Chassis design - Modelling and FEA
Discussion
I've been wondering about this recently and started looking into what's involved with designing a chassis, modelling it in some sort of 3D CAD tool to check everything (i.e. the driver) fits, etc. Then using an FEA tool to refine the design for best stiffness / weight compromise, strength of suspension pickup points, etc.
So, of I go searching the web for FEA tools or 3D CAD tools for mech. eng. and boy do I feel out of my depth.
Does anyone have any recommendations on how to read up on this stuff? Are there any other subjects that would make things a little clearer if I were to read up on them first?
So, of I go searching the web for FEA tools or 3D CAD tools for mech. eng. and boy do I feel out of my depth.
Does anyone have any recommendations on how to read up on this stuff? Are there any other subjects that would make things a little clearer if I were to read up on them first?
For chassis design, the "bible" is a book by Mike Costin published in the late '60s. It covers all aspects of spaceframe design that you will need to know, and don't worry about the age of it. All there is to know about spaceframes was tried and tested on Formula cars in that era and nothing remarkable has been found since. Aside from that, spaceframes have been done to death over the years and it's generally known who has built good ones and who hasn't, so a little internet research should also give pointers and cautions about what works and why.
Having done a complete CAD design and packaging exercise on a new kit design, I can tell you that there is a hell of a lot of work involved if the results are to be of any great use to you. You WILL need a CAD package that can perform dynamic analyses in order to visualise suspension and steering movements or there is a high chance you will come unstuck somewhere! Without dynamic analysis on your CAD package I think it is honestly better spending more time in the workshop with the parts in hand, and less in front of the PC - no personal slight intended, this is from hard-won experience!
A lot of the work is in modelling the mechanical parts you intend to use, and the more accurate these models are, the closer the chassis and bodywork can be, resulting in no wasted spaces, minimal package size, etc. But there is the law of diminishing return, and you have to decide how compact and efficient you want the package to be and therefore how accurate your models need to be.
The FEA analysis is useful to see where you are compared to other idealised designs but for real value from the analysis, you must take care to replicate the real-world realities of the chassis construction on your CAD model or you will get an optimum result not a real world figure you can replicate in the workshop. For example, where a large diameter rollbar meets a smaller diameter chassis member, there is usually a stagger between the centrelines of the tubes, and it's easy to ignore this on the model. The resultant torsional forces are not likely to reult in failure but the inaccuracy of the load case makes the model irrelevant...
I would suggest modelling the mechanical layout fairly simply at first, and accomodating the cockpit area. I built a wooden buck to make sure the cockpit size & shape and control positions were correct before going any further. The chassis is there to link all the mechanical components together and preferably offer some impact protection, so from your mechanical layout start the chassis design in principle, making reference to racing and kit car best practices, and I would recommend using the FEA (if you have it) to fine-tune the chassis. Examples like: which orientation to run diagonal bracing in; whether to use central longitudnal bracing down the length of the chassis; what size of tube and what wall thickness for critical areas (suspension mounts, engine mounts, seat and harness points) and what size roll bar to fit, if any. FEA won't help that much in locating main chassis members - they go where they have to go - but it can help in showing what other factors affect the stiffness of the structure.
Don't let me put you off - designing your own car is incredibly rewarding (even more so is actually driving it!) but I would caution against too much CAD and FEA work if you are not (a) familiar with these systems because it takes time to get proficient at them, and (b) confident that you can get useful information and data out.
But if you feel confident about your conceptual design abilities and can pick up the CAD skills quickly, there's a lot to be gained from using the PC as an aid.
Having done a complete CAD design and packaging exercise on a new kit design, I can tell you that there is a hell of a lot of work involved if the results are to be of any great use to you. You WILL need a CAD package that can perform dynamic analyses in order to visualise suspension and steering movements or there is a high chance you will come unstuck somewhere! Without dynamic analysis on your CAD package I think it is honestly better spending more time in the workshop with the parts in hand, and less in front of the PC - no personal slight intended, this is from hard-won experience!
