[Agent-X]

I've been researching the free, available Linux CAD programs out there, and none really strike my fancy to a professional level. FreeCAD (SourceForge.net: FreeCAD) looks like the best one at the moment, but it isn't fully developed. It seems to be in an alpha stage at the moment. I'm aware of the U.S. military program, but if I understand it right, it doesn't use parametrics.

The reason for creating this thread is because I'm curious about what mechanical/architectural engineers/designers think about the employment of parametrics within programs.

If I understand correctly, parametrics allow a person to use mathematical equations to create and compile geometric/algebraic objects.

Let's say a program does not allow a person to use parametrics. How can a person get around this? What is an alternative?

For what I read, those who are making custom design would be in favor of using parametrics, because this allows quick changes incase the workplace or materials change. Or does it have more to do with certain features changing when variables change? I don't fully understand all of it.

If I understand correctly, programs without parametrics are bound to a person using points within the 3D plane to make an object. Is that right?

So, would it be best to first plot something on paper using equations, then extract the exact points in 3D that would be used, and then plot those points for objects within a 3D CAD program?

[Oxygen]

At school we use 2D design, though that's for Windows quite unfortunatly.

[bidi]

There's a distro called CAELinux, or something like that, that is all about CAE. I can't remember what their website is, but it shouldn't be hard to find.

As for your other query, I teach a CAD for Engineers class (Engineering Graphics) which uses AutoCAD as the software to teach and I gotta say it's not fun (AutoCAD is based on points on the cartesian plane). Any software using parametrics to define features, such as ProE and SolidWorks, are 100% easier to work with. They are, however, harder to work with at first if you have no 2D/3D experience, and this is why Engineering Graphics is taught using AutoCAD, just so students can get their feet wet before going into more advanced CAD software.

From experience, there's really no way around the lack of feature on the software. If you make a mistake, or need to redo a certain part of your drawing in something AutoCAD it's usually easier to draw everything from scratch than try to fix it. Not to mention real 3D CAD software usually keeps a list of features in each design (lets say you create 3 cylinders, it'll show you a list with the 3 cylinders and their parameters), which isn't possible with other, half a**ed implementations (read AutoCAD).

With that in mind, comparing software like AutoCAD with software like SolidWorks and ProE is like comparing water with beer.

As you can imagine, I'm a little anti-AutoCAD, but that's only because it's so useless. Yet people insist on trying to make it work.

[Agent-X]

So, with that in mind, would you say it's better for a person to create a parametric model first? Afterwards, would you suggest that a person manipulate its mesh if other features were desired?

I am thinking of things in an engineering sense. I've seen things that require mesh modeling, and I think Linux has plenty of programs that allow a person to manipulate that kind of stuff. I haven't seen any that are fine-detailed, though.

One page I saw described a breathing device that must be made custom to the person using it. That was primarily made with a mesh. I suspect some concepts of physics were involved but not too many.

I looked into this, and supposedly a person could manipulate the variables inside of a program with a spreadsheet or programming language. Parametric AutoCAD drawings

I'm thinking this would control and adjust input and output variables. I figure that a Linux user would have to sort through all the code and possible learn a new language if things get too deep.

[bidi]

I don't really understand your latest questions!

I generally attribute a mesh with FEA software, which are separate from CAD software (even though some are packaged together).

The breathing device spreadsheet you describe seems like just a numberical way of doing the problem. The same can be done in ProE or SolidWorks and see what the object looks like. Sometimes this is considered a waste of power though, which is probably why they created a spreadsheet instead.

[Agent-X]

Originally Posted by bidi

I don't really understand your latest questions!

I generally attribute a mesh with FEA software, which are separate from CAD software (even though some are packaged together).

The breathing device spreadsheet you describe seems like just a numberical way of doing the problem. The same can be done in ProE or SolidWorks and see what the object looks like. Sometimes this is considered a waste of power though, which is probably why they created a spreadsheet instead.

The breathing apparatus is here: Matt Writes » Blog Archive » Is parametrics history?

It doesn't really relate the idea of changing variables with a spreadsheet.

