Exhaust System CAD Modeling

[BenB]

I was wondering what everyone's CAD modeling strategy is for exhaust systems (if you model them at all)

I defined points in space for my exhaust header and my collector and then make a 3D sketch (solidworks) to try and make a pipe route that will work.

I seem to get a lot of errors with my method and its basically guess and check to get the pipe lengths correct. Its proving to be a pretty inefficient way to do this.

Is this pretty much what everybody does or is there a better way to do this?

[BenB]

I was wondering what everyone's CAD modeling strategy is for exhaust systems (if you model them at all)

I defined points in space for my exhaust header and my collector and then make a 3D sketch (solidworks) to try and make a pipe route that will work.

I seem to get a lot of errors with my method and its basically guess and check to get the pipe lengths correct. Its proving to be a pretty inefficient way to do this.

Is this pretty much what everybody does or is there a better way to do this?

[drivetrainUW-Platt]

I have not done it..but from a manufacturing standpoint, it would be easiest to make them as an assembly of bends, straights, reducers etc so that way you can actually build it. Pretty hard to take a soldworks 3d sketch with splines and curves and make a real part out of it.

[jeff-NIUMS]

I use Solidworks to model our exhaust system. Using 3D sketches and sweeps it's obviously not ideal but it does work. Too make sure that we can manufacture it I make all bend radii the same as bends on the market.

[ad]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by jeff-NIUMS:
I use Solidworks to model our exhaust system. Using 3D sketches and sweeps it's obviously not ideal but it does work. Too make sure that we can manufacture it I make all bend radii the same as bends on the market. </div></BLOCKQUOTE>

there are heaps of ways of doing it, but the 3d sketch is as good as any, keep the bend radii, OD and angles to standards, unless you can get any bend angle or unlimited funds for new mandrels.

also you could use the weldment feature which would also give you the end profiles of tubes.

[BenB]

Has anybody ever used a CNC tube bender for exhaust? It seems like that MIGHT solve the manufacturing problems associated with 3D sketches, but as complex as an exhaust pipe can get I wouldn't be surprised if you start running into tool interference problems.

If a CNC bender would work it might be cheaper to bend up a single length of pipe how you want it instead of trying to weld multiple pieces together. Any thoughts? I have pretty limited experience with CNC benders.

[ad]

Yes we have used CNC mandrel bending in the past, however it was limited to single bends angle only. i.e. tolerancing and spring back.

Did not really help with the sketching process, only added a bit of flexibility to the angle of bends that we could use.

[chavez]

I've designed a couple exhaust systems using Solidworks. The method I use involves making every section have a bend and a straight section. This does 2 things, keeps the number of welds down and allows easy manipulation of the sections. I also created a jig to ease manufacturing, that the U-bend fits in and has a protractor mounted to it. Once you determine the length of the straight section and the bend angle you measure it off of the protractor and you have the section.

[Nilsen]

There are a few things to keep in mind for modeling.

The first thing you are going to have to do is obviously define the engine points. With your chassis in the model, play around with different shapes for your desired primary length. Then you will want to use either a swept or weldment feature to check interferences. Accounting for the size of your collector, rinse and repeat with either your secondaries or tailpipe.

If you are using a vendor who professionally does tube bending, the main thing they will care about is the tangent lines to your bends. If you are doing the bends yourself, the protractor method is probably about as good as you can do without highly specialized equipment. A fairly good rule of thumb is to estimate that 4 diameters is the tightest bend which you can make.

[Discretely elite]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by chavez:
I've designed a couple exhaust systems using Solidworks. The method I use involves making every section have a bend and a straight section. This does 2 things, keeps the number of welds down and allows easy manipulation of the sections. I also created a jig to ease manufacturing, that the U-bend fits in and has a protractor mounted to it. Once you determine the length of the straight section and the bend angle you measure it off of the protractor and you have the section. </div></BLOCKQUOTE>
I have done the exact same thing for this years turbo exhaust manifold and it turned out very easy to assemble as a result. Also the jig we made ensured the key points in SolidWorks (head ports and turbo bolt holes) are preserved.

[BenB]

Great! Thanks for the advice. Modeling as a straight and a bend should be a lot more robust way to model this and it looks like thats how Burns SST likes to sell exhaust sections.

