Zero10

As much as most of us hate Honda, this guy has a good point. Perhaps it's worth investigating what specific things have been done to their engines to allow perlonged high-revs. Seems to me I've seen a few honda engines with a redline around 8 grand. I'm sure with some slight mods, they could rev well past that.
Doesn't the S2000 rev up to 9200 or something crazy like that?

AznDrgn

Look at the bore, stroke and displacement on a honda. If I'm not mistaken they have small bores and long strokes or maybe it's the other way around. Either way the combination makes for a high revving motor that doesn't produce a whole lot in the way of torque, just take a look at any honda dyno chart and you'll see what I mean. Looking at the 3.0L it has a HUGE bore and a long stroke which makes mad torque (ask anyone who has used an NA tranny on a 3.0L) and a decent amount of HP. Now I'm not saying that the 3.0L isn't capable of being usable at 8500RPM I'm just saying that it's not going to be easy and it probably won't be reliable as a daily street car. Good luck with the mods.

Zero10

Ahhh, yes. I forgot about the whole torque thing.

Duke

The 3.0 liter 968 engine is an extremely pleasant engine and very comfortable to drive around town. But it's not "racey" at all.

In order gain HP in N/A engines there's really only two ways to go. Increase displacement or make it rev higher.
And going the displacement way isn't a feasable solution. Maintaining the original hp/liter output would only raise a 3,2 liter engine HP to 256.

No matter what you think of Honda, they make some of the best engines in the world.
The S2000 has it's power peak at 8200 rpm with 120 hp/liter (same CR as the 968). And that's on a completely stock engine.

A 968 engine revving 8000-8500 rpm with a matching powerband should theoretically produce around 100-120 hp/liter putting it somewhere along 300-360 hp. All theory though!

IMHO a regular tuned bottom end should be able to handle what you're looking for. The challenge probably lies within the valvetrain.

Another option is as mentioned to destroke it to lower piston speed. But I wouldn't go that route before you really have tested a 8000+ rpm 3 liter solution.

How about the pistons, the stock ones are quite heavy. Too bad there's so few options there.

Danno

"Doesn't the S2000 rev up to 9200 or something crazy like that?"
I've seen numerous ones at the dyno with 12,000rpm redlines.

"If I'm not mistaken they have small bores and long strokes or maybe it's the other way around. "

Large bore, short-strokes make for a high-revving engine for numerous reasons:

1. air has to travel less distance from valves to bottom of stroke (lower velocity, less stretching of air-column).
2. lower piston speeds, less friction
3. larger valves can be used, to move less air, again lower velocity and better chamber filling

Part of the problem we have with RPMs is that the valves has to move through 750cc of air per intake-stroke while the S2000 has to only fill 500cc. The valves on the 968 are only 20% larger than the S2000, yet has to move 50% more air. It's the uneven change in square vs. cubic functions. As a result, VE goes down on the 968 much faster with RPM.

So we need to make various and different mods to the 3.0L engine in order to rev it, you can't just take an S2000 engine and enlarge all the parts and dimensions. The way to make a 9000rpm 3.0L engine would be to add two more cylinders and keep everything the same as the S2000.This is linear increase. But to make a 3.0L 4-cyl., the exponential (3rd power) change, will need to have different mods..

M758

It takes effort, but you can rev a 3.0L motor 18,000 rpm and make 800-900 hp. Really out of our price range however.

PorscheG96

Hey Danno, how you been lately? Give me a call when you get a chance or launch me an e-mail.

Zero10

Hehehe, somebody has been watching too much F1 racing, 18,000rpm...
Just get a rotary at those speeds. I've seen one rev up to over 20,000rpm, sounded crazy. Eventually it starved for air somewhere around 22,000rpm.....

Fishey

Wow, The only reason a honda can hit its RPM numbers is because of there piston/rod weight as its only a 2.0L ... Even when honda went to the 2.2l they found that 9,000rpm was very unreliable proof that it all comes down to piston weight. Thats one reason the F1 cars use such small pistons!

Danno

Yeah, big pistons weigh a lot more. If you double the bore, you get double the displacement, but the pistons will weigh about 3x more. That puts a lot more strain on the con-rods, which will then need to be about 3x heavier as well. Which means the crank and counterweights have to be heavier, so you end up with a cascading effect and all that weight makes for a lazy-revving engine. At 7000rpm, the pistons are under 10,000G's of force, which is then exerted on the rods & crank.

The V10 F1 engines usually have a bore-to-stroke ratio of about 1.75 using slipper pistons. They're ultra low-profile to save weight, the distance from top to bottom of skirt is about 1" and that's it. The large bore allows for the largest valves possible which only has to breath enough for a very small volume due to the short stroke.

