911T Head Flow Rates

[Rich911E]

I have spent some time looking at the data you present and I would like to see if you believe I am interpreting this head flow data correctly.

The following is my understanding of the valves and ports your are evaluating:

2.2T stock = 46mm intake valve; 32 mm intake port
2.4S stock = 46mm intake valve; 36 mm intake port
3.3 turbo stock = 49mm intake valve; 32 mm intake port
3.3 turbo ported = 49mm intake valve; ????? intake port (presumably larger)

Flow at .450 lift (i.e., S camshaft profile intake lift)
2.2T stock = 200
2.4S stock = 202 (interpolated)
3.3 turbo stock = 228
3.3 turbo ported = 239

So, from this data, I make the following observations:

Comparing the 2.2T stock to the 3.3 turbo stock: an increase in valve size from in 46 to 49 mm (a 13% increase in valve cross sectional area) increased the flow by almost 15% with the same size port. Almost a 1:1 ratio. In contrast, by comparing the 2.2T stock to the 2.4S stock, an increase in port size from 32mm to 36 mm (an increase of 26% in cross sectional area) produced no increase in flow with the same size valve at this lift. My interpretation of this data is that the port sizes are just fine but that the valves are too small.

Second, I interpret this data as suggesting that putting a higher lift camshaft in any early motor as being relatively pointless. Intake lift on cams stock cams is : T = 0.387; E = 0.408; S = 0.450. Your data suggests that the benefit of the larger ports in the S motors is primarily observed at lifts above 0.450 lift. This would suggest that an S head could support a cam with a lift greater than .500”. Could you post flow data on these heads up to .600”. I would really like to see what happens up here. I know that one successful 2.0 liter engine builder uses camshafts with lifts up at .500 intake lift with 36mm intake ports.

My take home lessons: If the 3.3 turbo 32mm intake port and 49 mm intake valve combination can flow 15% more than a 36 mm port and 46 mm valve, I think that we should be looking at this port configuration as a great model for our cylinder heads. Putting larger valves in 2.2 T heads should be a perfect streetable performance modification -- retaining the low RPM intake gas velocity for efficient cylinder filling and drivability with little loss in top end performance. Additionally, as you observe, if you want more power you need more duration and more RPM not more lift. So a longer duration cam with a relatively low lift to minimize the possibility of valve float in a 2.2 head should make a killer combination retaining both midrange power and response with really good high RPM performance.

This tells me to put in bigger valves like the size of the turbo valves. Since Porsche valves are ridiculously expensive, I could use Chevy valves. Manley offers a service to make whatever size valve you need – really nice stainless steel and even titanium. Chevy small blocks used a 1.94” (49.276mm) intake valve which could easily be cut down slightly to the size of thet turbo 49 mm intake port. Second, Chevy also used a 1.625” exhaust valve (41.275 mm) which is almost identical to the 41.5 mm turbo exhaust valve. Matched with get an E camshaft (.408” intake/.393” exhaust) but reground to extend the duration to something closer to an S camshaft (perhaps the early Solex) should make an ideal flexible engine.

[snowman]

There several things to consider in head flow, the maximum flow is not really useful for street engines as they have to live for a reasonable ammount of time and need low end torque. To live long a low lift is needed, ie about 0.35 to 0.4 max. Therefore only flows at these lifts need be looked at. Next the highest flow at the lowest lift (with reasonable port velocity, ie small ports) is what is needed. These are stock T heads, with maybe larger valves. If you look at the head there is not much room for a larger valve, will a 49mm fit? I don't know. BUT WAIT! The stock 911T head flows enough air to support all the power the rest of the engine can stand, even with improvments.

In other words to soup up a 911 with T heads on it, does not require any head work for a street engine. Its only if we wish to run at above 7000 RPM to we even need to look at increasing port size. Some minor porting, with a couple of additional angles on the valve job, ie a 5 angle valve job, champering the front of the exhaust valve, possibly swirl polishing the intake valve ( I don't think it lasts very long), and taking any sharp edges out of the inside radius of the intake is all that is needed to improve air flow at lower lifts. Even this is of questional value for the cost.

