JDL Header & Exhaust System Development

[JDLAutoDesign]

Add Lees info the the first post

[Tansey86]

So there is no way to use the EL header with the up and coming turbo kit?

[FR-S Matt]

This is the entire reason why I'm not jumping on Borla's UEL header. JDL is doing some serious collector work and when you compare both collectors, I think the dyno results will speak for themselves.

Definitely continuing to eyeball this thread till the development finishes on the UEL. Nice to see great progress!

[Jesse@JDLAutodesign]

Originally Posted by Tansey86

So there is no way to use the EL header with the up and coming turbo kit?

No the turbo header is quite a bit different...thick wall 1.25" stainless runners and twin scroll design. Its a nice piece in itself Everything else in the system is modular.

[Captain Insano]

Holy shit I am excited.

[OrbitalEllipses]

Originally Posted by Lee@JDLAutoDesign

Sorry for the wait guys! We saw an opportunity to do some analysis and hopefully improve our merge collector design and then leverage this design for both the equal and unequal length headers. Doing some of these design iterations analytically ultimately reduces the amount of building and testing we have to do so the end result is better products making their way to the customer sooner.

In my mind, the merge collector can make or break the success of the header design. If the collector design is effective, it can reduce the upstream exhaust pressure (for a given flow rate) and thus reduce parasitic losses associated with a cylinder having to "pump" out it's own exhaust. Obviously harmonic tuning plays an important role in cylinder evacuation, but the strongest exhaust harmonics are primarily tuned by runner length. A good merge collector can also minimize the mixing losses associated with 4 out of phase cylinders entering a common passage.

I'll spare most of the analysis details unless people are particularly interested, in which case I would be happy to elaborate. The gist of it is, I designed several collector profiles and then compared the resulting flow fields to the "baseline" collector design that has already been built and tested. I repeated these CFD runs for a range of exhaust flow rates to see if the benefits I found were across the board or only at a specific operating point. When I say a "benefit", I'm referring to reduced pressure losses across the collector. This benefit will be realized via reduced backpressure to the exhausting cylinders (and hopefully more power!).

Here is my model of the baseline collector:

[IMG ]http://i249.photobucket.com/albums/gg237/Leebro61/JDL/Baseline_UVEL_zps361f84fa.png[/IMG]

The figure shows contours of X-Direction Velocity. The most obvious loss sources here are the large separation regions (dark blue) downstream of the collector throat and the wake that forms in the mid passage from the boundary layer of the inner wall.

An obvious way to reduce the pressure loss is to reduce the wall slope on the expanding section of the collector. Here is a model of the baseline collector with half of the expanding wall slope:

[IM G]http://i249.photobucket.com/albums/gg237/Leebro61/JDL/LongDiffuser_UVEL_zpsa9bf32f2.png[/IMG]

Overall, these are very similar results. The longer collector with the shallower wall angle has roughly ~2% less pressure loss than the baseline with no other changes. It also has a more uniform flow field in the center of the passage, caused by a reduced diffusion rate and less separation blockage... so definitely a step in the right direction.

I have ran several other simulations that I am still post processing, but I think we have already seen enough to influence our next collector design. I'll edit this post very soon with the results of the remaining designs

Edit - I thought I would add this one also. In this model I am starting with the baseline collector and the only thing I have done is added a 1" long constant diameter gap between the converging and diverging segments of the collector. The idea behind this analysis is that gains could be had by "turning" the flow in two steps as opposed to a single, large change in wall slope. In an ideal world, I would be able to use curved walls and fine tune the collector design... but such a design would have to be cast and would be a long lead, expensive item. By adding an additional segment to the diffuser we are trying to accomplish a similar effect...

[ IMG]http://i249.photobucket.com/albums/gg237/Leebro61/Baseline_wGAP_UVEL_zps3462aad0.png[/IMG]

As far as losses are concerned, this model provides about a 3% pressure loss reduction as compared to the baseline... so even better than just reducing the wall slope

@Dimman would eat this up.

[Acree]

Kudos for doing your own CFD. It does however strike me as strange that work like this with collector/merge analysis hasn't already been done and isn't widely available. Very interested to see the final design.

-Acree

[Lee@JDLAutoDesign]

Originally Posted by Acree

Kudos for doing your own CFD. It does however strike me as strange that work like this with collector/merge analysis hasn't already been done and isn't widely available. Very interested to see the final design.

-Acree

Very good point. I'm sure it's available... somewhere. I also wouldn't be surprised if a lot of the work that has been done is either proprietary or under non-disclosure agreements between universities and automakers.

I'll do some more digging and see if I can find and link some relevant papers. The other difficulty with finding and interpreting other people's CFD results is that often time you are left to wonder how they ran the case they are presenting results for (grid quality, solver scheme, boundary conditions, etc.). In this case I wrote the solver and grid generator myself so I know EXACTLY what it is doing.

[Dimman]

Originally Posted by OrbitalEllipses

@Dimman would eat this up.

CFD and header design. Together. Pretty sweet.

However this is more of just a visual presentation on stuff that's already established in fluid dynamics. Entrance loss, dealing with vena contracta, and exit loss dealing with turbulence and detachment.

[Lee@JDLAutoDesign]

Originally Posted by Dimman

CFD and header design. Together. Pretty sweet.

However this is more of just a visual presentation on stuff that's already established in fluid dynamics. Entrance loss, dealing with vena contracta, and exit loss dealing with turbulence and detachment.

You're absolutely right, this is far from novel stuff... but I would argue it's still worth looking at.

I think anyone with any sort of fluid mechanics background will know that in a converging/diverging passage you are going to have issues with loss/separation, etc if the inlet Mach number is high enough. One could even go to any fluid mechanics text book and find pressure loss charts based on wall slope, area ratio and inlet Mach number (Sovran & Klomp comes to mind). I'm trying to go just a little bit further and figure out if the quantitative benefit that I see in my analysis is worth retooling fixtures and jigs to incorporate in the design. The other thing that is fractionally different here from the textbook case is that the flow upstream of the contraction is turning slightly (due to the merge). This has some impact on the thickness of the boundary layer upstream of the throat and it will also drive a vertical pressure gradient that could have a slight impact on flow migration toward the endwalls (affecting the separated region), but again, this is probably small potatoes. Thanks for the feedback

[JDLAutoDesign]

Pulled the car apart to start the UEL and snapped a shot of the EL. Flex bellows are in route for those of you wondering about them

[dabocx]

One of the twin scroll BorgWarner EFR series turbos would be a blast on this car.

[Jeff@Racer X Fab]

...

[Lee@JDLAutoDesign]

Originally Posted by racerx1715

Performing CFD with just the collector will give entirely different results than performing it with the entire manifold. In my opinion you would really need to study this analysis with the whole manifold to have a good idea of how effective the collector is. In addition you would need to run a dynamic flow simulation which would provide the actual data to get you in the ball park. From the manifold pictured above you would have flow seperation on the bends before entering the collector and this affects the data provided.

Sorry for the brief response but im typing on a cell phone from the rolex 24 @ daytona...

The runners upstream of the collector certainly can affect the pressure and velocity profiles entering the collector but the acceleration upstream of the throat is so strong that it really suppresses these effects. Ive actually done a study similar to what you describe and the impact on performance was very small... Certainly not enough to drive me to a different collector design.

Regarding the unsteady cfd... I dont have nearly enough computer resources to even attempt to simulate that (and in my personal experience most industrial cfd design work isnt even done unsteady).