Introduction & Product Background
This project
will focus on reject reduction, optimization and improving efficiency of the Subassembly
& Blister Pack Assembly Line. Current reject rates are at 12% and if
successful 50% to be decreased in scrap without increasing budget. Total
project saving expected of €11.400 per year and cost of increase of production
yield to be determined.
Eight different components are used across four stations on
the Subassembly Line and the Blister Pack Assembly Line to form the final
packaged product. The assembly process include a welding station, 100% online
automated test machine, manual assembly, packaging and printing.
Define Phase
The aim of the define phase is to determinate what are the
rejects and the efficiency level of the assembly lines by reviewing historical
data. My project is in the area which I complete daily quality tasks as a part
of routine production. Overall the define phase was more time consuming and
challenging then I expected.
To begin with I organized a meeting (department of quality,
engineering and production member) where we clearly set the project structure
and project plan. After the meeting I was able to create Project Charter,
reference in Figure 1 below.
 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Figure 1 Project Charter
Next, I mapped the process and create the process flow -
SIPOC Diagram to ensure I understood the core process and all different stages
of the assembly lines, reference in Figure 2 below.
 Figure 2 SIPOC Diagram
To identify the most common rejects and its location on the line, I collected data from six shop orders manufactured during four months and created Reason for Failure Check Sheet, reference in Figure 3 below.
Figure 3 Reason for Failure Check Sheet
Based on information gather from the data listed in Figure 3,
I created Pareto Chart (reference in Figure 4 below), which conform to some extent
80/20 principle. Base on this principle the top three reject accumulated to 74.9%
of a total failures. The test machine is the most frequent reason for failure
with 777 rejects found (31.7%). The Pareto Chart was very useful tool to show
what areas are to be targeted as a part of the project reject reduction.
Figure 4 Pareto Chart
To show the efficiency of the production lines I calculated
the Performance Rate from the formula below.
I collected data from six shop orders manufactured during
four months and calculated performance rate at 101% (Project Proposal). Since
the project started all the data was reviewed in more detail and indicate that
the Ideal Cycle Time is set incorrectly as new performance rate was at 108, refer
to Figure 5 below.
Figure 5 Performance Rate Check Sheet
Finally base on the information gathered during the Define
Phase I created a project plan – Gantt Chart to help me visualize the project
tasks and management timeliness, reference in Figure 6 below.
Measure Phase
The Measure
Phase includes the following sections:
Section 1 - Scope
The Measure Phase was more in
depth than expected before I started the work. During this phase I have learned
a lot about the assembly & test process from an engineering point of view.
Base on the information gathered
in the Define Phase it was decided to concentrate on the three areas of reject
reduction:
To gain an understanding on why
the current process in each of the three areas is causing rejects a number of
data sets were collected, samples tested and studies completed.
Also, reviews of the Performance
Rate inputs were completed as a part of improving efficiency & optimization
of the lines.
Section 2 - CTQ Tree, Process Mapping & Data Collection Plan - Operational Definitions
I started with creating the Critical
to Quality (CTQ) Tree to provide clarity around aspects of the Voice of
Customer. Refer to Figure 7 below.
Figure 7 CTQ Tree
Post the CTQ step I completed the
detail mapping of the real process for the Subassembly & Blister Pack
Assembly Line to help bring clarity to complex processes and highlight the
Bottle Neck station. Refer to Process Mapping below, Figure 8.
Figure 8 Process Mapping
A Data Collection Plan –
Operational Definitions has been created to specify how much data will be
collected and how often. Refer to Figure 9 below.
Figure 9 Data Collection Plan - Operational Definitions
Section 3 - Rejects Reduction
Three areas for reject Reduction:
100% Online Test Machine, Welder & Visual Issue – Gate Stringing.
The measure phase builds upon the
existing data available in order to fully understand the historical behavior of
the process. This was very important step in the project to establish the baseline
in the terms of level of rejects and where the data falls within the control
limits.
This was achieved with collecting
the data outputs from the Test Machine – Station 2 over five days production.
There are 5 different functional test
completed on the Test Machine:
From the reject quantities for
each individual test I was able to create a Pareto Chart. Refer to Figure 10 below. As we can see from the chart below two tests are contributing to 96.5 % of the rejects.
