short shot simulation, holding/packing phases, Tmelt, DOE

[rwoodyn]

Dear gents/ladies,

I'm working on classical ordinary injection molding floor and recently I was thinking about some "traditional" problems so I was thinking it might be good to try to post some of these to actully see what is the opinion of other people.

I have more questions, but I streamlined them into 4. They are completelly diffent, more like examples of questions from different groups of problematics.

Would you be great if you can comment/give advise or just note something. Thank you in advance!

1. Simulation of "short shot" in simulation?

- is it possible to reasonably simulate/emulate in software what is happening with plastic melt during "short shot" in real mold? I mean that when plastic is actually compressed, then screw stops, there is some residual pressure with in the melt so melt front must be still moving a bit even though screw is stopped. Can Moldflow simulate the same? To say for example, inject only 80% of full cavity volume and to simulate how melt expands for the moment in partially filled cavity, etc. when filling stops?

2. Continental vs US vs...  understanding of HOLDING vs PACKING phase terminology

Does anyone have any reference to some "recent book" or who knows what person in industry where it would be clearly defined what is actually exactly "packing phase" (normally described as compensation for shrinkage...) and "holding phase" (normally described as isobaric cooling)?

I mean in books you have normally 2 things:

- picture of the pressure versus time as measured pressure inside the cavity of tool and on such a picture, there is normally some pictures showing that pressure rises till v/p switchover, then still rises (that till max. pressure is reached) (= "packing") and then follows "holding phase" where pressure in cavity is going down actually.....

- some picture showing pressure of melt in machine (say in nozzle) showing that during filling the pressure in nozzle (in barrel...) is normally rising and then there is v/p swithover to control futher injection by pressure till max pressure is reached (= packing phase) and then holding pressure is applied. This holding pressure could be profile but for simplicity, let say that in book it is shown as constant (meaning for example 10 s of 80% of max. filling pressure).

... but then.... why in "Moldflow Software", the "Packing" phase is what is in "these" books considered as "holding phase"????

It is really confussing for me, these 2 therms are used somewhat loosly or in vague way.

So does anyone have clear (maybe on example with numbers...) clarification as to what

a) machine actually does (from start to fill to v/p swithover, then from v/p swithover to max. pressure, then from that point to "start" of applying holding pressure, etc.).

b) how this "what is happening really on current machines" is actually "considered"/ "implemented" in Moldflow "filling/ packing" analysis?

-myabe adding reference to some accessible articles/books where this terminology and its definition is clarly described...

3. Magic word - "measure melt temperature"

Normally, we see it everywhere (moldflow specialists, some "quality specilists", books, magazines, etc.) - everyone is saying "you have to know what is plastic variable - like Tmelt) NOT whatever barrel zone setting temp.

And we see picture, advices, etc. to actually make air shot/purge and to simply measure with pyromether/thermomether the actual melt temperature.

Normally, the is for purpose of putting this value "regulary" to for example part setup sheet or sometimes for pursposes of some DOE/optimization, etc.

But.....

I don't know why there is actually very big resistance of any molding shop floor practicioner to do that?

I mean, I might be wrong.... but isn't it simply that "simulation guy" is missing some real world experience and simply "does not take into account some things that are simply making production less efficient".

I just try to list some:

- when you process PA 66 GF 30% and change barrel (say middle) temp by +4 oC, you click "OK" or "SET" on your machine.... then what? Do you know how long you have to wait? 20 minutes easily, but could be also 30 minutes....to be sure... But this half an hour is something that is quite some significant time on typical molding floor.... how do you account for that in normall injection practice? By "that" I simply mean that molder has some "business model". So simply... if you go at certain "car garage" where they do only "limited" number of repairs (standard change of oil, break pads, etc.) they will simply not start to repair if something is wrong with electrics on your car... why... beucase it does not fit into their model as to how they make money. Simply... they will not go into some details because... even they know that "there migh be something on it", they know that they can manage only some simple operations and thats it.

- Tmelt is such a quantity that is "never steady", if you measure it for purpose of putting it to setup sheet for particular molded part, then... well, logics is that you would have to measure it really regulary, maybe not every morning.... but every month? or so? Otherwise how do you ever use it actually? How do you put it in SPC chart maybe....?  I mean melt temperature is easily changing simply becuase of natural batch-to-batch variance of raw material! So if you put barrel temps into you part setup sheet... well then.... it is simply very rough control and you never ever can actually find out that there is a problem on molding becuase Tmelt changed.... or for purpose of Moldflow vs reality correlation

- another point - contradictory reaction of Tmelt. For some plastics, easily it can happen, that if you change say rear & middle temp by +5 oC, then actually measured Tmelt can be still constant or EVEN can decrease by 1-2 oC!!! Simply.... because barrel heatter bands are giving some 5-30% of heat to plastic to melt, rest is shearing.... so when plastic is "more molten" on certain part of barrel, consequence is easily that it can flow better, thus having less shearing heat generated -> thus Tmelt after all does not have to rise at all ....  

4. Sig sigma word - "DOE"

I know.... it is possible to change few of these 4 main "plastic variables" (say Tmelt, Tmold, hold time, cooling time...) and to set up "study" to "find" best combination of say these parameters for certain molding to have, for example, critical dimensions well with in tolerances....

But....

My questions are:

- do you know if somebody is actually doing it in real form of DOE study (if you consider 20 - 30 runs, at each making say 5 parts.... measuring then 20 x 5 = 100 parts.... by CMM ....) ?

