LocostUSA.com

Sure hope so. I have put hundreds of hours into the modeling and learning these tools. Actually, I hope it's more then just me that builds these - it's been too much work for it to be worthwhile for just me. I hope this stuff is going to make other people's path easier. To build copies of these cars and to use the models to make their own cars...

I don't want to go into this in depth in this thread. We can talk about this in the "messy shop" thread, and it's slowly getting to be time for these cars to get their own threads, I suppose.

Next time someone brings up aluminum frames, we have some models to make more educated guesses with now.

_________________
Marcus Barrow - Car9 an open design community supported sports car for home builders!
SketchUp collection for LocostUSA: "Dream it, Build it, Drive it!"
Car9 Roadster information - models, drawings, resources etc.

Can you provide a pointer to that thread?



Aluminum frames? Wow, that's a place few go to these days. It's a different world all together. Like you say, it belongs in another thread.

Cheers,

_________________
Damn! That front slip angle is way too large and the Ackerman is just a muddle.

Build Log: viewtopic.php?f=35&t=5886

One of easiest ways is to click on "View active topics" at the top of the page. Scroll down to "my messy shop" thread. . I LOVE the "view active topics" function to keep track of active threads.

_________________
-Keith

Here it anyway. "Active Topics" is a great help. Things sneak past "New Posts" somehow...

My messy shop:
http://locostusa.com/forums/viewtopic.php?f=32&t=12980

_________________
Marcus Barrow - Car9 an open design community supported sports car for home builders!
SketchUp collection for LocostUSA: "Dream it, Build it, Drive it!"
Car9 Roadster information - models, drawings, resources etc.

I do not use GRape so can not do it myself, but may I suggest something..

why don't you try cross bracing the two leftover sections, the dash and the driver/passenger compartment... You will not be able to drive it like that.. but at least you will see how much stiffness you loose there...
Then, you can think about getting some of it back in a different way..

_________________
Building a single seater middy BEC with GSXR 1000 power
build log: http://locostusa.com/forums/viewtopic.php?f=36&t=5899

day job: http://smg.photobucket.com/albums/v723/turbolimac/portfolio/

I haven't really followed who is using GRAPE and who is using LISA, but is anybody using both programs yet such that the outputs of the two programs can be compared directly against each other when using the same structural designs and loading conditions?

_________________
-Justin

"Orville Wright did not have a pilots license." - Gordon MacKenzie

Hi Kiki. I have been doing this as I go along for the reasons you mention, find places where a perfect brace helps and then do what I can to make a practical brace. Here is a set of results for doing as you suggest. First, on it's own a crosswise brace under the dash does make a difference. Most of the difference goes away though when a brace from the firewall to the corner by the passenger is installed.

It looks like the car might be driveable with such a brace, you would need a short shift lever though, it would pass over that.

Here are the numbers, the lateral deflection is first and the vertical deflection second. I'll post in millemeters, just for practice and hope my countrymen don't track me down and coat me in tar and feathers...

5.76 4.70 model as posted above
4.62 4.01 model with dashboard diagonal
3.48 3.40 model with firewall thru passenger diagonal
3.81 3.56 model with reasonable firewall / passenger diagonal

You can see the flex is reduced by %25 using the "reasonable" solution. That wold be suitable for track use, but no passenger. In this case I run it from the driver side firewall to the midpoint of a crosswise bar under the dash and then to the passenger corner of the main hoop. It is also braced from the mid point of the dash bar to the dash hoop at the passenger floor, This makes a very large and well triangulated section.

The cost for the above change in weight is 7 pounds, 3.2 Kilograms.

The really fun and fascinating part is watching the colors change on the display when these changes are made. It is a considerable amount of additional inforation and ultimately leads to a deeper understanding of the model. Takes some effort though .

We had a very nice picture posted today of a Locost with side pods. I have avoided this because I didn't know how it would work ( look good I suppose I mean ). THe one time I tried this in the model it did not work as well as I hoped, but I think I did not stay with it as long as I should have, It should produce good stiffness numbers. I have been trying to get the stiffness thru tall sides as opposed to wide section sides... Basically just because the diagonal bars above the upper rail are required and I thought they should carry their weight, so to speak.

_________________
Marcus Barrow - Car9 an open design community supported sports car for home builders!
SketchUp collection for LocostUSA: "Dream it, Build it, Drive it!"
Car9 Roadster information - models, drawings, resources etc.

I should have something in a few days. I was also curious about how they compared.

_________________
-Andrew
Build Log
Youtube

On LISA v Grape ....

But is either one 'correct' and what is 'correct'?

A better approach might be a comparison of trends in stiffness or stress variation with geometry change: does either do anything that is non-intuitive, but even then, non-intuitive results might be modelling rather than software issues.

If 'validation' (whatever that really means) is the goal, it might be better to revisit the Aussie Mods thesis and rerun that geometry in LISA and Grape. At least the Aussie mods thesis provided physical test data, assuming of course that you can model the constraints.

