Abstract
This paper tackles a difficult problem: How
to install a new generation of process control in existing chemical plants. The issues
of: is it really needed and what's it going to take, both technically and managerially,
are developed in depth. It concludes with specific gains that should be expected.
The goal of this paper is to demonstrate the validity of a structured, repeatable
approach to such a daunting problem.
The author was the external interventionist in an earlier installation of a new
generation of process improvement methodologies. The managerial template which
structured that installation is applied here to the case of process control.
PRESENTED AT: The 1993 Du Pont/Honeywell Users
Conference in Nashville, Tennessee on May 13, 1993.
Introduction
Back in the 1950s when Du Pont was
pressured to license nylon, it became clear that there was going to be a competitive
nylon market of enormous proportions. The technical management back then realized
that Du Pont could lose its dominant market position because the existing batch
processes simply could not meet this market's demands.
The response to the impending business crisis was to develop and implement a new
production paradigm organized around the principle of continuous processing. This
resulted in the ETF project, named for the collaboration between Du Pont's Engineering
and Technical Functions. In 1957, they set up a prototype plant in Seaford, Delaware
which was running smoothly in under two years. By the mid-1960s, the transition from
batch to continuous processing was essentially complete, including the advanced
Bulked Continuous Filament (BCF) process for making carpet yarn.
Today, Du Pont's nylon production is a similar business crisis. These same
continuous production plants that have served our businesses so well are being
challenged in the marketplace. The crisis we now face is escalating customer (and
societal) demands in increasingly competitive markets. And we must meet these
demands with fewer people using existing facilities at just the time when
technological progress seems to be increasingly difficult.
The thesis of this paper is that again the response to Du Pont's business crisis
should be the adoption of a new paradigm. In this case, it is not how we do
production, but how we do process control. As in shifting from batch production to
continuous production, we need now to shift from our top-down, hierarchical,
centralized control paradigm to a new bottom-up, organic, decentralized control
paradigm. Only after this shift occurs will our Nylon business achieve the low costs
and high levels of customer satisfaction that will inevitably result from the
empowerment and engagement of all technical and manufacturing people.
This paper begins with the business factors which press us to consider a change as
drastic as a paradigm shift. Dr., "Doc," Blanchard, former president of
Du Pont, has gone on record as saying that our competitiveness critically depends on
a step change in the rate of our process improvement. This new process control
paradigm is intended to leverage such a step change by transferring into Du Pont
existing technical know-how from non-traditional disciplines (e.g., aerospace) and
new managerial models from leading organizations (e.g., Toyota, U. S. military).
Next, an actual precursor to process control's next generation is discussed as
it exists today at Du Pont's Camden, South Carolina nylon plant. Its greater than
90%" decrease in experimental cycle time and up to 70%" reduction in
breaks (i. e., production interruptions caused by thread line breaks) demonstrates
the fertility of its new paradigm. The section concludes with the portrayal of the
actual paradigm shift that was implemented.
The paper segues into a section about the next generation of process control.
The section suggests how the technical seeds of the next generation are actually
evident and working in a Camden's BCF nylon spinning test cell. This new
generation's paradigm shift is followed by some discussion and concludes with the
criteria for selecting suitable applications. The compelling domain for this new
generation of process control is process steps which are both fast changing and
complex.
The actual conceptual and managerial factors that go into the implementation of
a paradigm shift are discussed next. This is followed by specific suggestions for
anyone who might consider an initiative in this direction. The conclusion ends with
speculations about performance gains which may seem exaggerated, but actually are
based on over a dozen such paradigm shifts.
With such precedents in mind, it seems only appropriate to reflect that there is
ample evidence (take rayon, acetate, Orlon, Dacron, for example) that the life cycle
of our major fiber innovations is about 40 years. If you consider ETF and BCF as our
last major nylon innovations, then barring new major nylon innovations, this may be
the last decade of Du Pont's crown jewel --- nylon. This paper is about a major nylon
innovation --- a whole new process control paradigm so that nylon might make its
100th anniversary as the Du Pont jewel!
