Down pipe design

[CowsWithGuns]

Hi all,

I'm working on installing a 6boost highmount and an EFR7064. Having my first crack at stainless and more intricate exhaust work. I have no idea what I'm doing so 'the best way' is guesswork at the moment.

I mocked up a bunch of cuts into a new dump pipe. My aims are to keep the exhaust close to the manifold, and as far from the firewall as possible (might put AC back in the future, also keeps distance from heater hoses and ABS if that ever gets returned). I also wanted to minimize total bends for whatever fractional flow gains I can feel good about. Also, less bends = less surface area to radiate heat into the bay.

When looking online most fabricators seemed to take a different approach: pictures I've seen so far show a straight from the turbo, a ~90 down tucked close to the firewall, and then further tucking it along the chassis line (note the ~45 + 45 in reference, instead of another 90)

I've drawn on my approach versus reference here:

 

I'm starting to doubt my approach (perhaps theres an issue I will run into later following further assembly), and would like to hear other thoughts on what the 'ideal' dump pipe should be. At the moment I think it would look cleaner if it was following the chassis line... but is there any other reason to run a straight from the turbo then ~90 down?

 

[GTSBoy]

The ideal dump pipe is closer to the reference and further from what you were wanting to do. The important thing to understand is that it is nowhere near possible to get a "good" dump pipe into almost any car. So it just becomes a matter of how bad you are willing to accept it being.

The ideal is for the turbine outlet to be circular (ie, no integrated wastegates here) and for the dump to expand at a nice shallow (say, 7-11°) cone up to the very largest pipe you can get. For a ~2" turbine outlet, a 4" big end is nice. And that should be straight, obviously, because it's very difficult to make the sort of conical expansion being described here also be a bend. And having a bend in the middle of the conical expansion is likely to interfere with the way that the cone should work anyway. And then any bends after the end of the conical expansion should be as small angle and/or long radius as possible, so a pair of 45s is defo better than a single 90, but there's only dust and grains of difference between those anyway.

But if you do the maths/sketching, you will see that such a cone takes up a huge amount of length. So unless you want to punch through the firewall and drop the dump between the passenger's knees....you have to do something else.

So the "best" compromise is usually somewhere between what you want to do and the ideal, and that is why a lot of aftermarket dumps go straight toward the firewall and then straight down. They are prioritising the exit development length over the rest of it. Well, so long as they actually implement some sort of conical expansion in there.

This is why a good turbo installation should also consider whether to angle the turbo upward a little so that the exhaust points down. Makes all this a little easier to fit - at the cost of perhaps making the inlet side less pleasant to plumb.

[TurboTapin]

I run a 6boost twin gate manifold with a 4" downpipe and retained AC. I went with 2x 45's as well but it hugs the firewall a lot more then what you have pictured. I'm roughly 10mm from the rubber air conditioning drain but with heat wrap on the downpipe, it's fine. 

[CowsWithGuns]

GTSBoy, your engineering knowledge is impressive, thank you for explaining this. To be honest, I didn't consider going for a larger pipe for the dump, as I would be necking down to 3" afterwards (perhaps just before the cat).

I have some (many - sorry) followup questions

• Is a larger dump pipe than the rest of the exhaust desirable purely because the gasses are hotter upstream and more pressurised?
• What are the specific negative consequences of using an aggressive expansion cone like found 'off the shelf' (e.g. 30-45 degrees)? I assume it's still much better than having no expansion at all even if it promotes turbulence
• If one is using no expansion (3" turbine outlet to 3" pipe), is there still a benefit to having a straight from the outlet followed by bends (rather than bending immediately out the outlet)?
• Realistically, is it actually worth spending another few hundred and remaking what I have with 3.5 or 4" piping and an aggressive cone? This is the smallest turbine on the B2 frame so my gut tells me it doesn't really matter... still interesting though!

I have an urge to learn a CFD software and have a play with various setups so I could have a quantified reference of changes... on which note, GTSBoy is there any CFD software that's beginner friendly you would recommend? All I played with at uni was 1D modelling (hysys) and you can stick it for cars (and my work).

TurboTapin, that's some good intel. If you could fit two gates and a 4" without burning everything, then I must be putting too much weight to my heat concerns.

