Silicone Rubber part, behavior and wall thickness 3

[maxvl]

Dear forum members,

I am working on a silicone part for a new product. You should be able to push it down, and then it should stay there, until you pull it back up. I have no clue how to choose the exact shape and wall thickness, so that it is easy to do this with one hand, maybe I should have a variable wall thickness or other features for the desired functionality. I chose silicone rubber as a material, because of the fact that is will be incorporated in a "kitchen appliance" and it should withstand the dishwasher.

My explanation might not be clear enough so I have included images.

Maybe there is some way to simulate this phenomenon, but I reckon it's too complicated and I'm better off with an expert's opinion from someone who works with rubber daily.

Feel free to ask for more information.

Thanks you,

Max

 

[maxvl]

image 1

 

[maxvl]

image 2

 

[maxvl]

Is there no one who can help me with this, or did I use the wrong forum?

Max

 

[GrahamBennett]

I think you've posted it in the right forum Max but there's no-one able to help. I suggest you start by talking to your silicone supplier's technical department.

 

[btrueblood]

Yeah, don't know of any forum that would work any better. The phenomena you describe is called "snap through" buckling, and it is a pretty non-linear problem to work on. Personally, I don't think your proposed shape will work, I think the outer edge has to roll over for the deformed condition to be stable (unless the bottom gets "stuck" to something?). The "poppers" that my kids played with looked similar to what you drew, but they were not stable in the buckled position, and would pop back after several seconds, hopping into the air when they did so.

The good news is, silicones are easy to mold. If you have access to a lathe, you should be able to bang out prototype mold halves (top and bottom) and tweak the wall thickness and outer edge detail (you may want a thick "bead" at the o.d. to control shape, but again, I'm just guessing). Silicone mold compounds are readily available in small quantities. Check Mcmaster-carr.

 

[maxvl]

Thanks for the replies. I had contacted a silicon supplier, but they're not very helpful, so far. I'm going to try and make some mock-up models with a sugru substitute (silicone chaulk and corn starch). And later maybe some higher quality parts with casting silicon. Thanks for the description of snap through buckling, it seems as though I might need to add some "fold lines", or look at collapsible silicone tableware.

Max

 

[btrueblood]

You're welcome for what little help we can give, Max. I think your fastest route forward is to play with real objects, so you are going the right way. Do be careful, silicones are available in a wide variety of stiffnesses/durometers and you can change the way your part behaves by changing compounds. But, with enough experimentation, you may come up with some design rules of your own on how the shape needs to change to compensate for material variations...

Good luck.

 

[acspain]

Max,

My suggestion would be to use Finite Element Analysis. You will need a non-linear package such as Abaqus and some tensile data on a material.

An experienced analyst would be able to evaluated multiple design concepts in a short amount of time.

My experience is that analysis is faster, cheaper, and provides more detail than going straight to prototypes.

Good Luck

 

[btrueblood]

"and some tensile data on a material."

I like FEA, but predicting non-linear snap-through buckling on non-linear material? You're talking pretty cutting edge stuff there, people are writing their PhD theses on these types of topics.

You need a lot more that just a simple pull test to even approximately characterise rubber. And every rubber formulation is different, so you need test coupons for each one. So, he would have to buy samples of rubber and cast coupons, and send them to a lab, and do a series of FEA runs on several geometries. What's a seat of Abaqus or ANSYS cost, with full non-linear solver packages, upwards of $20k per seat?

"My experience is that analysis is faster, cheaper, and provides more detail than going straight to prototypes."

I could build a prototype mold for less than $1000...I'm not sure the analysis approach is the best way for this project.

 

[acspain]

btrueblood,

Your right there is a lot to it and maybe the cost is too high. I was simply throwing out an option that I have used as a product engineer (at a cost sensitive manufacturer) and have expertise in as an analyst.

To answer your points:
1. The goal, as I see it, isn't a deadnuts match between analysis and product but a series of simple analyses (2d, axisymmetric) to point in a direction. Then onto to prototyping.
2. I have modeled parts that would go through a wall collapse or snap behavior. If a static run didn't work then onto quasi-static dynamic runs.
3. Abaqus has a material model called Marlow that takes one tensile data set to characterize the material. The Marlow model does as well as or better than the three expensive and difficult tests recommended by literature and used by other material models.
4. Yes, each elastomer is different. The goal is to get to a part behavior and you could do that with one material. Prototype to confirm and maybe learn enough so that if you have to switch materials then another material test and analysis may not be needed.
5. If the op does not have analysis resources already in house then he would have to contract it out to make the cost palatable. You wouldn't buy a seat nor spend the time needed to learn it for one project.
6. At my previous job, the prototype molds cost well beyond a $1000 a piece. If you could prevent building a couple prototype molds along with the time savings of waiting for the mold to be built, waiting for press time and then figuring out what to do next when things didn't work out, and repeat, well, analysis ending up saving time and money.

 

[maxvl]

Hi all,

Thanks for the in-depth discussion.

I've messed around with some silicon and made a quick and dirty prototype that luckily suggests that the desired behaviour is quite easy to obtain. I'm going ahead and order a prototype for my total product. I can make some final adjustments afterwards if needed. It helps that the product is not extremely cost-sensitive. Maybe I'll even ask my Abaqus trained friends to help me out with some additional simulation.

Thanks again for your thoughts!

Cheers, Max