Business Factors
The situation at Du Pont Fibers, as with most
of its industry, is that it is under increasing pressure from market trends at the same
time that its rate of technical progress seems to be slowing. Just when we in Fibers
seemed to have triumphed in making enormous volumes of product at a low cost (as in the
1970s and the 1980s), we look up from our tasks only to find that our competitors are
catching up. Furthermore, our customers (sometimes seeming unappreciative) are demanding
a higher variety of offerings, more stringent performance in all dimensions, and higher
levels of even more responsive service. If that weren't enough, they are backward
integrating while expecting increased "openness." Meanwhile our society
(including ourselves and our families) is becoming ever more critical of emissions
and waste. All this while the financial markets seem to have shorter and shorter
memories and no appetite for investment levels of the good old days.
In a memo dated May 3, 1990, to the Fibers Plant Managers, Ray Johnson, vice
president of Fibers' manufacturing, says in eloquent Du Pont-speak:
"Over the last two years, the Corporate
Manufacturing Committee has identified sizable opportunities
for earnings enhancement through benchmarking studies and
other opportunity analyses. The opportunity in improved
process control, resource sharing, simplified and integrated
computer systems, best maintenance practices, and continuous
flow manufacturing total about $[the size of which is both
large and proprietary, but would raise the price of the common
stock by over ten dollars per share]. No other effort, whether
it be new products or new ventures, provides a comparable
opportunity for this company. Because of the magnitude of the
stake and the need to move significantly faster than we have
thus far, E. P. Blanchard has stated that we need to make a
step-change [Ray Johnson's emphasis] improvement in our
rate of progress, if our businesses are to be competitive
throughout the 1990's."
Such a daunting technical challenge
seems all the more formidable precisely because so much is at stake. But this is
exactly the time for calm reflection. The approaches associated with any domain
will, quite understandably, after decades of yielding useful discoveries lose
their fertility. During those very same decades, whole new approaches built on
new technologies and applied productively in different domains have come into
existence carrying on quite a separate life. The calm reflection is to allow
the mind to contemplate which of the many approaches of the intervening decades
might inject rejuvenation to a seemingly worn down domain.
We're at just such a juncture today, I contend. Over the past thirty years we
have seen computers advance and proliferate beyond imagination. During this same
period, the aerospace industry developed spacecraft that performed high speed,
precision, maneuvers such as docking and reentry. In our two major military
engagements during this same period, we as a Nation saw the disastrous effect of
an outmoded command structure followed by the success of a new paradigm of control
(known as maneuver warfare). These are the proofs of principle for the next
generation of process control.
Next Generation's Precursor
With a measure of inspiration from Ray Johnson's
memo above, about two years ago a team of us in Du Pont's Nylon Fibers set out to design
and implement a paradigm shift in how we did process improvement. After about eighteen
months, we actually experienced the punctuated progress that we predicted an appropriate
paradigm shift should yield. At our Camden BCF spinning test cell, test series that took
a week just two years ago, now are routinely done in a couple of hours. The progress on
thread line breaks, due in good measure to additives and finishes, is brought on-line
and understood in weeks rather than months and manifests in one third the breaks.
High volume products that were thought to have been optimized are being
re-investigated and are also experiencing similar gains.
We believed that we could improve yields to the extent of running three times as
long between breaks. This is because an economical, novel (to most fiber producers)
and effective new paradigm is now made possible by coupling the dramatically
beneficial economies of personal computing (software as well as hardware) with the
maturity of sophisticated aerospace technologies. Our reasoning was that breaks occur
on-line, in milliseconds, as the result of any number of factors (usually working in
concert). Therefore, we proposed doing our process improvement by using real-time,
on-line, millisecond, sampling and processing of a large number of variables (64
initially) using technology common to the aerospace industry (Fourier Transforms,
Kalman Filtering, Nyquist Sampling, Bode Plots, phase diagrams, ...) on a flexible,
inexpensive, personal computer platform. We also felt that we wanted an approach
with such alacrity that ideas would just be tried, rather than predicted, debated,
and approved before being tried. We thought results would be richer and more
conclusive, if we worked under the watchful, objective, and helpful eye of
manufacturing.