Edited December 27, 2023 by CowsWithGuns

[GTSBoy]

53 minutes ago, CowsWithGuns said:

Is a larger dump pipe than the rest of the exhaust desirable purely because the gasses are hotter upstream and more pressurised?

Yeah. Most people don't worry about it, and the difference may not be big enough for most people to put any effort into it. But the simplest rule for turbines is that they work on pressure difference. The more volume you have downstream to expand into, the more the gas can and will expand, and that is good for both response and total power. If you have to neck back down to a smaller system after the main length of the dump, it will still be "better" than only expanding up to the final size of the exhaust. It is really really hard to quantify what that difference is though, and it is probably different for all possible combinations. Might be insignificant on smaller turbos, for example.

56 minutes ago, CowsWithGuns said:

What are the specific negative consequences of using an aggressive expansion cone like found 'off the shelf' (e.g. 30-45 degrees)? I assume it's still much better than having no expansion at all even if it promotes turbulence

Gases don't like to expand into steeply increasing cone angles. You get separation from the wall, causing recirculations back along the wall (ie, the turbulence of which you speak, but I prefer not to refer to it as turbulence, preferring to describe the actual flow phenomena being observed, because all automotive flows are in the turbulent regime). These recirculations waste energy and reduce the effective cone angle anyway.

Having said all that, a steep cone angle is better than either starting at the final size (ie, 2" outlet dropping straight into a 3" pipe with a big 1/2" step all around) or never growing beyond the outlet size at all (ie 2" outlet to 2" pipe all the way down the dump). You need the extra volume more than you need to avoid "turbulence".

Some will argue, and they are probably somewhat correct, that the flow exiting the turbine is a rotating flow and so it will fling out to occupy a wider cone angle than the "ideal" small angles near 10° anyway. And....that's probably at least part of the reason that you can get away with reasonably poor exit arrangements. Good old Corky Bell's book Maximum Boost showed the test results (from about a million years ago) where they definitely got better results with nicer expansion cones than without. It's worth putting in whatever effort can be put in.

1 hour ago, CowsWithGuns said:

If one is using no expansion (3" turbine outlet to 3" pipe), is there still a benefit to having a straight from the outlet followed by bends (rather than bending immediately out the outlet)?

Yeah, but probably only a little benefit. But this is likely to be a bad design simply because a 3" turbine outlet is really really really going to want at least a 4" dump, preferably even bigger.

1 hour ago, CowsWithGuns said:

Realistically, is it actually worth spending another few hundred and remaking what I have with 3.5 or 4" piping and an aggressive cone? This is the smallest turbine on the B2 frame so my gut tells me it doesn't really matter... still interesting though!

Yup. 4". 3.5 would be a waste of time cf a 3". OK, maybe slightly better. But of you're going to the effort to fab an expansion, it might as well be for something worthwhile.

1 hour ago, CowsWithGuns said:

is there any CFD software that's beginner friendly you would recommend? All I played with at uni was 1D modelling (hysys) and you can stick it for cars (and my work).

Nah. I don't rate any CFD software unless it is driven by a PhD expert in modelling. We use Ansys Fluent at work, which is right at the very apex of the CFD heap, and it is a pile of shit that needs to be massaged and cajoled and occasionally beaten to get it to work. It breaks with nearly ever version release. Things that used to work stop working, etc etc.

Having said that, we push Fluent to the very edge of its capabilities in many areas (much combustion, radiation heat transfer, multi-phase flows, etc). Our new machine is 128 cores with umpteen gazzillion gigs of RAM, costs $50k or so and chews a kilowatt all day every day. And our sims take days to converge, and that's just for steady state modelling. We don't even consider doing time-varying simulations. They add about 2 orders of magnitude to the solution time (ie, we could be at a single one for months!).

I say all that because I suspect that to model the outflow of a turbine you'd probably want to be doing a non-steady state sim, because the rotating turbine is a dynamic element in the mesh. But the fact that it spins at >100k rpm is something confounding to me. You'd have to do ridiculously small time steps to capture anything out of it. I'm probably teh wrong person to be thinking about this though, because I am not the PhD CFD engineer. I'm the practical hands on aerodynamicist/process eng who has to talk to the client on the one side and the modelling geeks on the other.

