Six Sigma and Simulation – Complementary Solutions
Executive
Summary | What Is Six Sigma? |
Six Sigma Methodology | Case
Study | Conclusion | Project
Reviews
Executive
Summary
Six Sigma is a disciplined program
or methodology for improving quality in all aspects
of a company’s products and processes. The primary
concept is to define quality metrics CTQ’s (Critical
to Quality) that are important to a customer and then
understand the relationships between the inputs to
the product or process and the outputs (metrics).
Simulation is just one of the
many six sigma software tools used in a Six-Sigma
initiative; however it is one of the most important.
Within the Analyze and Improve stages of a DMAIC project,
or the Analyze and Design stages of a DMADV project,
simulation is a powerful six sigma software tool because
of the following value and benefits it provides:
|
Click here to
view or download the Six Sigma Webinar |
| 1)
Simulation
takes into account process variances, uncertainties
and interdependencies
2)
Simulation
can test many alternative solutions quickly and easily
3)
Models
can be developed with little risk and no disruption
to existing processes
4)
Simulation
takes the subjectivity and emotion out of decision
making
5)
Animation
features make simulation a good tool to help “sell”
others on the best solutions
6)
Reusable
models encourage continuous improvement
7)
Impact
on upstream or downstream customers/operations/processes
can be considered
|
|
|
What is Six Sigma?
In simple terms Six Sigma is a disciplined program
or methodology for improving quality in all aspects of a company’s
products and services. In many ways it represents that latest step
in the evolution of the Total Quality Management movement begun by
W. Edwards Deming in the 1950’s. The Six Sigma program is credited
to Dr. Mikel Harry, a statistician who is co-founder and a principle
member of the Six Sigma Academy in Scottsdale, Arizona. Early corporate
adopters of the program include Motorola in the 1980’s, and other technology-based
firms such as General Electric, Texas Instruments and Allied Signal.
The central theme of Six Sigma is that product
and process quality can be improved dramatically by understanding the
relationships between the inputs to a product or process and the metrics
that define the quality level of the product or process. Critical to
these relationships is the concept of “Voice of the Customer.” In other
words, quality can only be defined by the customer who will ultimately
receive the outputs or benefits of a product or process.
In mathematical terms, Six Sigma seeks to define
a transfer function, y = f(x1, x2, …, xn),
between the quality metrics of a product or process (e.g. the life
expectancy of a product, or the % on-time delivery for a fulfillment
process), and the inputs that define and control the product or process
(e.g. tolerance of a physical dimension or the number of resources
available to service customers). The focus of Six Sigma then is two-fold;
1) understand which inputs (x’s) have the greatest effect on the output
metrics (y’s), and 2) control those inputs so that the outputs remain
within a specified upper and/or lower specification limit.
In statistical terms, Six Sigma quality means that
for any given product or process quality measurement, there will be
no more than 3.4 defects produced per 1,000,000 opportunities. An "opportunity" is
defined as any chance for nonconformance, or not meeting the required
specifications. By comparison, typical quality levels for manufactured
products today achieve about 4 Sigma, which translates to about 6,000
defects per 1,000,000 opportunities.
The diagram below shows the results of two poorly
performing processes. On the left, the process is off-center, producing
outputs that are mostly below the target value. All observations below
the lower specification limit (LSL) are considered defects. On the
right, the process produces outputs that, on average, are on target,
but with great variability. All observations below the LSL or above
the USL are considered defects. The goal of Six Sigma is to both center
the process and reduce the variation such that all observations of
a Critical to Quality (CTQ) measure are within the upper and lower
spec limits.
Structure of a Six Sigma Initiative
To ensure success, a Six Sigma initiative must receive
complete buy-in and continuous support from the highest level of a
company’s leadership team. In addition, a rigorous training program
and dedicated staff positions will require the best and brightest minds
that can be allocated to the initiative.
Roles and Responsibilities
“Champions” are high-level executives, including VP’s
or General Managers, who communicate the business case for the initiative,
and provide the resources and the motivation necessary to sustain it.
“Master Black Belts” are change agents who lead
the initiative in each major division or function of the company. Their
primary role is to train and support the Black Belts who will be leading
major projects. They also track the success of each project in terms
of customer satisfaction and other corporate goals.
“Black Belts” are full-time resources who are dedicated
to leading major Six Sigma projects for a period of 18-24 months. They
receive rigorous training and mentor Green Belts on the use of Six
Sigma tools and methods.
“Green Belts” are resources who are trained on
the concepts of Six Sigma and support the Black Belts in projects involving
their own functional areas. Green Belts also lead one or more small
projects each year.
Project Selection Criteria
Successful projects share common elements. The
following criteria should be considered to ensure a successful project.
- The
project clearly relates to the customer and their quality requirements.
