10 Case Studies

10 Case Studies
In more than thirty-five years of experience in this
business, I have seen all sorts of parts, customers, and
problems. Anyone in this business who has not been
visited by an inventor with a “wonderful” idea that he
wishes to implement in plastics must be in a position
where he does not deal with the public.
Here comes a bright-eyed inventor. Perhaps the (abbreviated) conversation goes like this.
I NVENTOR: “I wanna make this here whatsit outa
plastic.”
MOLDER: “What kind of plastic?”
INVENTOR: “Uhhh……Hard plastic?”
MOLDER: “Do you know that this invention will
require molds that cost thousands of dollars?”
INVENTOR (IRATE): “You tryin’ to rip me off? I know
plastics are cheap!” and he stalks out in an
angry huff.
Then there is the guy responsible for the case described in Sec. 10.5, who cannot be taught or warned
about our less scrupulous brethren.
This chapter deals with a variety of molding experiences, problems, and solutions that stand out in my
memory. Even though this book’s emphasis is controlling shrinkage and warpage, these “personal experiences” deal with other things as well. I hope you find
them interesting and informative.

10.1 Unexpected Housing Shrink

Figure 10.1 shows a housing that contains a rotor
with very little clearance between the rotor and the inner bore. This part was in a family mold with other
simpler parts. To avoid three-plate or hot-runner expense, this part was gated at each of the ears marked
“G.” The center core pin at “V” was inside a sleeve
ejector, and the clearances between the pin, sleeve, and
cavity were generous to provide venting. The material

was 30% glass-filled nylon.
The mold builder realized that the material flow
and fiber orientation would be predominantly radial
and assumed a shrink factor higher than published to
allow for cross-flow shrink.
The inner diameter shrank about twice as much as
expected, but the outer dimensions were right on spec.
What was the cause?
Notice the many sharp corners and changes in direction between the gates and the vent. When the part
was placed in an oven and the plastic burned away,
there was a significantly lower concentration of glass
fiber in the inner cylinder than in the outer cylinder.
Two things contributed to the shrinkage. First, of
course, was the reduction in fiber concentration. It
seems that each corner effectively combed out a small
percent of the glass fibers. Second, each corner caused
energy loss and a reduction in effective holding pressure in the plastic in the inner diameter.
Adding radii to the various corners would have
helped the situation, but ultimately it was necessary to
open the bore and increase the core size to provide the
clearance for the rotor.

Sometimes strange things happen during molding
operations. This case is a good example.

Figure 10.1 Rotor housing.

© Plastics Design Library

Ch. 10: Case Studies

140

10.2 Changing Materials Triggers
Warpage
Figure 10.2 shows a part that was originally designed in ABS. Part of the runner and gate are shown
to indicate where the material was introduced into the
mold.
A need for better chemical resistance dictated a
change to nylon after the mold was built. In order to
maintain size, a glass-filled nylon was chosen that had
the same published shrink rate as ABS. To the molder,
it seemed like the solution was easy. Using a material
with the same shrink rate should yield an identical part.
Unfortunately, that was far from the reality.
When the first samples were shot, they looked
something like Fig. 10.3. Since flatness was a primary
concern, parts warped like this were not satisfactory.
The molder had to find a solution—fast!
The nylon supplier was consulted and he explained
the phenomena of differential shrinkage based on fiber
orientation. He drew a picture like Fig. 10.4, showing
the approximate flow paths in the part with the edges

folded up to form a flat pattern. It can be seen that the
flow path on the gate side is essentially parallel to the
long edge and symmetrical on that edge above and below the gate, so the fiber orientation is predominantly
along the long axis of that side. On the opposite side,
the flow is predominantly vertical across that edge.

Since shrinkage in fiber-filled materials is significantly
greater across flow than it is along the flow, the long
side opposite the gate was shrinking significantly more
than the gate side, causing the warpage shown in Fig.
10.3.
Fortunately, the solution to the problem was fairly
simple. By moving the gate to the top center of the
narrow end of the part, the warpage was reduced to a
satisfactory level.
This problem occurred long ago when glass-fiber
reinforcement was relatively new and before talc- or
flake-filled materials were available. It is possible that
a simple substitution of a talc- or flake-filled nylon
for the glass-filled nylon would also have solved the
problem.

Figure 10.2 Views of a part designed for ABS plastic.

Figure 10.3 The same part molded in glass-filled nylon.

Ch. 10: Case Studies

Figure 10.4 Flat pattern of the part showing flow directions.

© Plastics Design Library


141

10.3 Creep in a Water Heater Stand

When building codes required that water heaters
be raised eighteen or more inches from the floor to inhibit ignition of heavy flammable vapors, one enterprising entrepreneur proposed to make a plastic stand
to lift the tanks.
Figure 10.5 shows the design of the stand. The
molder and the mold builder cautioned the customer
that there needed to be reinforcement or metal pads
under the water heater feet. Initial tests indicated that
the stand would support many times the weight of an
80-gal. tank. The customer was sure that his product
was fine. After all, look at all the reinforcing ribs and
the enormous test loads that had been sustained.
After about four years, the water heater feet started
sagging or breaking through the top of the stand. Creep
had struck again. A plastic part can withstand very
high stresses for a short time, but sustained, high
stresses cause major deformation or failure. This failure could have been delayed or prevented if the point
loads at the feet of the water heater had been spread
over several square inches of area. A 4–6-inch diameter steel or aluminum disk under each foot would have
been adequate.

