Developing a Proper Suspension Model

[cdrazic93]

@RBbugBITme, by trade your a fluid dynamics engineer correct? A little OT but what level (400 series, post grad?) did you get into your fluid dynamics?

I'm assuming @Shankenstein, you've also gotten your hands wet with CFD or atleast FD.

(My physics puns aren't that good, sorry)

[RBbugBITme]

Just a simple BS in Mechanical Engineering with a focus in Thermo and fluids.

[pseudo]

Correct me if I'm wrong, but fluid dynamics from a theory side of things is pretty simple, right? It just breaks down to solving the Navier Stokes equations.

Everything else is just using software to model the forces.

[fika84]

Originally Posted by pseudo

Correct me if I'm wrong, but fluid dynamics from a theory side of things is pretty simple, right? It just breaks down to solving the Navier Stokes equations.

Everything else is just using software to model the forces.

Simple compared to what? Multi-body dynamics is difficult enough and fluid dynamics is even more difficult IMO with so many partial derivatives and long convoluted equations (also have a BS in Mechanical Engineering). We were always taught to NEVER trust software if you can't do a "simple" hand calculation to back it up. Software makes things easier/quicker to not only get answers, but to get WRONG answers if you messed up.

[pseudo]

Originally Posted by fika84

Simple compared to what? Multi-body dynamics is difficult enough and fluid dynamics is even more difficult IMO with so many partial derivatives and long convoluted equations (also have a BS in Mechanical Engineering). We were always taught to NEVER trust software if you can't do a "simple" hand calculation to back it up. Software makes things easier/quicker to not only get answers, but to get WRONG answers if you messed up.

I didn't mean anything offensive by it, all that I meant was that from a theory perspective, you have an initial state, a set of governing equations (navier stokes), and all you do from that point on is solve them. I can certainly appreciate the practical complexities of that last step, and I didn't mean to trivialize it in any way.

As an applied mathematician I can sympathize with having a certain level of distrust for numerical methods, especially when applied to chaotic non-linear equations. But I am surprised to hear that as an engineer you actually solve your own PDEs.

If you have done any work with applied FD, how do you go about solving problems? Do you actually try to simplify navier stokes and solve them in closed form? I assumed it was more like, throw rk4 at it and call it a day. Or even, throw 'software' at it and call it a day.

[fika84]

Originally Posted by pseudo

I didn't mean anything offensive by it, all that I meant was that from a theory perspective, you have an initial state, a set of governing equations (navier stokes), and all you do from that point on is solve them. I can certainly appreciate the practical complexities of that last step, and I didn't mean to trivialize it in any way.

As an applied mathematician I can sympathize with having a certain level of distrust for numerical methods, especially when applied to chaotic non-linear equations. But I am surprised to hear that as an engineer you actually solve your own PDEs.

If you have done any work with applied FD, how do you go about solving problems? Do you actually try to simplify navier stokes and solve them in closed form? I assumed it was more like, throw rk4 at it and call it a day. Or even, throw 'software' at it and call it a day.

Sorry if I came off as harsh. I didn't mean it to be that way (Engineers and social skills.... ha!).

I personally only took one FD class and didn't like it much or my teacher (probably more teacher). My favorite part of the class talked about making miniature prototypes of objects and using the power of FD to adjust the fluids used in the tests. But in essence I think you're right about the Navier Stokes equations and getting them right. I just remember it getting confusing FAST with multiple fluids and difficult geometries for the fluids to travel through.

When I speak of having to solve things by hand I'm referring to when I had to do it for Finite Element Analysis. Our teacher would made us do matrices that took 4-6 pages taped together by hand before he would let us use Solidworks. Of course you can simplify FEA problems by reducing the number of nodes to do a "back of the napkin" calculation and make sure the software was setup properly. It's very easy to over fix geometry and get false results in FEA.

I also can't speak for all Engineers, this is just what I was taught and live by.

[pseudo]

Originally Posted by fika84

Sorry if I came off as harsh. I didn't mean it to be that way (Engineers and social skills.... ha!).

It's all good, mathematicians aren't exactly known for social skills either!

Originally Posted by fika84

I personally only took one FD class and didn't like it much or my teacher (probably more teacher). My favorite part of the class talked about making miniature prototypes of objects and using the power of FD to adjust the fluids used in the tests. But in essence I think you're right about the Navier Stokes equations and getting them right. I just remember it getting confusing FAST with multiple fluids and difficult geometries for the fluids to travel through.

Miniaturization of prototypes sounds pretty cool, I know that's a big part of Formula 1 engineering. You basically just scale the reynolds number of a fluid to compensate for the smaller size, right?

Originally Posted by fika84

When I speak of having to solve things by hand I'm referring to when I had to do it for Finite Element Analysis. Our teacher would made us do matrices that took 4-6 pages taped together by hand before he would let us use Solidworks. Of course you can simplify FEA problems by reducing the number of nodes to do a "back of the napkin" calculation and make sure the software was setup properly. It's very easy to over fix geometry and get false results in FEA.

I had the exact same experience working with stupidly large matrices by hand in one of my applied linear algebra classes. Those teachers are plain sadistic.

I'm curious how geometries come into play when doing FEA by hand. I don't have much(hic. any) experience with engineering. What do the matrices represent? Do you basically make a matrix that represents cartesian point approximations of the properties of a fluid, and then evolve the matrix based on neighboring cells? (if that makes any sense)

The advantage and disadvantage of being a mathematician is that everything makes sense at a theoretical level, but I have no practical understanding of how things are done in the field

[fika84]

Originally Posted by pseudo

It's all good, mathematicians aren't exactly known for social skills either!