Let's say I wanted to make *thinks of stuff* .. a gun. Why not.

Wouldn't it be a good idea to design it according to parametrics in case something with the design were to change, such as the barrel and the bullet or the clip? Afterwards, much of the design could be fancied with mesh modeling?

Perhaps I'm not too experienced with CAD/CAM/CAE software. However, I have in mind that most device would start with a basic, rigid design. This would allow a person to calculate its internal and external relationships, such as what volume of gun powder could fit into a space. Yet I think changing the internal design through mesh modeling would damper any rigid, ideal volume.

[bidi]

Originally Posted by Agent-X

Let's say I wanted to make *thinks of stuff* .. a gun. Why not.

Wouldn't it be a good idea to design it according to parametrics in case something with the design were to change, such as the barrel and the bullet or the clip? Afterwards, much of the design could be fancied with mesh modeling?

Perhaps I'm not too experienced with CAD/CAM/CAE software. However, I have in mind that most device would start with a basic, rigid design. This would allow a person to calculate its internal and external relationships, such as what volume of gun powder could fit into a space. Yet I think changing the internal design through mesh modeling would damper any rigid, ideal volume.

That's usually the case, you start with a 3D model, which isn't actually rigid it's just nothing, and then use the FE method to solve the mechanics problem. You only really need the mesh because that's how you solve FE problems.

One thing that you need to keep in mind is that the 3D model and the meshed model are two completely different things, in some softwares they're not even the same file on the hard drive. You'd generally create a mesh out of a 3D model, but you don't need the 3D model to create a mesh. Most basic designs are actually not even modeled in 3D since it would just be too much extra work.

The link you posted actually goes a little into this. This guy states, on the third paragraph, some applications that are made easier with parametric based software (which is what I'm used to since I'm a Mechanical Engineer).

He mentions Rhino on the article which, even though I have never used it but have heard of it, uses a mesh to create 3D shapes, as opposed to FEA software which uses a mesh to define nodes for the FE problem.

At this point, I think I might have confused myself (and maybe you) so can you please clarify a little what exactly are you looking for?

[Agent-X]

Well, what I'm looking for is a program where I change change variables to change a part of the whole, which in the end influences the whole.

For example, let's say I wanted to build this.
Come on, who doesn't?

And let's say I wanted to change its hands. Now, I figure that if I have some kind of constraints and mathematical control along with a way to test its mechanical abilities, I can make sure the fingers don't lock up or jam when they move. I suppose that's more of a biomechanics issue. What I would want is that if I increase the length of one finger, then all fingers would increase in length. Not only would that happen, but the dimensions of the palm would also increase in ratio. Also, the underlying hydraulics/mecahnics within the forearm would be mathematically dependent upon any changes made to the fingers and palm. If a finger increases, then all five fingers increase, and the palm increases, and the mechanics also increase.

Something like that.

I don't see that happening with AutoCAD. I'm sorry. With AutoCAD, I just see a program that allows a person to set coordinates and some dimensions. I guess people can use it for CNC, but in my mind, I assume I'd want to port anything made inside of it immediately to SolidWorks.

Heck, I don't think I'd want to build something like that inside of AutoCAD in the first place. It's the idea of mathematical control that appeals to me. I don't see AutoCAD giving a lot of control. And for that reason, I don't see a lot of the current Linux CAD programs giving a lot of control. Most of them don't seem to have a lot of mathematical control. They seem to be more of a mock of AutoCAD than solidworks.

But, I guess things like the spikey hair don't matter too much. Well, balance is an issue in biomechanics and so forth, but I don't think I would care to put any mathematical constraints on that. I've have to be super picky about what kind of environmental factors would play in. As a prototype, I wouldn't care. And it probably wouldn't be an immediate worry. So, for the hair, I could use a mesh modeler.

I'm thinking what matters first would be the underlying mechanics. Afterwards, a person could apply a shell/cover over whatever parts to hide what's inside. I don't see why FEA can't be used against solid objects such and cylinders made out of aluminum. Why exactly do I need to make a cylinder mesh?