The open, trusting environment earned the enthusiastic participation of all
manufacturing people involved. Lo and behold, manufacturing people were beginning to
feel like full partners in the collaborative pursuit of break causes. Another
constituency, new product developers, actually get excited solving problems this new
way. And, finally, having sufficient resourcing to pursue systemic causes of thread
line breaks gives those involved a real sense of accomplishment.
The cumulative result is a shift in process improvement performance as evidenced
in the monthly reports of new products coming on-line rapidly and with low break
levels. Most dramatic is the new first pass yield record set by Camden BCF in April
of this year which is a full 3%" higher than the previous record!
We went about implimenting this by setting out to induce the following paradigm
shift:
TECHNICAL SHIFTS
(process improvement)
Off-line, Static |
---> |
On-line, Dynamic |
Multi-second Sampling |
---> |
Millisecond Sampling |
Low Dimensionality |
---> |
High Dimensionality |
Analog (hard wired) Processing |
---> |
Digital (programmable) Processing |
Hardware Intensive |
---> |
Software Intensive |
Table 1
MANAGERIAL SHIFTS
(process improvement)
Internal Technology |
---> |
External (and Internal) Technology |
"Know then Try" |
---> |
"Try then Know" |
Ignore Manufacturing |
---> |
Include Manufacturing |
Table 2
To most this may not seem as consequential a
paradigm shift as when the continuous processes replaced batch processes in the early
1960s. However, back then it was the ever conspicuous reconfiguration of equipment
(i.e., the means of operating continuously) that made that paradigm shift obvious.
This latest paradigm shift is a dramatic reconfiguration, not of equipment, but of
where process improvement is done, how process improvement is done, and by whom process
improvement is done. It is no longer done in the laboratory or in the technical area,
but rather on the factory floor. It is done not in isolation of manufacturing people,
but in collaboration with them. And, it is done dynamically and spontaneously, with
much less premeditation (e.g., planning), authorization, and consensus.
Next Generation's Paradigm
Camden BCF's spinning test cell discussed above
is on an operational production spinning machine and sometimes actively controls it during
actual production. Already there are ongoing test series (particularly those investigating
bulking) which require control of the winding tension and the post bulk tension. If some
of the controls (e.g., draw ratios and finish roll speeds) can be decoupled from the
other thread lines, then a much wider range of tests can be run and this spinning
machine will be even more autonomous. In these cases, this test cell is or will be
acting like an alternate control of the spinning process. When it is controlling the
spinning process, it does so right at the spinning machine (not from the centralized
control room) by the people at the "bottom" with an intelligent or organic
awareness and concern for the whole. This isolated instance of a shift in control
contains the seeds of the next generation of process control.
The paradigm shift being proposed here for the next generation of process control
will include the following:
MANAGERIAL SHIFTS
(process improvement)
Off-line (analyses) |
---> |
On-line, At-line |
Statistical Inspection |
---> |
Continuous Inspection |
Method Availability |
---> |
Method Hiding |
Uniformity |
---> |
Performance |
Table 3
Some of these Technical Shifts are not self explanatory.
For example, "Method Hiding" is meant to suggest that when an operator runs a process,
how he or she goes about it is not available for inspection or criticism without their explicit
permission. If it were otherwise, operators would not feel free to try new approaches.
By being so freed, operators will discover improvements resulting in improved
performance, at which time it would be incumbent upon them to share their methods with
their peers.
MANAGERIAL SHIFTS
(process operation)
Centralized Control |
---> |
Decentralized Control |
Supervisor Control |
---> |
Operator Control |
Control and Direct |
---> |
Support and Trust |
Minimize Costs per Pound |
---> |
Maximize Margin per Minute |
Compliance Oriented |
---> |
Mission Oriented |
"Know then Try" |
---> |
"Try then Know |
Train (to eliminate options) |
---> |
Educate (to create options) |
Ignore Marketing |
---> |
Include Marketing |
Table 4
Managing Change
I have been told of many step changes in
performance experienced by Du Pont technical people at some time in their career.
What is most notable about the Camden BCF paradigm shift (and concomitant step change
in performance) is not that our ambitious goals were achieved, but that we set about
doing that programatically, in a way that is repeatable rather than episodic. It is
the highlights of that programmatic structure that I want to share with you.