- to which extent it makes sence to spend 1-3 days with this on molding floor and to do that same in Moldflow simulation? I mean, I know ... simply mold can be cut with some wrong dimension, etc... so after all you "have to" actually always to consider any "optimization" mainly with real tool, etc.... but anyway... anyone who was seeting tools, process etc., knows that time that is needed to do such a (even very simple DOE...) study in practice is significant. .... So if you do that... what next day? How do you know that your Tmelt (with another batch of raw material where viscosity is changing easily very much...) is same as when you yesterday optimized your Tmelt, Tmold, .....???

- so I mean.... doesn't it after all.... in practice GOES BACK TO SOME PERSONS (SETTER).... who actually CAN PROVIDE FOR correction in most efficient way? Because...... how otherwise in practice this is done?

I mean immagine that "next day" you have problem... what do you do???

You call you QA / 6 sigma specialist with whom you plug in "best" parameters to machine and what you do now?

Will he take a look in his "Minitab" to see what? That hold time has "highest" effect on certain dimension?

Will he tell you as correction what? That "Hmm.... I see in my minitab that you should incrase it by 1 second" ... and that should improve this dimension???

Or you should take your pyromether and to start checking if Tmelt is same as yesterday? But it will take 30 minutes.......

My point is: is actually in current "world of injection molding process" COST EFFECTIVE to use this "scientific molding" knowledge? I mean becuase of many administrative/personal/formal problems.... how on the Earth this (in this simple above example) should work in "cost effective way"....

I mean setter "knows" or "should know" that changing certain dials on machine (like increasing holding time by 1 s...) will change this dimension.

I know... yes... he will not write it maybe anywhere.... compared to "theoretically" pure example of DOE application that took 3 days to do and now "only" thing is to plug in these few parameters into machine.

But... setter will adjust it in few seconds... compared to what....days? Hours to do what? Call the QA or 6 sigma guy to do what? Go to some meeting room to take a look into Minitab.... but then 6sigma specilist must "do" something otherwise he would feel useless/not utilized.... but what is that this "someting to do"? I mean machine is waiting "now".... at this minute.....

Any comments, thoughts on this "time gap" or simply fact that people talking too much on DOE....does not simply understand that TIME is what counts and somehow magically they don't understand that machine is idling... or simply on that "contradictory" thing:

- setter is chaning something with in seconds, few shots, minutes....

vs

- QA specilists, 6 sigma specilists, whoever who's action must take "10-20 minues???" to "days"??? to "do" something becase they assume that they know "more" or at least that DOE they have done 3 days ago is source of some "higher knowledge" and they would fidle with their DOE study in XLS sheet or Minitab and would give some advice like "OK, increase holding time by 1 s... and see"........ if you really think about that.... Do they / can they really know better then setter? And.... is their "minutes, to days" duration of their action actually automatically worth compared to immediate action of setter?

Thank you and best regards

Robin

[mtessier]

Hi Robin,

You have some really interesting points here.  I'll try my best to answer them.

1.

You should be able to do this using a 3D Fill analysis.  Set Process Settings>Fill Settings>Pack/holding control>Hydraulic pressure vs time>Pack/Holding Control Profile Settings to some hold time long enough for the flow front to freeze at a hydraulic pressure of 0 psi.  Also, in Process Settings>Fill Settings>Advanced Options>Solver Parameters>Fill + Pack Analysis make sure to check the Simulate inertia effect checkbox.

2.

a) A machine uses velocity-control to fill most of the cavity; then the ram reaches the v/p switch-over point at which time the machine switches to pressure-control to fill the rest of the cavity; then the machine maintains pressure control for "compensation for shrinkage"; then the machine maintains pressure control for "isobaric cooling".  On the floor, reported fill time is often actually time to reach v/p switch-over (you would need a sensor at end of fill to report the true fill time).  On the floor, the difference between packing and holding is not really considered because the only difference between pack and hold is that for pack the material is still entering the cavity, and for hold the material is no longer entering the cavity but the pressure is still applied (from the machine's perspective, nothing changes).  The only way to differentiate between these would be to use a sensor to track the ram position, but both refer to simply controlling the process through pressure.

b) A Moldflow Fill analysis ends when the last node is filled, and it includes both the velocity-controlled phase and part of the pressure-controlled phase (until the cavity is first filled).  A Moldflow Fill + Pack analysis considers all of the pressure-controlled phase including what you call packing and holding.

3.

Your point is well taken.  Yes, it is difficult to accurately measure Tmelt.  Yes, it often seems like the Moldflow user is asking the floor to do work that they wouldn't normally perform.

However, the Moldflow user is trying to solve a very complex mathematical model.  Without the correct inputs, the simulation results are going to be wrong.

If may be frustrating, but if you want the analyst to provide meaningful results, he needs the correct data (Tmelt, coolant inputs, etc...)

4.

This question follows the same line as #3.  Moldflow will always take longer to set up than the setter modifying a machine setting.  Moldflow is most useful for testing virtual prototypes, for testing mold modifications, and for investigating complex input-output relationships.  But remember, once an analysis is set up, you can change any process setting you want and rerun the analysis overnight or over the weekend.  IMO, it is sometimes easier to perform DOE on theoretical results due to the changes batch-to-batch that you are talking about.  How can you know whether a DOE performed on the floor now will be valid after running for 10 hours?  Or next month?

If the setter knows intuitively how to solve the problem, he should solve it.  But if not, Moldflow allows you to dig into the problem further, and come to understand it.  It basically allows you to have pressure and temperature sensors at all points in the virtual mold.

Hope this helps.

Matt