Homebrew Head Porting: Attempting the dark art .....

[304065]

Originally Posted by Gertvr

Hey Kenik, you mentioned the velocity #s can you share those? It would be interesting to compare them to the stock heads once i get the flowbench rigged up.

Have you already completed the engine?

Thanks
Gert

That material should be shared privately and not on the board.

[Gertvr]

Totally agree

[kenikh]

I never got velocty numbers - those are rightfully kept confidential by the builder. I did get volumes at lift.

[jluetjen]

A lot of my thoughts on the subject are encompassed in this thread from a few years back. In fact there is lots of information collected on this BBS over the years. Here's another thread that I think you will find contains a lot of good data.

As far as getting 300 bhp out of a 3.0 SC engine, a couple of questions would help me to understand the scope of the project better...

1. Are you staying with the 3.0 liter capacity?
2. What sort of induction system are you using? Carb? CIS? To be honest I doubt that you'll get to 300 HP in a 3.0 liter engine using CIS no matter what you do to the heads.
3. What sort of cams are you going to be using?

As I mentioned on the earlier threads, I think that Porsche actually did pretty well with their head design. As long as your peak torque intake velocity doesn't go above 75 meters/second, your porting is most likely fine. The happy range seems to be from a low of 60 meters/second up to 75. Low end of the range is defined by engines such as Porsche's early 2.0S, the 906 and a few other early models that generally suffered from "peaky" engines. As time went on Porsche tended to push the intake velocities up which most likely helped the overall driveability of the motors.

Here's a chart of some head flow data that I've collected over the years.



A few things to consider...
1) Valve Size
2) Port Size
3) Port shape

Valve size seems to have a much bigger impact at lower lifts. So putting bigger valves in an engine without touching the ports can make the engine act like it has a bigger cam, except it will give away some peak RPM HP compared to an engine with a bigger cam and bigger ports.

Port size has two impacts..
1) It determine the peak RPM HP by limiting the ultimate amount of air that the engine can take in.
2) If the ports are too large (in the quest for big HP numbers) you can wind up with a big loss at the low to mid-RPM performance.

Port Shape is far more subtle. Keep in mind that compared to a Detroit V8, 911 ports are almost perfect. They are not compromised by uneven intake lengths (Until the advent of CIS and later injection systems) nor the intrusion of push-rods or water jackets. So the ports that you see in a 911 are pretty much how Porsche's engineers wanted them. In general Porsche's intake tracks tend to be conical, so they have the largest diameter at the tip of the intake trumpet, and then they tend to get smaller until just before the curve in the port before the valve. The reason for this appears to be that it improves the harmonic performance of intake track, which helps to minimize reversion and flat-spots. It also insures the maximum turbulence by having the highest port velocities as the mixture goes past the valve.

To sum it up, I think that the motto of the medical profession bears consideration... "Do no harm". If you can avoid messing things by making changes, you've accomplished 90% of the battle.

[Dynamohum]

I am going to try and stay away from new technology and stick to purely "home brew " porting that anyone should be able to do with a dremel or die grinder and some hand tools.

Me to!

Dynamohum: did your labor of love yield any positive results? and yes i will flow test your AFM.

To answer your question "did your labor of love yield any positive results?"

I'm still building the engine, so nothing to report yet and the heads were not flow tested prier, so really there is no base line to go by.

The work done on the heads can be classified as a Street Port. Cleaning and removing casting flaws, cleaning up the areas around the valve guide boss for a smoother transition. There was no hogging out or enlarging anything. The majority of the work was all done by hand, with lots of sore fingers. I used a dremel where I could (with the guides out, which helped allot) but still found the best results were by hand sanding.
The intake ports will be left as a sanded finish and the exhaust ports are going to be flow coated with a slick Thermal Barrier Coating (TLHB) from Tech Line Coatings, Inc.

And Yes of course, in True Spirit of the Hard Core DIY I'm doing all the Coatings myself!

I Have all the equipment, and nobody has a greater vested interest in giving A 110% to do it right, so why not!

Your Flow bench is coming along nicely and thanks for the offer to test a AFM

[Gertvr]

Hi John,

Thanks for all the great information, my head wants to explode from all the information from the last months research, on this forum, Speed-talk, books (like Dalton, Vizard etc) basically everything i can get my hands on.

