Honda / Acura K20a K24a Engine Forum banner

1 - 8 of 8 Posts

·
Registered
Joined
·
4 Posts
Discussion Starter #1
First time poster, spent far too long reading lots of interesting threads on intake manifolds.

However times have moved on and I thought you might be interested in my little project.

My car is an S1 elise with a K20a2, it currently has the PRB manifold and the RRC wont fit without cutting the bulkhead.

Then I discovered the RSP, not only does it look okay on paper but more importantly it is a modular design so I can take it apart to tweak quite easily.

I have seen several posts on here where people have cut the back off and welded it up and done similar to the trumpets, even seen people CNC machining new trumpets but all of that is either beyond my skill set or budget.

So this evening I set up a mini photo studio



Resulting in 109 photos of the trumpets, example below



I then imported those into 3DF Zephyr and after an hour or two of custom making image masks I was able to generate this splodge:



The rings of blue dots are the camera positions and the image mask process basically cuts out the block of wood the trumpets are balanced on. The wood is on a lazy susan (turn table) on an oversize clear plastic protractor. After each photo I rotated the part by exactly 10 degrees and took another photo, after 36 photos I rotated the part.

The splodge is a point cloud, the software has compared all of the photos an identified specific points that it can see in all (well some) of the photos and it uses these points to understand the 3d model.

Next step is to refine the point cloud:



And then generate the model



This can then be exported into your modelling software of choice. Because I am on freeware that would be Blender:



Its pretty lumpy and bumpy.

Next step is to basically trace it in the modelling software so that the whole thing is tidied up and then tweak it so it matches the dimensions of the actual part. Place particular emphasis on bolt holes etc.

I've found this to be a nice easy way to get to a 90% model that can be refined because freehand measuring things like curved surfaces can be tricky.

So why bother, where is this going?

First step will be to 3D print a new plenum back plate that does away with the resonance chamber, I can run this a while to check/prove reliability as it will be easy to see if its damaged.
Second step will be to print a set of 20mm trumpets with nice elliptical edges to open up the top end.
Third will be to experiment with different trumpet designs / angles to see if flow can be evened out / improved.
Then try and find a way to add a dual plenum to the back plate to feed the plenum.

Once the design has been tested (including dyno time) and reliability proven the last step will be printing a custom flange and runners as part of a full working printed inlet manifold to maximize.

I'm expecting all of this to take some time but i'll keep this updated if there is interest :)
 

·
Arouse the DAMPFHAMMER!
Joined
·
6,979 Posts
Welcome to K20a.org mrluke!

First step will be to 3D print a new plenum back plate that does away with the resonance chamber, I can run this a while to check/prove reliability as it will be easy to see if its damaged.
Second step will be to print a set of 20mm trumpets with nice elliptical edges to open up the top end.
Third will be to experiment with different trumpet designs / angles to see if flow can be evened out / improved.
Then try and find a way to add a dual plenum to the back plate to feed the plenum.
Amazing what can be done out of digital pictures. I am interested in the quality. Did you measure the bolt pattern in the model and compare it to the real bell mouth bolt pattern?

A very good basis for your design approach is this paper here: RET_Bellmouth_Sept.pdf (profblairandassociates.com)
 

·
Registered
Joined
·
4 Posts
Discussion Starter #3
Welcome to K20a.org mrluke!

Amazing what can be done out of digital pictures. I am interested in the quality. Did you measure the bolt pattern in the model and compare it to the real bell mouth bolt pattern?

A very good basis for your design approach is this paper here: RET_Bellmouth_Sept.pdf (profblairandassociates.com)
Thank you and thanks for the paper :)

The accuracy is really very good, you're talking about fractions of a mm really. The biggest issue is my photo quality isn't great so the software struggles to find the edges of the dark grey object against the slightly darker grey background. I scaled the model to an overall length of 250mm and checked the widest diameter across the trumpets, 64.6mm in the model and 64.6mm in real life.

The picture below shows my model and the cad scan overlayed so you can see the differences



As you can see its mostly around the edges with little pockets on the surface finish. I think I may have more success using a green screen as the background rather than black so will try that in the future.

And here's the model looking slightly prettier. Renders aren't my forte, I just make and print the parts lol.



There's a bit more work to do before its a finished part, the edges need tidying up so that it looks smooth rather than faceted. But for a couple of hours in CAD i'm pretty happy with where I have got to.

I'm not going to spend much more effort on this design because I don't want/need to print trumpets that duplicate the originals. This is good enough to get the dimensional accuracy, next step is to create an elliptical profile in both stock height and shortened lengths.
 

·
Premium Member
Joined
·
10,456 Posts
I like it. Really neat approach instead of fancy 3d scanning and proprietary software.
 

·
Arouse the DAMPFHAMMER!
Joined
·
6,979 Posts
...64.6mm in the model and 64.6mm in real life.
Seems you rises a bit more of my interest with that result. Just the radii weren't build up correctly, especially at the bell mouth. What did happen to the parallel cut on the bolt bores? De-rendered?