10.4 Oversize Part Injection-Molding
Alkyd Thermoset
While thermosets are not addressed in this book,
there is one experience that I would like to share with
you.
When injection molding of thermosets was new,
my company was asked to bid on making a thermoset
box as shown in Fig. 10.6. The material was to be
alkyd, and the material supplier assured us that it was
a simple matter to mold their brand new injection-molding grade.

We completed the mold and began mold trials. The
part is gated between the two mounting feet at the top
of the picture. The walls are about one-tenth inch thick.
We could not get the part to fill. The material would
set up before it filled the cavity, leaving a void in the
wall at the bottom of the part in the figure.
We tried everything. We changed mold temperature, material temperature, injection pressure, and injection rate without success. We opened the nozzle orifice, the gate size, and the runner size. The material
supplier came to the plant and basically shrugged his
shoulders, “We thought it would fill.”

Figure 10.5 A plastic hot water heater stand.

© Plastics Design Library

Ch. 10: Case Studies


142

Figure 10.6 A thermoset box.

We were ready to try anything. That was when
our molding shop foreman asked, “Why don’t you lubricate it?” He suggested adding just a touch of zinc
stearate.
Well, it worked. In fact it worked too well. The
first parts molded after we added the lubricant filled
and packed the mold so well that the part did not shrink.
It actually came out of the mold and cooled larger than
the cavity. That little bit of lubricant allowed the part
to fill so easily that we compressed the plastic more

than the shrink rate, causing “negative” shrink. By reducing the lubricant loading and the injection pressure,
and modifying the injection rates and mold temperature, we were able to produce thousands of good parts.
We never told the customer or the supplier how we
made such good parts.

10.5 Inadequate Baby Dish Mold
A young man walked into my office, many years
ago, with an idea for a baby dish that would not spill.
It would be clamped onto a highchair tray so the baby
could not push it off or turn it over. The dish was to
look something like Fig. 10.7. The gentleman looked
and acted in every way like a frugal person. But unfortunately he was not very wise.
I priced the tool. He advised me emphatically that
he could get the mold built for half that amount. I delicately inquired who the builder might be and cringed
at his reply. The proposed supplier was infamous in
our area for building the cheapest molds possible and
for making a profit on the initial 50% down payment.

Figure 10.7 Proposed baby dish design.

Ch. 10: Case Studies

© Plastics Design Library


143
Because I was concerned about possible liability, I
could not tell this young man that this particular mold
builder produced junk. Rather I tried to educate him
about mold construction, showing him some molds that

we had built. I advised him to get references and to
look at some of the other molds this guy had built and
compare them to our molds.
My education efforts were to no avail.
Some months later, very late in the day, the young
man came into my office struggling to carry what appeared to be a stack of rusty, flame cut square plates. I
realized that this stack of junk represented what the
other mold builder had produced. Over my objections,
he set the plates on my desk and told me his sad tale.
The other mold builder had not been able to produce even one part from the mold. There was a show
in less than a month for which that the young man absolutely had to have parts.
Figure 10.8 approximates the construction of this
mold, but it does not show that the plates were not
ground flat nor the edges finished after being flame
cut. Major deficiencies are visible in this sketch:
• The core pieces were surface-mounted on
an unground plate and positioned with
two dowel pins in each core piece and
were retained against the plate with four
bolts in each core piece. Plastic was injected into the mold at the “ear” on the

right side of the cavity (shown at the right
in Fig. 10.8). The viscosity of the plastic
was so high that the forces pushing the
core pieces sideways were sufficient to
cause the dowel pins to distort the holes
in the core pieces and in the mold-support plate, allowing the cores to move out
of position.
• There was inadequate plate thickness on
the ejector (left) side of the mold under

the cores. The plate under the core was
flexing so that injected plastic was flowing under the cores, effectively trapping
the molded part on the mold so that it
could not be ejected. In a good quality
mold, the plate containing the cores will
have at least an inch of thickness under
the cores and the cores will be integral
with or pocketed into that plate. In addition, there is normally an additional plate
under the core plate that adds additional
support and stiffness to the assembly.
• Part of the problem with flex in the support plate was the total lack of support
pillars. Even with the much thicker plates
normally found in a quality mold, support posts or pillars are necessary in
molds having a span between the side rails

Figure 10.8 Sketch of a poorly constructed mold.

© Plastics Design Library

Ch. 10: Case Studies


144
of greater than 6 to 8 inches. Injection
pressures are typically in excess of 5,000
psi and may be as high as 20,000 psi or
more. This dish had an area of about 40
in.2. With high injection-pressure, the
bending force on the core plate could be
as high as 800,000 lb. Certainly this is

sufficient to cause significant deflection
or bending of the core plate.
• There were no leader pins or bushings;
instead, dowels were placed in drilled
holes through both plates. The dowels
were allowed to “select” the side on which
to stay, depending on the fit and friction
at that moment. The dowels were about
the same size as the dowels that were supposed to position the cores, so there was
less area resisting the side force caused
by injection pressure at the nominal leader
pin location than there was at the core.
Therefore, the “leader pin” holes were
stretched out-of-round, which aggravated
the moving cores.
• The unground plate surfaces resulted in
irregular gaps between the two plates. The
gaps were large enough for plastic to flow
into them, causing flash.
I doubted if that poor excuse for a mold could be
made to work properly and I was right. After grinding

the plates, adding pillars, leader pins and bushings,
and adding large keys to position the cores, the best
part that could be produced is shown in Fig. 10.9. There
was some flash all around the part, especially at “A”
and “B,” but the largest was at “C.” The young man
was satisfied because he was able to trim the flash away
and sand the parts smooth enough for demonstration
at his show.