If the customer doesn’t feel the improvements resulting from the project,
then much time and effort has been wasted.
- The
problem and goal statements are clearly stated and understood by everyone
on the project team. Effective problem and goal statements are SMART: Specific, Measurable, Attainable, Relevant
and Time Bound.
- Definitions
of a “Defect” and an “Opportunity” are clearly stated and understood.
- The
project does not presuppose a solution at the outset.
- The
project aligns with the business strategy.
- The
project makes effective use of the Six Sigma tools.
- The
project is data driven.
Evaluation of Project Performance
Just as the level of product and process quality
can be defined in measurable terms, so can the success of a Six Sigma
project. Projects are evaluated in terms of their:
1) Perceived
benefit to the customer
2) Improvement
in output performance (Z-value)
3) Financial
contribution to the company
Six Sigma Methodology: DMAIC vs. DMADV
Six Sigma has two branches depending on the focus
of improvement efforts. For existing products and processes, the DMAIC
methodology applies. For new products and processes, the DMADV methodology
applies. The first three steps in each case are similar, Define, Measure
and Analyze. For DMAIC, the last two steps focus on Improving and Controlling
existing product or process inputs. For DMADV, the final steps focus
on Designing and Verifying the future product or process inputs.
The DMAIC methodology for existing products and
processes consists of the following steps:
• Define – Identify the project goals and
objectives and the customer’s CTQs. Create a team charter that lists
the roles and responsibilities of the team members including, of
course, the customer. Define the boundaries of the project, including
the products and processes to be examined.
• Measure – Select the characteristics of the
product or process to be measured (i.e. the outputs). Define the
performance standards for those outputs. Create a data collection
plan for gathering data on the outputs. Perform a measurement system
analysis to determine the capability of all measuring devices to
accurately measure the outputs (Y’s).
• Analyze – Establish the capability of the
current product or process in statistical terms (i.e. what is the
current sigma quality level). Identify a transfer function that relates
the inputs to the outputs. Identify sources of both random cause
variation and special cause variation in the inputs (X’s).
• Improve – Use Design of Experiment techniques
to determine which inputs should be the focus of improvement efforts.
Determine the effects of improved inputs on the outputs by performing
a sensitivity analysis. Establish acceptable operating tolerances
for the inputs.
• Control – Perform a measurement system analysis
to determine the capability of all measuring devices to accurately
measure the inputs. Implement statistical process control measures
on the inputs to ensure that they remain within acceptable operating
tolerances.
Steps 4 and 5 of the DMADV methodology consist
of the following activities:
• Design – Generate and verify system and/or
subsystem models, allocations and transfer functions. Optimize X’s
through statistical analysis of variance drivers. Generate purchasing
and/or manufacturing specifications and verify measurement systems
on the X’s.
• Verify – Statistically confirm that product/processes
match predictions. Develop mfg and supplier control plans. Document
product/process capabilities and transition to full production rates.
Each step in the DMAIC and DMADV methodologies
requires the use of specific skills and tools to achieve the desired
results. For example, in the Define stage the Critical to Quality parameters
(CTQ’s) are identified using a tool called Quality Function Deployment
(QFD). In the Measure stage, a calibration process called Gage R&R
is used to determine the capabilities of the output measuring devices.
And in the Control stage, Statistical Process Control Charts are used
to ensure that system inputs remain within acceptable tolerance levels.
Simulation is a tool that can be used in multiple
stages of a Six Sigma project. However, it is typically applied in
the Analyze & Improve or Analyze & Design stages. A simulation
model becomes a transfer function that relates the critical X’s (inputs)
to the Y’s (outputs). Experimentation with the X’s leads to a better
understanding of process capability. The next section shows how a simulation
model was used to predict the future-state performance of a manufacturing
process.
Case Study: Cycle Time Reduction for Generator Field Machining
Background:
A
manufacturer of industrial power generation equipment produces five
types of generators. A key component of each unit is the rotating
Field that turns inside the generator. Current planning cycles for
Field machining range from 73 to 124 business days, depending on
product type. The Field machining area has been experiencing rising
WIP levels and inflated cycle times due to capacity constraints in
the system. The process is expected to be stressed even further due
to increased volumes in the coming year. A solution to achieve the
required throughput and maintain acceptable cycle times must be found.
Define:
In
this stage we answer the following questions: who is the customer,
who is on the project team, and what are the boundaries of the process?
The results of the Define stage are as follows:
Customer – Generator Final Assembly Production Manager (Internal
Customer)
Team – Black Belt, Machinists & Shift Foreman, Production
Control Staff, Forging Supplier
Process Boundaries – All Process Steps from Receipt of Forging
through delivery to Final Assembly
The
process consists of 9 steps, with bridge crane movements between
each operation. Also, each operation may be performed at multiple
locations in multiple areas, which increases the demands on the cranes.
