Deflection of plate with hole

[civicmadness]

Hi, i would like to do an analysis of a deflection of a plate with hole using Ansys. I would like to minimize the deflection of the hole by adding stiffeners around the hole, but i am not sure how to go about it. I am not sure what to define as my design parameters, objective function and state variables. How should i start? I found a few articles on the internet on design optimisation of beams and they used the volume as an objectibe function. How do I obtain the max deflection at the hole? And how to minimize the deflection with the location of the stiffener? (for example, radial location of a circular stiffener)

 

[SWComposites]

How is the plate loaded?? Seems like the objective function is the deflection, which is defined where?? in what direction?? You need to be more specific re the problem that you are trying to solve. The design parameters are what you want to vary.

 

[reeves111]

you can use plate theory to calculate the deflection which will correlate very closely with anysys results, this I can tell you from experience.as far as using stiffeners, that would require an ansys analysis but try crunching some numbers before diving into a lengthy modeling process.

 

[civicmadness]

There are 3 types of loading:torsional, pressure at the central hole and symmetry loading at 4 corners of the plate. Since it's symmetric I work on only a quarter of the plate. The aim is to minimise the deflection at the central hole in the Z direction and optimise it with siffeners. The design variables should be the location of stiffeners(ie circular, rectangular and epllise stiffener placed at a distance around the central hole). The problem is to do optimisation, how do I use the *GET command to retrieve a value and store it as a scalar parameter? And is that parameter the deflection?

 

[flamby]

Based on my view of the system, I do not think circular stiffener around hole will be of any use. Because the whole plate is bending, the plate should be ribbed both ways, in x and y direction in a grid pattern. Actually, FEA is the lousy way to go about it as the whole problem boils down to design of a T section, easily done by hand.

 

[Qshake]

I also agree with Falmby's suggestion. Look to Roark's or basic mechanics of materials before embarking on a FEA of this element. Often times we're too quick to disregard perfectly good analysis techniques and jump right to numerical analysis.

Regards,
Qshake

Eng-Tips Forums:Real Solutions for Real Problems Really Quick.

 

[sdm919]

Although I’m not an ANSYS user, I do use optimization. Some definitions might be useful if you are new to optimization:

Objective function: The thing you want to minimize (or maximize). This is usually a single thing, and for structures is most often weight. If only a single material is involved, minimizing volume is the same as minimizing weight. For a beam of a single material and constant length, minimizing the cross-sectional area is the same as minimizing weight.

Design variables: The things you will allow to vary to meet the objective. There can be many of these and they are usually things like area, thickness, location, dimensions, etc. You should put upper and lower bounds on the values of these to keep them within realistic limits.

Constraints: The things you want to put limits on, such as stress, deflection, vibration frequency, buckling load, etc. You should be sure to include all the important criteria in your problem. If you only put limits on deflection, for example, the result may (or probably will) exceed some other criteria, such as stress, unless you put a constraint on stress as well.

Although I don't totally understand your problem, I do not think you should try to “minimize” the deflection. As the deflection goes to zero, the weight of the part will go to infinity! There must be some small value of deflection you are willing to allow, and that should be the “constraint” not the “objective”. Weight should be the objective, since that is a measure of how efficiently you are keeping the deflection down.

You should be careful about constraining the deflection of a single point. The optimizer might meet your requirement at that point, but exceed it elsewhere, so you may have to constrain several points. If you are talking about keeping the hole circular within a certain tolerance, then you are talking about the “relative deflection” of the points around the hole. Is buckling a possibility in your problem?

You should probably constrain some stresses as well, or at least check the stresses in the final result. If you have too many design variables and/or constraints, it will get bogged down numerically, so you must understand your problem well enough to know what’s important and what’s not. The “optimum” will be a “local” rather than a “global” optimum. It’s the closet optimum found near your starting point, so if you don’t like the results, you could try again starting with different initial values of the design variables.