10.6 Gas Entrapment in Baby Dish
Mold
There is a sequel to the story. More months passed
and the young man showed up again. This time he had
another mold. This one was far better but it had a little
problem. The plastic flowed around the outside of the
part faster than it flowed across the part, and trapped
some air approximately at point “A” in Fig. 10.10. As
the air was compressed by the plastic, it was heated to
the point that the plastic around the hole was charred
black. This time the solution was more direct and the
mold was salvageable. By thickening the bottom of the
part by removing some material from the tops of the
cores at “A” and “B,” and by making the rib between
the dish pockets between “A” and “B” thinner, the plastic flowed across the mold more easily than it did around
the edges. So we could mold a good dish, as shown in
Fig. 10.11.

Figure 10.9 A part from a reworked junk mold. Note the “flash” at locations A, B, and C.

Ch. 10: Case Studies

© Plastics Design Library


145

Figure 10.10 A new baby-dish mold with air entrapment.

Figure 10.11 The final, good, baby dish.

10.7 Warpage in a Molded Spool

most obvious was the gate design with respect to the
part-wall thickness. Note that the wall is almost twenty
times greater than the gate thickness (0.37 in. vs 0.02
in.). The gate length (0.05 in.) is more than twice the
gate thickness (0.02 in.). These two errors resulted in
an almost immediate freeze-off at the gate as soon as
the mold filled. This left the spool with only a thin
wall, perhaps less than 0.06 in., which solidified while
the remaining mass was molten. Furthermore, the thin
gate resulted in a significant amount of shear heating
of the material at the gate, which further raised the
temperature of the molten mass in the cavity. The higher
temperature resulted in greater thermal contraction than
would have been experienced with a cooler melt.

The part represented in Fig. 10.12 contains many
of the classic mistakes that are made by part and mold
designers. Only pertinent dimensions are shown. The
molder had reported “A little trouble with warpage and
shrinkage.” In fact, the shrinkage was about twice the
published shrink rate and the internal bore had ballooned out a considerable amount. Almost certainly,
the customer and part designer had indicated a desire
for clean ends, without gate marks. The molder wanted
an easy-to-remove gate and an inexpensive mold.
These restrictions led an inexperienced mold maker
to make several gross errors in the mold design. The

Figure 10.12 Heavy-wall, glass-filled nylon spool. Dimensions are in inches.

© Plastics Design Library

Ch. 10: Case Studies


146
Although it is difficult to analyze, it is possible
that the jetting effect of the thin gate encouraged the
material to flow to the far side of the cavity and then
flow down the length of the part. This would result in
cooler material on the side of the cavity away from the
gate and warmer material near the gate. In addition,
there is a greater heat load from the shear heating at
the gate and on the gate side of the center-core pin.
This means the mold is warmer on the gate side than
opposite the gate. The turbulence from the gate jetting
discourages glass-fiber orientation, so there is not likely
to be much, if any, fiber orientation to affect shrink
rates. Therefore, the likely primary causes of excessive shrinkage, other than the small gate, are the mold
and melt temperature differentials.
Because nylon is a semicrystalline material, it has
a higher rate of shrinkage than amorphous materials,
and the heavier the wall section and the higher the melt
temperature, the greater the percentage of crystalline
material in the molded part. Higher crystallinity translates into higher shrink rates. Higher temperatures on
the gate side of the mold cause the part to shrink more
on the gate side and assume a shape somewhat like

that represented by the dash-dot-dot lines. These lines
indicate some concavity in the outside walls in addition to the bending effect.
The inside walls, as molded, are represented by
the dotted lines. The central core was not adequately
cooled and as a result encouraged the ballooning of the
inside bore to compensate for the shrinkage in the solid
mass around the core. There was little foaming or voids
in the mass, as would be expected in an unreinforced
material. The glass fibers helped prevent voids forming in the mass. The hot core encouraged the inner wall
to sink away from the core to make up the shortage of
material that resulted from the high shrink-rate.
In such a case, the first action to minimize shrinkage is to increase the minimum dimension of the gate
to at least 50% of the wall thickness. It may be necessary to increase the gate to as much as 70% to 80% of
the wall thickness. This will allow material to flow from
the runner into the cavity, as the material in the cavity
cools and shrinks, for a longer time. It has been established time and again that longer, effective holding/
packing time reduces shrinkage.
Notice that the runner and sprue are smaller (0.23
in. to 0.25 in.) in cross section than the part (0.37 in.).
The material in the runner is cooling faster than the
material in the cavity. The runner is surrounded by cool
mold-plates, while the part is cooled effectively from