If a move is made from one bay to another then two crane moves will
be required in addition to a rail car move.
All
machines have preventive maintenance schedules in addition to random
breakdowns.
All
work is inspected before moving to the next operation in order to
eliminate additional rework steps.
Setups
are required at each step in the process whenever the previous and
current product types differ.
Measure:
In
this stage we select the characteristics of the product or process
to be measured, we define the performance standards for those outputs,
and we perform a “CTQ Flowdown” to understand the relationship between
the inputs and outputs.
Primary
CTQ – Cycle Time, from receipt of forging through delivery to Generator
Final Assembly
Performance
Target – 0 days late from Master Schedule plan (Span = 0)
CTQ
Flow Down:
The diagram above shows how our primary CTQ, Mfg Cycle, is
a function of parameters which are under our control, such as machining
times, changeover times, downtimes and crane availability. Other
factors, such as product mix and holiday schedules must also be taken
into account.
Another
tool used to show the relationship between inputs and outputs is
a fishbone diagram like below.
Analyze:
In
this stage we seek to establish
the capability of the current process in statistical terms, i.e.
what is the current sigma quality level (Z-score). Then we seek to
develop an accurate transfer function that relates the inputs to
the outputs, so that realistic delivery dates can be determined and
on-time delivery of each Field can be assured. Finally, we identify
sources of both random and special cause variation in the inputs,
and perform experiments to understand their effects on the outputs.
The
current method for determining delivery dates is a manual process,
using project management-type software to sequence and schedule each
customer order through every step of the process. While this method
provides a good estimate of the expected completion date, it does
not account for the realities of both random and special cause variation
in the process. The typical way to account for these delays is to
add a cushion factor to each operation time, resulting in an extended
cycle time.
In
addition to the inherent inaccuracies of the current planning method,
the process is time intensive. Each time a major disruption occurs,
like an unplanned downtime at a bottleneck machine, the planning
process must be repeated.
In
order to determine the current process capability, we must have a
transfer function that accurately relates the inputs to the outputs.
For a simple system, with minor process variation and dedicated resources
(no interdependencies), a spreadsheet calculation can be used to
create the transfer function. However, the complexity of most manufacturing
systems requires a more robust model that takes into account these
realities. The picture below shows the complexity of the process
routing, where each arrow represents a potential move from one machine
or queue to another, across two manufacturing bays.
The
process is made even more complicated by the fact that most machines
are shared between Field mfg and other components. In this case,
steam turbine rotors are also produced in the same area. Adding random
machine downtimes and setup times that are dependent on the current
and previous part types, it is easy to see that no spreadsheet calculation
will adequately reflect the complexities of the Field and Rotor machining
process. Therefore, a simulation model was built to represent the
current-state process and predict future-state performance under
several proposed process improvements.
The simulation model included all operations in the Field
and Rotor machining processes. However, before the model could be
used to test future-state scenarios, it had to be validated against
actual historical production data. This task was performed and some
refinements were made until the model outputs were very close to
the results from the actual system.
The outputs of the simulation provide both total cycle times
and deviations from planned cycle times for each Field and Rotor
processed. However, our primary concern in this model was Field cycle
times, so they are the focus of this analysis. A first view of the
Field data is seen through a histogram showing the distribution of
the deviations from planned cycle times. The chart below shows an
approximate normal distribution with a mean value of 7.15 days early.
The standard deviation of 10.47 days and the range of 31 days early
to 12 days late reflect a process with great variation.
The
next step in the analysis was to create a Run chart showing the number
of days early or late for each Field over the 75 week production
schedule. A run chart provides insight beyond the summary statistics,
such as trends over time. The chart below shows that most of the
late deliveries would occur between week 10 and week 45. The improvement
after week 45 is due to additional machinery coming on line.
Beyond the summary statistics and the run chart there are
several excellent statistical tools available to perform a complete
Six Sigma analysis on process data. The charts below were created
using Minitab®. They show that our current process has
a Z-value of only .89! In practical terms, this means that nearly
25% of all observations were greater than our target of 0 days late.
It is obvious from the run chart above that something must
be done to prevent late deliveries in weeks 10 through 45. From our
CTQ Flowdown, we know that we have some degree of control over machine
cycle times, changeover times, downtimes and crane availability.
However, at this point we do not know which of these factors will
have the greatest impact on overall cycle time.
Improve:
In this stage of the analysis we used Design of Experiment
(DOE) techniques to determine which inputs should be the focus of
our improvement efforts. A discussion with the project team identified
a set of experiments to help understand the sensitivity of the cycle
times to each of three control parameters: changeover times, machine
downtimes and crane move times. Machine cycle times were considered
to be too difficult or too costly to reduce.
