6σ Student Case Study UMICHCE - Oil Fill Level

7/31/2019 6 Student Case Study UMICHCE - Oil Fill Level

1/17

Oil Fill Level Project

Solving Problems with Labor and Maintenance Costs

Prepared by:

DR


2/17

Oil Fill Level Project DR 2

Executive Summary

Problem Statement

Reports of incorrect oil level at assembly plants and dealers have been reported as a significant issue. In response to

these reports, management implemented two inspection points to protect the customer.

The focus and business need of this project is to eliminate hard cost associated with packout inspection. This expense is

$130,000 annually. Secondary benefits include the opportunity to rebalance operation OP400 and reduce internal

repair costs. And because inspection is not 100% effective, a reduction in DPM will directly improve quality to the

customer.

Problem Solving Approach Utilized

The DMAIC process was utilized throughout this project. A cross-section of both qualitative and quantitative tools was

used. Some key methods were ishikawa diagrams, run charts, hypothesis testing, and distribution analysis. Pp and Ppk

values were the primary indicators of performance.

Major Project Results and Recommendations

After review of the current state data, the focus of this project was oil level variation. The key contributor was the cold

test stands. This was highlighted by a change in strategy to focus on oil weight, instead of oil level, as our response

variable. As you can see in the main body of the report, the other main contributors, such as location of the oil and

component variation, were eliminated.

A design flaw within the oil delivery circuit was the primary issue responsible for the variation. The oil circuit had a

bleed off circuit that provided an internal leak path. This bleed off circuit was the safety mechanism used to relieve

pressure in the oil delivery system during maintenance activity. In other words, this meant that oil counted as entering

the engine was being returned to the storage tank. To correct this issue a normally open solenoid valve was installed.

As a result of this project and the changes mentioned above and in the body of the report, the 100% inspection was

eliminated at packout and OP400. The packout inspection removal resulted in a $130,000 savings by redeploying two

team members. The removal of OP400 inspection resulted in a (3) second reduction in work time. And more

importantly, the housekeeping and safety issues associated with oil on the floor at OP400 are no longer a concern.

Another potential benefit was realized during this project. It was found that all engines are being filled above nominal

by approximately 230ml per engine. This amounts to $65,000 per year savings potential.

Control mechanisms were implemented to maintain the current condition. An audit process was implemented to

provide ongoing variable data. This data will be used to provide real-time feedback about the oil delivery system health.


3/17


1.0 Improvement Opportunity: Define Phase

1. 1 Discussion of the process being examined and problem addressed

Reports of incorrect oil level at assembly plants and dealers have been reported as a significant issue. In response to

these reports, management implemented two inspection points to protect the customer. One inspection point was

added to the final assembly operation at OP400. A second inspection point was added in packout located after the

engines are racked for shipment. See the diagram below.

OP10

Line Start

OP400

Line End

123

= Inspection Point

Packout

Inspection

Area

Cold Test

(Oil Fill Equipment)

Shipping

The measurement method used by assembly plants and dealers is based on a visual inspection of the dipstick after warm

the engine and allowing five minutes for oil to drain back into the oil pan. Specification limits at the assembly plant are c

defined as 10-24mm above the bottom of the safe zone. See the diagram below labeled as CSA Audit Range.

Differences in engine conditions (angle and temperature) and measurement technique exist between the engine plant an

vehicle assembly plant. At the engine plant, a graduated hand gage is used for measurement. At vehicle assembly a

production dipstick is used. Mapping was completed to align the measurement values between the two sources. A follo

hypothesis test yielded a high p-value. This indicated no significant difference exists between the values obtained using

two measurement methods. In other words, we could now take measurements using the engine plants gage at time of

predict the assembly plant measurements. See below for details.


4/17


5/17


Stand1

Stand2

Stand3

Fill pt 1

Fill pt 2

Initial Fill

(~ 3.5 quarts)

Oil Test & Running

Torque

(~ .7 quarts)

Top Off

(~ 1.5 quarts)

1

2

3

Step Fill Pt

1/2

2

1

Oil

Supply

Totalizer

Orifice

Engine

1.2 Key measurements defining project successo Reduction in common cause variation and achievement of 1.44 Ppk.o Removal of packout inspection and the associated expense.o Removal of operation OP400 inspection needs for 1st time-pass engines.

1.3 Project scope

The focus and business need of this project is to eliminate hard cost associated with packout inspection. This expense is

$130,000 annually. Secondary benefits include the opportunity to rebalance OP400 and reduce internal repair costs.

And because inspection is not 100% effective, a reduction in DPM will directly improve quality to the customer. The

vehicle assembly plant is referred to as the customer for this project. The DMAIC process was followed.