_________________
Warren
Isuzu Pickup/SR20DE, +401 COLD frame
Build Log: viewtopic.php?f=35&t=11601

They're both going to be wrong to some extent. From doing analysis in ANSYS for FSAE chassises, the real world test results and the ANSYS predicted results were usually within 5% or so but there are other factors such as modified frame geometry during assembly, table attachment stiffness, etc that will alter the numbers. Seeing as ANSYS is one of the industry leaders, I don't think a free download will beat them.

It still really depends on the accuracy of the model and how the chassis is built. Space frames are pretty simple structures for a computer to analyze so a good model has a high likelihood of being accurate. Heck if someone wanted to create an Excel spreadsheet, it would be possible to analyze a chassis that way but creating the spreadsheet would be worse than water boarding.

I would be very interested in creating identical models in GRAPE/LISA and ANSYS and seeing how the results differ - that would be the real test.

_________________
-Andrew
Build Log
Youtube

It sounds like you may have read the Wesley Linton analysis from Cranfield too. There was physical data in that case (a real chassis), but it was used to bring the computer software results in line with the reality of the physical, full size example as tested on an accurate, instrumented test rig. It's reference here might just be cautionary in nature because few will ever have the means to test real examples in metal, although it had some excellent ideas, which I used in my own chassis. What will most likely come from LISA or GRAPE is just and indication of "better" or "worse" and the numbers generated can't be taken as absolutes, just indications of improvement or worsening.

Also, unless you do FEA all the time, it's probably easy to make subtle errors in modeling. That doesn't mean you shouldn't do it, just don't take a 3% improvement as true "gospel" and an absolute measurement, just a "maybe" IMHO.

The Linton Analysis (actually, his Master's thesis?) is available on the Internet and there is another Aussie study (separate from the well known Aussie Mods) done in FEA software available too, but that one did not have a corresponding physical model to test. I can either post references or upload them although both may be already be here on LocostUSA already. I haven't checked.

_________________
Damn! That front slip angle is way too large and the Ackerman is just a muddle.

Build Log: viewtopic.php?f=35&t=5886

Regarding some real numbers on the passenger area diagonal:
The bar doesn't interfere with the shift lever as it goes through to the top of the firewall cross bar (see pic). This bar then continues through the engine bay down to the lower frame rail at the front suspension plane. The measured torsional stiffness of this chassis is 3600 ft-lb/deg.

Plus the passenger is well protected in the aluminum box! You can barely hear the screams

I was thinking it wouldn't be too hard to actually measure that, how did you land up doing it? I was thinking if you used a dial gauge you wouldn't have to bend it very far and it would still be an accurate measurement.

Does your car respond well to anti-roll bar adjustments?

_________________
Marcus Barrow - Car9 an open design community supported sports car for home builders!
SketchUp collection for LocostUSA: "Dream it, Build it, Drive it!"
Car9 Roadster information - models, drawings, resources etc.

The measurement was done as follows:
The front of the frame was clamped to a blocked-up 2x3 steel beam running under the shock pickup plane. The rear was supported in the middle (the jacking point is a rounded-head 1/2" bolt directly below the IRS diff). A beam was attached across the top of the rear frame at the shock mounts with the beam extending to the right side such that a weight (80 lb engine block) could be hung 5.3' from the chassis ctr. A separate steel beam was attached to the back of the cage, extending to the left side. The deflection measurement was taken from this beam with a dial indicator. An average of 0.105" deflection was measured.
Yes, rear bar adjustments are effective.

You're a guy who checks his work. I am impressed.

One of the questions I've wondered about is how stiff is stiff enough? Keep in mind here that I'm asking and not telling. A deflection 0.105 does not seem like much to me, but I guess the question is relative to how much force is applied in a true dynamic case and I don't see that any of we Locost amateurs will be in a position to answer that kind of question without building some kind of hydraulic "shake table." So, we are going to have to guess and perhaps measure the static case.

I've seen figures for one of the French Le Mans teams where their sports racer tub of carbon fiber has a static stiffness of approximately 10,000 ft-lb (~its Imperial equivalent) per degree. That is hugely stiff. The Lotus Elise has pretty high numbers too (can't recall exactly) for its aluminum extrusion/bonded tub. I've actually driven one and it's fun to drive and not at all uncomfortable even on bumpy pavement, so they have their stiffness/spring thing all figured out.

But, what's a good enough target for a street Locost? 3,000 lb-ft/degree? 4,000? 5,000? More? It's really hard to know. I do suspect there is a number above which no practical benefit is delivered for any extra stiffness added. With a race car you would expect to need a higher number than for street applications. Anyone know a real chassis engineer we can ask?

Cheers,

_________________
Damn! That front slip angle is way too large and the Ackerman is just a muddle.

Build Log: viewtopic.php?f=35&t=5886