Figure 10 Pareto
Chart – Test Machine
The data from this tests were
subjected to normality tests and Process Capability Sixpack using Minitab
Version 17 to help visually assess
To understand correlation between
the Online Test Machine and the Offline Test Rig a total of 8 days production
rejects from the Online Test Machine has been collected and tested on the
Offline Test Rig. The results from this tests are
displayed on the check sheet below (refer to Figure 11). 56% of the failed
rejects from the Online Test Machine failed on the Offline Test Rigs.
Figure
11 Check Sheet for Rejects from the Online Test
Machine
The ultrasonic welder – Station 1
is a manually operated equipment and is used to weld the Valve Lid to the
Housing with a Silicone Disc in between these parts. There are three control outputs on the welder: Weld Energy, Weld Time and distance RPN after holding time (RPN+).
The welder machine was set to
record data outputs over 5 days production. From this data, I was able to
analyze which outputs are contributing to highest level of rejects. Refer to
Pareto Chart below, Figure 12.
Figure 12 Pareto Chart - Welder
A Minitab analysis was performed
on the welder output data. An expected overall
performance was checked and gave us prediction of the proportions that will
fail the specification limits. A graphical summary of the data was generated and Process Capability Sixpack performed for the three control outputs to help visually assess capability of the process.
All the rejects collected on the
welder have been tested on the Online Test Machine and the Offline Test Rig. Refer
to Figure 13 for the summary of the results. Only one part has failed both
online and offline Valve Seal Integrity Test – Pressure Decay at the Top
Sealing Ring. This tells us that 2.3% of the weld rejects are functional
failures and give as a scope to work on the for rejects reduction.
Figure
13 Check Sheet for Rejects from the Welder
Weld flash/string is present on
the subassembly after welding process (Station 1) but it’s not a visual issue
as remains intact against weld joint. It’s become a visual issue when is being
pulled up.
A control study – 50 parts were assembled
and tested as per usual process. From this study all parts have been inspected
to investigate at which step the weld string is being pulled up. The results are summarised in Table 14 below.
Figure
14 Check Sheet for Weld Stringing Study
I started with measurement of the
Cycle Time during production run over 5 days. Each day I measured time for the 10
final assemblies. The average time (theoretical) for completion of the one unit
was 57 seconds and included operators waiting time for parts and minor
stoppages (work overload at Station 4 – Bottle Neck Station).
Based on this information I
organised a brainstorming session where we revised the all parameters
contributing to performance rate calculation. At the end of the session we were
all confident that the Operating Time includes non-productive time and this is
contributing to the issue with high Performance Rate. We estimated that on the
average 14% of the time each day is non-productive on the assembly lines.
Section 5 - Gantt Chart update
The Gantt Chart was updated to reflect changes
made to the project during the Measure Phase, refer to Figure 15.
Figure 15 Gantt Chart
Analase Phase
The Analyse
Phase includes the following sections:
Section 1
The aim of Analyse Phase was to:
Section 2
To begin with we completed a Control Study with Open Limits to understand where the outputs are
falling when the welder stops once one of the output is not met. The specification limits have been selected by
Engineering Leader:
A total of 100 parts has been made and data
outputs collected. The data from this study were subjected to normality tests
and Process Capability Sixpack. All samples have been tested on the Online Test
Machine and Offline Test Rigs.
Next we completed the Verification Run based on the information gathered
during control study. New specification limits for three outputs have been
suggested:
A total of 500 parts has been made and data
outputs collected. Also, all samples have been tested on the Online Test Machine. A graphical summary of the data and Process Capability Sixpack was generated. An expected overall
performance was checked and gave us prediction of the proportions that will
fail the specification limits:
Summary of the information
gathered during verification run with new suggested limits:
Finally we completed a functional test and checked weld strength - Valve Lid Weld Tensile, specification limit ≤ 200 Newton. 50 samples have been selected
randomly and independently from 500 parts made during verification run. 25 samples have been taken with at least one of output at
lower end of the specification and another 25 samples with output at higher end
of the specification. All samples passed specification limit and confirmed that parts are functionally acceptable. Refer to Box Plot below, Figure 16.