The heads stay as is, increase compression, cam like an E cam, or a little wilder in duration, oil squirters, external oil cooler, and you should have a good 170 HP engine. Any more requires more than 7000 RPM and all the changes that implys. Even then you should be able to get 300HP with the stock T head at 9000 RPM. I don't know all the reasons Porsche did not actually acheive that power, maybe they did not feel 9000 RPM was sustainable, or there was enough cooling to maintain 300HP. I suspect the cooling had a lot to do with it as Porsche engineers have told me that anything above 300 HP requires water cooling for acceptable durability.

[Bill Verburg]

I looked into custom 52mm/43mm undercut ss or Ti. valves for my 3.8 RS. The universal feedback was go w/ the PMS valves. They are expensive but hard to beat even at their exorbitant prices.

[snowman]

Heres the plaster copy of the 911T head.

It flows exactly the same as the real thing. The only improvement I can see to try is to remove the sharp edges on the short radius. These are not present on all the intakes. Otherwise the only thing to do is to open up the port. to say 36 mm. I did not choose to do this because I have found data on 911 engines with 32 mm intake ports with over 200 HP and I only want about170..

[jluetjen]

Guys - That is really great work that your doing and definitely a cut above the usual "I want to add an XXXXX header to my car - how much HP will I make type question."

- John

[Rich911E]

I am glad to see that somebody else has a copy of the book "Practical Gas Flow" by John Dalton. This is one of my all time favorite books. He details how to build a flow bench and make the plaster models, etc. I think it is out of print now, but you could probably find a used copy on-line. The publication information is:

Title: Practical Gas Flow
Author: John Dalton
Publisher: Motor Racing Publications Limited
ISBN: 0-947981-33-0
First Published: 1989

Rich

[Rich911E]

Second idea. I noticed that you stopped your flow data at .450 or .500 lift where the larger ports started to come on. Perhaps we have some power here following the lead of other recent engine designs.

When the small block chevys and ford 302s were modified to meet emissions and such but still put out really good power, one thing they swapped everything over to a roller cam set-up. These camshafts had very high lift (.560 for a stock cam is a lot) but short durations and some had asymetric lobe designs. The roller cam can deal with a much more agressive ramp on the camshaft so that you can lift the valve higher in the given amount of time. More lift with same duration.

It seems like it would be relatively easy to make 911 cam followers to have a roller tip where it contacts the camshaft rather than the equivalent to a "flat tappet" style that the iron camshaft followers have. Could also have a roller tip on the valve contact point. This would permit high lift camshafts without the long duration that kills driveability. This would mean more power and torque across the RPM band.

I once put a .560 lift flat tappet camshaft in my Chevy 302 (Z/28) motor and it killed driveability and had no idle vacuum but ran like crazy on the top end. However, I swapped it for a .500 lift street roller Crane cam and have almost 18 inches of idle vacuum, great throttle response and really nice power.

When the factory made the 906 race cars, they went with more duration and big ports to make power up to 8000RPM. However, in 1960's, valve spring material was a lot different. When they raced the Z/28s, they used a lot less cam lift than they do now because the valvespring materials couldn't handle it for long races. However, the AASE valve springs are really nice, and they should be able to support a lot more lift than the old springs.

I think that we can make equivalent power down lower in the RPM range if we can use more agressive camshafts profiles.

I love this thread!

Rich

[jluetjen]

The trick here will be to also keep an eye on the valve accelerations. I haven't had a chance to map out the cams in question (E/S/906/GE80 etc), but it would be interesting to see what sort of accelerations the stock cams have and how much room you'd have before they start to get out of control.

[Rich911E]

John:

I don't think the valve accelerations would be much of a problem given the fact that they are about the same size as small block Chevy valves and should weigh about the same. A decent Chevrolet racing engine will run a lift of 0.750 and 8000 RPM. If the valve springs are not breaking with this sort of lift and acceleration. In any event, it would not be difficult to grind an asymetric camshaft profile with a slightly milder closing rate would seat the valve more slowly and avoid the valve bouncing off the seat but still give a lot more area under the curve.