But for a couple of hours in CAD i'm pretty happy with where I have got to.
I assume this is a test case for you to investigate the approach for further projects, isn't it? Wouldn't you be faster by to measure and draw it into 3D with that bell mouth flange? We used a 3D laser scan tool (80 kEur) to transfer a head into 3D CAD, the effort to render it correctly would so high we skipt it, maybe we had the wrong technology. But on free form shapes ghost differ, there the real advantage lays to dig out with such technologies. Would nice to see how it works there, e.g. on a more complex runner design of the RSP IM.

I'm not going to spend much more effort on this design because I don't want/need to print trumpets that duplicate the originals. This is good enough to get the dimensional accuracy, next step is to create an elliptical profile in both stock height and shortened lengths.
Very good, I hope the Blair paper helps on that.

Do you have an idea how long you want to have the runners, for the runner length tuning?
 

·
Registered
Joined
·
4 Posts
Discussion Starter #6
I like it. Really neat approach instead of fancy 3d scanning and proprietary software.
Thanks :) much cheaper than a scanner but I am very jealous of what they can do with them, the real time scanning is pretty cool. Not very accessible for a hobbyist though.

Seems you rises a bit more of my interest with that result. Just the radii weren't build up correctly, especially at the bell mouth.
The software struggled to work out where the edges of the trumpet were which reduced the accuracy.

What did happen to the parallel cut on the bolt bores? De-rendered?
The model has two parts, the first is the underlying geometry and the second is a texture file. On the image its the texture file that hasnt been compiled 100% correctly, its overlapped the holes slightly to make them out of round. Again this has been from the software struggling to identify where the metal stops. Ultimately this is easily overcome because we know that the holes need to be round, we can measure the diameter from the part, When I add the holes to the model I make sure there are all at the same height, the scan will indicate the horizontal position and then its just a case of double checking against the actual part to make sure they are perfectly placed. But again we are into fractions of a mm here.

I assume this is a test case for you to investigate the approach for further projects, isn't it? Wouldn't you be faster by to measure and draw it into 3D with that bell mouth flange?
I'm not quite following. The new bell mouths I will draw free hand and add to the model as my first tweak but things like matching the curvy flange are much easier to do with this technique than from the part with a ruler :).

The plan is to 3d print a functional set of trumpets that can be installed on the car. This will be by outsourcing the printing to a large company that have the latest tech available at affordable prices.

We used a 3D laser scan tool (80 kEur) to transfer a head into 3D CAD, the effort to render it correctly would so high we skipt it, maybe we had the wrong technology. But on free form shapes ghost differ, there the real advantage lays to dig out with such technologies. Would nice to see how it works there, e.g. on a more complex runner design of the RSP IM.
I can see a head being somewhat difficult to render if you are trying to include all the cooling galleries etc. If you are just looking for airflow then it shouldn't be too bad, depending upon whether you want to accurately recreate the variances in each cylinder/port or just do one and mirror it 4 times till you get a good fit.

Very good, I hope the Blair paper helps on that.
I'll use their profile for the bellmouth

Do you have an idea how long you want to have the runners, for the runner length tuning?
From what I have seen there are multiple ways to calculate ideal runner length and the outputs vary enormously. I have come to the conclusion that the calcs will only land you in the ball park and dyno testing is required to get the perfect length.

My thought is that the issue with the RSP manifold is that runner #1 blocks the TB opening. Fine at lower rpm but at the top of the rev range it makes the flow path too difficult. Hence why people cut them down to a 25mm runner length and regain the top end.

If I can add a dual plenum onto the back of the RSP and cover up the existing throttle opening then maybe the stock length trumpets will show a benefit over the shorter lengths. Given my use case i'm probably aiming for a peak HP at around 7.5k.
 

·
Registered
Joined
·
4 Posts
Discussion Starter #7
So I have modelled the elliptical profile from the paper:



However they aren't going to fit the RSP manifold.



They are also 1 diameter high so circa 50mm which means they will obstruct the TB opening as well.

Once I have got all of the elements scanned I will start looking at CFD to narrow down the options.
 

·
Arouse the DAMPFHAMMER!
Joined
·
6,979 Posts
So I have modelled the elliptical profile from the paper:
However they aren't going to fit the RSP manifold.
They are also 1 diameter high so circa 50mm which means they will obstruct the TB opening as well.
Welcome to the engineering world. Theory and Practice, here the engineering began to span the bridge between both: clearance limits and theoretical recognitions to implement. If you read carefully the paper you will find the that Blair has a heart for engineers, as he already foreseen issues and made general design rules to follow. I would follow that.

I am no friend of 3D CFD, as the results are too depended on modeling approaches of boundary layers and mathematical requirements to get it started or to be able to calculate it. It doesn't include the fuel droplet spitting and it's effect of vaporization during that back and forward process. So I wouldn't mind about doing 3D CFD. BTW, most CFD codes look into stationary conditions, but the conditions here are by far not stationary. Modelling this correctly will cost some days or even weeks to calculate one single inlet event with respect to Plenum pressure and mass movement :D.

Regarding the length approaches. Here simple approaches like Speed of Sound wave front calculations up to 1D-engine simulation are rational. The interesting is, the first differ not too much from the last and every misses the reflection flexation and efficiency. The last was also not included by Blair, he just looked at the flow properties, not into the oscillation properties. Something, many approaches misses. What engine speed is in your specification?
 
1 - 8 of 8 Posts
Top