Figure 16 Box Plot - Valve Lid Tensile Test
Section 3
To begin with the verification run was completed to find out the operators
influence for weld string creation. A total of 500 parts were assembled and
tested as per usual process. Two operators have been involved with visual
inspection of the parts. The results are summarised in Table 17 below.
Figure 17 Check Sheet
for Weld String Verification Run – Operator influence
Next I organised a meeting with project team to review all weld string rejects collected during operator influence
study. The weld string is located on the
different areas around weld joint. The size of weld string varies between 0.2
cm – 1 cm. Base on the size of weld string and review of customer acceptance requirements, a
limit sample has been signed off to provide more detail instruction during inspection, refer to Figure 18 below. With new limit sample in place, only 10 out of 23 parts would failed during weld string verification
run and suggest improvement of over 50%.
Figure 18 Weld String Acceptance Sample
Section 4
I organised the brainstorming session
with the project team to identify the potential root causes of the poor productivity /
non-productive time on the assembly lines. Base on the
information gathered during this session I was able to create the Fishbone Diagram to
visually display all potential causes, refer to Figure 19 below.
Figure 19 Fishbone Diagram
Improve Phase
The Improve Phase includes the following sections:
Section 1
The aims for the improve phase
was to:
Section 2
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Based on the information gathered
during Analyse Phase a pilot study was completed on the assembly lines with new
outputs on the welder and improvements made for the weld string rejects
reduction. This controlled trial confirmed the effectiveness of
the proposed changes on the Assembly Lines before full implementation.
To begin with I organised the meeting with project team to plan a pilot study where we clearly set the scope, time frame
and data collection plan. Refer to Figure 20 below.
Pilot Study Plan
|
|
Scope and area of the study (Where?)
|
The scope of the study is to complete controlled
trial on the Assembly Lines: Station 1 – First Stage Assembly, Station 2 –
100% Online Test Machine & Station 3 – Manual Assembly Cell.
New welder outputs from the Verification Run will
be set up on the welding machine (RPN+ - 0.33mm-0.40mm, Energy – 45J-61J,
Time – 0.15s-0.22s).
All parts will be tested on the 100% Online Test
Machine.
A visual inspection of the weld string will be
completed at each station with new limits sample in place
|
Time
frame (When?)
|
A minimum of 8 hours’ shift – 500 part to be made
during this study.
|
Data collection plan (How?)
|
All parts will be number from 1 to 500.
First parts will be assembled on the manual
operated ultrasonic welder, tested in order on the 100% online automated test
machine and visually inspected for the weld string.
All good parts and rejects will be collected by Operators
and results recorded.
Printout of the outputs from the welder and test
machine will be collected by Engineering Leader and review by project team.
|
End
of Table
|
|
Figure 20 Pilot Study Plan
- 7 sample failed during the welding process – this give us 1.4% reject rate
- 2 samples failed for weld string above the acceptable limit sample – 0.4% reject rate
- 19 samples out of 500 failed on the Online Test Machine. These failures are typical of the process rejects on the Online Test Machine and were within the welder parameters
On the analysis of this data a significant reduction of rejects was observed for the welder and weld sting. I created reject reduction charts to display improvement made. Refer to Figure 21 below for the Reject Reduction Chart for the Welder which shows a saving of 2.4% from the baseline figure of 3.8%. Refer to Figure 22 below for the Reject Reduction Chart for the Weld Stringing which shows a saving of 0.78% from the baseline figure of 1.18%.
Figure 21 Reject Reduction Chart - Welder
Figure 22 Reject Reduction Chart - Weld Stringing
Section 3
An pFMEA risk analysis tool was
used to evaluate changes made on the welder outputs. Analyses of the process
key outputs, potential failures and consideration of the effect of process
failure on the product were targeted. Refer to Figure 23 below for analysis up to RPN scoring. I have gained an enormous amount of understanding about the tool and practical experience while working on the completion of this risk analysis.
