Is Six Sigma a Prerequisite to Design for Six Sigma?
No! It’s a common misconception though.
Design for Six Sigma (DfSS) is perceived to
be more complex than Six Sigma and often ‘too hard’.
Before we address this, just what is DfSS?
DfSS is a methodology that revolves around
the ability to be able to predict the
capability of a process or product to meet its specifications.
If you can’t do this then you can’t really ‘do’
DfSS.
If one can make such predictions then this
gives the opportunity for optimisation of the process or product design before the process or product is
realised.
Unfortunately this does not happen often
enough, resulting in having to (over)use product/process improvement techniques
like Lean Sigma!
Surely it makes more sense to put the
effort into the design?
Once the process or product design is
frozen it is very difficult to change it to such an extent that would give a
step change in capability – most Six Sigma projects, for example, don’t get close
to Six Sigma! Typically such projects
give valuable, but incremental improvements relative to those that could be
achieved at the design concept stage.
In DfSS for process design we should have the ability to ‘play’ with different
process designs in the virtual world. A
discrete event simuation package is very useful for this and several exist on
the market (Simul8 being one such
package).
Using such a package one is able to predict
lead times, staffing, work in progress, inventory levels and many more ‘lean’
metrics under as many different process design scenarios as you wish – quickly,
and for free!
In DfSS for product design we should have the ability to predict the design
attributes of interest (e.g. lifetime, emissions, vibration levels, fluid flow
characteristics etc.)
Some companies working in fields of advanced
technology (automotive and aerospace being two such sectors) have sophisticated
simulation tools that are routinely used during the design process for such
calculations. It is but a short step to
using these for optimising the ‘robustness’ of the design; making it
sufficiently insensitive to uncontrollable factors.
Designed Experiments
In sectors with less analytical simulation
capability DfSS requires the use of Designed Experiments (DoE) on existing (or preferably
prototype) hardware to generate mathematical models of the outputs of
interest. This sounds difficult but
doesn’t need to be – although it will certainly be time-consuming, and this
will certainly cause significant stress in the New Product/Process Introduction
process if not adequately factored into the work breakdown structure.
DfSS requires very strong and disciplined
project management to work effectively.
DfSS requires very strong and disciplined project management to work effectivly. DfSS isn’t just about prediction and
optimisation, however; it is a methodology,
just as Six Sigma isn’t just about the Improve phase!
Unlike Six Sigma, however, DfSS does not (yet)
have a standardised (e.g. ISO) methodology.
There are several common methodologies,
however.
Examples are:
- DCOV (Define, Characterize, Optimize, Verify)
- DMADV (Define, Measure, Analyze, Design, Verify)
- IDOV (Identify, Design, Optimize, Verify).
I have used these over the years but have
found shortcomings in all of them.
This led me to invent my own methodological
phases of Define, Design, Quantify, Optimize, Verify and Capture; DDQOVC, or D2QOVC
(pronounced DEE SQUARED QUO VEE CEE).
Want to know more
about D2QOVC? click here.
Admittedly, this doesn’t quite roll off the
tongue but I find it a more appropriate sequence of phases.
The flowchart below shows these phases and
the typical steps within each phase:
The flowchart below shows these phases and
the typical steps within each phase:
The importance of a strong Define phase is
evident!
There is no point optimizing a process of
product whose requirements are poorly understood or poorly articulated.
This phase is heavily reliant on the
broader team, whilst the Design, Q, and O phases obviously place focus on the
design community. Quality Function Deployment (QFD) is a good
vehicle for managing many of the activities in the Define phase but it is not
essential. (Indeed QFD can be counter-productive if not handled carefully.)
The Capture phase is vital to DfSS. There is no doubt that adopting a DfSS
approach will initially take rather longer than the new product/process
introduction process usually allows.
This is a cause of tension if expectations
are not managed from the beginning.
However, by capturing the predictive models
and all other pertinent information to enable easy retrieval and
contextualisation, the process of DfSS can be significantly ‘sped up’.
Indeed, if the Capture phase is done well
you should find that adopting DfSS for process or product design takes up less time than traditional approaches,
and gives more conformant processes and products, with less reliance on having
to improve them after the fact.
Coming back to our earlier question: “Is
DfSS hard?”
Well, yes it is – it requires a good deal
of training and commitment by many people (not least of which senior
management) over the whole design and realisation process.
DfSS is not simply a focused well-bounded
stand-alone 3-6 month project like typical Six Sigma projects – the scope of
DfSS is by definition much wider and some tools and techniques will be new to
people, so it also requires experts in DfSS to help with training and application,
at least initially.
The real question should be “Is DfSS
beneficial?”. The answer is emphatically
“Yes” – it can be a real game-changer.
Don’t wait until you’ve ‘done’ Six Sigma (or Lean Sigma), whatever that means!