Ch. 10: Case Studies

the outside but not from the inside. This means that the
runner will solidify a significant amount of time before the part does. Therefore the runner also needs to
be increased in size to approximate or even exceed the
cross section of the part.
Note the 0.12-in. diameter cylinder of material at

the end of the sprue. Here we have a molding-machine
nozzle of 1/8-in. diameter feeding a part that is 3/8 in.
thick. The material in the nozzle can freeze and stop
the flow of makeup material before the material in the
cavity solidifies. Furthermore, with a larger gate and
runner cross section, shear heating will occur at the
nozzle instead of at the gate as earlier discussed. Therefore the largest nozzle that can be used without drooling (assuming good material drying) should be chosen.
More effective cooling of the core and a resulting
decrease in cycle time and ballooning around the core
can be achieved by use of a bubbler or cascade, or a
heat pipe, within the core. The small core requires a
very small feed tube and clearance around it for a bubbler. Any water contamination or corrosion would likely
block effective flow, resulting in hot spots, or would
revert to inadequate cooling. Probably the best solution is to have the core built with an integral heat pipe.
The back, or base, of the core would require a heatpipe extension into a water channel that is about the
same length as the heat-pipe exposure to molten plastic, that is, about 3.5 in. Fins can be added to the rear
extension-tube to reduce this length requirement.
The region of the part near the gate is always affected by the variation in flow and packing at the gate.
Depending on conditions, the gate effect could cause
either higher or lower shrinkage than the more remote
areas. Therefore, it is unwise to place a gate in an area
where warpage is a concern. For this reason, the gate
should be moved to the end of the part, perhaps gating
axially parallel to the center bore and on both sides of
the core to balance pressure from one side of the part
to the other.
Finally, a hot runner or hot sprue should be considered. The massive gate and runner could be entirely
eliminated with significant savings in waste or reground
material. Even with a single-cavity mold, a hot-sprue
design that centers the core while providing material

on two or three sides of the center-core pin should be
possible. The hot sprue would also reduce the size requirement for the gate in that the heated sprue would
help keep the material molten at the gate to provide
longer, effective packing time.

© Plastics Design Library


147

10.8 Daisy-Wheel Breakage
When the daisy-wheel printer was new, the wheel
was made with a plastic hub, and molded letters were
connected together by a stamped-steel spring. The
spring had one arm for each letter or character. Analysis indicated that a wheel that was all plastic, eliminating the steel spring, was feasible. A one-cavity mold
was produced but when the new, all-plastic daisy
wheels, as shown in Fig. 10.13, were tested, the spring
arms would fail when struck repeatedly (for example,
when the underline character was used to create a line
across the paper).
Samples of the molded part were analyzed and it
was discovered that the physical properties of the glassfilled nylon were significantly below what they should
have been. The question then was, “What is causing
the material degradation?”
Right away it was discovered that the plastic was
not being adequately dried. A simple, glass-slide test
indicated moisture in the material in the feed throat of
the molding machine.
The glass-slide test requires two glass slides and a
hot plate. The hot glass slides are placed on a hot plate

and heated to just above the melt point of the plastic
being tested. When the slides are hot, two to four dried
plastic pellets are placed on one slide, spaced about
one-half inch apart along the centerline of the slide.
Tweezers are used to position the pellets, and to place
the second slide on top of the pellets. The slides are
pressed together with the edge of a tongue depressor or
popsicle stick. When the pellets are thinned so that they
are translucent, each pellet is about one-half inch in
diameter or a bit more, and will be translucent, even if
highly pigmented. If there is any moisture present, it
will appear as bubbles in the flattened pellets.

Adequately dried plastic was molded with some
improvement of properties, but still significantly below what they should have been.
The part was center-gated with a 0.040-in. diameter gate. The gate had a cross-section area that was
less than twice the area of even one of the many spring
arms. The mold was difficult to fill even when the mold
and the material temperatures were at the upper recommended limit. Often the plastic would freeze before
the spring arms were fully filled. It was taking several
seconds to fill the part even when it would fill at all.
We theorized that the gate was so small that the
part could not be filled fast enough to finish filling before the material would freeze in the spring arms. Attempts to fill faster with increased injection pressure
caused the material to get even hotter from shear heating in the gate. This additional heat was causing heat
degradation of the material.
We doubled the diameter of the gate and were able
to lower the mold and material temperatures and still
fill the part in under two seconds. The physical properties improved to the levels expected and the parts no
longer failed under test.

10.9 PVC Part-Flashing Problems
A large part with a projected area of 240 in.2 (10
in. × 24 in.) was being molded on a 730 ton clamp
molding machine. The material was rigid PVC. The
molding machine shot capacity was about four times
the required shot size. The average wall thickness was
0.200 in.
The molder had been plagued with PVC degradation in the barrel of the molding machine. When PVC
degrades, it can break down into something almost like

Figure 10.13 An early prototype of a daisy-wheel print disk.

© Plastics Design Library

Ch. 10: Case Studies


148
a thermoset, that is, a powdery solid. Because of this
history, the molder was wont to keep the barrel temperatures as low as possible and use a slow injectionrate to avoid overheating the PVC. The barrel temperature settings were essentially flat at the minimum
temperature to melt the PVC. The injection rate was
set to take a minimum of eight seconds to fill the mold.
Toward the end of the fill cycle, the injection rate slowed
due to viscous back-pressure from the mold, increasing the total time to over ten seconds.
The problem was that under these conditions, it
took over ten seconds with injection pressures equating to over 13,000 psi to fill the mold, even with the
sprue centrally located as shown in Fig. 10.14. Over
half the time the part would flash dangerously near the
sprue, even when the part did not quite fill.
If even half of that injection pressure translated

into separation forces, the pressure trying to force the
mold open would be 780 tons. The clamp pressure setting was less than 700 tons. No wonder the mold was
flashing. Examination of the mold found that the support pillars were essentially the same height as the side
rails, or perhaps 0.001 in. less. The mold builder increased the height of the pillars so that they were
preloaded about 0.003 in.
The molder was persuaded to increase the frontzone barrel temperature to the maximum recommended
by the material manufacturer, with each previous zone
lower so that the feed zone was at the minimum-recommended temperature. The maximum fill-rate was
more than doubled, reducing the fill time to less than

Figure 10.14 A large part that was having flash problems.