A lot of the work is in modelling the mechanical parts you intend to use, and the more accurate these models are, the closer the chassis and bodywork can be, resulting in no wasted spaces, minimal package size, etc. But there is the law of diminishing return, and you have to decide how compact and efficient you want the package to be and therefore how accurate your models need to be.
The FEA analysis is useful to see where you are compared to other idealised designs but for real value from the analysis, you must take care to replicate the real-world realities of the chassis construction on your CAD model or you will get an optimum result not a real world figure you can replicate in the workshop. For example, where a large diameter rollbar meets a smaller diameter chassis member, there is usually a stagger between the centrelines of the tubes, and it's easy to ignore this on the model. The resultant torsional forces are not likely to reult in failure but the inaccuracy of the load case makes the model irrelevant...
I would suggest modelling the mechanical layout fairly simply at first, and accomodating the cockpit area. I built a wooden buck to make sure the cockpit size & shape and control positions were correct before going any further. The chassis is there to link all the mechanical components together and preferably offer some impact protection, so from your mechanical layout start the chassis design in principle, making reference to racing and kit car best practices, and I would recommend using the FEA (if you have it) to fine-tune the chassis. Examples like: which orientation to run diagonal bracing in; whether to use central longitudnal bracing down the length of the chassis; what size of tube and what wall thickness for critical areas (suspension mounts, engine mounts, seat and harness points) and what size roll bar to fit, if any. FEA won't help that much in locating main chassis members - they go where they have to go - but it can help in showing what other factors affect the stiffness of the structure.
Don't let me put you off - designing your own car is incredibly rewarding (even more so is actually driving it!) but I would caution against too much CAD and FEA work if you are not (a) familiar with these systems because it takes time to get proficient at them, and (b) confident that you can get useful information and data out.
But if you feel confident about your conceptual design abilities and can pick up the CAD skills quickly, there's a lot to be gained from using the PC as an aid.
Hmmm - tricky one for me to answer.
I can't really comment about mechancial aptitude or knowledge because I've been into cars since age 3 and up to my armpits in them since age 12! I can't visualise being anything other than an automotive engineer and designer (and mechanic in my spare time).
It is a huge job to design any car from scratch, but it's also possible to make life a hell of a lot easier by taking other peoples' best ideas and adding your own flavours to the mix. See most Seven clones as a case in point!
Re the SVA test, if you can design a car you can design it to pass SVA - the test is there to weed out the horror stories and those kits that were designed with the corners already cut. Designing with the SVA manual at hand should result in no fears about road registration.
Anyone who wanted to design their own car but felt unable due to aptitude or experience could do no better than build a couple of kit cars, with increasing complexity in each build. There is nothing better for seeing what works well and what could be improved than trying out someone else's design for yourself.
Once you have got a couple of builds under your belt I bet you would have a stronger idea (a) what ideas you have to improve on existing designs, (b) roughly what constitutes a sound design, (c) how to achieve an SVA pass, and (d) whether you really, really wanted to go down the route of doing your own design. If it took you a year to build a reputable kit, for example, count on 3-5 years to do your own car. The additional labour input is vast.
Lastly, the bodywork on a kit is horrendously difficult and time-consuming to do on your own, and even more horrendously expensive to get someone else to do for you. It can be fudged to some extent with ally panelling and so on, but on anything other than a Seven loads of flat panels look cheap and nasty - beause they are. Bodywork is the main thing that stops designers from realising more of their own designs, and ensures there's loads of replicas in the kit industry!
For bodywork, you either need a large budget (think around £15k for a pattern maker to do an average set of bucks and moulds) or a lot of bodywork skill and patience. For preference, you'd want a kind relative with lots of bodywork skill and patience!