Personally, I think we have been too conservative with these engines expecting them to last the 100,000 miles on a rebuild. If we are talking racing engines, more frequent rebuilding is a fact of life. The 911 engine with 8 main bearings, steel nitrided crank, and lighweight pistons and rods should easily sustain 8,000 RPM. If the racing engines could sustain this RPM in 1966 in the 906 with available valvespring technology, we should be able to make them last at that RPM today. It is not terribly uncommon to see normally aspirated emissions legal street engines producing 100 horsepower per liter. A racing 911 engine should be able easily to sustain this level of power.

There is no way to get a lot of horsepower out of a small engine without winding it up with more air and fuel which is going to require more area under the camshaft curve. To avoid a complete lack of low speed driveability (since we don't have DOHC and variable valve timing), a more agressive camshaft opening and closing profile would seem to be required. If this is the case, the stock camshaft followers would not seem capable of this without significant side loading which would lead to premature failure. Therefore, a roller cam follower would permit more aggressive camshaft profiles within the design of the SOHC system.

Rich

[Rich911E]

Here is a picture of what I mean. Honda uses it with the SOHC VTEC engines and you can see that this design would work very well with Porsche SOHC motor. It is obviously reliable since Honda has a great reputation with their VTEC motors.



Rich

[TimT]

Ive wondered myself about using roller rockers in a 911 engine. Finding the ratio of the rocker arms would be easy enough, you may even find some aftermarket chevy roller rocker that are dimensionally similar to the Porsche arms. A bit of machining and off you go. Or you could have them manufactured ( maybe Jack could do this?). Id have to check whether aluminum is suitable for this app.


For the 911 engine to survive long high RPM periods special bearings are needed, and the oiling system must be optimised, cross drilled crank etc. Some rather succesful race teams use special Clevite bearings for sustained high RPM use ( the nascar and irl boys e.g..) A number of Porsche engine builders have these clevite bearings in the engines they build.


I know these ideas go against the whole mantra of this BBS but perhaps some should think "outside" the box.

[snowman]

Guys,
The only t rouble is that the head dosen't flow any more air at higher lift!!!

It already flows more than a ported small block chevy head, stock. the only thing left to do is up the compression and RPMs.

[jluetjen]

Just an extra data-point. In the era of the 917 (The 5 liter 917 motor has the same size cylinder as a 2.5 liter 911), apparently the engines were happy for 24 hours at 8000 RPM, and would even spin up to 8700 RPM in an emergency. But I guess spinning a motor past 9000 killed the engines pretty routinely. (Corrected numbers edited in on 3/19)

The cam shafts had the same lift and duration as the 906, but were set-up with more overlap by closing the cam lobe angle sum which seemed to make the torque curve peakier. The valve train was also a twin overhead cam. I think that they dispensed with the rocker-arms starting with the 916 race engine, and continued this through the 908 into the 917. I'm pretty sure that there is a cross-sectional drawing on this BBS somewhere. If you can't find it I have one at home which I'll post later today.

[Rich911E]

Tim: Aluminum would be just fine for the rocker material, but the alumunium ones do have a tendency to not last as long as the stainless ones. I know the drag guys used the aluminum ones but they are pushing 10,000 RPM out of a small block in a few seconds. For longevity, stainless steel is the way to go. Harland Sharp used to make really nice stainless steel roller rockers for small blocks and perhaps they could be convinced to make some roller rockers for the 911. The design would be very similar to the shaft mounted rockers that Mopar used and even with the FoMoCo racing small blocks. It would be a good deal for them as I am sure almost everyone would use them if they had the chance.

The roller tipped followers would open up an entire new world of camshaft choices, especially for the emission controlled cars and especially the CIS cars that can't handle large cam overlaps because revserion pulses in the intake make the metering plate go wacky. Crane already makes cams for the 911 and they used to make a nice line of roller rockers as well. A combination of a nicely matched roller cam for the 911 with the followers would be wonderful package. Plus, the oiling requirements with roller cams are substantially diminished. I would buy the package if it were offered to me.