You can combat weight and improve the strength-to-weight ratio of the bottom end with exotic costly materials. I think the main limitation on the 968 engine in revs is the head and its configuration. The intake-ports needs to be opened up and the valves made bigger. The cams sucks, they're designed for low-RPM torque and the VarioCam gives a compromised setting for high-RPM flow. Better to get something with about 30-50 degrees more duration and 15-30 degrees later intake-closing (depending upon redline).

Laust Pedersen

I would start with an engine simulation program such as EnginePro , initially entering standard 3.0L engine parameters and then change camshaft and shorten manifold runner lengths.
Custom chips and solid lifters sound like a must.
Since there is more oil on the (inside) block and cylinder wall at high rpm some attention should be paid to baffle and oil level.
It may also be a good idea to put a strobe light on the belts to see if there are unwanted resonances at those rpm's.
Those heavy pistons and con-rods will generate some serious inertia forces, which only costly analysis or experimentation, which the parts may or may not be able to handle.

Enough hand-waiving, I am certainly interested in whatever you come up with.

Bengt Sweden

Does anyone here have info on pinning the girdle both ends without re-machining? I am thinking of using conical pins after assembly?

Bengt

NineMeister

Sorry Danno but I have to disagree with you completely. In my opinion the problem with the 968 head is the opposite, the ports are so large that the engine does not build up sufficient intake velocity, and without a higher velocity it will be impossible to achieve the VE (volumetric efficiency) required to get to the 100+bhp/litre that we are looking for. I doubted this theory until I built our 3.82 RSR engine with small valves and immediately achieved 103bhp/litre on the first attempt.

I have run many computer simulations on different programs and this theory is backed up by the results every time. In a nutshell: VE increases in direct proportion to the flow whereas VE increases by the square of the velocity; therefore even if you sacrifice flow to achieve a higher velocity (provided that the valve is open long enough to allow the air time to get in) you will increase VE and thus increase bmep (cylinder pressure) for more torque and power. The finite limit appears to be around 0.6 Mach (600ft/sec) with conventional head designs.

I do agree over the piston weight, but the usual trick of fitting stronger rods & bolts should keep the stock piston in check up to 8500rpm ish that we are looking for.

As an aside since you mentioned slipper pistons, the 9m 3.8 race engine runs our own design of 103mm slipper piston that is 5mm shorter than stock with a correspnding 5mm longer rod. Apart from a little slap under very hard load below 2000rpm the piston runs clean up to its 8000rpm oiling limit, proving that this kind of design can also be made to work in more conventional engines.

Danno

Colin, what software are you using? On the package I have, it took me a while to accurately model the stock 968 motor. But I finally got it to output similar dyno-sheets compared to an actual 968 engine.

With the VarioCam set for low-RPM operation, there's a tremendous drop in torque after 4500rpm, similar to the S2 engine. With it retarded for high-RPM, there's less low-end torque, but the peak is 1000rpm higher. Combine the two settings into gone graph and you get a nice wide torque-curve. But there's still a sharp drop-off after 5500rpm.

Examining the cam-profile, I find that it's one ground really for low-RPM torque operation, not high-RPM HP and the VarioCam was a compromise to get some decent high-end HP out of it. I would've prefered to grind a cam for high-RPM operation, and use VarioCam advance to compromise the cam for low-RPM torque generation while preserving its native specs for high-RPM flow. Ok, with a properly profiled cam, the 968 can hold a flatter torque-curve to the upper-RPMs and get decent HP. But it's not symmetric and drops off much faster than it builds.

What kind of gas velocities are you measuring at 4000rpm? and 8000rpm? And where are you measuring it? I may not have been specific enough in my claim. The ports on the outside where it mates up with the intake-manifold is fine, but it narrows down to the valve-pocket much too quickly and tightly. You may find that velocities there are easily 2-3x faster than at the entrance. Problem is at 8000rpm with stock valve seats, guides and valves, the velocity well exceeds your mach-0.6 figure. Increasing cam-lift has no effect on lowering this velocity at all. Some tuners actually reduce valve-lift on their high-RPm cams. I just think the 968 head can use slightly larger valves for 8000rpm operation with careful detailing of the seat and its transition into the port.

Another issue, although I've only seen it taped on high-RPM 911s, the 968 is probably similar. That's oil getting trapped inside the valve-springs at high-RPM. The quick opening and compression times don't allow oil to fully drain out and some gets trapped inside and artifically increases the spring-rate. This adds friction and wear to the cams. Don't have any specific numbers, but some people feel it's worth 20-30hp, similar to going with roller rockers. Anyway, teflon-coating the valve-springs helps to prevent oil getting trapped at high-RPMs.

BTW - all my simulations also are for using 15psi of boost as well.

Ken D

Bump.....this is a great thread...glad to see ninemeister around. I'm learning a lot!