Strosek Ultra

A tribute and many thanks to Rob Edwards who helped me getting a bottle of manometer fluid for the flow bench. The guy who used to make the flow benches over here passed away years ago. I did not know where to get more gage fluid having the right specific gravity. Searching online I found the right kind of fluid available to be ordered from Amazon but shipped out from Dwyer Instruments. When ordering Amazon replied that Dwyer do not ship to Sweden. Unbelievable - what to do? I asked Rob and he was pleased to order a bottle of fluid to his address and then ship it on to Sweden. I am so grateful he helped me out of this awkward dilemma.
Thank you Rob.
Åke

Strosek Ultra

I have always wanted to know how accurate my air flow numbers measured at 10" of water and corrected to 28" are compared to numbers taken on a large flow bench working at 28" of depression. Tuomo has been so very kind and provided me with flow numbers for an all stock S4 intake port measured at 28". The flow test was carried out by Don Redmon at RMI. Compared to the result of the RMI flow test I can see my numbers are a bit low. See below.
Åke

Valve lift: 050 100 150 200 250 300 400 500
RMI: 41.7 104.7 149.2 198.3 239.2 269.6 296.9 301.0
Åke: 44.0 92.0 141.0 188.0 237.0 268.0 292.0 299.0

GregBBRD

The conversion math used is very accurate.

Most of the differences in flow are from small differences in zeroing the manometers and slight inaccuracies in the valve opening measurements, temperature correction, and changes in the actual suction that the machine can develop (supply voltage variances). 3-5 cfm from day to day, much less from machine to machine is very good!

Testing at 10" has always seemed more realistic than testing at 28", to me. The internal combustion engine that can "pull" 28" of vacuum at low lifts is non-existant. And in my humble opinion, gains in lower lift air flow and velocity are the only thing that helps the 928 head. The stock head flows enough air at higher lifts to produce over 500hp. Trying to increase airflow at high lifts is only reducing velocity and thus cylinder filling effiency throughout the rest of the lift range.

There were some recent airflow results posted, which were hilarious, and showed the complete lack of fundimental understanding of what is needed to increase the horsepower from an internal combustion engine. The high lift gains were completely moot. The increase in airflow would only be usable at rpms the 928 engine could never reach! Big valves and huge ports that provide gains at high lifts just reduce velocity through the ports...actually hurting power output!

All of my head work is targeted at lower lift airflow increases....increasing the ability of the engine to draw more air on the smaller lifts. A 10%-20% gain at those lifts, with increases in velocity, results in significant horsepower gains!

Increased displacement engines have the ability to draw more air....since the swept volume of the cylinder is increased. This does absolutely no good, if the airflow past the valve, in the usable lift areas, is increased, at the same time!

Go back and look at the lower lift airflow results for the recently posted big valve CNC ported heads....Zero increase in airflow. Only increases at the higher lifts (why they tested the heads above .450" of lift....when there is never going to be a 928 cam with this lift, shows the basic lack of understanding...a complete waste of time and money!)

Before anyone makes changes to an internal combustion engine, in an attempt to increase power output, they have to understand the basic facts of engine design. These heads show a complete lack of even a tiny bit of that understanding.

A completely stock cylinder head, with more attention paid to the valve and seat area would be much better than these heads....velocity through the port would be higher, lower lift airflow would be increased....and the stock head will allow 500hp to be produced at higher lifts.

^^^^^^^^^^^^^
The above is just my opinion, backed up with years and years of R&D, years and years of study regarding airflow, and the building of hundreds of naturally aspirated engines, with the impressive results from those engines. I bought my own personal flow bench (and have virtually worn it out) when I was 17 years old....almost 50 years ago!

V2Rocket

this is about 2 years too late (post# 238), but you used the wrong number for "rod length" in this simulation...your rods were 8mm short in this simulation. could affect piston speed and port velocities and stuff.

ptuomov

I agree with much of this.

The square-root formula for correcting the flow bench depression is very accurate as long as the the flow velocity is under 0.3 Mach and the flow is the turbulent regime. The second is only an issue at low valve lifts; in my opinion, one may want to up the test pressure a bit at low lifts since the bench can deliver that easily.

