SMED IN THE PROCESS INDUSTRIES
Improved flow through shorter product changeovers
By Peter L. King
The need to reduce changeovers - SMED and its origins
For any manufacturing step which must process a variety of product types or materials,
determining the total campaign length for each product type is a key operations management
decision. If product changeovers (set-ups, transitions) are long or costly, there is a tendency to
run long campaigns before changing to the next product, to minimize the overall penalty incurred
with changeovers. This drives higher inventory, in WIP, finished product inventory, or both. It
also tends to create other forms of waste: movement waste may be created if the extra material
must be conveyed to a remote storage location instead of flowing to the next step. Yield losses
may be increased if any of the material is out of specification; it may take longer to get to the
point in the process where defects are recognized, so more out of spec material will be produced
before the problem is discovered. And since material is being produced for which there is no
immediate downstream need, overproduction waste is created. Long campaigns also reduce
operational flexibility, and make the process less responsive to changes in customer needs. And,
of course, long changeovers steal capacity away from profitable production. For all of these
reasons, it is critical that product changes be accomplished as quickly as possible, so that short
campaigns are feasible, and valuable capacity can be used to meet customer needs.
Toyota recognized this in the early 1950s as the Toyota Production System was beginning to
evolve. One of the most time consuming changeovers they faced was the replacement of the dies
on the large presses used to stamp out auto body parts, which was taking several hours. Shigeo
Shingo, an Industrial Engineer consulting with Toyota, developed a methodology for examining
all set up operations and modifying the set up process to reduce the overall time. Using Shingos
techniques, Toyota was able to shorten the die changes from 3 hours to 15 minutes by 1962, and
to an average of 3 minutes by 1971. In recognition of this tremendous accomplishment,
Shingos methods and techniques have become the standard for changeover reduction and have
come to be known by the acronym SMED, for Single Minute Exchange of Dies.
Changeover and campaign length are even more of an issue in process industry plants, plants
that produce materials like paints, paper products, foods, beverages, personal care items, and
fibers and apparel rather than assembled products such as refrigerators, cell phones, or
automobiles. The equipment found in process plants is often large enough and expensive enough
that dedicating a line to each specific product or product family is not economically viable. So a
given process line must be shared by a wide variety of product types, often with vastly different
operating conditions and settings.
Changeover complexity is another challenge for process lines. In addition to the mechanical
and electrical changes that may be necessary when changing to a different product, there are
often chemistry and physics changes that are required as the process is reaching equilibrium on
SMED article for IE Mag
12/4/2009 11:39 AM
the new product and reaching specified properties. Thus the traditional SMED methodology
must be broadened and extended to address these additional changeover challenges.
SMED Concepts
Four of the fundamental ideas that SMED promotes are:
1. Identify external tasks and move them outside of the changeover time. Recognize
that some of the tasks normally done during a changeover can be done before the
equipment is turned off and production stopped, or after the process is running
successfully on the next product. These are called external tasks and include
things like bringing all of the necessary tools and new parts to the equipment. As
the set up is nearing completion, moving any parts or assemblies removed from
the equipment, housekeeping, and clean up are tasks which can often be
performed after the equipment is turned back on. These external tasks can
consume a lot of time so moving them outside of the time window when the
machine is not producing can shorten changeover time dramatically.
2. Determine if any of the remaining internal tasks can be modified so that they are
done as external tasks, such as pre-assembly of any apparatus or tooling required,
and any required preheating of new components which could be done before
being installed on the equipment.
3. Simplify the remaining internal tasks. Use dowels, locating pins, fixtures, and
visual marks to speed the time required to get new parts in place. Standardize
bolts where possible to minimize the number of wrenches required. Use quickdisconnect fasteners where possible. Use Poka-yoke (mistake proofing)
techniques to insure that apparatus cannot be installed incorrectly.
4. Where feasible, perform internal tasks in parallel. If two operators can perform
tasks concurrently, the time can be reduced without increasing the total labor
content of the set up.
So the key to SMED is analyzing everything that happens during a changeover to understand
what can be moved to be done outside the changeover window, and how the remaining internal
tasks can be simplified, shortened, and perhaps done in parallel. Developing a detailed process
map and timing diagram is a good way to start the SMED analysis. Video recording also
provides a valuable view of what actually happens during the changeover. Point-to-point
diagrams (spaghetti charts) can highlight opportunities to reduce the time operators spend
walking, by relocating tools, parts, and even control panels. Cross-functional process maps
(swim lane charts) are another effective tool to clarify changeover tasks, and particularly to see
where tasks can be done in parallel.