Ch. 10: Case Studies

four seconds. (The sprue was quite large.) These
changes allowed the molder to cut the injection pressure nearly in half. The discoloration seen in the far
end of the part in Fig. 10.14 is residual discoloration
from earlier shots when the barrel temperatures and
fill rates were lower. (The mold builder rarely gets the
best parts.) The front barrel temperature could probably be even higher, because the plastic melt rarely
raises to the barrel temperature. The pillar height adjustment was probably not necessary.
The higher melt temperature reduced the viscosity.
The steep temperature gradient in the barrel compensated in part for the relatively small shot-size compared
to the maximum shot-size. The large sprue allowed a
rapid fill-rate without significant shear heating.

10.10 Polycarbonate Switch Failure
When polycarbonate first came out, some folks
thought it was the answer to every plastic problem.
One company decided to use polycarbonate in a switch

inside an explosion-proof housing in an oilfield application. The switch required two of the (A) parts and
two of the (B) parts in Fig. 10.15. Only one switch
rotor (C) was required. By rotating the contact leafsprings, the five contacts could be either normally open
or normally closed.
Figure 10.15 shows the two polycarbonate parts
positioned properly. Two more parts assembled in the
same way and inverted completed the polycarbonate

Figure 10.15 Partially assembled polycarbonate rotary
switch.

© Plastics Design Library


149
parts. The gray part was a thermoset part to better
resist any electrical arcing as the switch opened and
closed. It was placed between the electrical contact bars
in the center hole of the end pieces.
The parts worked very well in testing. Unfortunately, the initial testing was relatively short and did
not include an adequate exposure to the wide variety
of aromatic hydrocarbons that are present in a petroleum-producing environment. After about six months,
some of the switches began to literally fall apart from
cracking and crazing. No one had studied the chemical
resistance of polycarbonate to aromatic hydrocarbons.
The solution was to change the material to oldfashioned, glass-filled nylon, which has a very high
resistance to aromatic hydrocarbons.

10.11 Square Poker Chip Tray,
Inadequate Shot Size

The very first problem I encountered when entering the injection molding industry was one of maximum shot-capacity of available machinery. I was working with a start-up company that had only two presses.
One was a 3-oz. shot-size, Van Dorn plunger-type
press. The other was a 450-ton clamp, 24-oz., Reed
Prentice screw-injection machine. My predecessor had
quoted and accepted a contract to mold square poker
chips and a poker chip tray somewhat like the one
shown in Fig. 10.16, which was to be available in a
variety of colors. The difference was that the tray was
to have 1/4-in.-thick walls. The one shown has a wall
thickness of about 1/8 in.

Figure 10.16 A square-poker-chip tray with some chips.

© Plastics Design Library

Apparently, my predecessor had a habit of ignoring or guessing at part weight. When the mold was
complete the customer was present for the initial test.
The molding machine setup man set the machine for
its maximum shot size, expecting to reduce the shot
size later and hoping for a full part on the first shot.
The injection unit bottomed out, pushing a full 24
oz. of plastic into the mold. That surprised the setup
man. When the mold opened and the part was removed,
you can imagine the expressions of shock, anger, and
dismay when the part was little more than half formed.
Later calculations indicated that a full part would weigh
about 36 oz.
At that time in that company, no one was aware of
how to use a foaming agent when injection-molding
plastic. That would have probably formed a satisfactory part. Other molders with larger-capacity machines

would not mold the heavy part with the longer-thanexpected cycle times for the quoted part price. Did I
mention that the customer gave the impression that his
business was on the shady side? The molder was left
with only two options: cut down the mold to reduce the
wall size, or else. As you can see, they chose the former.

10.12 Problem Ejecting Square Poker
Chips
Part of the poker chip deal was that there were to
be no visible ejector-pin marks on the 1/8-in.-thick
poker chips. They were molded of crystal polystyrene
with flecks of aluminum. They were molded with radii
all around and were formed on both sides of the mold
with the parting line in the center of the part thickness.
They really looked nice. The same individual who failed
to consider shot size was sure that the poker chips would
fall right out of the mold with the runner. In fact they
did, but the formidable customer was dissatisfied with
the pinpoint parting-line gate. Besides, there was a fair
amount of labor removing the chips from the runner.
The parting-line gates were plugged and tiny tunnel gates were cut into the 1/16-in.-deep ejector side of
all forty-eight cavities, as shown in Fig. 10.17. That
solved the problem of gate blemishes, but then fewer
than half of the poker chips dropped from the mold.
They usually stuck to the ejector side of the mold.
This time the solution to the ejection problem was
to modify the radius of each chip on the injection side
of the mold opposite the gate, to create a small undercut, as shown in the upper right part of Fig. 10.18.
This caused the poker chip to pivot around the gate

Ch. 10: Case Studies


150
and out of the ejection side of the mold as the mold
opened. The ejection system in the runners severed the
gates, leaving the poker chips virtually hanging in thin
air. The sound of those chips tinkling into the collection chute was the sound of money.

