Wood, 8020 extrusion sandwich modeling

[laviniu]

Hi,

I'm a total amateur in both finite element modeling and structural analysis so please bear with me. I'm trying to model a sandwich table structure composed of 8020 Inc aluminum extrusions between two medium density fiberboard(MDF) sheets, 0.75" thick.

I'm building a CNC router for hobby purposes and I need to calculate the deflection of the table once I put a heavy loading representing the machining stock on it. Ideally I need to be under 0.01" for the heaviest load I need to machine. I've already modeled the structure by using beams and thick shells that share common nodes and I have a non acceptable deflection, like 0.0189" so I thought about adding another layer of MDF underneath the aluminum beams which are 3" thick. This souds logical because of the increased moment of inertia the thickeness of the beams gives to the MDF sheeets.

I have an issue with adding the second layer of thick shell elements. I don't know exactly how to model those. Considering everything symmetrical, the nodes from the the bottom MDF sheet should somehow be "in sync" with the correspondig nodes of the upper sheet in the sandwich. I'm using EMRC NISA software because it's cheap and I don't know exactly what construction should I use. I don't like (or I don't completely understand) the rigid links or coupled displacements. What I miss is some sort of contact elements. But no such thing with NISA.

What I was thinking about is actually creating small beam elements that have very small cross section area and superstiff low density hypotetical material to "link" the nodes from the upper and lower sheets and to account for the thickness of the beam. Would that be a reasonable aprroximation or something stupid? You see I have a couple of limitations here: non-standard cross section beam and no contact elements. How would you guys model this in Nastran?

I'm throwing this question but I don't know if someone will answer it. So be it. I have the gut feeling the sandwich will take care of the deflection but I don't know how to prove it, plus the roughing doesn't need to be that accurate and it will remove a very big part of the stock so the loading decreases significantly and the finishing pass will be more accurate.

Thanks,
Laviniu

 

[GregLocock]

First things first, if the MDF is glued to the previous MDF sheet then it isn't a contact problem.

Have you considered just incraesing the offset of the lower plane by .375, and doubling its thickness?


Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[laviniu]

My mistake. I wanted to say that I want to model the sandwich structure not that I've already modeled it. Initially I had just a simple table, one sheet of MDF with aluminum beams underneath connected through screws. That was easy. They share common nodes and I made the beams centroid's offset from the nodes.

Adding the lower layer posed some issues. I don't know how to add it correctly. That was my question. I did it two ways and they both ended up with almost the same results (within 4%). One way was to consider the beam as two halves and then each half beam will share nodes with one of the MDF sheets and then link the corresponding lower and upper nodes by rigid links. I've verified the results on a simple beam case and it concurs in deflection with 8020's deflection calculator (less than 2%). That made me trust the model. The other way was to consider the beam as a whole and share nodes with the upper MDF sheet and then link these upper nodes with the lower MDF corresponding nodes with rigid links. I didn't have a way to verify this though. But it ended up in almost the same results as the 2 half beams method. I guess that kind of solved my problem although I don't know for sure this is 100% correct. Also I noticed a decrease in deflection (as expected but not by much): 0.014" instead of 0.0189". I guess I should just go with an aluminum table if I want that much precision. But it's expensive.

Thanks for help and the tip with contact elements. So it's about glueing...I don't think NISA has such a thing.

Laviniu

 

[GregLocock]

Glueing is the same as welding - that is, you can treat it as ideal joint. So, one way of meodelling it would be to model the beam, then at each node on the beam put in a rigd element out to the mid section of the MDF, and then join all those nodes with MDF.

does that make sense?

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[laviniu]

Thanks for the tip. It makes complete sense. I'm going to try it.

Thanks,
Laviniu