All in all, no it ain't easy no matter who you are, because aside from all else it's mentally and physically demanding to do both the design and construction jobs. You certainly don't have to be an engineer or a mechanic to do it well, but a logical and patient approach is essential. Last point on this - one of the UK's most respected clubman racecar and suspension designers is Allan Staniforth, whose books are excellent. He has no formal mechanical training but has produced some great and innovative designs simply by considering the problem from a layman's perspective. There must be something to be said for that...
I can't really comment about mechancial aptitude or knowledge because I've been into cars since age 3 and up to my armpits in them since age 12! I can't visualise being anything other than an automotive engineer and designer (and mechanic in my spare time).
It is a huge job to design any car from scratch, but it's also possible to make life a hell of a lot easier by taking other peoples' best ideas and adding your own flavours to the mix. See most Seven clones as a case in point!
Re the SVA test, if you can design a car you can design it to pass SVA - the test is there to weed out the horror stories and those kits that were designed with the corners already cut. Designing with the SVA manual at hand should result in no fears about road registration.
Anyone who wanted to design their own car but felt unable due to aptitude or experience could do no better than build a couple of kit cars, with increasing complexity in each build. There is nothing better for seeing what works well and what could be improved than trying out someone else's design for yourself.
Once you have got a couple of builds under your belt I bet you would have a stronger idea (a) what ideas you have to improve on existing designs, (b) roughly what constitutes a sound design, (c) how to achieve an SVA pass, and (d) whether you really, really wanted to go down the route of doing your own design. If it took you a year to build a reputable kit, for example, count on 3-5 years to do your own car. The additional labour input is vast.
Lastly, the bodywork on a kit is horrendously difficult and time-consuming to do on your own, and even more horrendously expensive to get someone else to do for you. It can be fudged to some extent with ally panelling and so on, but on anything other than a Seven loads of flat panels look cheap and nasty - beause they are. Bodywork is the main thing that stops designers from realising more of their own designs, and ensures there's loads of replicas in the kit industry!
For bodywork, you either need a large budget (think around £15k for a pattern maker to do an average set of bucks and moulds) or a lot of bodywork skill and patience. For preference, you'd want a kind relative with lots of bodywork skill and patience!
All in all, no it ain't easy no matter who you are, because aside from all else it's mentally and physically demanding to do both the design and construction jobs. You certainly don't have to be an engineer or a mechanic to do it well, but a logical and patient approach is essential. Last point on this - one of the UK's most respected clubman racecar and suspension designers is Allan Staniforth, whose books are excellent. He has no formal mechanical training but has produced some great and innovative designs simply by considering the problem from a layman's perspective. There must be something to be said for that...
I've done it.
I do FEA for a living and looked at several kit car types and several basic designs of my own to find out what worked and how it worked.
If you've not done it before and you decide to have a go I'd suggest that you compare the results of a Lowcost chassis with the results I got. If your model agrees roughly with mine then you know that you can do a simple spaceframe and can therefore take on something of your own design.
Have a scroll through my posts on the Lowcostbuilders.co.uk forum or on the Yahoo Lowcost theory forum. I've posted a lot of thoughts and results on this subject so it might be usefull for you.
Don't get stuck in the rut of assuming that existing designs must be good or that spaceframes are always better than ladder frames. I did and it was a while before I got the message. Most kit car spaceframes are actually less stiff for their weight than well designed ladder frames. Only a few are actually stiffer for their weight.
You could do worse than copy the Lotus 23 or Caterham 7 chassis. Take a look at the Project LMP chassis. This build is comming along nicely and it looks like a stiff design. A very interesting site is Ultimav12.ca. Lots of info on how to stiffen up the Ultima chassis.
Good luck with your plans. What kind of car are thinking of making ?
I do FEA for a living and looked at several kit car types and several basic designs of my own to find out what worked and how it worked.
If you've not done it before and you decide to have a go I'd suggest that you compare the results of a Lowcost chassis with the results I got. If your model agrees roughly with mine then you know that you can do a simple spaceframe and can therefore take on something of your own design.