Figure 10.17 A square poker-chip cross section showing
the tunnel gate and modified radius on the opposite edge.

that the material could be placed directly into the cavity without compacting it. Most compression-molded
parts enter the compression mold as preformed blocks.
The preforms are usually formed slightly smaller in
diameter than the cavity into which they are placed.
There are machines that take standard pellets and compress them into preforms. The machines can reduce
the volume of the raw material to about 25% to 50%
of its original density. This material had to be compressed to less than 10% of its original volume to make
the preforms. It was reduced to about 3% of its original volume in the final product. That is a lot (this is a
bit of a “stretch”) of shrinkage. We had to design and
build a special preforming machine to complete this
contract.
The second reason this stands out in my memory
is that the same guy who forgot to calculate the shot
size of the poker-chip tray did not realize that this special machine was necessary. And he forgot to include
the cost of the raw material in the quoted price.
Did you ever try to get out of a government contract?
Have you heard the concept, “If you are loosing
money on every part you mold, you have to make it up

in volume?” It does not work. The profit on this job
was shrinking too. Too much!

Figure 10.18 Edge detail of the square poker-chip.

10.13 Military Cup Material
“Shrinkage”
I don’t know what the military uses for dinnerware now but, at one time, they had coffee mugs, serving trays, and cups made of a linen-reinforced
melamine. One of the cups is shown in Fig. 10.19. This
little project stands out in my memory for two reasons.
The first is that this material, as received, was less
dense than the cotton stuffing often found in an aspirin
bottle when first opened. It was light and fluffy. With a
little bit of effort, one could pack enough of it into a
large coffee can to make a single molded cup. Most
plastics come in a granulated form and are reasonably
dense.
Once again I have strayed into the thermoset field
(sorry). The cup was molded in a four-cavity compression mold. All the material for a part had to be placed
in the cavity before the mold closed. There was no way

Ch. 10: Case Studies

Figure 10.19 A drinking cup once used by the U.S. Army.

© Plastics Design Library


151

10.14 Core-Deflection Problems
Steel is rigid, right?
Consider the part shown in Fig. 10.20. It appears
to be a straightforward molding problem. The part is
not very big and the core is almost 5/8-in. thick. There
are no side cores. How simple a mold can you imagine? If you imagined that you are wrong.
As expected, the mold was built with a freestanding core from the parting line at the open end of the
box. The part was sprue-gated in the center of the closed
end. Ejection was by stripper bars across the long sides
of the core. The base of the core was about 2-in. wide,
6-in. long, and almost 2-in. thick.
The first test-shots resulted in large voids near the
gate/sprue on one of the large flat sides. The plastic
was flowing down the opposite side and the ends to the
base of the core, then around the base, trapping air
near the closed end on the opposite side. Drat, and other
expletives. Some misbegotten toolmaker had obviously
made a mistake grinding the core or cavity off-center.
Inspection revealed that in fact this was so, but
only by a very few thousandths of an inch. Not nearly
the amount that would be required for the wall thickness variation at the closed end. The closed side was
over 0.06 in. thicker than it should have been opposite
the air-entrapment void. We puzzled over that awhile
and decided that the core to retaining-plate fit was too
loose, allowing the core to pivot in the retaining plate.
The core was precisely centered, and the retaining
plate-core fit was adjusted to a tight shrink-fit. The
core might as well have been machined from a solid
block of steel.
The next molding trial was even more frustrating.

The void moved from side to side. Some shots would
have the void on one side of the core and others would
have the void on the other side. We could scarcely be-

Figure 10.20 A proposed electronics case.

© Plastics Design Library

lieve what was happening. The core could not be moving that much without breaking, but it was. The core
was flexing almost 1/16 in. from side to side, each side
of center.
We finally figured out that, on a random basis, the
plastic flow would start down one side or the other of
the core. As soon as the flow started down one side,
the injection pressure would flex the core slightly, encouraging flow down the thicker side and inhibiting
flow down the thinner side. It was a vicious circle:
• Thicker wall, easier flow.
• Easier flow, more pressure.
• More pressure, more flexure.
• More flexure, thicker wall.
Okay, we can solve this problem. We sought and
obtained permission to put a couple of alignment dowels between the core and cavity. The electronics were
to be potted into the case and the potting material would
seal the holes. We placed a 0.250-in.-diameter dowel
near each end of the core on each side of the gate. The
dowels would keep the core centered.
The dowels lasted an average of less than a dozen
shots before breaking.
The pressures in an injection mold are sometimes
almost beyond belief. Consider that the injection pressure was in the vicinity of 10,000 psi. If one side were

to be mostly filled with molten plastic before the other
side started to fill, the pressure on the filled side would
be 4 × 4 × 10,000 lb. That’s 160,000 lb! That’s a bending moment of about 320,000 lb-in. or 16 ton-inches.
No wonder the core was flexing. The dowels were only
about 0.05 in.2 each, and if each carried one-quarter of
the side load, that translates to a shearing stress of
40,000/0.05 = 800,000 psi. No wonder the dowels were
breaking.

Figure 10.21 Modifications to keep the core in the center of
the box.

