Tip of the Day 185: Sensor Placement: “Put the Bandage where it Hurts” †

When I was working with a customer on a part similar to that in tip #183 we were planning sensor locations. Of course we jumped first on the idea of last place to fill.

As would seem logical we thought, “To find a short shot, put the sensor where the flow fronts join together furthest from the gates.” Of course there are four possible spots and we might need to pick only one. But it seemed like a good idea at the time.

As you recall from #183 it turns out, becasue of freezing, that the thin flange makes the part fill last on the flange opposite the gate like this…

Shot – no goal.

So let’s put an ejector pin on the flange near the gate.

But then we found out that short shots are NOT the problem. In fact, the problem is under-packing near the gate. This causes a sink in the thick section and a bad seal with a rubber gasket. Another shot – no goal.

If we had located a sensor at the edge of the flange (as above) it would be frozen too early to “see” the under-pack condition in the thick section. Instead we selected a position near the first gate to freeze as the monitoring sensor location.

What do we learn from this? Here are some general principles for locating the “bandage” (an in-cavity sensor) to score a goal; i.e. to get the most value for effort.

1. First, ask “What hurts?” That is, define the problem you are trying to solve.^‡
   
    


2. Place the appropriate sensor near the location where the problem is so that it can „see“ the changes that cause the defect.
   
    


3. Sensors can be used for V->P transfer control (switchover) to improve consistency. They should be in the „area of influence“: A place that the machine can still drive plastic at the time the switchover is to take place.

Control sensors do not guarantee quality everywhere in the cavity. More to come on control sensors later.

 † The title for this one comes from Denis Fecci, charter employee of RJG France.

‡ Some quality problems and the data we need to see them:

• Pressure Related (cavity pressure sensors):
  
  Sink, short, flash, void, dimensions, weight, chemical resistance, warp or in-molded stress (pressure difference across the cavity)
  
   


• Flow Rate Related (timing between cavity temperature or pressure sensors):
  
  Paint or chrome adhesion, blush etc.
  
   


• Temperature Related (cavity temperature sensors or back-slope on pressure):
  
  Cooling circuit variation, imbalance or blockage => warp due to semi-crystalline shrinkage behavior, improper melt temperatures.

Tip of the Day 184: Thin to Thick Wall: Molder’s Misery part 2

Now let’s reverse the situation for mixed wall thickness as discussed in tip 183. We often see this situation in such parts as idler wheels, gears, impellers and so on. These parts have a thick outer rim with an inner “web” that has been made thin to save weight and cost. Typically they are semi-crystalline parts.

Being semi-crystalline, a substantial amount of flow occurs during hold. The material shrinks dramatically as it crystallizes (see tip # 178). Sometimes up to 10% of the total material required can flow into the part during hold.

But, at the same time the thin rib section is also cooling. With less mass and thickness it cools much faster, closing the flow channel. As flow decreases, the heat supply keeping the rib “open” decreases. This rib acts like a gate that freezes before the thick outer rim has enough material in it. Like so…

The thick rim that has flow “choked off” becomes a low density area. In fact, the flow through the thin section ceases before even before the gate freezes.

It follows, then, that this low density section will continue shrinking as it cools and create a void (in this case). The strength of the material (acetyl) is able to hold the outer wall rigid while the inner section shrinks. This makes a void instead of a sink.

What to do?

• Part Design?
  
  
  If the outer wall could be made thinner its freeze time would more closely match that of the inner web. Or perhaps some ribs from the gate to the outer wall would stay open long enough to feed the shrinkage in the rim.
  
   


• Differential Cooling?
  
  
  As before, we are trying to match the plastic variable(s): Cooling Rate and Cooling Time. We might be able to slow the cooling of the thin section by allowing the mold in that area to run hotter. This could keep the „pipe“ open to feed the full shrinkage of material into the core of the rim.

Note: If the  material is fiber filled much of the shrinkage behavior will be controlled by fiber orientation rather than the quantity of polymer in the final part.

Tip of the Day 183: Thick and Thin Wall: The Molder’s Misery

Many of our readers have read or been taught how important it is to design parts with constant wall thickness. I asked some of our consulting team, „When did you last see such a part?“ Apparently they are becoming rare. We are now fining many more parts with varying wall thickness.

The molding quadrant and the discussion of  thin and thick wall now leads us to this “mixed” wall thickness problem. We’ll start here with thick-to-thin. The next tip covers thin-to-thick.

