Simulate Early, Simulate Often... In Rhino
I need Your feedback about correct methods on chair testing. I have no experience with CAD analysis and simulation before.
Could I validate this test?
Or, any other factor I might skipped during the test? Also, were there any incorrect number (or method) in the test?
There are two .html reports attached: 03---SnS---Report.html (loaded with scalar force: 1,000 Newtons) and 04---SnS---Report.html (loaded with scalar force: 10,000 Newtons).
On second report, there were no danger level indicated.
Many thanks. :)
The .html reports seem to have not come through with formatting or images since these files are external to the html report itself.
I will attempt to comment based on the images included in your post.
In Scan&Solve a scalar force acts into a surface along the surface normal as if a pressure is being applied. The pressure is determined as follows: Pressure = Scalar_Force / Face_Area. If the loaded face happens to wrap around as you might see with a cylindrical surface, the effect is similar to squeezing the cylinder in your fist -- as opposed to pushing or pulling the cylinder to one side. I only mention this because the arrows used to annotate the scalar force are shown wrapped around the upper part of your chair back. Perhaps this is your intent, perhaps not.
To apply a pushing/pulling force on a face that wraps around (e.g. a cylinder), a vector force should be used. The force might be specified using components (Force_x, Force_y, Force_z) or by picking the Direction through the UI and specifying the force magnitude.
Does this help?
It does, thanks. :)
It should be a pushing force on y direction.
I was not aware of vector components (x, y, z) when applying certain force.
I have updated applied force at sides and top of the backrest. Force components (force_x, force_y, force_z) = (0, 2000 (equal as 200 kilograms on y direction -not per face, correct?), 100). Newtons.
Report indicates an overall safe of the back. Although, there are two danger spots shown:
Report file in .pdf attached. :)
Thanks a lot, Michael.
If you do not select the per face checkbox in the Boundary Load Creator dialog then the load you apply is distributed among the selected faces in proportion to their areas. For more details see the updated documentation.
I'm curious, are you always getting two legends inside "tall" Rhino viewports or does it only do that in the report?
Thanks. I think in this case, it should be correct to assume forces are distributed proportionally in surface area.
We would have this design manufactured very soon. :)
By the way, do You also provide printer friendly documentation in .pdf?
Two legends only appear in report, and not in rhino viewports. :)
For the US standard on chairs you need to look up Ansi-BIFMA, X5.1 (backrest overload test, see attached). This is also pretty much a global standard.
The BIFMA load criteria (406mm from seat, etc) will probably not match anything in your model but that is easy to get around by recalculating the moments (Force1xDistance1=Force2xDistance2) and loading the top bar. Just remember that some energy will be absorbed in the upper region of the frame which will give you slightly better results than the mid-back loading (so make sure everthing well out the red).
PA+GF30% is a safe choice. We are increasingly using PP+GF because of the lowered cost, no pre-treatment of polymer, less shrinkage, etc. We are at an altitute of 5000ft and yes, polyamide can shrink.
We have learned from experience also to check the chair with a reversed load to accomodate Friday afternoon chair races. Chair backs that have been through cycled fatigue testing have failed (with expensive an embarassing results) because this was not checked (fortunately, not my design).
I have used SnS extensively with chairs and it is very effective, but I must qualify that by saying I take a morbid interest in product failures. That is important because you can then mentally link what SnS is 'saying' with what happens in practise.
Don't forget about assembly stresses either - if a part doesn't quite fit it will be knocked in with a hammer/mallet/pipe. Don't forget about shrinkage stresses either - keep wall thicknesses constant, etc. I sound like an old fish wife but I can show you examples (developed in pre-FEA era) that cost us every month. Reliable products sell well.
Ian (Dauphin Human Design Group)
Thanks for the tips. :)
Your extensive knowledge and experience would be very helpful for me.
Unfortunately, we don't have such experience in plastic design. There are some critical points from your note:
We would have a prototype soon enough, but it would be made from ABS, not PA. I think we could only study the proportion, and see if all parts could be assembled together. :)
Many thanks. :)
(Sorry for taking so long to reply... a bit hectic here). It's a good thing to be testing with SnS now but there is still some road to travel. I am making the assumption you are serious about going into production.