Ch. 10: Case Studies


152
After much anxiety, tears, and prayer, someone suggested changing the core design as shown in Fig. 10.21.
This had the effect of a diaphragm gate that would not
be removed. It would tend to force relatively uniform
flow to both sides of the part. In addition, if the core
tried to flex, it had the effect of closing the gate slightly
on the thick side and opening it on the thin side. It was
what is known as a “negative feedback” system.
It worked. We molded thousands with only minor
wall-thickness variations. A sample part is shown in
Fig. 10.22 Notice the classic hourglass shape of the
open end, caused by higher shrinkage at the hot corners of the core as compared to the cooler cavitytemperatures in the corner. As usual, I only get to keep
rejects. See the missing letter “D” in the phrase
“FRAGILE _O NOT DROP.”

elevator, one on each side at the top and bottom of the
elevator car. I understand that the nylon gibs were direct replacements for cast bronze gibs; perhaps that
explains in part the massive cross section. The gibs
were held in cast and machined iron brackets that restrained the gibs on all surfaces except those in contact with the guide rail.
The molding problem related to this part that I want
to mention is the warpage that is shown in the drawing
and shaded images. The sidewalls are supposed to be
parallel, but because of slower cooling and higher
shrinkage at the end and corners of the mold core, the
sidewalls were drawn toward one another as shown.
Our best efforts to cool the core more rapidly were
of little help. The only way to control the warpage was
with a shrink fixture. Two pieces of steel about ¾-in.
thick, 2-in. wide, and 4-ft. long were machined on a
taper, reducing the ¾-in. thickness, with the narrow
end of the taper the same width as the bottom of the
groove in the nylon gib. The sides of the steel bar were
machined with a taper so that the gib was held with the
groove slightly wider at the open side than the closed

Figure 10.23 Elevator guide gib, approximately 1.5 in.
across, 2 in. tall, and 9 in. long.
Figure 10.22 Photograph of the molded box with a “flexible”
core.

10.15 Elevator Gib Warpage
Most personnel elevators are guided as they move
up and down the elevator path (it can hardly be an
elevator shaft when it is outside a building), with rollers or wheels that roll against two guide rails, one on
each side of the elevator.

At one time, and perhaps in some cases today, the
U-shaped nylon gib shown in Fig. 10.23 and Fig. 10.24
replaced the rollers. There were four of these on each

Ch. 10: Case Studies

Figure 10.24 Shaded image of the elevator gib, as molded.

© Plastics Design Library


153
side when the gib was placed on the bar. The two bars
provided cooling space for eight or ten gibs. After each
one-cavity mold cycle, the coolest gib was removed
from the bars and replaced with one fresh from the
molding machine. This allowed about fifteen minutes
of cooling time out of the press and on the shrink fixture. If the nylon gibs were slightly wider at the open
side than the closed side, it was felt that the iron holding-devices on the elevator would hold the sides parallel when they were installed.
The holes in the nylon gibs were simply to remove
mass. When the gibs wore to the point that the holes
were exposed to the guide rail, it was time to replace
the gibs. There were some cored areas across the closed
outside edge of the gib as well. These are not shown

10.16 Sucker-Rod Guide Brittleness
Oil-well pumps are usually several thousand feet
below ground. Steel rods called sucker rods, that extend to the pump jack on the surface, drive them. The
drilled oil-well hole and the pipe that lines the hole are
not completely straight. However, the sucker rod, being under tension, tries to assume a straight line. This

causes the sucker rod to rub against the pipe that lines
the hole. Over time, the sucker rod can wear out or
wear a hole in the pipe. Then, instead of pumping oil
to the surface, oil is pumped into the strata where the
hole is worn. This is not good from both an economic
and an environmental standpoint.
To avoid wearing through the pipe, nylon guides,
like the one shown in Fig. 10.25, are placed on the
sucker rods every so often. The nylon wears faster than
the pipe by far, and also distributes the wear over a
larger area.

Figure 10.25 One of several designs for a sucker-rod guide.

© Plastics Design Library

These rod guides are supposed to be installed by
placing the groove around the sucker rod and driving
the rod guide against the rod so that the rod snaps into
the circular center section of the rod guide. The grip
against the rod by the rod guide holds the rod so that
the rod guide moves with the rod as it moves up and
down. Installation requires a BIG hammer.
We made several test installations when we first
tried the new mold, entirely successfully. The rod guides
were still warm. The molding problem we encountered
was that when the customer came, he performed the
same test, with his big hammer and tested parts that
had cooled overnight. The nylon rod guides may as
well have been glass. They shattered into a dozen pieces.

The customer explained that the rod guides are often
installed in the arctic. They have to be tough enough to
be taken from a deep freezer and installed while cold.
The cure for the problem was to boil the rod guides
for several hours. This forced them to rapidly absorb
water. Once they were thus properly moisture-treated,
they could be installed while cold, using a 10-pound
sledge, without breaking.

10.17 Bottle-Cap Thread Distortion
Sometimes, if threads are shallow enough, or if
the plastic is flexible enough, undercuts such as threads
can be stripped from a core. Just such a scenario was
planned for the bottle cap shown in Fig. 10.26.
The desired thread profile (which was the profile
cut in the mold core) is shown in Fig. 10.27 (A). The
rounded thread had a sharp corner where the thread
contacted the wall of the cap. During the stripping operation, the threads were distorted so that they looked
something like the profile shown in Fig. 10.27 (B).