Example: A ring with a flange

You may easily imagine what will happen to the material flowing into the flange vs. the ring body. Let’s walk through it.

1. The pressure loss in flowing plastic is always the same from the gate to the flow front. The flow front is ~0 pressure and the gate is just one point (effectively).
   
   
   
    


2. That pressure loss is relative to flow rate, length AND to the inverse of wall thickness CUBED! 1/T³ or 1/(T x T x T).^†
   
    


3. In order to have the same pressure loss in the thin flange as in the thick ring section, something must balance to get equal pressure loss. Thus the material in the flange flows a much shorter distance in the same time the material in the ring flows significantly further. This means a slow progression of flow front in the flange.
   
    


4. The flow in the flange is SO slow that the heat supply keeping the “pipe” (flow channel) open is not enough to prevent freezing. As mentioned in thin wall parts, we hit the “time barrier” in a thin section where freezing occurs.
   
   
   
    


5. Of course, as cooling occurs the material becomes even harder to flow and soon shuts off flow entirely in the flange opposite the gate. With the material stalled, the flow front wraps back around itself from the knit line creating air traps inside the flow front (note the one on the right side). This creates the odd scenario where the last place to fill is a point right near the gate.
   
   
   

What to do?

• Don’t BUILD parts like that!
  
  Of course we rarely have a choice. So…
  
   


• Raise the melt temperature?
  
  Perhaps. But the added heat load will make it harder to cool the thick ring.
  
   


• Fill faster? (3.5 x faster shown here).
  
  
  
  
  This does result in some improvement in the back-flow. But we can see that the last place to fill is still opposite the gate. The flow front keeps moving, at least. But the flow front opposite the gate only progresses as pressure builds up around the ring.
  
   


• Prevent the flange from cooling as quickly?
  
  A tricky bit of tooling, probably. Still, remember that the 4^th plastic variable we are trying to match is Plastic Cooling: rate and time. It would be nice if the flange froze at roughly the same time as the thick ring. To do that with different wall thicknesses would require a hotter mold temperature under the flange than the ring.
  
   


• Other ideas?

† For those following this algebraically pressure loss is proportional to…

 

Thus a 3.5 times reduction in thickness must be balanced by a 43 times (3.5³) reduction in flow rate, flow length or a combination of the two. This is to get the same pressure loss to all points on the flow front.

This does not count viscosity vs. shear, flow front width and frozen layer thickness, of course. But it helps conceptually. The actual power of thickness seems to emperically fall around 2.

Tip of the Day 182: Stored Energy in Thin-Wall Filling

When pressing wine grapes with a bladder press, why do we pressurize it with water instead of air? If you were standing near the press when it burst, which would you rather have in it? 120 gallons of compressed air or water? I would go for the water. Why? Stored energy. With the air at say 100 psi you a standing beside a bomb.

Think of what we do in a mold manifold and injection barrel during fill when we compress plastic to say 25,000 psi (1724 bar). Like air, plastics are compressible, though not as much. Of course the pressures are much, much higher. Once it is compressed, that energy must go somewhere after transfer: either into the part or back into the barrel.

This problem becomes ever more significant as we move the left side of the molding quadrant . Parts with very thin walls have very little volume. The runner system (usually hot runner) that delivers the material has both area (manifold branches) and depth (drops). This can add up to quite a lot of volume relative to the parts. Commonly we see manifold volumes 3 to 5 times the total shot volume (all parts), sometimes even more (12?). Think of it as molding parts with a cushion at 500% of shot size.

So why not just slow down to reduce stored energy? As discussed in the previous tip, we have no choice about fill time. The “time barrier” forces us to get the parts filled before they freeze.

Let’s walk through an example. Here are some data from a 6-cavity, thin-wall packaging mold. We find that the end of cavity cannot be packed much after about 0.5 seconds. So we end up with a 0.2 second machine fill time and 25,000 psi injection pressure to do it.

How can we lower this stored energy? If we transfer earlier or decrease filling speed, then the peak injection pressure will be lower. But then more of the part would be filled in the decompressing phase, “coasting” toward a full part. This would cause even more freezing and less ability to pack. And more variation as shown here.

Here is one method that worked in this case. Instead of transferring earlier or filling slower we added a second speed to slightly slow the screw just before transfer. This is a little like decoupled 3, though in this case transfer occurs before the cavity is full. Also, the machine did not have external transfer so we used position for transfer.

