean 2.6   Mean 35.0

St. Dev 18.5   St. Dev 18.67859   St. Dev 17.4   St. Dev 17.29501

What drives schedule variation? Look at the next fishbone diagram and remember to use graphical and/or mathematical verification.

Possible Solution: Schedule Variation is affected by Parts, End of Quarter, and Planner.

Parts:

The schedule variations were sorted by parts and their means and standard deviations were calculated.

  A B C DAverage -2.8 1.1 -3.0 28.2

Standard Dev 21.1 21.7186 19.96461 20.25956

As can be seen in the table, parts A through C have an average schedule variation close to zero, meaning that on average the parts come in on time, and similar standard deviations. Part D, however, has an average schedule variation of 28.2, meaning that it is always significantly late, and could account for a lot of the schedule variation in the parts.

Measurements Materials Men

Environment Methods Machines

Parts

Planner

Schedule Variation

End of Quarter End of Quarter

Sales input

Potential X’s

End of Quarter:

The data for schedule variation were sorted by whether the deliveries occurred at the end of the quarter or not. The average and standard deviations were calculated.

EOQ? Average St. Dev.Yes -17.1 21.45522No 13.1 20.39629

The calculations show that, while standard deviation is similar at either time (the delivery times always vary by the same amount), the average time of delivery at the end of the quarter is 17 days early, while the average delivery time otherwise is 13 days late. This indicates that whether or not the delivery occurs near the end of the quarter can cause great variation in the delivery schedule.

The Planner

The data for schedule variation were sorted by the planner, John or Mary. The average and standard deviations were calculated.

Planner Average St. Dev.John 20.2 18.91597Mary -9.8 19.62879

The calculations show that, while standard deviation is similar at either time (the delivery times always vary by the same amount), the average time between planners is very different. John is, on average, 20 days late, while Mary is about 10 days early on average. This indicates that the planner can cause great variation in the delivery schedule.

Looking at the internal process below, which of the three manufacturing “Gates” appear to have the greatest influence on TAT? Prove using the Excel file Data File.xls and making regression plots (scatterplots) between the Gates and TAT and looking at the correlation.

Process Map

Possible solution: Gate 2 has the greatest influence on TAT. The correlation is the strongest (0.816) for this Gate. See scatterplots below.

Gate 1 vs. TAT

R = 0.416

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

0.0 5.0 10.0 15.0 20.0 25.0 30.0

Gate 1

TA

T

Receive Clean Inspect Machine Clean Inspect Ship

Gate 1 Gate 2 Gate 3

Gate 2 vs. TAT

R = 0.816

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0

Gate 2

TA

T

Gate 3 vs. TAT

R = 0.406

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0

Gate 3

TA

T

Using the “Family Tree” concept below, fill in the X’s based on the information gathered so far.

Possible Solution:

*Note: EOQ = End of Quarter

Customer Delivery

Supplier Delivery TAT

X X

X X X

X X X

Summary of Variation Sources

Customer Delivery

Supplier Delivery TAT

Schedule Variation Supplier

Parts *EOQ Planner

Gate 1 Gate 2 Gate 3

Summary of Variation Sources

Improve and Control: Based on your findings, create an Improve plan complete with recommended actions and

a Control plan for each of your corrective actions that will need to be implemented to control and minimize the variability in the customer delivery problem. Support the Improve plan with your data analysis as you will be trying to convince the finance department and other groups to support your actions.

Possible Solution:

Improve/Control: TATGate 2

Since Gate 2 has the highest influence on TAT, the company needs to improve their performance in the inspect and machine areas.

Improve/Control: Supplier DeliverySupplier

Kansans has the least variability of all the suppliers. They are also closest, on average, to the mean date and to the range of delivery times for customer delivery. The company should either use Kansans more or to ask the other suppliers to reduce their schedule variability.Institute a penalty if the suppliers do not deliver their parts within an acceptable range of times.

Schedule Variation

PartsImprove the efficiency of manufacturing part D so that it can be closer to on-time delivery as the other parts are.

End of Quarter

The parts coming in at the End of Quarter are too early. Minimize variations due to the changes that occur near the End of Quarter.

Planner

Because both planners have about the same variability, the company should use one planner for each customer depending on whether they prefer their parts early (Mary) or late (John). There should be no rewards for bringing parts in early and penalties for bringing parts in late. The planners should be rewarded when their average in schedule variation is close to 5 and their standard deviation is close to 7.5.

With the newly implemented changes, customer delivery has improved. Use the new data in Improved Customer Delivery.xls to determine if you have reached your statistical goals. Show your results using graphical and statistical methods. Has the customer delivery improved enough that 95% of the deliveries occur within the given tolerance?

Possible Solution: Develop a histogram in Excel using Improved Customer Delivery.xls

Improved Customer Delivery

0

50

100

150

200

250

300

-15 -10 -5 0 5 10 15 20 25 30 35 40 45 50 55

-20 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 50

Customer Delivery

Fre

qu

en

cy

94.4%

We can see that the company is now very close achieving their goals and meeting their customer’s needs. The deliveries occur within the acceptable time frame 94.4% of the time, an increase of 39.1%. The mean has been reduced to 6.05 days late, and the standard deviation has been reduced to 7.95. This is very close to the ideal situation which prescribes an average of 5 days and a standard deviation of 7.5. The company has done a good job and the customer is happy.