Have a scroll through my posts on the Lowcostbuilders.co.uk forum or on the Yahoo Lowcost theory forum. I've posted a lot of thoughts and results on this subject so it might be usefull for you.
Don't get stuck in the rut of assuming that existing designs must be good or that spaceframes are always better than ladder frames. I did and it was a while before I got the message. Most kit car spaceframes are actually less stiff for their weight than well designed ladder frames. Only a few are actually stiffer for their weight.
You could do worse than copy the Lotus 23 or Caterham 7 chassis. Take a look at the Project LMP chassis. This build is comming along nicely and it looks like a stiff design. A very interesting site is Ultimav12.ca. Lots of info on how to stiffen up the Ultima chassis.
Good luck with your plans. What kind of car are thinking of making ?
Hi Jay,
I don't think we have talked before, but I chat with Hugh quite often, and Steve now and then.
Basically, I agree with all your points...especially when there comes the time that it is better to get away from the screen and start building...
I think it's worth adding that although useable soild modellers don't cost too much, a decent FEA package does...or am I missing out on some bargains somewhere?
I have Mechanical desktop...know any cheap FEA that would work well with it? (yeah, I know it has it's own built in...but it doesn't seem much use)
Cheers,
Alan
www.desicodesign.com/meerkat/
I don't think we have talked before, but I chat with Hugh quite often, and Steve now and then.
Basically, I agree with all your points...especially when there comes the time that it is better to get away from the screen and start building...
I think it's worth adding that although useable soild modellers don't cost too much, a decent FEA package does...or am I missing out on some bargains somewhere?
I have Mechanical desktop...know any cheap FEA that would work well with it? (yeah, I know it has it's own built in...but it doesn't seem much use)
Cheers,
Alan
www.desicodesign.com/meerkat/
Alan,
We haven't spoken directly I think, when I was with Coram I spent more than all my time concentrating on designing and developing the car... been following your car though and have to say that it's looking impressive these days!
I haven't got access to an FEA package (for that matter I'm struggling to get proper access to a decent solid modeller) so if I hear of a cheap package I'll be in like flynn... to be honest you may have gathered from the above that although I appreciate the subtleties of a good FEA analysis, at the level most of us can aspire to I think they're best left for optimisation of a design when it's close to being finished, as an educational aid, and for comparison to other idealised designs. I can't honestly say that when designing the LMP I had a stiffness target in my head and then had to update the design to achieve that target; what we got was very much what we got
We haven't spoken directly I think, when I was with Coram I spent more than all my time concentrating on designing and developing the car... been following your car though and have to say that it's looking impressive these days!
I haven't got access to an FEA package (for that matter I'm struggling to get proper access to a decent solid modeller) so if I hear of a cheap package I'll be in like flynn... to be honest you may have gathered from the above that although I appreciate the subtleties of a good FEA analysis, at the level most of us can aspire to I think they're best left for optimisation of a design when it's close to being finished, as an educational aid, and for comparison to other idealised designs. I can't honestly say that when designing the LMP I had a stiffness target in my head and then had to update the design to achieve that target; what we got was very much what we got
Thanks for all the advise guys.
I've currently got a load of books on order with the library to read through to pick up a few ideas.
My plan is to build a few kits, (see my thread on part built kits) in which I hope to get a better idea of the different designs and see if I'm cut out for this lark.
I work in IT, have A Levels in Maths, Further Maths and Physics so hope to be able to pick up the software and theory where necessary so long as I can find the right teaching materials.
As for the plan for the car, I'm thinking very long term. The sort of idea I have in the back of my head at the moment is a composite spaceframe, perhaps carbon/aluminium honeycomb, all round wishbones, in-board push rod suspension, mid engined.
Not sure about powerplant but maybe something different - perhaps give one of these new style motorcycle V8s a go (made from two motorcycle inline fours on a common crank).