John: I agree with the RPMs. Last night I looked up the RPMs used on all the race motors in Lothar Boschen and Jurgen Barth's book and as far back as 1955 and they all went to 8000+RPM with a 66mm crankshaft (2.0, 2.2, 2.5). The Type 547 carrera engines had Hirth roller bearing crankshafts, but the bottom line is that the valvetrain could stand it with no problem.

Snowman: I am confused that this port flows as much as ported small block Chevy head. Ported small block Chevy cylinder heads flow up to .700 lift where the data you present shows the Porsche falls down above .450. Here are two charts on the flow numbers for two ported small block Brodix heads that you can buy off the shelf from Brodix:





These Chevy heads are flowing a lot more CFM at the same lifts as the Porsche head and continue to show increases in CFM at higher lifts. The Brodix exhaust ports are flowing as much as the Porsche intakes. Do you have data on the Porsche heads above .450 lift under the same conditions to compare? I am very curious about the larger S-ported 2.4S heads at higher lifts.

I simply cannot believe that the Porsche factory increased the port sizes on every single one of its racing engines without the dyno numbers to back it up.

Rich

[Rich911E]

I replotted the graph above and superimposed the data from the Brodix SBC heads below. Note that the CARB legal heads out of the box (unported) are roughly equivalent to the 3.3 Turbo ported and unported heads. I was amazed at the improvement in flow when these CARB legal heads were ported at the low lifts. I know that there are variations between flow benches making it hard to compare but a change from 170 to 255 CFM at .400 lift is not all due to flow bench variation.



Just some more stuff for consideration.

Rich

[snowman]

I am refering to stock chevy heads, older ones like in 70s, 80s and even early 90's. . Without porting some of them don't even make 200 cfm at any lift. The high performance heads, and Vortech heads excluded.

I think Porsche increased the port size to reduce the intake port velocity at higher RPM. They also may have had some other problems in the intake system that reduced the air flow so they made up for it in the intake ports. I think there is some data on Porsche race engines with 32mm venturi in the carbs that made over 230 HP. Using the same size for the intake port may present some problems.

I can see why Porsche stuck to long duration, lower lift cams. They flow all they need to at 0.4" lift to make the power they were after. They have some very very agressive cams already and they work ok with the present design.

The larger ports in the S head do flow more at higher lift but not much. The reason I didn't take data at higher lifts is because the flow didn't increase any, and in fact at about 0.6" starts decreasing. The reasons have to do with how the valve is angled and the closeness of the cylinder wall and the angle of the port to cylinder.

I first chevy flow graph shows negligeble flow increas above 0.5" lift, ie only an additional 6 cfm for the extra o.2" lift or about 12 HP. The second one flows real well to 0.650" lift and the last 0.050" lift gives only about 5 cfm or 10HP.
The question here is can you do better by increasing the compression by reducing the valve lift, ie smaller valve pockets in the pistons, and the piston can come closer to the head or increasing the cam duration, ie valve chasing the piston to close.

Some drag racer flowed one of his custom chevy heads on the flow bench. He was getting almost 600cfm! IN ONE CYLINDER! He was also using one inch valve lift and very very BIG valves. THat means he could make over 1200 HP normally aspirated, on pump gas, but of course he has a blower and uses nitro. Hes one of the 300MPH guys.

Anyway back to Porsche. My understanding is the limiting thing on an engines RPM is bearing velocity, and ring velocity. If you exceed a certain velocity the oil no longer works and the metal parts come togather. This is fixed by the basic design, ie stroke, size of the crank journals , rod length. So given a particular engine you can only rev it to a certain point, no matter how much you do to it (other than redesigning it) I guess you can make some small improvements, like changing rod length or bearing width. I have been told that NASCAR is using Honda rod bearings. Aparently using Red line oil allows them get away with this. It works good in Porsche engines too. It works at much higher temperatures than any other oil, including Mobil 1, and is ashless when it burns, leaving no residue to clog up the oil system.

Interesting info on air flow and HP. There is an equation that predicts the HP of a normally aspitated engine and the RPM needed to achieve that HP. I simplified it a great deal and came up with the following which gives a conservative number. It is possible to get as much as 10% more power with the latest and greatest of everything and all the tricks.