I will have to start formulating objectives for various parameters like inlet tract lengths, port CSA, port to valve ratios, flow and flow ratios, velocities, anti-reversion etc, based on the information that i have and put the theories and objectives out there for scrutiny, and i hope that both, here and at Speed-talk, the knowledgeable folks will confirm, reject and add what is missing before i even start doing anything.

To answer the first few questions

1) Yes, i need to keep it below 3.0l for Historics racing
2) Induction: 46 PMO (40 or 42mm) venturis on their standard lenght manifold
3) After consultation with John D, DC62 Cams with 106 lobe centers
4) Equal length tube headers 1 3/4 ( i know its a little big but that's what i have) with Burns SS merge collectors on reverse cone mega-pones or muffled 6-2-1 config for sound restricted tracks and street.

Where did you get the peak tq velocity intake numbers, are they measured dynamic on a running engine, if so how would that relate back to a static test at 28" H2O?


Cheers

Gert

[jluetjen]

Originally Posted by Gertvr

Hi John,
To answer the first few questions

1) Yes, i need to keep it below 3.0l for Historics racing
2) Induction: 46 PMO (40 or 42mm) venturis on their standard lenght manifold
3) After consultation with John D, DC62 Cams with 106 lobe centers
4) Equal length tube headers 1 3/4 ( i know its a little big but that's what i have) with Burns SS merge collectors on reverse cone mega-pones or muffled 6-2-1 config for sound restricted tracks and street.

OK. That sounds like the right approach without going through the details. Ask John D that the expected peak torque engine speed is for a 3 liter with your cams. You can also use the rule-of-thumb equations in my "How to pick a camshaft" thread to calculate this for yourself. You will also need to know the lift so that you can ensure that valves and ports will flow throughout the valve's entire lift. Since I don't have any SC head flow data, you'll still need to flow them. I believe that your 78 SC should have the bigger 39 mm intake ports. They should really big enough to get you to 290+ HP. I wouldn't make them much bigger then 40 or 41 mm at the largest.

An example, CStreit had a 290 HP, 3.0 911 race engine that he made which had 39 mm intake ports and generated it's peak torque at about 6000 RPM. This equated to an average intake velocity of 83.5 meters per second at 6000 RPM. Prior to some tuning he was developing his peak torque at 5600 RPM, which worked out to about 78 meter/second.

Where did you get the peak tq velocity intake numbers, are they measured dynamic on a running engine, if so how would that relate back to a static test at 28" H2O?

Looking back at my notes, I said something incorrect earlier. The smallest diameter in the factory air-cooled 911 intake is at the intake manifold/head mounting plane. So that is the area which is going to be gating the maximum airflow. From there they generally open up at about 1 mm for every .5 inches that you move in from that plane. This is to slow the mixture down a little prior to making the corner behind the valve. In order to calculate the peak torque intake velocity, you need to do the following:

1) Know your engines expected peak torque engine speed in RPM.
2) Know your engine's stroke
3) Know your engine's bore
4) Know the diameter of your intake port at it's smallest point.

Then use this equation: =(Cylinder Stroke * (peak torque engine speed))/30000 *(Cylinder Bore/Intake Port Dia (mm))^2

This actually is a calculated average intake velocity rather then an actual measurement. It's fine as a "rule of thumb" for modeling. In reality the actual peak intake velocity will higher since the mixture is only flowing when the intake valve is open, which is a much shorter period of time. I haven't spent the time to understand the relationship between flow data and this number. If we were talking about different engines with different intake port configuration this number might not be as meaningful, but since we're talking about comparing one air-cooled 911 engine to another, in which the ports are all pretty similar in shape (but not dimensions), it seems to work close enough.

If you really wanted to map the relationship between the flow bench and reality, you would need to look at the actual amount of time that the valve is open. I've already done this for a number of factory configuration 911 engines and noticed that in general the engines' torque peak generally coincided with the engine speed when the intake valve was open .00015 seconds. Using some calculus, you should be able to sum up the total flow available (from seat to seat) based on your cam and the engine's capacity. That will get you close, but won't take into account such dynamic factors as the crank-angle and piston speed during that time as well as the harmonic tuning aspects of the intake and (in the case of cams with overlap) exhaust. Hopefully you'll find that you can flow enough air in .00015 seconds to fill up your desired cylinder capacity.

