Choosing kinematics software?

[EdDanzer]

We are doing more product design utilizing hydraulic cylinders to drive structures that must be light weight and robust. I have used Working Model and COSMOSMotion both, neither have a way to input acceleration rate, deceleration rate, maximum velocity and force into one element to simulate a simple linear actuator.
What is the best software to do this type of kinematics with so the pin forces and direction can be accurately be input in to the FEA model?

 

[gfbotha]

Greg, I did not look into this but if you could align a slot with the prime actuator, WM2D allows you to model friction via the script menu (by simply specifying the coefficient).

I guess specifying a friction FORCE (rotating with body = ON) will give the same problem you desribed above.

 

[GregLocock]

A damping element of type -c (as opposed to -c.v) looks like friction, but seems to apply just a constant force, it doesn't oppose the direction of motion. I'll have a dig around for a better friction model, it is a bit interesting to me.

Cheers

Greg Locock

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

 

[EdDanzer]

GregLocock,
The problem with adding mass for limiting the max acceleration is you change the acceleration forces, output forces at different positions, and inertia when stopping.

I talked with Hari Padmanabhan with Cosmos today about what I was trying to do and how to implement it. The short answer is I want to see the resultant actions of a solid model driven by an off shelf drive energy source. Examples of an energy source are a servo motor, electric motor, hydraulic motor, air motor, linear electric motor, air cylinder, electric cylinder and hydraulic cylinder. This would be like testing a car with different engine power to now how it would affect the frame or other components.
It appears that current Kinematics software does not model power available. You can assign part of the power components to a model but not all in the same study at the same time.

 

[GregLocock]

"It appears that current Kinematics software does not model power available."

Well, I don't really agree. For example, in my powertrain model (which plugs into my standard vehicle model) I'd have an engine torque vs rpm curve. Then my driver model applies a certain throttle %age. The solver looks up the current rpm, and delivers that % of that torque to the crankshaft. This then goes through the rest of the driveline model, via inertias, springs, damping, gaps and gears and whatever, to the contact patch.

That's overkill for most vehicle dynamics work, but it is there.

So the problem seems to be that the kinematics software bundled with Cosmos is less powerful than the standalone products. I don't think that is surprising - using ADAMS less than 2 days a week would be frustrating, and I normally edit the code by hand, in parallel with the GUI.

"The problem with adding mass for limiting the max acceleration is you change the acceleration forces, output forces at different positions, and inertia when stopping."

Sure. All of these models work better if you don't trick them, but model the actual process in the mechanism. So if your pump has an inertia, put it in. If the fluid in the lines has an inertia, put it in. If there is a pressure restrictor, or a one way valve, put it in. Replacing a 5 element block, as suggested above, by one transfer function is /possible/ but is hard to debug.

As another example my power steering model has a simple pressure vs engine rpm vs flowrate delivery map to represent the hydraulic pump side. That's not really good enough for what I want, so the next step is to put the hoses and so on into the model, as I am interested in elasticity and resonant effects. I could cheat and add some arbitrary first order systems in there to represent those resonances, but ultimately it is more robust to model the real system than abstract it down to a very complex equation.




Cheers

Greg Locock

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

 

[fkmeyers]

At the risk of upsetting Garland (Garland, I am apologizing in advance here) I would say just be careful before buying Algor. I will say no more on the subject. You have been advised.

I do not use motion software myself but have found the nonlinear transient analysis capability in Nastran to be adequate for my needs. I would recommend the below motion package that a colleague is using and swears by...

http://www.fedem.com

Check them out and lets us all know

 

[GBor]

Come on, Frank, surely you don't think I get upset just because you disagree with me? Actually, I didn't bring Algor up...I hope you noticed that

I'm not convinced that I understand the problem and Greg seems to have a working solution. I do know that Algor has a function coming out that may be appropriate, but this problem seems like it should boil down to a well-planned time-displacement curve to handle the velocities and accelerations. I apparently don't understand the force issue and really haven't taken the time to try and get it...I called Ed Simmons to make sure he understood.

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.
Lower Alabama SolidWorks Users Group

 

[EdDanzer]

Greg,
The problem with tricking the system is sometime you trick yourself into thinking you have done the tricks correctly. That is what I have been doing and probably increases the total design time by a factor of 2.
I need to do 1 or 2 studies per month maximum. If the ability to input values for motors and cylinders are not similar to what you read in the manufactures specifications then you have to translate them. Like translating languages or units of measure, the chance of making a mistake goes up especially if you don’t use those methods regularly.