Historical records related to out-of-specification claims were not available. The quantity and condition of non-

conforming engines sent to the customer were unknown. In addition, engines found at the internal inspection points

were not being recorded. We had little more than subjective opinions about the current status at the beginning of this

project.

This project will focus on 1st time pass engines only. Reworked engines are viewed as special cause situations due to the

human interface and are removed from the data analysis. Additionally, turbo engines have different specifications and

will not be included in the study. Lessons learned from this project will be applied to these cases.

2.0 Current State of the Process: Measure Phase

2.1 Current performance level

As mentioned above, there was no objective data available to assess the current state. This required initial data is

gathered to facilitate problem definition. Measurements were taken at three inspection locations; OP400, packout, and

stable. Data from the stable condition was used as the current state and provided a baseline for the project.


6/17


2422201816141210

LSL USL

LSL 10

Target *

USL 24

Sample Mean 19.5446

S ample N 258

StDev(Within) 1.71902

StDev(Ov erall) 2.02296

Process Data

C p 1.36

C P L 1. 85

CPU 0.86

Cpk 0.86

P p 1.15

P P L 1. 57

PPU 0.73

Ppk 0.73

C pm *

O verall Capability

Pote ntial (Within) Capability

P PM < LS L 0.00

PPM > USL 11627.91

PPM Tota l 11627.91

Observed P erformance

P PM < LS L 0.01

PPM > USL 4773.10

PPM Tota l 4773.11

Exp. Within Performance

P PM < LS L 1.19

PPM > USL 13817.46

PPM Total 13818.65

Exp. O verall Performance

Within

Overall

Baseline PCA for stable, -rework, -turbo, -cf

The current state data was stratified to analyze stand-to-stand and model-to-model differences (see below). Hypothesis

testing confirmed that not one stand or model was significantly different than the others. But from a practical

standpoint, cold test stand 2 had the highest standard deviation. Focus was placed on this stand for testing and

improvements. Lessons learned would be shared to the other two cold test stands. In addition, the three locations

were analyzed to understand the impact of time on the oil level readings. Below are results.

3

2

1

3.002.752.502.252.001.751.50

coldteststand

95% Bonferroni Confidence Intervals for StDevs

Test Statistic 8.87

P -V alu e 0. 012

Test Statistic 3.35

P -V alu e 0. 037

Bartlett's Test

Levene's Test

Baseline TEV, stable, al l stands, -rework, -turbo, -cf

CH

CG

CE

CA

BA

AA

3.02.52.01.51.0

model


Test Statistic 4.03

P -Valu e 0. 545

Test Statistic 0.79

P -Valu e 0. 557

Bartlett's Test

Levene's Test

Baseline Test for Equal Variance for stable by model, -rework, -cf


7/17


2422201816141210

0.20

0.15

0.10

0.05

0.00

oil level adj

Density

10 24

16.69 2.262 258

18.56 2.112 201

19.54 2.023 258

Mean StDev N

OP400.1

packout

stable

location

measure

oil level

Normal

Baseline Histogram for all locations w/ updated specs, -rework, -turbo, -cf

Below is a bulleted list that summarizes the project findings before progressing to the analyze phase. The analysis tool

and conclusion is listed for each. Some of the supporting evidence is shown in the appendix.

- The Engine Plants hand gage measurement technique is acceptable.o Analysis method paired measurement scatter ploto Conclusion this method can be used to measure oil level throughout the project

- Difference between the engine plants readings (stable) and assembly plant readings are insignificant. (This istrue because the engine plants gage was adjusted to correlate properly.)

o Analysis method hypothesis testing f-test, p-value .683 t-test, p-value .418

o Conclusion The Engine Plant can use the same specification limits as the assembly plant 10-24mm

- Engine rack angle has an impact on the oil level readings.o Analysis method distribution analysis & square root of sum of squares

1 = 2mm Estimated s impact of .25mm

oConclusion oil level measurements will be taken while engine is mounted on pallet prior to beingplaced in engine rack

- No cold test stand is significantly different than the other two stands. From a practical standpoint stand 2 isworse than the others.

o Analysis method hypothesis testingo Conclusion testing will be completed on stand 2 and read across to the other stands

- No model is significantly different than all the others.o Analysis method hypothesis testingo Conclusion any model can be used for analysis

- Oil level readings change significantly from the time of fill to the time of shipment.