You're welcome to figure all of that stuff out, but I suspect that you'll discover that you know less about the process then you need to in order to build a solid model. In that case you may conclude (as I did) that my empirical model gets you close enough to avoid fouling things up too badly.

[Gertvr]

" Looking back at my notes, I said something incorrect earlier. The smallest diameter in the factory air-cooled 911 intake is at the intake manifold/head mounting plane. So that is the area which is going to be gating the maximum airflow. From there they generally open up at about 1 mm for every .5 inches that you move in from that plane. This is to slow the mixture down a little prior to making the corner behind the valve."

(From SF Flowbench manual i think), this tapers down as you mentioned earlier but deeper into the port and the divergence happens closer to the valve seat creating a venturi passing the valve(tulip valve flow pattern?) The port approach to the valve in this case seems to have a higher angle maybe the reason for the later divergence. Ratios are interesting to note.

[carslutt]

good collection of data!.... in about 45days i hope to start down this road on my own heads

[JFairman]

With a 930 head the smallest diameter in the ports is where it joins up with the injector bocks and intake manifold.
They are 32mm at that point, and some people open them up up from 36-38mm for more power during boosted rpms.

[jluetjen]

Gertvr; The intake port that you showed is significantly more of a downdraft port then a 911 has. The configuration shown is what the Super-Touring engine tuners do when they would update 1 2-liter sedan engine into a Super-Touring race engine. (Keith Duckworth did the same thing with some of the Ford engines he worked on in the 60's and 70's.) Basically they'd weld up the intake ports on a street head which generally make a 90 degree turn at the valve, and open the angle up so that they came straighter down valve. That would be about the only major change to a 911 port that I think would make a significant difference to the head's performance. But since a 911 combustion chamber is a hemi design, shrouding of the valve isn't really an issue and so I'm not even convinced that changing to a straighter port would make a huge difference.

Anyhow, the factory heads that I've looked at measured out like this...

2.4TK head: Distance from manifold face...
- 0.0" 30 mm diameter
- 0.5" 31 mm dia.
- 1.0" 32 mm dia.
- 1.5" 35 mm dia (this is almost at the leading edge of the valve)

2.2E head:
- 0.0" 32 mm
- 0.5" 33 mm
- 1.0" 34 mm
- 1.5" 35 mm

2.2S head:
- 0.0" 36 mm
- 0.5" 38 mm
- 1.0" 38 mm
- 1.5: 40 mm

BTW - Does your nickname suggest that you are a TVR owner???

[Gertvr]

Originally Posted by Dynamohum

Me to!
And Yes of course, in True Spirit of the Hard Core DIY I'm doing all the Coatings myself!

I Have all the equipment, and nobody has a greater vested interest in giving A 110% to do it right, so why not!

Cool, you are way more adventurous than i am.

[kenikh]

John, from what I have seen from the port designs of gurus like Xtreme cylinder heads and Dick Evelrude, they tend concentrate on two things:

Improving turbulence
Making the intake tract a true, tapered venturi

My understanding is that they build velocity by ensuring the entire port is tapered from the port opening to the valve. Every infinitely small increment behind the intake port is an aperture when you cross section a head. Performance is about port design which maximizes velocity as a function of air volume, not a static diameter reading. For example, my 2.3L heads measure at any given diameter on the intake port between 39mm and 40mm. The air flow was modeled to my specific displacement (2.3L, 85mm x 66mm), camshaft (DC44 on 102 lobe centers) and top engine speed (8000 RPM). I provide this example as a counterpoint to the notion of "small" ports: it ain't the ports as much as what is behind them (between the port and the valve).

This what you get when that kind of methodology is applied:



Also notice the ridges from the CNC work - that adds additional turbulence to the mixture.

[jluetjen]

kenikh; I didn't mean to come across as saying that there is nothing to be gained. I'm sure that there is. My point is that if there is 5%-10% of a performance improvement to be gained in a SBV8 from porting the heads, there is most likely only about 1/3 of that available within a 911 head since it is already far more optimized then most other heads due to the fact that it is not compromised by push-rods, water jackets and uneven port lengths.

You bring up a good point which I missed earlier -- namely that there are really two reasons for the ports opening up as they get closer to the valve:
1) To slow down the charge prior to making the 90-degree turn behind the valve.
2) To maintain an even cross-sectional area in spite of the valve boss and guide intruding into the port.

I suspect that the second is more important then the first.