So the best we can do in very thin-wall molding is to try to control the crash.

In the next tip we will cover some adjustments in our thinking that can help interpret these graphs.

Tip of the Day 181: Thin Wall Molding: The Time Barrier

If you saw the film “The Right Stuff,” you may remember their concept of the “demon that lived out beyond mach 1.”^† In thin-wall molding it may help to think of that “demon” as TIME. Time to freeze during fill or shortly after during pack.

Parts made for packaging are often around 1 mm ( .040” ) thick or less. Even though the flow lengths may not be very long, any hesitation in flow will allow the wall to freeze so quickly that packing will not be possible. Or, if there is no hesitation in flow, a full and packing part has such slow flow that the part freezes very quickly after transfer.

The “time barrier” is the time beyond which you cannot influence much of the part. If you can find that time then you know that all filling and packing must be completed before it.

There are at least two ways to test or observe the freeze time using cavity pressure. One is to make adjustments in hold time or steps in hold pressure. Then observe the effect on the sensor in which you are interested. This graph is a hot runner and so there is no gate seal.

In the graph above, the “End of Cavity #Ring” sensor cannot be influenced much beyond about 1.8 seconds after the cavity is full. Even at 1.3 seconds after fill, dropping injection pressure (using hold time) has a minimal effect. The gate end still has some influence but even that is quite small after 1.8 seconds.

The second method is to observe the difference in the pressure slope for EoC relative to the injection or post gate pressure slopes.

In this graph you can see that the slope of the post gate is steeper than the end of cavity. The part is freezing so fast that the influence over end of cavity is getting weaker as it fills. In fact, the part barely fills at all and then does not really pack much after it is full.

In the graph below we have finished filling and packing the same part before 0.35 seconds instead of out at 0.85 seconds. While it is hard to pick out, you can see that the pack slopes are similar.

The slope of the EoC pressure during fill is NOT similar to the PST. This can’t be helped. There is some freezing during fill which accounts for the different slopes. And we could not fill faster because of pressure limits on the machine.

If you want to set up a decoupled 3 process, as we did above, your fill AND cavity pack time must be shorter than the wall freeze time. The point is, you need to know that time for your working melt and mold temperature.

What makes a part „thin-wall?“ (The left side of the molding quadrant) Here is a working definition: a part in which packing (and maybe filling) is significantly influenced by freezing.

In thin-wall molding, a leisurely fill or pack will rouse that demon that lives out beyond the time barrier.

^† The “demon” in the Bell X1 was the instability as the sonic shock wave built up. I was always intrigued by how they later discovered that the supersonic cross-sectional area profile had to include the wings as well as the fuselage. Hence the “coke-bottle” shape of modern super-sonic jets. 

Tip of the Day 179: Volume Curves: Shrinkage or Leakage?

In tip #178 you saw how the screw moves forward as a thick, semi-crystalline part continues to shrink after “packing”. The action of this shrinkage can be studied by making the volume curve scale larger to see the detail after the cavity pressures reach their peaks.

The crystallization of the melt in the cavity allows the screw to move at an ever decelerating rate until the cavity freezes. The curve looks like an arc because the molten core in the part continues to get smaller as the walls thicken, reducing the flow rate throughout the shrinkage phase. Think of the slope of the volume curve as the flow rate. A steep slope is faster flow. A horizontal line indicates no flow. When the part has “frozen”, the volume curve flat-lines showing that the screw has stopped.

Consider the following process where the screw does not stop at the freeze time.

Here the screw continues to move forward after the part is no longer accepting material during crystallization. If the material did not go into the cavity, where did it material go?

Through the check ring^‡. In fact, if left long enough the screw will go all the way to bottom.

Take-away:

If the screw continues to move after the part (or cold sprue) is solid, then the check ring is leaking. Leaking screw movement usually appears a straight line on the volume curve instead of the arc that you see during the shrinkage phase.

Aside:

For those curious about the rise in post gate pressure after the EoC cools, see tip #146.

Technical Notes

The fluid flow equations tell us that, with a fixed pressure, flow is proportional to pressure, viscosity and the size of the flow channel.

Since there is no flow into the cavity after freezing, the only place for the material to go is back through the check-ring. If the check ring has a crack or does not seal, it forms a flow “channel” of a fixed size.^† Since the size of the channel is not changing, the viscosity is constant and the pressure is constant (hold) then the flow is at a constant rate.