Are you saying that people begin with solid objects, but FEA wants a mesh so things such as impact damage can be shown across the objects?

My view would be...

1. Sound mechanics through proper geometrical construction using mathematical variables and constraints.
2. Other objects can be added as mesh or solid objects. (take efficiency and time cost into consideration)
3. Turn into mesh and give physical properties to test the designs integrity.

[carlosponti]

for the cad i do i use qcad which is in the ubuntu repositories. there are several cad programs now available. for 3d stuff i am playing with google sketchup. probably not the kind of 3d models you would have used in other programs but it works for me.

[bidi]

Originally Posted by Agent-X

Well, what I'm looking for is a program where I change change variables to change a part of the whole, which in the end influences the whole.

For example, let's say I wanted to build this.
Come on, who doesn't?

And let's say I wanted to change its hands. Now, I figure that if I have some kind of constraints and mathematical control along with a way to test its mechanical abilities, I can make sure the fingers don't lock up or jam when they move. I suppose that's more of a biomechanics issue. What I would want is that if I increase the length of one finger, then all fingers would increase in length. Not only would that happen, but the dimensions of the palm would also increase in ratio. Also, the underlying hydraulics/mecahnics within the forearm would be mathematically dependent upon any changes made to the fingers and palm. If a finger increases, then all five fingers increase, and the palm increases, and the mechanics also increase.

Something like that.

Well, like I said before, i don't have any experience with software like Rhino, but I know SolidWorks or ProE would have no trouble doing this. Your biggest issue would be finding those equations.

I think you're thinking of that a drawing and mechanics of a part as the same thing, which they are not.

I don't see that happening with AutoCAD. I'm sorry. With AutoCAD, I just see a program that allows a person to set coordinates and some dimensions. I guess people can use it for CNC, but in my mind, I assume I'd want to port anything made inside of it immediately to SolidWorks.

Heck, I don't think I'd want to build something like that inside of AutoCAD in the first place. It's the idea of mathematical control that appeals to me. I don't see AutoCAD giving a lot of control. And for that reason, I don't see a lot of the current Linux CAD programs giving a lot of control. Most of them don't seem to have a lot of mathematical control. They seem to be more of a mock of AutoCAD than solidworks.

AutoCAD sucks for anything but 2D.

I'm thinking what matters first would be the underlying mechanics. Afterwards, a person could apply a shell/cover over whatever parts to hide what's inside. I don't see why FEA can't be used against solid objects such and cylinders made out of aluminum. Why exactly do I need to make a cylinder mesh?

I'm really confused here. FEA is used to analyze stress and strain (for the most part). For example, cylinders are generally modeled as beams, which are 1D parts, the only time a cylinder would be modeled as a shell is if you're modeling something like a tank.

Are you saying that people begin with solid objects, but FEA wants a mesh so things such as impact damage can be shown across the objects?

Actually, people usually start with hand sketches, just to give them an idea what they want. The bulk of the calculations are done on those sketches. You need to know what has potential to work before you divulge 3 days on a solid model.

FEA models are only used to prove or disprove your own simpler hand calculations. When you solid model something, you already have a pretty good idea of how it's going to react to impacts and what not, you just need confirmation.

When I did my senior design project, we started with hundreds of possible solutions to the problem, and slowly narrowed them done, before ever modeling anything. From those hundreds we did some very quick (hand) calculations to narrow it done to a few. At this point we redid all possible designs and did some more hand calculations. We ended up with 3 designs that would all solve the problem and cost roughly the same. At this point, we still had no computer models of anything. Generally, in real life, you'd start modeling the three designs and prototype them, but that wasn't an option for us since we had a limited budget. So we narrowed it down even further to 1 design, that we thought would be the most efficient. With 1 design, we modeled it and confirmed everything that our hand calculations had already told to us.

Anyway, what I wanted to point out with that little side bar is that when you do a computer model of anything, you already have a pretty good idea of what the answer will be. Nothing is done 100% in front of a computer, even though some people might think that's how designing is done.

I hope this is what you were looking for as I'm not exactly sure what it is.