Professional Documents
Culture Documents
November 2017
Summary
At a Glance . . .
This case study details a Six Sigma project that occurred at a polypropylene extrusion facility. The
• A polypropylene extrusion facility had a strong emphasis on maximizing overall plant yield; however, the largest extrusion line
facility had 13 extrusion generated an unacceptable quantity of scrap. The project aimed to streamline operations and reduce the
lines, each with varying
scrap ratio from an average of 1.19% to .60%.
output levels. The largest
line accounted for almost
30% of the plant’s total Case Study
production output. It
was also the biggest Matthew Mateos worked at a polypropylene extrusion facility that placed a strong emphasis on maxi-
contributor of scrap
mizing the overall plant yield. During his tenure, there were 13 extrusion lines in the plant, each with
material, accounting for
almost 740,000 pounds varying output levels.
of scrap material—close
to 30% of the total scrap Each of the extrusion lines generated scrap material every time the extruder was deactivated and
material produced by the restarted due to a product changeover or a process disturbance. Twelve out of the 13 extruders could
facility—in one year.
be activated at the required operating output or a rate that was slightly below the operating output.
• A Six Sigma project These 12 extruders were less than 50% the output of the largest machine. They had significantly lower
team was assembled to
implement a solution that horsepower and torque levels. As a result, they could be activated at the desired rate without ramping
would reduce the scrap up. The 13th extruder, which had to be ramped up in stages, was the plant’s largest unit and a critical
ratio from an average component of daily operations.
of 1.19% to 0.60%.
• The team reduced the total During Mateos’ tenure, this line accounted for almost 30% of the plant’s total production output.
scrap produced on the It was also the biggest contributor of scrap material. In one fiscal year, it accounted for almost
main extrusion line by 55%,
resulting in an annual cost
740,000 pounds of scrap material; close to 30% of the total scrap material produced by the facility.
savings of over $300,000.
Due to the high output levels, its original startup process consisted of several stages that required the
production operators to manually increase the extruder screw speed and output. The length of time
spent at each stage affected the amount of scrap material produced during startup.
The plant’s engineering department specified how much time was to be spent at each stage of the
startup process for the largest extruder; however, under the manual startup process the operators were
required to manually type in the extruder speed, output, and formulation percentages. While doing this
they also had to monitor the feeder deviations, feeder motor speeds, extruder screw speed, and extruder
torque levels.
75
70 with a natural logarithmic pattern.
65 Mathematically, the rate of change of
60
55 natural logarithmic functions will rise
50 and then stabilize with respect to the
45
40 y-axis over time. As the logarithmic
35 function stabilizes, the rate of change
30
25 will decrease. This can be observed by
20
15
calculating the derivative of the equa-
10 tion. The derivative of the equation
5
0
can then be used to predict the rate of
change with respect to time.
devices and control the state of output devices, The derivative of a natural logarithmic equation can be derived through the fol-
based on custom programming. lowing method:
Several grades of product were compounded I df(x) f(x + h) − f(x) aLn(x + h) + b − aLn(x) − b
f (x) = = Limith➝0 = Limith➝0
through the largest extrusion line. Each prod- dx h h
uct had a distinct temperature profile that must
be adhered to in order to avoid inadvertently I a[Ln(x + h) − 1n(x)] a[Ln((x + h) / (x)]
f (x) = Limith➝0 = Limith➝0
h h
degrading the product. Since the startup and
operating temperature profiles must be treated
I aLn(1 + (h / x))
as a monument with respect to the auto start f (x) = Limith➝0 = Limith➝0 aLn(1 + (h/x))1/h
h
project, the extruder load would have to be gov-
erned by the ratio of the mass flow rate (m) and At this point the following substitution should be made: u = h/x. Through this
extruder screw speed (ώ). identity, the following relationship can be derived: h = ux. By applying the pre-
ceding identities, the following equation can be derived:
m/ ώ
I a 1 a
A sequence of multiple stages, with each stage f (x) = aLimitu➝0 Ln(1 + u)1/ux = Ln[Limitu➝0 ] = *Ln(e)
x (1 + u)u x
being a specified ratio of mass flow rate to
extruder screw speed, would be used to ramp After simplification, the final equation for the derivative is generated:
up the extrusion line in a safe, efficient, and
I d a
controlled manner. Taking advantage of the f (x) = =
dx x
excellent data collection system that was already
in place, Mateos and the team were able to iden- This type of a mathematical system is illustrated below:
tify mass flow rate (m) to extruder screw speed
(ώ) ratios that appeared to be acceptable.
60
On the new auto start screen, operators 50
specify the material grade that is about 40
to be produced. When initially activating 30
the extruder after a product changeover,
20
the production operators are required to
purge through the die face prior to pellet- 10