[Philly]

Hi all,

I am at the CAD modelling stage also now. I wanted to know what is the best way to make sure all primary and secondaries are of equal length(4-2-1 layout).
Obviously each pipe is going to be angles and routed differently. Is there a way of say drawing 4 equal straight lines from the exhaust ports, and then bending each at specific points. This will ensure that the lengths are always equal regardless of any manipulations....?????
Then use the sweep feature over these lines.
Is this the best way?
Any help?

Cheers,
Phil

[Tintin]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by Philly:
... Is there a way of say drawing 4 equal straight lines from the exhaust ports, and then bending each at specific points. This will ensure that the lengths are always equal regardless of any manipulations....?????
Then use the sweep feature over these lines.
Is this the best way?
</div></BLOCKQUOTE>

I'd be tempted to say a plain no, that doesn't sound right regardless of the CAD package. I'm only used to Catia V5 but there it would be relatively easy to produce centre lines of equal lengths made from straights and radii by means of parameters if the pipes would later be build by welding pre-formed pieces which is probably the most common method in FSAE. You'd specify a common radius for the bends, build a skeleton and parameterise all the straights and angles with one of them depending on the total length you aim for and the sum of the rest. If you really get deep into the subject you could probably also do an automatic packaging if the parameterisation is good.


Tim

[Nishant Jain]

With the bends and straights specified on CAD alright, how is it that everyone tends to import it on software like Ricardo? Is it ok to go for simple angle deviations as asked for in the duct panel? Or is it more complexity required for greater accuracy?

[Discretely elite]

I've never worked with Ricardo (GT-Power instead) but you should be fine plugging in the correct CLR, angle, wall thickness, and/or length. Also use the correct material in the model.

[oz_olly]

I am also very interested in peoples strategy for achieving equal length primaries using a 4-2-1 system. I used CATIA and it was worse than a rubix cube. I set planes at the exhaust port which I could rotate about a line perpendicular exhaust port face and then proceded to make sketches using known bend radius and varius angular segments. In previous years we used to just leave space between the engine and the roll hoop, take the car to a handy bloke at a local exhaust shop and say 'can you fill the gap between the engine and the muffler'. I haven't ever actually measured one of those pipes. Ron Tauranac reckons using ping pong balls is a quick and easy way to measure a pipe for comparison purposes.

I could not get the four primaries to line up in the same place so I will have to use the secondaries to adjust the positioning so they fit nicely into the final collector.

Cheers

Olly

ACME Racing
UNSW@ADFA

[DonMolina]

Using SW is pretty easy to do a full exhaust design (at least, it seems easy after the 5th iteration or so http://fsae.com/groupee_common/emoticons/icon_smile.gif )

Personally I take the chassis-engine assy to have the free space layed out in front of me. Then I start with sweeps going to the place I want the primaries/secondaries to meet. Depending on the iteration I had each time it was a square of different dimensions. Getting the lengths equal is a pain in the ass, but with SW you can define the length of the center-lines, after fixing the start/end points. That would give you a first idea about the form of the lines. Then delete the dimension defining the sketch and start playing with the control points of the splines to get it manufacturable.

If your protractor (or your lab/workshop/whatever) can't manufacture your design (pretty possible) you can also model the angle pieces used by most (if not all) protractors, the ones that are welded together to make the exhaust. Then assemble them in a jig fashion and try to squeeze everything right.


Basically, there is not real difficulty in terms of ideas. It's just lots of houes in front of a screen experimenting with a zillion positions of a trillion control points or a billion of puzzle pieces...
So just keep on trying with your 3D software and you'll get there http://fsae.com/groupee_common/emoticons/icon_smile.gif

[Nishant Jain]

Getting the same primary lengths for all 4 headers in a 4-2-1 is beginning to look impossible for us. Ours is a side-emerging exhaust system, and like most of you people we're knocking on the doors of local mechanics and getting them to come bend it for us.

I'll put it all onto our CATIA model to fit it in. With extra length, it can be made sure that it at least doesn't anti-resonate and spoil everything else. Right? In the case that you dont get your exact length.

I'm thinking of having one set of 2 primaries of one length meeting at a Y junction. Then another set of another length. To make it not too odd torque-wise, I may make one twice the other.