Figure 23 pFMEA - Welder
Section 4
Based on the information gathered
during the previous phases of the project about non-productive time on the
assembly lines and its potential causes, I organised the brainstorming session with
the project team and production personnel. In order to reduce time wastes at
the bottleneck neck station (Station 4) decision was made to remove visual
inspection point for the loose particulates/black threads. This
additional step was introduced to the line over eight months ago. All quality
and production data from the last 8 production lots have been reviewed (First
Piece Approvals, In Process Inspection Reports, Attributes Control Charts &
Final QC Inspection Reports). There was no issue found during this period with
loose particulates/black threads. Based on this information, it was decided to
complete control verification study during the next production run before full
implementation of the proposed change.
Control Run requirements:
- Temporary changes to be made to all production work instruction and paperwork to remove inspection for loose particulate/black thread at the end of Station 4
- Additional QC inspection to be completed during this run for loose particulate/black thread
- All operators and QC Inspectors to be trained on the required changes
- Data to be reviewed on the daily basis at the morning meeting by management
From the data gathered during first four day of the control production run we estimate improvement of non-productive time from 14% to 7%-5%.
Control Phase
Results:
Result: The goal was achieved and potential annual saving of €11400 .
Result: The production yield has increased by 5.7% from a baseline of 88% to 93%.
Control Phase
The Control Phase includes the following
sections:
- Section 1 – Scope of the Control Phase
- Section 2 – OEE Rate Metric & Stop Even Log
- Section 3 – Goals summary
- Section 4 – Lesson Learned and My Contribution
Section 1
The aim of the control
phase was to ensure that the improvements that have been implemented become
embedded into the process, maintained high quality level and sustained after
the project has been closed.
Section 2
To control the
rejects level and the performance rate together with availability & quality
rate I drafted the template for the OEE Rate metric calculation. Refer to Figure 24 below. I found this metric very informative as overall efficiency of the assembly line will be controled and reviewed for each lot.
Figure 24 OEE Rate Metrics
Also, I created a Stop Event
Log (refer to Figure 25) to record all downtime
occurred during production in order to get accurate values when calculating the
OEE Rate and control non-productive time.
Figure 25 Stop Even Log
Section 3
The project has
achieved the majority of its goals and proved that the reject reduction &
optimization of the assembly lines was possible.
The following
were the goal of the project:
1. Overall
reject reduction on the Subassembly Line and the Blister Pack
Assembly Line from baseline 12% to 6%.
Results:
Results:
- Welder – Proposed changes to the welder outputs show a substantial reject reduction of the 63% from the baseline of 3.8%.
- Weld String - the goal was achieved and 67% of the reject reduction has been made from the baseline of 1.2%.
- Reducing level of the reject on the welder and minimizing level of the weld string show that overall reject reduction of 5% from the baseline of 12%
Results:
- Process flow on the lines have been improved due to elimination of the time wastes at the bottleneck station.
- Non-productive time have been reduced by minimum of the 50% and significant increase of the output has been noted from the average of 450 parts being made per shift to minimum of 500 parts.
3. Overall financial saving after scrap reduction and optimization of the assembly lines -
potential annual saving of €11400.
Result: The goal was achieved and potential annual saving of €11400 .
4. Increase
Production Yield
Result: The production yield has increased by 5.7% from a baseline of 88% to 93%.
Section 4
Lesson learned
Using DMAIC approach through the project showed
the systematic and informative method of addressing the issues. Far better
understanding of the processes, exact rejects causes & use of the data
driven quality approach are one of the few things learned during the project. All
future projects (e.g. reject reduction on the moulding machines) should be
addressed by using DMAIC methodology.
The project was conducted between the Blister Pack Assembly routine production runs which meant that the time for the
control studies/verification runs was very limited. Time constrain for the submission of the
project together with limited access to the lines were very difficult to
overcome. It is recommended for the future projects to be scheduled at times of
the year when production is not running.
The project focused primarily on the reject
reduction and optimization of the assembly lines. All other areas were outside
the scope. During the analysis of the data outputs from the rejects testes on
the 100% Online Test Machine and Offline Test Rigs (measure phase) a new
information has been highlighted that the correlation study should be completed
between the the test machine and test rigs. This issue is being currently addressed by
engineering department.
My role and Contribution
The project was
95% my own work with assistance from the Quality & Training Manager and
Engineering Leader. I
drove this project through to completion and successfully reduce reject rate
and optimized the Assembly Lines.