For a V8 engine simply double the number of cfm that a single cylinder flows. E.G. if it flows 150 cfm then you can get 300 HP out of the engine. Never more than 2 HP per cubic inch though, whichever number is lower.

For a 4 cylinder the HP will be the same number as the flow of a single cylinder. EG if a cylinder flows 150 cfm you can get 150HP.

For a 6 cylinder use 1.5 times the flow.

Additional assumptions are:
Exhaust flows 75 to 80% of intake

Entire intake and exhaust system flows the same as the cylinder (or real close to it).

The engine is normally aspirated.

You can actually turn the RPM needed. It may take 20,000 RPM or more but typically will be 8,000 to 10,000 RPM.

Remember this ain't no real calculation, for that you need all the details, bore, stroke, etc. Its just an estimate that can be as good as 5% on many race engines that can turn 8,000 plus RPM.

[TimT]

Some interesting stuff here I spent alot of time on my 2.2 improving the oiling. 930 oil pump, cross drilled crank grooved 2&5 bearings, by pass mod etc. Spent time on the crank increasing the chamfer of the oil passages on the crank. Rods were prepped by a local drag shop. I think the bottom end I built will live at 8000, but I didnt pick the right cam( the way I had it timed it ran out of steam at 6800). Im going to get a GE-80 and time the cam for top end.

(and hope that someone makes some roller rockers for the 911)

[Rich911E]

Regarding piston speed, A.G. Bell provides these guidelines for piston speed in his book, Performance Tuning in Theory and Practice:

Stock Motor (cast crank, stock rods and cast pistons): 3,500 ft/min Heavy Duty Motor (forged crank, shot-peened rods w/good bolts, forged pistons): 4,000 ft/min
Drag Racing Motor (forged crank, aluminum rods, lightweight pistons, etc.) = 5000 ft/min

The formula for piston speed at a given RPM is (once you convert the units) is:

Piston speed (ft/min) = (RPM x stroke (inches))/ 6

Solving for RPM;

RPM = (piston speed in ft/min x 6)/stroke in inches

The 911 uses two primary strokes, 66 mm (2.60 inches) and 70.4 mm (2.77 inches). Using the “heavy duty motor” guideline of 4000 ft/min max piston speed in the above equation, you get:

4000 ft/min with 66 mm crank = 9230 RPM
4000 ft/min with 70.4 mm crank = 8664 RPM

Just for a reality check, a piston speed of 4000 ft/min in a 350 Chevy with a 3.5 inch stroke would provide a maximum RPM of 6857 RPM (this is commonly attainable with relatively common high performance sportsman type 350 motors). A 4,000 ft/min piston speed in 302 Chevy engine with a 3.00” stroke correlates to 8000 RPM which seems about right for a road racing 302 that should have good longevity. I found an interesting note that the piston speed of the new production BMW M3 at its 8000 RPM redline is 4740 ft/minute – about the same as that of the Williams-BMW F1 powerplant at 18,000 RPM. So, it says we should be able to spin our little 2.0 and 2.2 liter motors up in the high 8000 to near 9000 RPM range without too much an issue with piston speed but that the 2.4 and 2.7 engines should stay down around the lower 8000 range.

So, with those piston speeds what bearing speeds do we get? I couldn’t find a similar guideline for bearing speeds, but assuming that what works for a Chevy should work for a 911 at the bottom end, a Chevy 302(68-69)/350 uses the same 2.45” bearing diameter. The formula for bearing speed in ft/minute is:

Bearing speed (ft/min) = (pi x diameter (inches) x RPM)/ 12 (in/ft)

Which at 7000 RPM for a small chevy crank (2.45”):

Bearing speed (ft/min) = (3.14 x 2.45 x 7000)/12 = 4487 ft/min.

Similarly, at 8000 RPM (302), the bearing speed is 5129 ft/min.