Figure 10.26 A bottle cap that is stripped from the core.

Ch. 10: Case Studies


154

Figure 10.27 Bottle-cap thread profiles.

The bottle had buttress threads on the neck, with

the flat side of the buttress thread toward the bottle.
The distortion of the threads in the cap caused interference between the bottle threads and the cap threads.
The cap was very difficult to thread onto the bottle and
defied reliable installation by automated filling equipment. No amount of tinkering with the molding machine conditions or mold temperature would resolve
the problem, although adding a small radius, similar
to that shown at Fig. 10.27 (D), helped.
Ultimately, we had to accept that some distortion
of the threads was inevitable. Therefore, new cores were
made with a profile that would have looked like the
profile in Fig. 10.27 (C) and (D) if the part could have
been removed from the core without distortion. We
found that the radius shown in Fig. 10.27 (D) was particularly important. It apparently reduced the unit stress
during ejection, made it easier to initiate the stripping
action, and provided a smoother surface over which
the threads slid.
After being stripped, the threads looked much more
like the intended profile as shown in Fig. 10.27 (A).

stop-sign posts about 2.5 in. in diameter to power-line
poles that were 10 to 14 in. in diameter performed superbly.
Several stop-sign posts were installed in a small
Oklahoma town in the summer. By the end of the summer, the signposts were leaning north by 20° to 45°.
The heat and the persistent south Oklahoma wind conspired to maintain a high enough load from the south
to cause the plastic pipe to creep and unload the internal compressive forces and allow the pipe to “lean north
with the wind.” But you should see some of the Oklahoma trees. In some cases, the leaves and branches are
all north of the trunk. Maybe the plastic didn’t perform so badly after all.

10.19 Excessive Shrinkage of GlassFilled Nylon
A four-cavity, three-plate, center-gated mold for a
cup-shaped part, shown in Fig. 10.28, was built using

published shrinkage data for 50% glass-filled nylon.
The gate design was a bit unusual in that it was a ring
gate around a central core-pin that extended into the
drop tube from the runner level of the three-plate mold.
This allowed the gate to break at the closed surface of
the cup, leaving a center hole as required by the drawing.
At mold trial, the height of the part was just fine
but the outside diameter was undersize. While the cavity was drafted so that the open end of the part should
have been larger than the closed end, the part was actually smaller at the open end than at the closed end.
The flow pattern of the mold oriented the fibers radially from the center gate and then parallel down the
side walls.

10.18 Plastic Post Creep
It is politically and environmentally correct to divert used rubber and plastic products into secondary
uses. One such attempt was to make posts from groundup scrap rubber and plastic. The particles were mixed
with a bonding agent and packed into PVC or polyethylene pipe. The theory was that the compressed rubber inside the pipe would keep the outside structural
pipe in tension and thereby make it stiffer and stronger. Initial tests yielded great results. Everything from

Ch. 10: Case Studies

Figure 10.28 Glass-reinforced nylon cup.

© Plastics Design Library


155
The radially oriented fibers in the closed end did
not allow significant shrinkage on the diameter at the
closed end. However, there was nothing to inhibit
shrinkage at the open end. The parallel flow down the

side walls inhibited shrinkage on the height of the cup,
but the fibers oriented along the height of the cup did
little to prevent circumferential shrink. The fiber orientation caused the radial shrinkage to be at or below
the published shrink rate, but the circumferential shrinkage at the open end of the cup was above the high-end
published shrink rate. The net result was that the differential shrinkage from the top to the bottom of the
side of the cup was so great that the open end of the
cup was below tolerance and the closed end of the cup
was at or above the maximum tolerance.
Fortunately, the only critical dimension was the
open end of the cup, and it was possible to correct the
problem by increasing the diameter of the flange at the
open end of the cup by grinding the mold cavity.

10.20 Preventing Warpage in Thin
Molded Lids
Thin container-closure lids are often a very exacting and difficult molding operation. Preventing warpage
caused by differential shrinkage requires special attention. The previous example discussed differential
shrinkage caused mostly by fiber orientation. In lid
molding, the differential shrinkage is mostly due to differential pressure from the center gate to the periphery

© Plastics Design Library

of the molded lid. The center of the lid is exposed to
much higher pressure than the outside edges of the part.
Most of the differential pressure is caused by the increasing viscosity of the plastic as it flows away from
the gate and is cooled.
It is well nigh impossible to control the temperature and pressure differential. The next best alternative is to design the lid with a level offset to provide a
flex ring to absorb the radial shrinkage variations. Figure 3.13 shows a flex-ring design. The figure is reproduced here as Fig. 10.29 for the reader’s convenience.
The further from the gate, the greater the shrink
rate. Figure 10.29 shows that this lid has “toe in” that

is caused by greater shrinkage at the open edge than at
the closed edge of the cylindrical section. The disk section attached to the closed end also restricts shrink.
Internal snap rings can be stripped from the core when
molding polyethylene provided they are not too deep
and are well-rounded. See the molded, internally
threaded part in Sec. 10.17. Snap rings of 0.030 in.
are common. If the depth of the ring exceeds 0.05 in.,
it may not be possible to strip it.

Figure 10.29 A molded lid with a flex ring and a stacking
ring.[3] (Reprinted with permission of Voridian, Division of
Eastman Chemical Company.)

Ch. 10: Case Studies