8/17


o Analysis method run chart, distribution analysis, hypothesis testing p-value 0.000 (t-test) OP400 (~5 min after fill) = 0 (baseline) Packout (~1 hr after fill) = +2mm Stable (~24 hrs after fill) = +3mm Mean differences are significant, variation differences are not significant

o Conclusion target oil levels must account for change based on measurement location, the total change level between fill and shipment is approximately 3mm

2.2 Identification of key variables

The diagram below was developed to break the potential causes into (4) distinct categories. A fishbone shown in the

appendix lists individual potential causes along with the rationale for elimination throughout the project.

Oil Level

Variation

@ Dipstick

Oil Supply

Location of Oil

Components

Measurement

2.3 Identification of target performance levels or project goals

Based on the current state data, this project is primarily focused on variation reduction. The project goal is to identify

the primary sources of variation affecting oil level and therefore eliminate the need for costly inspection. The

quantifiable targets are shown below.

o 1.94 Pp (s = 1.2mm)o 1.44 Ppk (accounts for 1.5s mean shift)o < 10 DPM

3.0 Analysis and Findings: Analyze Phase

In the measure phase above we realized stand-to-stand and model-to-model differences were insignificant. A look at thi

in time series highlighted drastic changes in oil level between consecutive engines.


9/17


12/5/

083:35PM

12/5/

082:37PM

12/3/0

812:46PM

12/2/0

812:45AM

12/2/0

812:43AM

12/2/0

812:26AM

12/2/0

812:12AM

12/2/0

812:10AM

12/2/0

812:04AM

12/2/0

812:03AM

12/2/0

812:01AM

12/1/0

811:56PM

12/1/0

811:53PM

12/1/0

811:36PM

12/1/0

811:15PM

12/1/0

811:04PM

23

22

21

20

19

18

17

16

15

14

cold test date/time

oilleveladj

Baseline Time Series Plot for stable, stand 2, CA, -rework, -cf

6mm change

in (20) minute

time period

The oil delivery cycle includes an air purge just before the coupler disconnects from the engine. A quick test of this

system was performed. The results showed this function was insignificant to oil level.

20

10

0

3.02.52.01.51.00.50.0

blowdowntime


Test Statistic 0.05

P -Valu e 0.978

Test Statistic 0.05

P -Valu e 0.954

Bartlett's Test

Levene's Test

Blow Down Test for Equal Variances at stable

In the diagram at the beginning of section 2.2, (4) categories of variance are shown. A strategy change was made at this

point in the project. The response being measured was changed from oil level to oil weight. This change provided an

opportunity to split the potential causes. Below illustrates the rationale.


10/17


Oil Level

Variation

@ Dipstick

Oil Supply

Location of Oil

Components

Measurement

Dictionary split can be

made between these by

using oil weight.

Addressed in

measure phase.

Consecutive engines were weighed before and after oil addition. Similar to the original baseline data, this test resulted i

consecutive engines having drastic differences. This focused our attention to the oil supply. In other words, location of o

components can not influence oil weight within the engine and therefore the variation must be caused by the cold test st

(oil delivery system).

As mentioned in section 1.1, oil delivery can be broken down into three segments. In effort to make another strategy sp

each segment was analyzed individually and the variation reviewed. No segment showed separation. In other words, alsegments contributed relatively equally to the overall variation. This indicated the issue was not segment dependent.

3a23a2321a

0.1

0.0

-0.1

-0.2

-0.3

segment

actual-r

ecorded

Individual Plot of actual-recorded by segment

.14 qt

.08 qt

.14 qt

.22 qt .09 qt

Points indicate the amount

of error between actual

volume delivered &

volume the stand believes

is present

Square root of sum of squares = .061 qts

(accounts for 51% of entire process)


11/17


Next we reviewed the oil delivery circuit and its components with the OEM of the cold test stands. Upon inspection of t

circuit it was realized a bleed off circuit offered a potential internal leak. This meant that oil counted as entering the e

could be redirected. The bleed off circuit was in place to depressurize the circuit during maintenance and eliminate sa

concerns. Testing of a blocked bleed off circuit was completed and this had a significant impact as seen in the f-test belo

using the square root of sum of squares, a 25% (.53mm) reduction in standard deviation was calculated. See below and

appendix for details.

123b

123

0.40.30.20.10.0

segment_1


123b

123

10.310.210.110.09.99.89.7

segment_1

oil wt_1

Test Statistic 8.27

P -Valu e 0.004

Test Statistic 14.27

P -Valu e 0.002

F-Test

Levene's Test

Test for Equal Variance of bleed off circuit affect

The totalizer was the last potential cause evaluated. As mentioned above, this is the device that pumps and counts oil

as it enters the engine. After inspection and testing it was found that this unit had no significant impact but it did have

some practical impact. In the control phase are the steps we took to address it. See the summary graphs shown in the

improve phase for the overall impact.