So, as you can see I'm not planning on doing anything much yet, hence my idea of trying things out in software so that I can just keep on refining my ideas while I'm building a few kits for experience.
Hey, at this stage I'm not even sure whether it'll be a track only race car, or more refined and road going. I did have one mad idea of designing it to be one platform with a shorter wheelbase race car, longer wheel base more refined road car.
Even if I get nowhere near commiting any prepreg to mould, I plan on enjoying learning the theory and trying out a few ideas electronically.
I've currently got a load of books on order with the library to read through to pick up a few ideas.
My plan is to build a few kits, (see my thread on part built kits) in which I hope to get a better idea of the different designs and see if I'm cut out for this lark.
I work in IT, have A Levels in Maths, Further Maths and Physics so hope to be able to pick up the software and theory where necessary so long as I can find the right teaching materials.
As for the plan for the car, I'm thinking very long term. The sort of idea I have in the back of my head at the moment is a composite spaceframe, perhaps carbon/aluminium honeycomb, all round wishbones, in-board push rod suspension, mid engined.
Not sure about powerplant but maybe something different - perhaps give one of these new style motorcycle V8s a go (made from two motorcycle inline fours on a common crank).
So, as you can see I'm not planning on doing anything much yet, hence my idea of trying things out in software so that I can just keep on refining my ideas while I'm building a few kits for experience.
Hey, at this stage I'm not even sure whether it'll be a track only race car, or more refined and road going. I did have one mad idea of designing it to be one platform with a shorter wheelbase race car, longer wheel base more refined road car.
Even if I get nowhere near commiting any prepreg to mould, I plan on enjoying learning the theory and trying out a few ideas electronically.
That's the one - I haven't got a copy either, and the only copy I've ever seen looked like it was a first edition
Dunno if you can try your local library (in theory they should try to get any book you ask for) or your local technical college. By this time I can't imagine there will be another print run done - shame...
Dunno if you can try your local library (in theory they should try to get any book you ask for) or your local technical college. By this time I can't imagine there will be another print run done - shame...
It seems to me that the biggest challenge is in defining the suspension layout and geometry, since this has a huge impact on the car's handling. The rest isn't by any means trivial but surely is largely a matter of packaging, strength/stiffness and crash protection.
Designing a suspension system is a mammoth job in its own right, and one that is virtually impossible for mere mortals to get right first time. Absolutely the best start you can get to help get this right is Warren Rowley's book on race car engineering, and Bill Mitchel's chassis simulation software.
Designing a suspension system is a mammoth job in its own right, and one that is virtually impossible for mere mortals to get right first time. Absolutely the best start you can get to help get this right is Warren Rowley's book on race car engineering, and Bill Mitchel's chassis simulation software.
LexSport said:
For chassis design, the "bible" is a book by Mike Costin published in the late '60s.
Would this be "Racing and Sportscar Chassis Design", written by Mike Costin of COStin/DuckWORTH?
If so, anyone know where the hell you can get it? It appears to be out of print...
I would have recommended Abebooks but the only copy they have is priced at $120
Maybe a specialist technical bookstore might have a copy?I know its a bit last century, but cars were being designed pre 3D modelling. If you have the visualisation and draughting skills, then 2D drawings can be a better design forum for your ideas - it is much simpler to get ideas down in the first place in 2D.
Many of even the large scale projects (not necesarily cars) around when 3D modellers came into their own were full of signatures due to the limitations of the software, or the patience of the designer attempting to drive it.
We design bespoke large scale machinery, and still lay out concepts in 2D for speed and the ability to cheat to get the ideas across.
On the FEA front, if you are only interested in the space frame, suspension members etc, then you only need an FEA package that models beam elements (all of the top end products use solid and shell (plate) elements, on top of a whole host of other fancy gismos). Futhermore, for a space frame model you will get much better results using beam elements, rather than a solid model of the actual space frame. Back to the point, packages such as STAD (which are aimed at structural rather than mechanical engineers) are much cheaper than the high end stuff.