I have attached a couple of pics of a nylon frame. Notice how it is shelled out to maintain a constant wall thickness. You will have to develop a seam line if you take a traditional injection moulding route. Alternatively, there is gas assisted (foam core) moulding. That has rapid cycle times (the gas cools the polymer in the mould) but is very expensive to set up. Also not so readily available.
The base of the frame looks solid but has a steel insert. This is the weak point of most frames - for the same reason that the tip of a bow is slender while the base near the handle is wider or thicker (and bows very often break near the handle). So take care with that area in particular.
SnS does not yet simulate multiple parts/materials so you will have to test them separately for now. Try a number of different ways - a vector load, rework that to torque, or swop the load/constraint and so forth. You may get slightly different results but problem areas will reveal themselves.
Our testing is done in a comprehensive lab in Germany. I have bought strain gauges to be able to do some local preliminary testing but I haven't got around to setting them up properly yet.
An ABS print is going to be expensive (but necessary). Before you invest in that though, you still have to investigate ergonomics. You will need that certification in addition to structural and environmental.
One way of testing ergonomics is to make 'focused' prototypes from wood/ply/hardboard. I have attached a photo of an example. These will of course not look like the final product, but must accurately reflect the shape the user will feel. This is important but be careful who you show that to - they may get the wrong impression and think you've designed a WWI biplane :-) Use your imagination to get the job done - you could use polyurethane foam on an aluminium frame, for instance.
Get it signed off by a surgeon or chiropractor. If you speak to them nicely, they might do it for free in exchange for the publicity. (Interestingly, we are getting more requests for chairs for disabled people).
I don't see adjustable lumbar support anywhere. Personally, I find it irritating and prefer a plain back, but the buying public demands it.
Simulate, prototype and test often. that's the nice thing about SnS - you can test/remodel quickly. Get the basic model sorted out before you start adding fillets and small details. Consider what could go wrong - can the frame be taken apart to be repaired? That is a problem with plastic frames - screws cannot be removed and replaced without a loss in strength around the thread, which can result in problems like upholstery pulling loose. So use more fasteners than you would on metal.
This is going to be an expensive to tool up for (Dauphin chairs typically cost US$1-million to put into production). Test until you can sleep at night. Blowing the R&D budget is much cheaper than scrapping moulds!
I am very sorry for late replying. We are waiting for our prototype. :) Should see it here: http://rendyhimawan.com/post/13157411558/prototyping
We have experimented with ergonomics before by using steel frame for "comfort test". We do not include lumbar and pelvic support yet.
What concerns me a lot was about thickness issue. This chair has various thicknesses in many sections. I could not avoid it, since this was the approved design. :(
As seen in illustration, I put an eight millimeters metal bracket to connect the back with chair's mechanism.
We have limited budget to get this project done. A supplier in China will inject the plastic back frame for us. Unfortunately, it is difficult to find a reliable (and affordable) supplier in Indonesia.
We will only have one chance for mold process, so I need to get it all correctly. :)
The prototype looking good!
8mm steel for back - no problem - if you run into difficulty with BIFMA look at moving to something like SAE1010 alloy (used by the auto industry for the same reason).
The problem with people who approve things is that very often they're not technically inclined :-) I'd suggest talking about the wall thicknesses with your toolmaker in China. I'm going through a similar process now with an aluminium casting. I will mill out a model on CNC and fly out to discuss with the sand casters early next year. That will be an expensive mold, even in China.
I have attached a photo of one of our backs showing assembly stresses. All it took to cure this was to ensure the reinforcement pin didn't exceed 14.0mm in width when cut out on a guillotine. If a bit too tight, it gets hammered in. And it's never hammered straight.
Looks like you're doing a great job there... keep up the good work!
Thanks a lot. :)
I'll remind my colleague in China about this thickness issue. Most probably, we will test this chair as soon as it finished to confirm BIFMA standards.
Good luck with your project. :) I really look forward to read and see it. :)