[cjanota]

Usually you connect the primaries of sister cylinders together.

[Nishant Jain]

Damn and I wanted to keep the engine only intensely masculine, what with reciprocative motion at 10000rpm and so much lube! http://fsae.com/groupee_common/emoticons/icon_smile.gif

But yeah you mean 1-4 and 2-3 for a 1-2-4-3 firing order. Correct?

[Wesley]

I've done both - modeled it in CAD, and duplicated it by sight, and just built a header that fit as I went.

The one by CAD was easier at build time, just because I had a reference for where to fit everything (plus it was a 2-cyl) whereas with the other one, I just built them as I went. I ended up within 1/8th inch on both of them for length.

I usually just cut one section at a time, tack weld, and modify from there. Takes about 2-4 hours to tack it all together usually.

[kapps]

That's how we did it as well. Big nastly 3D sketch in SW with only defined exhaust ports and collector location. We designed it with manufacturing in mind so all the bends could be taken off pre-purchased 180 degree bend sections. We made a jig to cut the angles precisely and were able to get all the angles off SW. It was a bit of a pain and time consuming but not too bad. It was pretty cool how everything fit together when it was all done, especially considering how many sections/welds we had to do.

[Math ze engineer]

we have a side exit exhaust and I was able to make all my runners equal length with a 4 to 1 config. I had my four outputs of engine in ma cad as well as the place where my collectore will be. Then I routed 3 pipes easily and noticed fourth was impossible to route so I tried many stuff to make it considering that I wanted a very compact header for engine to be closer to the firewall.

One thing I would do for a new header is to make cylindrical extrusions where the flange bolts. By avoiding tubes to pass through these cylinders you are sure that your bolts/nuts, box and ratchet will make their way to the flange (I have a tube bending right at the exhaust port and putting the nut is a pain in the ass)

Another thing I would do is a monoflange juste like V8 engines so your tubes won't spring back all the way around when you remove your header. another good thing about my design is that I can get engine in and out of the frame WITH header mounted on it (and this is not a multipiece bolted frame)

Hope that helps
Good luck

[Mikey Antonakakis]

Good thing about 8-inch exhaust runners is it's easy to take them out. It was a pain to get them equal length though. I used Pro|E to model it up beforehand. There weren't too many options for how to configure them and get them equal length, so they ended up protruding out from the engine pretty far. I hope to fix that this year, maybe with some custom bends. I just used 2-d sketches, made sure I used bend radii I could buy. It was a big pain, but I ended up getting them within a few thousandths of an inch in length. And boyyyyy did they sound good with no muffler.

[kapps]

It would be a good project to see the difference between equal length runners and runners that simply flow well. The ideal is probably a combination of the two. You can make equal length runners all you like but if you have 180 degree bends in them for packaging, the flow is going to be so impacted, it's probably not worth it. We had one runner this year that required a major bend. A professor here that has done a lot of work on engines took one look at it and said we're loosing power because of it. He had no problem that we were equal length within a 1/4 or 1/2 inch. Equal length effects a fairly small part of the revband but pipes that flow well should affect all of it.

[Gaanja]

This is how we did it:

We use a side exhaust config


The chassis with the engine and drive train is modelled first on solidworks. Then the collector and muffler are modelled and positioned favourably wrt the assembly (We r using a 4 to 1 collector) . 4 separate planes are created using the axes of each set of cylinder and pipe of the collector. For example Plane 1 is created with axis of port 1 and the axis of pipe 1 in the collector. On each plane a guide curve is drawn using standard bend configurations connecting the centres of each port and the respective pipe. The tubes are created using the sweep option. A similar method is used to connect the collector and muffler.

I got it right in about 6 iterations using 3 bends in each primary. the difference in primary lengths was about 10 mm at the most

[RenM]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by kapps:
You can make equal length runners all you like but if you have 180 degree bends in them for packaging, the flow is going to be so impacted, it's probably not worth it. </div></BLOCKQUOTE>

That very much depends on your backpressure level. If you have a backpressure level of like 200mbars , lowering the backpressure will have almost no effect. if its however like 300 mbars lowering the backpressure can have a positive effect.