A current state Value Stream Map provides insight into where SMED can have the biggest
benefit. This is not necessarily the steps with the longest changeover times, but may be the steps
with large changeover losses, especially those where changeover losses rather than time cause
SMED article for IE Mag
12/4/2009 11:39 AM
long campaigns and high inventory. In process plants, material lost during changeover is often
much more of a concern than is time lost.
PRODUCT A
PRODUCT A
CHANGE
OVER
CHANGE
OVER
PRODUCT B
CHANGE OVER
EXTERNAL
TASKS
EXTERNAL
TASKS
INTERNAL TASKS
PRODUCT A
PRODUCT C
PRODUCT B
PRODUCT B
INTERNAL TASKS
EXTERNAL
TASKS
EXTERNAL
TASKS
PRODUCT A
PRODUCT B
INTERNAL TASKS
EXTERNAL
TASKS
EXTERNAL
TASKS
PRODUCT A
INT TASKS
INT TASKS
EXTERNAL
TASKS
PRODUCT B
EXTERNAL
TASKS
Identify tasks
which can be
external
Move external tasks
outside the changeover
window
Simplify
Internal tasks
Perform
Internal tasks
in parallel
Figure 1 Major SMED Improvement Steps
After a successful SMED process, it is critical that campaign length be re-examined, and
shortened to take advantage of the changeover time reduction, so that all of the wastes described
above can be reduced.
Product Transitions in the Process Industries
In assembly plants, set ups generally consist of mechanical and/or electrical modifications to
the equipment, subsequent calibration and adjustment steps, and often creation of a test part to
check dimensions against acceptable tolerances. Changeovers in process plants also include
tasks of this nature, such as resetting the width of the die in a sheet casting process or changing
the extrusion die shape in a breakfast cereal extrusion process. However, it is frequently the case
that more of the time is spent in cleaning out the raw material feed systems and the processing
equipment to prevent cross-contamination. In many food processing plants, for example,
equipment is shared among several product varieties, which may or may not contain allergens,
such as peanuts. This can pose very stringent requirements for cleaning between products. As
another example, the tinting tanks used in paint manufacturing require thorough cleaning during
color changes. The tasks performed during these clean ups are very well suited to conventional
SMED analysis.
SMED article for IE Mag
12/4/2009 11:39 AM
But there are often additional changeover issues to be resolved in process operations. In
extrusion, sheet good, and batch chemical processes, much of the time lost is the time required to
bring the line to the appropriate temperature, pressure, speed, thickness, viscosity, etc., after all
the mechanical tasks have been performed. Therefore the SMED process must also deal with
these so that the total time for the changeover including the time for process conditions to
stabilize is reduced. Since these components of the transition are more dependent on the process
chemistry or physics than on manual tasks, more technology-related solutions are often
employed. These may include techniques like adaptive process control to speed up ramping
back to first grade product and thus also reduce the accompanying yield losses.
A summary of the kinds of tasks often required in a process line would include:
Getting tools
Getting replacement parts (gaskets, filters, guides)
Cooling down
Mechanical modifications
Calibrating, adjusting
Heating up
Discarding spent parts
Putting tools away
Getting temperature, pressure, and viscosity back to process conditions
Getting back within product specifications
CHANGEOVER
PRODUCT A
MANUAL TASKS
BEGIN PRODUCT B
PHYSICS AND CHEMISTRY REACH EQUILIBRIUM
- PROPERTIES ON AIM
FIRST QUALITY
PRODUCT B
Figure 2 Components of a Process Industry Change Over
Lets examine three different changeover situations typical of process lines:
A change over where all tasks are completely manual
Paper and film producing operations often include a slitting step, where the large master roll,
which may be ten to twelve feet wide, is cut to narrower widths to meet customer needs.
Changing knife positions on a slitter typically requires only manual tasks. When a new roll with
a different cutting pattern is placed on the slitter, all that needs to be done is to relocate the
rotating knives to the new positions along their shaft. Once that operation, requiring less than 5
minutes, is done, the machine can be restarted. SMED should be used to examine the specific
SMED article for IE Mag
12/4/2009 11:39 AM
steps in loosening the knives, repositioning them, and re-tightening them to decide if a different
locking technique would be appropriate, or if a second operator working in parallel would speed
up the task. If there is any possibility that knives can be positioned incorrectly, then a more
positive positioning mechanism, involving detents, pins, or precise markings should be
considered. Poka-yoke mistake proofing techniques could be employed. If economically
feasible, an automatic knife movement system with laser guided positioning may be considered.