This works for amorphous materials as well as semicrystalline in both hot and cold runner, thick-wall molds. In thin-wall, the screw motion during pack is so small that the arc may not be visible. However, if the check-ring is leaking during hold, the screw will still move forward at a constant rate after the cavity or sprue is frozen.

——–

^‡ “Check-ring” here is generic for any non-return valve.

^† The “D” is the effective diameter, also known as “hydraulic diameter” which is 4A/P.

Tip of the Day 178: When “Hold” Isn’t Hold in Decoupled III

Once upon a time a customer complained using Decoupled III “didn’t work.” After wiping some egg off of our faces we found out what he meant: That D3 did not control the part dimensions when viscosity varied. We can show that D3 does control the process in almost all cases better than D2. So why not the dimensions?

Let’s start with a look at the process graph:

You can see that the Decoupled III process control did a reasonably good job of controlling the cavity pressures, especially the peaks. Decoupled III is a process of filling fast to a machine position, packing with a steady machine speed to a cavity pressure and switching to hold to keep the material in the part until the gate seals.

You may notice that the hold time is quite long and the cavity pressure near the gate stays high for that whole time. This part is mostly thick-walled. Also, the reasonably sudden roll-off of end of cavity pressure tells us this is probably a semi-crystalline material.

Now let us look at the motion of the screw during hold ^‡. This is easier to see if we expand the scale, making the high edge of the graph window 1.85 in³ and the low edge 1.6 in³.

If D3 is supposed to pack to a cavity pressure and then hold the material in, why did the screw move 7.2%^† of the total volume required AFTER pack?

Shrinkage.

With semi-crystalline materials, the molecules “cuddle” together as they crystallize. As the material transitions from melt to solid the material very suddenly requires much less volume. This makes more room for more material to flow into the mold as the material freezes from the walls inward. This is sometimes called “compensation flow.” So the “hold” phase is not really holding but is still feeding material in as the freezing material shrinks. In amorphous parts this percent after pack is commonly less than 1%.

What happens when the viscosity increases?

If you apply a constant pressure to a fluid (plastic) to make it flow (hold pressure) and viscosity increases, then flow decreases. But the freezing rate stays the same based on heat transfer. When remote areas of the part freeze they may not have had time to accept all the molecules required to hold a dimension.

What does this mean to you?

• In this class of part, flow does NOT stop when the cavity is “packed” to a peak pressure.


• Cold gate designs must account for the required shrinkage flow and stay open as long as necessary.


• Valve gates must stay open long enough to account for shrinkage flow.


• If the gates close too early then voids and sinks are probable.


• If viscosity increases, fewer molecules fed in during hold can mean more shrinkage and a shorter part.


• To control dimensions, track viscosity variation and modify the process as necessary to restore them when viscosity changes.

Quiz question:

Which molding “quadrant” is this? (from tip 177). How does it differ from the opposite quadrants?

^‡ Some screw motion could be due to leakage of the check-ring (or non-return valve). Looking at these details requires a good seal on the screw. Leaking check-rings the the screw move forward forever under pressre, even with very long hold times (runner frozen) and they move with a constant speed: a non-changing slope on volume curve. The one in this example was good.

†

Tip of the Day 177: The Molding World in Quadrants

We often risk using a kind of “one size fits all” mode of thinking about plastic processes, their control and quality monitoring. I have found the following diagram useful for separating the different regions in the world of injection molding.

(For those who need the more complicated version, footnote^‡ )

Future tips will refer to this idea as we work though different strategies for quality and consistency. For example, you could use the quadrant idea to think of different applications in the injection molding industry like this:

Or you might imagine different areas of importance for control like this:

The main point is that each application should be evaluated based on its particular behavior. This basic, four quadrant model tries to cover the most significant categories.

† The term “thin-wall” can generally but thought of as an aspect ratio: flow length / wall thickness. Even thick-ish parts (2-3 mm), if quite long, can require fill times that allow freezing during fill. However, even short but very thin (½ – ¾ mm) parts can be called thin-wall because they can still freeze very fast.

^‡ If you like to complicate things you could add a “Z” axis for fillers: maybe fiber in one direction and filler in the other. You can also imagine parts with both thin and thick wall and blends or co-polymers of semi-crystalline and amorphous materials.