If we use these as our guidelines for survivability and solve for RPM with a 57 mm journal size:

RPM = (4487 ft/min x 12 in/ft)/(3.14 x 2.24) = 7655 RPM
RPM = (5129 ft/min x 12 in/ft)/(3.14 x 2.24) = 8904 RPM

So, the bottom line here is that our nice big 911 crankshaft journals are more of a limitation than piston speed but that we should still be able to spin these motors up above 8000 RPM. Although a lighter reciprocating assembly does not change bearing speed, it will permit higher bearing speeds to be sustainable. Given the fact that we have individual rod journals and a lighter weight reciprocating assembly than a Chevy, this should also permit higher bearing speeds.

Anyhow, the upshot of all this is that 8000 RPM should be sustainable for a 911 engine, particularly the shorter stroke engines, within generally accepted design guidelines for piston and bearing speed.

I would make one observation about the 1970 LeMans race comparing the 917 5.0/5.4 liter engines (70.4 mm crank) to the 917 4.5 liter engines (66mm crank) is that all the 70.4 cranked engines broke and the 4.5 liter engine won the race, despite incredible rains that year where the engines were likely not pushed to the limit. If you look at the finishing order in subsequent years between the 2.0/2.2/2.5 911s versus the 2.4 and 2.7/2.8 engined 911s, the shorter crank engineers were much less likely to suffer engine failure.

Just some more information to consider. I still think that these engines will survive up in these RPM ranges for quite a while and the GE-80 cam is the one I have heard recommended for 2.0 liter racing engines.

Excuse me, but those engines in my garage are suddenly calling to me!

Rich

[jluetjen]

Ok;
I'd agree that the bottom end of the 911 motor is pretty robust compared to the rev's that top end can endure. I doubt that the bottom end is limiting the rev's (with the exception of the 70.4 mm crank without the additional machining which Porsche figured out with the long-stroke 2.5 ST's).

Here's a cross-sectional drawing of a 917 motor (actually known as a 912).



Note that Porsche dispensed with the rockers all together and just had the cams working on bucket-followers. I do need to correct something which I said earlier. According to Ludvigsen's "Classic Racing Engines", the early (66 mm stroke) 917's made their peak HP at 8400 RPM and would withstand 8700 RPM. But 9000 RPM meant a broken valve. So this pretty much confirms everyone's suspicion that the bottom end piston speeds are not the limiting issue to making more HP.

So the question is, what sort of revs can the 911's stock style valve train withstand?

Another question would be: Why can't 911 motors make more HP through the heads. Obviously the engine's cylinder size limits the HP given a certain rev limit. 911 motors are not sloutches for their size when it comes to torque or HP. I wonder if the engine's Hemi design is also limiting the HP at lower rev's due to combustion issues. Either the CR will be low and you'll have an open chamber or you wind up with a combustion chamber shaped like an upside-down cereal bowl because of the large piston crown. This is certainly not an issue for the traditional small-block V8's from GM and Ford.

Here's my theory:
* T's have low CR, open combustion chambers, and so require short duration cams (and lower rev's) to generate usable torque.

* You can increase torque and HP some by increasing the CR, but there are dimenishing returns, especially at lower rev's because the combustion chamber becomes more and more obstructed. The static CR also goes up which can cause detonation when under load. Twin plugging can help.

* You can increase the HP by increasing the cam duration, but then low speed engine performance drops like a stone because of poor cylinder filling at low speeds. You can resolve this by increasing the CR. This was the strategy of the early rally motors. But given the comparatively small engine size, HP is limited by rev's.

Snowman; do you know what the volume is of the intake tract in the head? I wonder how that compares with the cylinder volume? In other words, how much of the air in the intake ports gets pulled into the cylinder with every cycle?

[Scott Clarke]

I wonder if roller rockers would constitute an advantage over the system shown for the 917 for motors not capable of 8000+rpm. If the roller allows for more agressive cam ramps, could that advantage outweigh the disadvantage of the presence of a reciprocating part (the rocker arm)? Are the aftermarket rockers for American V-8s cnc machined aluminum? If so, it would seem reasonable that a rocker manufacturer could alter the specs to suit 911 motors. This is a great thread. Let's keep it going.
-Scott