CHANGEOVER
MANUAL TASKS
PRODUCT A
FIRST QUALITY
PRODUCT B
Figure 3 A Change Over with only Manual Tasks
Changing the size of the bag or box in a cereal, snack food, or fertilizer packaging line is
another change over with only manual tasks; cleaning out pneumatic lines, loading the new bag
stock, and performing mechanical settings and adjustments are all that is required. Conventional
SMED methodology can have great value in improving these set ups.
A transition where the changes are completely in chemistry and/or physics.
Paper processing often includes a heat treating step called bonding, similar to annealing in
metal processing. The temperature on a bonder may require changing for a different product
type. For this changeover, the temperature change is the only task to be done, but it can take
considerable time, sometimes as much as an hour. The heat treating done by the bonder occurs
as the sheet is contacted by a large heated roll. The roll surface has been machined to very
precise tolerances to give very uniform heat transfer at high speeds. When the roll temperature
must be changed even by a few degrees it must be done very gradually to prevent warping the
roll surface. SMED in this situation may include a structured brainstorming workshop, with
mechanical engineers, physicists, mechanics, and operators, to conceive practical techniques for
more rapid heating and cooling.
CHANGEOVER
PRODUCT A
BEGIN PRODUCT B
PHYSICS AND CHEMISTRY REACH EQUILIBRIUM
- PROPERTIES ON AIM
FIRST QUALITY
PRODUCT B
Figure 4 A Change Over where only Chemistry or Physics changes take time
SMED article for IE Mag
12/4/2009 11:39 AM
In a food processing plant, products being baked as they are conveyed through an oven may
require different oven temperatures or different belt speeds for different products. Thus the only
time consuming change to be made during the transition is to raise or lower oven temperature;
belt speed changes can be done very quickly. So the SMED process should include people who
are very knowledgeable in heating and ventilating and in heat transfer, and in the physics and
chemistry of baking processes.
These structured brainstorming sessions have proven to be a very effective way to generate
novel concepts for resolving complex chemical and physical problems and are often done as part
of SMED in the more difficult situations. They are most effective when done as a Kaizen Event,
with everyone who has knowledge of the problem and potential solutions represented in the
event.
A transition where the changes include a combination of manual tasks and
chemistry/physics.
The process step which applies photosensitive emulsion in the manufacture of x-ray films is a
good example of the most complex type of process changeover. The base film for x-ray products
is typically cast and wound onto wide (8 to 12 feet) rolls, several feet in diameter. At some later
time, these cast rolls are removed from storage, unwound and moved through a coating
operation, where the emulsion flows through a long narrow die lip onto the base film. The
coated film is then rewound and stored for later slitting, chopping and packaging. The specific
emulsion differs by product type and end use: the photosensitive characteristics of films used to
x-ray welded structures to detect cracks are quite different from the properties required to x-ray
teeth and gums.
When product changes occur, the emulsion type must be changed, and the die lip removed
and cleaned. Once these manual tasks have been completed and the line restarted, the film must
be run for enough time to allow the emulsion flow to stabilize. Then samples must be collected
and taken to a lab to be tested to insure that the photographic properties are within specifications.
The lab procedure involves exposing the film under very carefully controlled conditions,
developing the film, and then testing its properties. This procedure can take several hours, and
during that time the film being produced must be put on quality hold. If any properties are out
of spec, the rolls must be scrapped. Only after receiving acceptable test results can the product be
released as first grade.
A SMED activity would, of course, address the die replacement and cleaning operations. It
may even be suggested that a second die be purchased so that a clean die is always ready to
install; the cleaning operation then becomes external to the changeover.
Since most of the change over time depends on lead time through the lab, SMED would also
focus very specifically on lab operations. What may then be discovered is that the lab has been
managing performance to insure quality, accuracy, and repeatability of test results, and to
optimize laboratory costs, with little regard for impact on flow or lead time in the manufacturing
process.
SMED article for IE Mag
12/4/2009 11:39 AM
When this is found to be the case, SMED focuses sharply on flow through the lab and on lab
scheduling processes. It may be appropriate to prepare a Value Stream Map of the testing lab,
including of course, the information flow. Opportunities would include improving information
flow between the coater area and the lab, prioritizing testing sequence, providing technicians for
lunch relief, increased staffing during critical periods, and perhaps buying more analyzers to do
more testing in parallel. It may be that samples are batched, i.e. gathered into groups before
testing, causing additional delays. In that case, single piece flow should be evaluated.