You can play with "free" CFD using on-line stuff. The stuff that SuperFastMatt uses for his car aero is probably as easy as it gets. The problem with CFD is that it is easy to set up a sim, run it, get results and believe them. but there are so many caveats and traps that the Colourful Flashy Diagrams are often just that. For example: The usual turbulence model people use is k-epsilon. That works for many situations, but it doesn't handle swirl at all well. It damps a swirling flow out where it would persist in the real world. You have to use k-omega for swirling flows. But when you do that, you then take on other compromises. And then you can use more sophisticated turbulence models to try to get around these compromises and then you end up using RANS or something that will take 10-100x as long to converge. And on it goes. There is so much skill involved in grid design choices, grid/mesh refinement, knowing how to set boundary conditions in the absence of good data, etc etc etc.

[CowsWithGuns]

OK - so be it - time for another spot of shopping to get me some 3"-4" reducers, 90s and all the other sources of joy. Hopefully my bandsaw can handle such a girthy chonk of a pipe... what better way to get peace of mind than to make it BIG!

Between your post and a writeup I found from a Garrett engineer on this topic (effectively echoing all your points) the nature of the beast is quite clear - much appreciated!

Regarding the CFD, I have heard my equivalent PhD expert at work discuss it and it is rife with places to stumble, fumble, and generally produce colourful stinky garbage. I will not use it to try learn ideal exhaust design 😛

[MBS206]

Chiming in regard CFD, especially dynamic CFD.

The really really good guys at CFD, typically have 15+ years of hands on experience in actually building the things they're trying to model.

They therefore have the real world experience to immediately call bullshit on results, and understand all the fun stuff GTS has mentioned about which areas you can simplify with specific assumptions.

 

I have a few things I'd love to model in CFD. I have enough knowledge to know I'd likely do a great job at generating rubbish simulations 

[CowsWithGuns]

On 12/27/2023 at 2:59 PM, GTSBoy said:

Good old Corky Bell's book Maximum Boost showed the test results (from about a million years ago) where they definitely got better results with nicer expansion cones than without. It's worth putting in whatever effort can be put in.

Ended up reading the whole book. Excellent read, I have learned many things and can already see mistakes I've done on the current build... I think you must have had a different book in mind however, as there was no test data comparing cone angles - no biggie though!

Since it's on topic to the thread though, here are Bell's recommendations for intercooler piping reducers:

 

Anyway, just want to ask if you could recommend some other books that offer good car molestation advice please

Edit/continuation:

I'm reading the last chapter and noticed this bizarre statement:

Also, the exhaust size guidelines imply a 3" pipe is suitable for 700+ bhp. Perhaps adequate but far from optimal I'm sure.

Since I started this thread and reading more, I've found plenty of dyno evidence that a larger dump to turbine outlet can gives substantial improvements, e.g. https://www.motortrend.com/how-to/modp-1102-exhaust-test-tech/

I guess it just goes to show even elementary guidelines from a well regarded textbook may not be accurate. I find this very... exhausting.

Edited January 1 by CowsWithGuns

[GTSBoy]

6 hours ago, CowsWithGuns said:

I think you must have had a different book in mind however

You're no doubt right. I haven't read Bell's book in >20 years. Coincidentally, the Garrett engineer you mentioned earlier was probably the guy behind the Lott0004 (or something very like it) username on Performance Forums about 20 years ago, and he definitely had things to say about turbine exducer angles, so...I probably conflated them in my mind.

And that would perhaps explain Bell's comment about keeping the dump pipe the same size as turbine exit all the way to the main exhaust pipe. It was written nearly 20 years before the Lott0004 files were compiled on PF.

The exhaust size guidelines were probably still influenced by the old NA world of trying to "keep velocity up" and the relative lack of availability of bigger exhaust pipe and mufflers for car applications at the time. We have since learnt that you can barely have enough exhaust size for a turbo. More is almost always better (output concerns prioritised over noise, etc). It's about getting the most volume to expand the gas into while still getting it out the back end of the car (otherwise we'd dump it out the bonnet/guard).

I don't have too many book reccos. I've only got some old Vizard stuff and the bible on modifying ALFAs.