We identified six critical factors needed to shift a paradigm: a Clear Purpose,
a Unifying Strategy, External Intervention, Sponsorship, Team Buy-In, and a Local
Champion. The specific example in the case of the BCF spinning test cell is shown in
the parentheses below.
CONCEPTUAL FACTORS
(general case)
A Clear Purpose |
A sharable, personally relevant,
business critical, concise, measurable purpose which can harness sufficient energy to
both overcome traditional turf issues and be powerfully coordinating
(e.g., reduce breaks) |
A Unifying
Strategy |
A pervasively relevant, technically critical,
concise, measurable strategy which can be used daily to help decide
and coordinate implementation issues (e.g., minimize experimental
cycle time) |
External Intervention |
A source (usually a person or persons) of new
technology, experience, and ideas which comes from outside the domain and
provides the requisite variety of possibilities to suggest new solutions
(e.g., an outside consultant with experience using on-line, real-time,
signal processing and applying the unifying strategy). |
Table 5
MANAGERIAL FACTORS
(general case)
Sponsorship |
A declaration of financial and moral support
from a senior member or members of the management who are respected by the
people that the paradigm shifting team is going to rely on (e.g., technical
director, technical group manager, and manufacturing manager jointly support
including a limited license to circumvent disabling procedures) |
Local Buy-In |
A consensus of support among those affected
daily which is achieved by an open, inclusive, trusting climate (e.g.,
evidenced by enthusiastic interest in the successful outcome) |
Local Champion |
A person with the energy, legitimacy, personal
skills and desire to be the hands-on team leader (e.g., the senior spinning
research engineer). |
Table 6
In addition to the above general factors, this paradigm shift to
the next generation of process control needs to incorporate three additional special
factors:
SPECIAL FACTORS
(process control)
Suitability |
Apply only to fast changing process
steps |
Sociology |
Include Marketing, Manufacturing and
Technical Staff |
Coexistence |
Before, during and after this control
paradigm is used, it is meant to coexist with the classical (centralized)
control paradigm. |
Table 7
As a practical matter, I advise complementing and leveraging
existing business systems (for resourcing, administration, coordination, networking,
etc.), except for the occasional use of the Sponsor as license to substitute a new
enabling system for an existing disabling system (e.g., some selected purchasing and
standardization policies).
The effect of this paradigm shift in process control is to enable manufacturing
people, often in conjunction with technical people, to continuously monitor, modify,
and improve the quality and scope of production. In turn, this will enable Du Pont
to retain premium commodity along with a wide range of specialty products (at least
as "old" plants go).
In turn, the high quality of the commodity product will permit premium pricing and
the low cost of the specialty production will permit premium margins. Slow changing
massive production processes have no imperative to shift from their current control
paradigm. Their mechanistic, centralized control can and should coexist with this
volatile, lean production. Further evolution of this control paradigm will show that
even more margin can be earned by doing scheduling by auctioning the transitioning
of machines. Understandably, such a future has these operators and supervisors (and
this lean production, in general) getting progressively closer to the customers and
their production and business needs.
As this new process control paradigm shift crosses Du Pont like a wave, we should
see the following performance gains:
PERFORMANCE GAINS
(step change)
First Pass Yield |
FPY should increase due to elimination of two
thirds of the scrap and rework |
Transition Times |
There should be a two-thirds decrease here,
too |
Product Variety |
The scope of the plant should increase at
least three fold. |
Table 8
By managing the Cycle of Progress
through each successive paradigm, we can experience the renewal of our nylon
business well into the next century!
Acknowledgments
The successes at Camden are largely the result
of the teamwork of both technical and manufacturing people so ably championed by Mr. Richard
Dommel and so enthusiastically supported by Mr. Craig Corey and Drs. Warren Easley and
Ted Sandukas.
The initiator and invisible director of this systematic step change in Du Pont Fibers'
renewal has been Dr. Richard Hess; Dr. Michael Geoghegan has provided a theoretical
framework for determining that we are "in league with the future." Dr. John
Hesselberth and Mr. Jay Tanzer insulated the process improvement paradigm shift
program during several reorganizations.
I want to thank Mr. Robert Wiecke for making the arrangements for me to present this
paper at this conference.
|