A final thought - I can run ANSYS and CATIA from the home laptop that I am typing on now - accessing the office licenses through VPN. This is £50k+ worth of software, plus horrendous monthly license fees, so I wouldn't suggest that you go out and buy it for a hobby project (unless you are bored with the fleet of dancing donkey's inhabiting your garage
). However, if you have a mate with a similar setup (I guess it is quite common), then that may make your decisions for you on which software to use.
Many of even the large scale projects (not necesarily cars) around when 3D modellers came into their own were full of signatures due to the limitations of the software, or the patience of the designer attempting to drive it.
We design bespoke large scale machinery, and still lay out concepts in 2D for speed and the ability to cheat to get the ideas across.
On the FEA front, if you are only interested in the space frame, suspension members etc, then you only need an FEA package that models beam elements (all of the top end products use solid and shell (plate) elements, on top of a whole host of other fancy gismos). Futhermore, for a space frame model you will get much better results using beam elements, rather than a solid model of the actual space frame. Back to the point, packages such as STAD (which are aimed at structural rather than mechanical engineers) are much cheaper than the high end stuff.
A final thought - I can run ANSYS and CATIA from the home laptop that I am typing on now - accessing the office licenses through VPN. This is £50k+ worth of software, plus horrendous monthly license fees, so I wouldn't suggest that you go out and buy it for a hobby project (unless you are bored with the fleet of dancing donkey's inhabiting your garage
). However, if you have a mate with a similar setup (I guess it is quite common), then that may make your decisions for you on which software to use.Making a 5th scale chassis model out of little balsa sticks glued together is a brilliant practical tool for assessing a design. If it's a reasonable spaceframe, the scale size of the sticks doesn't really matter too much as long as the joints are all in the right places. You don't need to profile the joints - the balsa glue is thick enough to fudge quite big gaps. Grab it by its suspension pickup points at each end and twist it - you'll soon see whaich members are doing most of the work and which aren't!
If you are looking for a free modelling tool for optimizing spaceframe chassis you could try GRAPE. Race Tech magazine (www.racetechmag.com/) had an article on using this in conjunction with a custom Excel program to optimize chassis. You can download the software from the race tech site.
For my project I used the old faithful balsa model approach. This is very good for analyzing stiffness. However, it doesn't really help optimizing the size and thickness of each individual tube. When I get time I plan on producing a GRAPE model of my chassis. Although the speed things are going this will probably be when I look at a MKII version of my car i.e. a long time away! If you are interested in my model you can check it out at www.projectlmp.com
For my project I used the old faithful balsa model approach. This is very good for analyzing stiffness. However, it doesn't really help optimizing the size and thickness of each individual tube. When I get time I plan on producing a GRAPE model of my chassis. Although the speed things are going this will probably be when I look at a MKII version of my car i.e. a long time away! If you are interested in my model you can check it out at www.projectlmp.com
T.K.E said:
jay w said:
For chassis design, the "bible" is a book by Mike Costin published in the late '60s.
Would this be "Racing and Sportscar Chassis Design", written by Mike Costin of COStin/DuckWORTH?
If so, anyone know where the hell you can get it? It appears to be out of print...
I would have recommended Abebooks but the only copy they have is priced at $120
I am also desperately searching for this book. I have been looking for weeks. The only copy I managed to find was sold at the shop the day I ordered it online (what are the chances?). If anyone has one that they'd like to part with or knows of one for sale for a reasonable sum i.e. not $120, I would be very grateful if they could get in touch.
Cheers
Avocet said:
Can't help I'm afraid. I don't have a copy but YES it is the same "Costin" that was half of Cosworth. It's a damned good book too. Have you tried Ebay?
I had a good look on eBay to no avail but it matters not because I managed to find a copy in New Zealand that has just arrived today
Gassing Station | Kit Cars | Top of Page | What's New | My Stuff