Fkmeyers,
I sent contact information to Fedem but no response yet.

GBor,
I received a response from Algor late yesterday and I’m not sure they (or maybe it is me) how your input power affects driving a structure.

If you look at setting up a servo drive on a CNC machine control your motor and amplifier limit the acceleration rate and maximum velocity of the motor. The control will work for small machines or big machines depending on how you program the drive parameters. If we have a CNC machine we need to run a heavier than specified part on if we reduce the acceleration, deceleration and maximum velocities we can run the part without damaging the drive system.

 

[Storen]

Ed,
I am a Fedem Employee, and would like to come with some enlightenment, if possible.

From what I understand of your problem, Fedem is quite capable of doing what you want in some way or another. Fedem has a simple control system module that lets you control most input parameters for the different elements. (Springs, prescribed motions, Forces etc.) A common thing is to actually model the actuators in the model (like electric motors) using the control system, and apply the resulting force to the appropriate part. Often including some speed-measurement etc.

In addition Fedem is designed to work directly with FE-parts,
making it possible to get the structural results directly.

But I am not quite sure that I actually understand what you are looking after. When you say "off the shelf energy source" dou you mean you have a fixed set of them, and want to test new designs with all of them ? Or do you think in means of a general parameterized energy source ?

Anyway I am quite sure that you can do it in Fedem, but I do not understand enough to say how easy it will be to build the model.

Jacob Støren
Fedem Technology AS

 

[EdDanzer]

Storen,
We do need to use “off the shelf energy source” item. With hydraulic cylinders we can obtain some bore and rod size combinations at a lower cost so we try to use them first. We start a design by making a basic structure with pin locations to obtain the range of motion the customer requires, then we add the customers load to see if the performance requirements are met. It is best to do 4 to 6 “what if” studies with different geometries and a few different cylinder sizes to be comfortable meeting the customer’s needs, manufacturability and cost of manufacture. The pin load information is used for FEA of the structure.

 

[Storen]

Danzer,

As I understand, your original question had to do with a simple way of modelling hydraulic cylinders/linear actuator based on accelerationrates, max forxe and max velocity.

Fedem has no component that does that out of the box, but it can be built using the control system and some standard elements. Which means that the parametres you want to enter can be entered in the model (and subsequent models) directly.

I would like to check out how to do this in our software if you would help me with the information on how it is supposed to work.

The main thing I am uncertain on is what "driving" input you want on this actuator?

The geomerty/hardpoint positions is very fast and simple to change in Fedem. You do not need any Cad models while experimenting.

The pin loads can be exported to ASCII, DAC and RPC3, but the best thing would be to use your FE models directly in Fedem to skip the export/import loads step, I guess.

Jacob Støren
Fedem Technology AS

 

[EdDanzer]

The last project I worked on had to be modeled with 4 different standard hydraulic cylinders. The product is a roll out bucket for a front end loader. The movement is 95° and the bucket with material weighs about 17,000 lbs. The goal was actuation at 2000 psi into the cylinder (14,000-25,000 lbs. force) .5 G max acceleration, and 5 seconds to move the 95°.
I will try posting pictures later today.

 

[EdDanzer]

Roll out bucket images.

http://picasaweb.google.com/EdDanzer/EngTip/

 

[GregLocock]

That mechanism is easy to analyse in WM or ADAMS. The thing that makes it hard is your desire for a simple black box to replace the hydraulic cylinder.


It could certainly be done - by writing a user subroutine in ADAMS. For example, our electronic stability control software is a black box, that gets called by the solver during the run, which just gives it the necessary state variables, and gets the black box output back.

Perhaps it would be easier to write a user subroutine that would generate the appropriate lumped model for you - so when you clicked on the cylinder it would ask you for the five parameters (or however many it is), it would then create the aprropriate geometry connectors and parts, and then you'd run the model in the usual fashion. That's how I build varaible ratio steering racks, the manufacturer has supplied a subroutine that cranks out the model, given a few user specified values.







Cheers

Greg Locock

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

 

[EdDanzer]

Thanks for all the feed back I have contacted all the vendors posted. When I have more information I will let you know.