Coating experts should also be included in the SMED process, to examine causes of quality
variation and to try to improve the physics of the coating application. If coating uniformity can
be improved, the time to get to first grade film will be shortened. Eventually, it may be
determined that the loading on the test lab can be reduced. In-line testing instrumentation is
another avenue that would normally be explored. Again, these are excellent candidates for
Kaizen Events using structured brainstorming techniques, with technical and operations people
participating.
SMED beyond Product Changes
Although SMED can be quite valuable in optimizing product transitions, it has additional
uses in process plants. In many process operations, equipment must be changed not because of a
product change but because some part of the equipment has become fouled, corroded, eroded, or
spent. Even in completely continuous chemical processes which produce a single product 24
hours a day, 365 days per year, and therefore never undergo a product change, the equipment
must be taken off line periodically to replace a catalyst bed, replace a corroded part, clean
residue off of vessel walls, or resurface precision equipment. For example, in extrusion
processes like plastic film casting, fiber spinning, plastic pellet forming, and cereal dough
extrusion, the extrusion head, die, or spinnerette can become constricted because of accumulated
material and require cleaning or replacement well before a product change is scheduled.
If not managed well, and executed in the minimum possible time, these changes incur most
of the waste that product changes do. Inventory (waste) is needed to maintain flow during the
scheduled outage, and to protect against the possibility of a longer than planned outage.
Bringing tools and replacement parts to the equipment creates movement waste. Yield losses can
follow the replacement as the process is getting back within acceptable performance limits.
SMED therefore has value in analyzing and optimizing all of these tasks, even though they are
not specifically related to a product changeover. In fact, this is the primary application of SMED
in many process plants.
SMED in real life
Many of the books and articles on SMED use a race car pit crew as an example of a set up
done very well. This has been used so often that it has become a clich, but the reason that things
become clichs is that they often have a high degree of truth and relevance, and the pit crew
analogy has both. Anyone who has watched a professional automobile race can relate to pit crew
SMED article for IE Mag
12/4/2009 11:39 AM
operations, to the precision, the coordination, and the purposefulness of everything thats being
done. It also provides a very strong visual image of SMED principles at work:
All tasks that could possibly be done externally are. The tools are ready, the new tires
are in place, and everything is prepared for the moment when the car enters the pit.
All tasks have been thoroughly analyzed, simplified, and structured to be done in the
shortest possible time.
All tasks are done in parallel. All four tires are replaced simultaneously, while the
gas tank is being filled.
Technology has been appropriately applied, for example to the mechanism used to lift
the car for tire changes.
Everyone understands their role, and has practiced it to get the time down to the
absolute minimum.
All pit stops continue to be timed, and there is an intense on-going effort to further
reduce time in the pit.
Everyone involved understands that these races are often won or lost in the pits, and is highly
motivated to contribute to the potential victory. The more that manufacturing teams understand
that operational success is likewise dependent on fast, effective changeovers, the more effective
SMED efforts will become, thus enabling shorter campaigns, lower inventories, and greater
responsiveness to customer needs!
Summary
Just as SMED has been very successful in dramatically shortening changeovers in assembly
operations, it has seen equal success in process operations. In fact, it usually brings even greater
value because of the complexity of process changeovers and thus the greater opportunity for
improvement. This is especially true in cases where there are significant material losses incurred
in getting chemical and physical properties back within specification. When structured
brainstorming activities, involving chemists, mechanical specialists, and process experts, are
added to the more traditional SMED process, dramatic breakthroughs can be achieved!
The Author
Peter L. King has extensive experience in the process industries, including 40 years with the
DuPont Company applying Lean and other manufacturing systems improvements to a wide
variety of product types. He has also consulted with food, carpet, and apparel companies. Pete
is currently president of the Process Industry Division of IIE. He is also the president of Lean
Dynamics LLC, www.LeanDynamics.us. He is the author of Lean For The Process Industries
Dealing With Complexity, Productivity Press, 2009, from which this article was excerpted.
A webinar on this material, sponsored by IIE, was presented in May 2009. An audio and
slide recording of that webinar, as well as details on upcoming Process Industry Division
webinars, can be accessed at www.iienet2.org/Landing.aspx?id=887 .
SMED article for IE Mag
12/4/2009 11:39 AM