The results of the DOE showed that total cycle times were
least affected by changeover times, while crane times had the greatest
impact. However, none of the parameters by themselves could achieve
the 90% target DPMO reduction that the team had established. This
meant that the process would require simultaneous improvements in
two or more factors. Considering this in Runs 3, 4, 6 and 8, we see
that both Run 6 and Run 8 could provide the required DPMO reduction.
At
this point in the Improve stage the project team had to consider
the feasibility of reducing both crane move times and machine downtimes
by 30%. It was determined that crane move times, which consisted
mostly of time waiting for a crane to be available, could be reduced
30% by changes in crane operator staffing schedules. However, it
was determined that machine downtimes could not be reduced by 30%
in the short-term. Therefore, additional scenarios were proposed
and those experiments run on the model.
In
the end, it was found that the combination of a 30% reduction in
crane move times, plus a 10% reduction in machine downtimes
could still provide the 90% DPMO reduction that the team required.
A Six-Sigma analysis on the final scenario (Run 9) is shown below.
The analysis shows a reduction in the mean cycle time of 8.65 days,
and a reduction in the standard deviation of 2.46 days. Therefore,
the improved state includes both a reduction in variation and a
shift in the mean.
The
final Z-value of the improved process is 2.67. Although this is much
better than the baseline Z-value of 0.89, it is apparent that achieving
a truly six sigma process will require continuous improvement!
Control:
The final step in this DMAIC project was to ensure that the
vital inputs (i.e. changeover times, machine downtimes and crane
move times) would stay within the limits defined in the DOE. It was
determined that the best way to accomplish this as quickly as possible
was to launch two separate DMAIC projects. The first project was
focused on reducing crane move times by 30%, while the second project
was focused on reducing machine downtimes by 10%. Both of these projects
used Statistical Process Control techniques that tracked the values
of the inputs listed above over time. Whenever, the inputs were determined
to be out of bounds, immediate attention would be given to correct
the situation and bring the process back under control.
Conclusion
Six Sigma is a disciplined program or methodology
for improving quality in all aspects of a company’s products and processes.
The primary concept is to define quality metrics (CTQ’s) that are important
to a customer and then understand the relationships between the inputs
to the product or process and the outputs (metrics). This is done by
determining a transfer function between the inputs, i.e. Y = F(X1,
X2, X3,…Xn). Once the transfer function is known, experiments can be
performed on the X’s to understand their effects on the Y’s.
Simulation is just one of the many tools used in
a Six-Sigma initiative, however it is one of the most important. Within
the Analyze and Improve stages of a DMAIC project, or the Analyze and
Design stages of a DMADV project, simulation is a powerful tool because
of the following value and benefits it provides:
1) Simulation
takes into account process variances, uncertainties and interdependencies
2) Simulation
can test many alternative solutions quickly and easily
3) Models
can be developed with little risk and no disruption to existing processes
4) Simulation
takes the subjectivity and emotion out of decision making
5) Animation
features make simulation a good tool to help “sell” others on the
best solutions
6) Reusable
models encourage continuous improvement
7)
Impact on upstream or downstream customers/operations/processes
can be considered
Project Reviews:
Here are a few specific examples from thousands of customers
who have successfully integrated ProModel solutions in their
lean manufacturing environments:
Lean Six Sigma Analysis to Improve Flight Crew
Equipment Discrepancy Reports Process
Situation:
United Space Alliance (USA) is the prime contractor to
NASA for space flight operations, responsible for all space
shuttle fleet and all international space shuttle processing
operations.
Flight Crew Equipment (FCE) is the name for all the items
produced or modified for Astronauts to take into space flight.
This includes: food, clothing, laptop computers, tools,
batteries, space suits, flight suits, etc.
Click
here to read the full six sigma project review.
Lean Six Sigma to Improve United Space Alliance
Solid Rocket Booster Element Documentation Scanning
Situation:
United Space Alliance must scan about 20 million pages
of documentation and convert them to electronic files in
the Solid Rocket Booster element. They also must determine
the resources and equipment needed to accomplish the task.
United Space Alliance must be able to predict completion
dates for scanning work based upon varying resource availability.
Click
here to read the full six sigma project review.
Lean Six Sigma Analysis to Improve Space Shuttle
Orbiter Tile Removal and Replacement Process
Situation:
United Space Alliance (USA) is the prime contractor to
NASA for space flight operations, responsible for all space
shuttle fleet and all international space shuttle processing
operations.
One of the key components of the outer surface of the shuttle
orbiter is the tile. An orbiter tile is a quartz fiber block
with silica coating. The tiles provide thermal protection
for the orbiter’s skin during ascent and re-entry.
Click
here to read the full six sigma project review.