4.0 Recommendations: Improve Phase

Based on the analysis above, improvement was made to the cold test stands to address the bleed off circuit. A

normally open solenoid valve was installed. This valve remains closed during oil delivery to close off the circuit and

eliminate the internal leak possibility. When the cold test stands are E-stopped prior to maintenance, the solenoid valve

will open and relieve pressure and the associated safety concern. See below.


12/17


OilSupply

Totalizer

Bleed Off Loop

Solenoid

ValveAdded Valve

Orifice

Engine

In addition, the totalizer for cold test stand 2 was replaced after finding an insufficient seal between the piston ringsand bores.

A summary of final results for model CG are shown below. It can be seen by the hypothesis testing and distribution

analysis that all three cold test stands have made the required improvement. Each of them is achieving a standard

deviation of 1.2mm or less. This standard deviation results in the required Pp. Adjustments can easily be made to

achieve the required Ppk.

2322212019181716

0.6

0.5

0.4

0.3

0.2

0.1

0.0

oil level adj

Density

21 0.8660 11

18.27 0.8148 31

Mean StDev N

0

6

sequence

Normal

cold test stand = 3

Histogram of seq 0 vs 6, CG, std 3

2322212019181716

0.6

0.5

0.4

0.3

0.2

0.1

0.0

oil level adj

Density

18.92 1.428 12

20.16 1.069 35

Mean StDev N

0

6

sequence

Normal

cold test stand = 2


2322212019181716

0.6

0.5

0.4

0.3

0.2

0.1

0.0

oil level adj

Density

18.43 0.9759 7

19.13 0.6734 43

Mean StDev N

0

6

sequence

Normal

cold test stand = 1



13/17


Impact of

Changes

Baseline

Original valve install

Totalizer swap

Encoder swap

Valve swap

The final step was to remove the inspection points at packout and OP400. This was completed on 3/1/10 and control

mechanisms were implemented. These are discussed in the control phase.

5.0 Monitoring and Control: Control Phase

As control mechanisms, the following items were completed. As mentioned in the define phase, this project was

targeted at 1st time pass engines. As a result you will notice 100% inspection remains on all repair engines. This

accounts for approximately 1% of the engines.

- Equipment drawings were updated to reflect the newly installed solenoid valve.- Information about the oil delivery circuit modification was shared with The Engine Plants new product launch

team.

- Preventative maintenance activity was added to monitor the condition of the totalizer.- An audit process was implemented requiring the OP400 team member to inspect oil level, on a frequency basis,

for 1st time pass engines and input values into the HMI. Repair engines remain at 100%. In both cases these

values are sent to the corporate server and can be accessed at any time to monitor the process.

- Improved the OP400 oil removal process by adding a coupler that locates on the dipstick boss and removes apredetermined amount of oil. This is used for repair engines.

- Improved the OP400 oil addition process by adding a graduated beaker with associated millimeter markings.This is used for repair engines.


14/17


6.0 Conclusion

As a result of this project and the changes mentioned above, the 100% inspection was eliminated at packout and OP400.

The packout inspection removal resulted in a $130,000 savings by redeploying two team members. The removal of

OP400 inspection resulted in a (3) second reduction in work time. And more importantly, the housekeeping and safety

issues associated with oil on the floor at OP400 are no longer a concern.

Another potential benefit was realized during this project. It was found that all engines are being filled above nominal

by approximately 230ml per engine. This amounts to $65,000 per year savings potential. All model types will be

adjusted to nominal over the next several months to attain these savings.


15/17


Appendices


16/17


Oil weight

variation

Man Machine

Material Method

Oil passages on block

Oil pan volume

Incorrect oil type

Crank oil hole volume

Rear seal

-

-

-

Incorrect gage used

Gage read incorrectly

A

A: checked against standard

B: completed MSA

C: inspected oil lines

D: would not cause variation

E: would not impact oil weight

F: test completed, no impact

H: test confirmed bleed off circuit

impact of .53mm sigma

I: test completed, some impact

Rationale

Incorrect gage design

Blow down

Oil drainback

Correction factor

Recycle

Accumulator

Oil circuit external leak

Totalizer (oil meter)

Oil heater

Valve sequencing logic

B

C

D

A D

Oil circuit internal leak

Parameter settingsA

E

E

E

E

H

F

Engine angle during checkE

I


17/17

6σ Student Case Study UMICHCE - Oil Fill Level

Documents

Transcript of 6σ Student Case Study UMICHCE - Oil Fill Level