Miniaturization of prototypes sounds pretty cool, I know that's a big part of Formula 1 engineering. You basically just scale the reynolds number of a fluid to compensate for the smaller size, right?

Pretty much that was the gist of it. That and changing the fluid from air to water so that you could imitate a faster speed of air. You can get equal reactions with different density fluids at different speeds.

Originally Posted by pseudo

I had the exact same experience working with stupidly large matrices by hand in one of my applied linear algebra classes. Those teachers are plain sadistic.

I'm curious how geometries come into play when doing FEA by hand. I don't have much(hic. any) experience with engineering. What do the matrices represent? Do you basically make a matrix that represents cartesian point approximations of the properties of a fluid, and then evolve the matrix based on neighboring cells? (if that makes any sense)

First, each node of your model represents a point where you want moment and force information. You can put these nodes wherever you want on the object, but usually they make the most sense to be at joints and edges and fixed points for starters, then more to increase accuracy in other areas. Since each one of these nodes represents 6-axis motion you start compiling 6x6 matrices per node with some things added and some things subtracting depending on orientation.. it gets large really quickly.. absolutely sadistic!

Originally Posted by pseudo

The advantage and disadvantage of being a mathematician is that everything makes sense at a theoretical level, but I have no practical understanding of how things are done in the field

This is one of the reasons I went into Engineering. Get to see some of both.

[cdrazic93]

Almost sorry i asked lol.



Almost. I wonder if vector calculus has any kind of benifit to FD.

[Shankenstein]

Originally Posted by cdrazic93

@RBbugBITme, by trade your a fluid dynamics engineer correct? A little OT but what level (400 series, post grad?) did you get into your fluid dynamics?

I'm assuming @Shankenstein, you've also gotten your hands wet with CFD or atleast FD.

(My physics puns aren't that good, sorry)

Officially, I'm a Mechanical Engineer but my studies have wandered.

As others have said, FEA and CFD are both useful... but only as much as you can guarantee the results.

If you have governing equations that are valid throughout the space, and a mesh that is well-defined around the objects of interest... then you can set the boundary and initial conditions to realistic. Spin the top and see what happens!

Your basic courses in fluid dynamics will teach you laws of conservation that are the base of most CFD. As the situation becomes more extreme, you'll need more flexible/complicated governing equations and boundary conditions. Typically, the extreme conditions are what you care about... which means you have to get "down and dirty" to get a meaningful solution.

Air/fluid flow, temperature (heat transfer), internal stresses, composite layer interactions, magnetic/electric field interactions, plastic/metal mold flow, fatigue/vibration stresses, etc. Each one is a different set of governing equations, boundary conditions, and initial conditions. Much of the analysis techniques are similar, but it's a different medium/model/environment.

[cdrazic93]

I will get into magnetic and electric fields next quarter, so simple fluids in disguise!

[plucas]

Originally Posted by Shankenstein

As others have said, FEA and CFD are both useful... but only as much as you can guarantee the results.

If you have governing equations that are valid throughout the space, and a mesh that is well-defined around the objects of interest... then you can set the boundary and initial conditions to realistic. Spin the top and see what happens!

Your basic courses in fluid dynamics will teach you laws of conservation that are the base of most CFD. As the situation becomes more extreme, you'll need more flexible/complicated governing equations and boundary conditions. Typically, the extreme conditions are what you care about... which means you have to get "down and dirty" to get a meaningful solution.

Air/fluid flow, temperature (heat transfer), internal stresses, composite layer interactions, magnetic/electric field interactions, plastic/metal mold flow, fatigue/vibration stresses, etc. Each one is a different set of governing equations, boundary conditions, and initial conditions. Much of the analysis techniques are similar, but it's a different medium/model/environment.

This is so very correct. Anybody who follows what I do on here know I am a huge proponent and user of cfd and fea. However, I am a proponent of using it to get meaningful solutions. I hate seeing young engineers or people in general who think that it is just pressing some buttons and waiting for it to spit out a result and look at colorful pictures. It takes a lot of work and knowledge to set up computational problems correctly with the right governing equations, boundary conditions, and initial conditions. It is also amazing that in many cases, you can get meaningful results with just some hand calculations (or my favorite is combining that with excel). You need to know the theory as well as the application.

[cdrazic93]

This just popped into my head (either its a super simple answer or complicated) if you match the perfect spring rate to the dampening force, you achive the critical damping ratio (~0.65). When you add adjustability, wouldnt that change the critical ratio? Or is the ratio more of a range thats acceptable given on road circumstances, i.e. higher ratio would be for track and a lower ratio would be on a street car.

[fika84]

Originally Posted by cdrazic93

This just popped into my head (either its a super simple answer or complicated) if you match the perfect spring rate to the dampening force, you achive the critical damping ratio (~0.65). When you add adjustability, wouldnt that change the critical ratio? Or is the ratio more of a range thats acceptable given on road circumstances, i.e. higher ratio would be for track and a lower ratio would be on a street car.

If only it were all that simple... here's a good read for ya. http://www.kaztechnologies.com/wp-co...m-Kasprzak.pdf. Mainly page 14 and on.. but all of it is a good read.