[kenikh]

Originally Posted by jluetjen

kenikh; I didn't mean to come across as saying that there is nothing to be gained. I'm sure that there is. My point is that if there is 5%-10% of a performance improvement to be gained in a SBV8 from porting the heads, there is most likely only about 1/3 of that available within a 911 head since it is already far more optimized then most other heads due to the fact that it is not compromised by push-rods, water jackets and uneven port lengths.

You bring up a good point which I missed earlier -- namely that there are really two reasons for the ports opening up as they get closer to the valve:
1) To slow down the charge prior to making the 90-degree turn behind the valve.
2) To maintain an even cross-sectional area in spite of the valve boss and guide intruding into the port.

I suspect that the second is more important then the first.

Your point is well taken - OCD guys like me tend to obsess on that last 5%.

[Gertvr]

Originally Posted by jluetjen

Gertvr; The intake port that you showed is significantly more of a downdraft port then a 911 has. The configuration shown is what the Super-Touring engine tuners do when they would update 1 2-liter sedan engine into a Super-Touring race engine. (Keith Duckworth did the same thing with some of the Ford engines he worked on in the 60's and 70's.) Basically they'd weld up the intake ports on a street head which generally make a 90 degree turn at the valve, and open the angle up so that they came straighter down valve. That would be about the only major change to a 911 port that I think would make a significant difference to the head's performance. But since a 911 combustion chamber is a hemi design, shrouding of the valve isn't really an issue and so I'm not even convinced that changing to a straighter port would make a huge difference.

Anyhow, the factory heads that I've looked at measured out like this...

2.4TK head: Distance from manifold face...
- 0.0" 30 mm diameter
- 0.5" 31 mm dia.
- 1.0" 32 mm dia.
- 1.5" 35 mm dia (this is almost at the leading edge of the valve)

2.2E head:
- 0.0" 32 mm
- 0.5" 33 mm
- 1.0" 34 mm
- 1.5" 35 mm

2.2S head:
- 0.0" 36 mm
- 0.5" 38 mm
- 1.0" 38 mm
- 1.5: 40 mm

BTW - Does your nickname suggest that you are a TVR owner???

John,

No I whish, I never had a TVR only saw them occasionally in Historic races, where I came from they are pretty scarce

Gertvr = Gert van Rooyen

I have taken some measurements of the 78 sc Heads and dont have a bore gauge so it might not be that accurate

0.0" 39.3mm
0.5" 39.8mm
1.0" 40.5mm
1.5" 41mm
2.0" 46mm

0.5" in from seat 47.5mm

I used your formula and came up with 81m/s or (276fps) @torque peak RPM but came up with a different number on the time available to fill the head. I came up with 0.01522s with an inlet duration of 274 @ 6Krpm which moved the decimal 2 places.

I think changing to a straighter port would minimize the SSR effect and the valve area utilization would increase for higher velocities.

Originally Posted by kenikh

John, from what I have seen from the port designs of gurus like Xtreme cylinder heads and Dick Evelrude, they tend concentrate on two things:

Improving turbulence
Making the intake tract a true, tapered venturi

My understanding is that they build velocity by ensuring the entire port is tapered from the port opening to the valve. Every infinitely small increment behind the intake port is an aperture when you cross section a head. Performance is about port design which maximizes velocity as a function of air volume, not a static diameter reading. For example, my 2.3L heads measure at any given diameter on the intake port between 39mm and 40mm. The air flow was modeled to my specific displacement (2.3L, 85mm x 66mm), camshaft (DC44 on 102 lobe centers) and top engine speed (8000 RPM). I provide this example as a counterpoint to the notion of "small" ports: it ain't the ports as much as what is behind them (between the port and the valve).

Kenikh, I agree with making the whole intake tract a taper venturi as on the picture above and i am going to aim for that as far as possible and a lot of literature points to that.

What do you mean with turbulence and how is it good for flow.

My impression is that turbulence is generally bad i.e. signifies that the flow is unstable and you are giving up port area at high velocities? I can see it being potentially good for slow moving air @ low lift as it could dynamically reduce the port area and speed up the flow in the remaining area for higher velocity and better cyl filling(BTW this is one of the concepts that i want to verify once I have the flowbench rigged up)

Gert

[Gertvr]

Originally Posted by jluetjen

kenikh; I didn't mean to come across as saying that there is nothing to be gained. I'm sure that there is. My point is that if there is 5%-10% of a performance improvement to be gained in a SBV8 from porting the heads, there is most likely only about 1/3 of that available within a 911 head since it is already far more optimized then most other heads due to the fact that it is not compromised by push-rods, water jackets and uneven port lengths.