Tip of the Day 161: How to Prove Material Change

Tip #160 triggered a surge of e-mails regarding material changes. Apparently this problem is more common that we realized. It causes aggravation for molders who are left with the burden of proof to show the material supplier that there is a documented change.

How can prove that the material has changed? In the previous tip we used process of elimination: The mold did not change nor did the process as measured from inside the cavity. The material was pronounced guilty on circumstantial evidence only.

Suggestion: During „Validation“ or „PPAP“ place a quantity of the original material in storage. You would, of course, save enough to allow a process to stabilize long enough to get valid parts for measurement. Perhaps more than once. Also, carefully document the in-cavity process and the cavity dimensions as built. Include with clamp style and tonnage.

Then, if you later discover a change in product that is not explained by changes in the mold or process^† you can re-run the validation with the original material. Run the exact same process and mold (i.e. cavity shape) with the original material. If the parts prove to be the same with the original material as they were during validation then you have the smoking gun: material. If the parts are different with the original material than they were during validation you need to look for mold (shape) changes or process changes.

^† Remember that the process is what the material sees in side the shape of the cavity (including mold deflection). For example, if the water temperature and flow are identical but someone put smaller water fittings on the mold you may have different cavity temperatures. To match the process you must match the in-cavity variables.

Tip of the Day 160: When a matched template… doesn’t

… match.

Suppose you were molding a 7” diameter dish-shaped part out of glass filled polyester. It has a bunch of holes and bosses on it. You have pressure sensors at both post gate and end of cavity in all 4 cavities as well as a temperature sensor in each. You did the designed experiment, found the best process center and adjusted the steel to get the flatness within the specification of ± 0.010″.

One day some parts are returned from your customer. Sure enough they are out of specification for flatness. The process is running Decoupled 3, controlled velocity pack with cavity pressure control. The peak pressure is the same and the cavity temperature sensors show the same mold temperature.

You try to match the template by adjusting the hold pressure.

But you can only match the post gate and even that only approximately. And when you match one curve the other goes into out of the alarm band. Why?

Step back for a moment and consider what makes the part. Here is way of thinking inside the cavity that focuses on three fundamental aspects of creating the part.

The “process” is what is going on inside the part cavities (shapes): The “4” plastic variables. It is driven from the outside by various equipment: the press, coolant control, drying, hot runner temperatures, valve gate control etc.

In our story we did not change the mold (same press => no deflection) and tried to keep the process as constant as possible. What remains in the eternal molding triangle that can change?

Material.

If the material changes and the process and mold remain constant then the parts are likely to change. In this case the material was the same grade number but the supplier had changed the flame retardant. This apparently caused a change in nucleation or crystallization rate and hence changed the tendency of the part to warp.

Moral:

Decoupled III molding and template matching cannot control the whole process that makes the part ^†.

 If the material changes significantly, especially in crystallization of semi-crystalline parts, you have a new and different material. You need to create a new process for that material using a new material in the eDART’s™ Job Setup. Then, after validating the parts, reset controls, templates and alarms for a new process to fit the new material.

Next Tip (# 161): How to Prove Material Change

Philosophical Phoot-notes

^† Decoupled III provides significant improvement in pack pressure control when material chain lengths change the viscosity. In the lab we compare processes by throwing a viscosity shift into a D2 and D3 process and comparing the peak pressure results. This is to demonstrate the kind of material variation typical of changes in lot. It is possible that our students get the impression that D3 controls everything under any conditions. It does not.

Likewise, just matching a template will not make the same parts with a different material. Each different material compound has different packability, cooling and shrink rates for a given process. This includes both the molecular structure of the compound and the additives. Therefore matching the process template for a different material is likely to produce different parts.

Decoupled III and template matching is not a panacea. They are tools to reduce process variation (without human adjustment) to and detect changes. Some material changes are large enough that they require a new process.

^‡ In some sense the mold as a whole could be considered an auxiliary. From the point of view of the plastic creating the part inside the cavity, all that it “knows” is the shape it is in and the plastic variables inside itself: Melt temperature, flow, pressure and cavity temperature. The water lines are an “auxiliary” designed to deliver the required process varaible „cooling rate“ efficiently. Support pillars deliver a more constant cavity shape. The runners and gates deliver the plastic into the cavity. But the process acting on the material inside the shape makes the part, regardless of what is going on outside.