You bring up a good point which I missed earlier -- namely that there are really two reasons for the ports opening up as they get closer to the valve:
1) To slow down the charge prior to making the 90-degree turn behind the valve.
2) To maintain an even cross-sectional area in spite of the valve boss and guide intruding into the port.

I suspect that the second is more important then the first.

From most of the research i did so far: a lot of effort goes to try and get the air to turn around the SSR or to do away with the SSR as you mentioned above. At high velocities it becomes a dead area and the air overshoots the turn and all tries to enter the cyl via the back of the port on the high flowing radius and this reduces the effective cyl entry area.

[JFairman]

Some porting designs widen and take material away from the high side radius around the valve guide to try and lift the airflow higher while it makes the turn around the short side radius.
By lifting and widening the air flow at the short side radius the theory is more air can flow through that area.
I've also heard of "eyebrow cuts" or grooves cut 90 degrees to the airflow along the top side or roof of the port just before the 90 degree bend starts to try and create a low pressure area there and lift the air flow higher away from the short side radius so more total air can flow through there.

You want to straighten out the air flow as much as possible before it hits the back of the valve so it flows out all the way around it as close to equally as possible. Removing too much material from the short side radius will make it so that doesn't happen as much and you can end up with the air hitting the back of the valve at an angle which will cause most or even more of the air flow to exit through the side of the intake valve farthest from the short side radius causing weird turbulance right at the valve opening and possibly reduceing air flow.

Anotherwords porting by line of sight rather than experince and a flow bench on an intake port with a 90 degree bend in it can make it flow less because of possibly creating bad turbulance.

I've looked at 962 heads and intake ports when they were sitting on a workbench and the factory ones I saw didn't look much different than a stock early SC head.

[jluetjen]

Gert;
Keep in mind that peak air flow really only matters from the peak-torque engine speed up to the peak-HP engine speed. More specifically it matters at the peak HP engine speed. Those are the times when the engine is not getting as much mixture as it needs. How much time in a lap do you spend at the peak HP engine speed?

Turbulence on the other hand improves the combustion quality, and this matters across the rev range, and will also tend to push the peak torque value up since it allows you to harness more of the energy available in the fuel. But turbulence is not a bi-modal factor. It's not like you have it or you don't. There are different forms of turbulence which can be effective...

- Swirl
- tumble
- random turbulence
- Combinations of the above.

The important point is that you want to do two things:
1) keep the fuel suspended in the incoming air charge. If the fuel falls out due to a lack of energy (aka: movement) in the charge, it will puddle or collect on the port walls, which will be fuel that doesn't power your car forward.
2) Make sure that the fuel is well mixed within the cylinder. If the mixture in the cylinders consists of a "clump" of rich mixture surrounded by larger areas of lean mixture, the chances of getting a good burn will not be good. The result will be fuel pumped out the exhaust which does nothing but heat the exhaust and not help to power your car forward.

The "art" of combustion engineering is designing an engine which maximizes the combustion quality across the whole rev range. Some engines are just not that good at this -- for example the original Ford Cleveland V8 Trans-Am motor. Other engines (such as most modern 4-valve, pente-roof designs) are actually very good at doing this.

As far as the inlet valve duration at the peak torque engine speed, you helped me find an error in my earlier calculations. Thanks! Here's a restatement of what I found...

911T: .00889 seconds at peak torque, .00644 seconds at peak HP.
911E: .00919s / .00636s
911S: .00856s / .00644s
906: .00757s / .00585s
Webcam 120/104: .00763s / .00592s

I suspect that some of the variation has to do with the flow capability. If you came up with 0.01522s, something still doesn't add up. Here's the formula that I used...

Intake duration in seconds =1/(360*Engine speed in RPM)*Intake duration in crank degrees*60

Did I miss something?

[Gertvr]

John,

No problem this is the formula that i used

Inlet duration on sigle power stroke= ((1/(number power strokes/s))/360 )* duration

i.e. ((1/(6000/120))/360) * 274 = .0152s Does this make sense or did i get somethig wrong?

At peak HP RPM it would be 0.0117s

Gert