What's realistic pressure differential in a running engine? My understanding is that a normally aspirated engine with headers tuned well will pull the highest vacuum during the valve overlap, and that can easily be over 50 inches of water. The four-stroke engine is like a crack *****, she lives off the pipe! After the exhaust valve closes, the next relatively high point in vacuum is shortly after the maximum piston speed crank angle around 75 degrees ATDC, but I'm guessing those numbers are closer to say 15 inches of water.

In my opinion, if one wants to most closely replicate the intake-valve conditions of a running engine, one should just set the flow bench to run the highest vacuum that it can very consistently run at that valve opening and then convert to any test pressure using the square root formula -- I don't see how one could go wrong there. But because the conversion formula is accurate, the test pressure shouldn't really make much of a difference at all on the intake side.

The number of 928 engines that can use the high-lift flow capacity of big-valve 928 heads are rare. Mike Simard's engine is one, and that's 7 liters and makes peak power near 7500rpm. When Ake finishes his large displacement engine, that'll be another. And cams have some significant lift in both cases.

The reason why the 928 platform is so seductive yet so treacherous is that the heads have flow capacity for big power with large displacement and/or high rpms, while the stock bottom end has relatively low displacement and needs a lot of new components to survive at high rpms. Showing high head-flow figures is tempting, and it's a natural temptation to try to get a high number there -- when in reality the binding constraint is the engine's inability to safely use the existing head flow capacity. Maybe we should all start quoting MCSA's and CFM/MCSA instead of just CFM to get the bragging rights better aligned with reality?

Another symptom of the intake port flow not being a constraint is that we aren't seeing normally aspirated stock displacement S4 responding to overall cam advance or retard much at all. If the intake port would be a restriction, my logic would say that the torque curve would rock with overall cam advance/retard. I'm not seeing much impact at all.

If I were to increase the displacement and/or rpms of a normally aspirated 928 engine, I would likely be trying to keep the minimum cross-sectional area in the port about the same as it is in stock S4, not increasing it. That's a lightly held opinion, since what do I know about that... (I have a stronger opinion about port sizes for turbo cars: I think that as long as the heads have relatively straight ports, I don't think turbo cars need any larger ports than normally aspirated cars making 1/3 the power!)

I think that the jury is still out there on whether high low-lift flow is good or not, if the cams can be freely chosen. It seems to me that more low lift flow makes the engine behave the same way as adding camshaft duration would. If the low lift flow is low, one can just increase the camshaft duration and one can theoretically get a more "digital" flow profile for the intake port. I think (but don't know) that the key issue for a street engine is how well the heads flow in the wrong direction when the overlap pressures aren't favorable. With stock cammed S4, my guess is that more low-lift flow is ambiguously good for high rpm power as it's kind of like swapping in longer duration cams.

Just in case this post is written in a way that appears (over)confident, I need to make sure that every reader understands that what happens inside the engine is largely just a total mystery to me. The above are (honest) opinions but not facts.

In any case, to take this to a more practical level, the ported heads on the blue engine that we're currently testing have stock size valves. The porting prioritized getting flow velocities to be even within the intake port, on all sides. The whole circumference of the valves was unshrouded slightly if there was an obstruction. The intake valve edges are as far from the exhaust valve edges as in the stock head. The throat-to-valve-diameter ratio is only slightly higher than stock, in fact very close. The port minimum cross-sectional areas weren't really increased on either intake or exhaust sides to keep the velocities. I'm not saying that I'd have them done exactly like this if I knew I could get any kinds of cams made, it's just that this is how they were done and it made sense to me at the the time. Low lift flow is certainly pretty high, whether that's good or bad.

All these design features attempted to produce a port that doesn't flow as much as it could at 12mm lift but flows very well at 8.5-9mm lift (the lift of the installed cams at 75 degrees ATDC) and _hopefully_ doesn't flow very well _in the wrong direction_ at couple degrees ATDC when both intake and exhaust valves are 1mm open -- if the turbo isn't cooperating... So far, it seems to run well anecdotally at 10psi boost, but of course many other things were changed at the same time so who knows why it runs strong?

ptuomov

Must have typed it wrong into pipemax, 150mm is 5.9055 inches. That makes some difference to the piston CFM demand curve in pipemax I think. I think the prescribed areas aren't that much impacted, but I'd have to check.

ptuomov

An old example of a higher flow bench CFM reducing power:

GregBBRD

Very good ideas in here!

I've added some notes, in blue.

ptuomov

On the exhaust scavenging effect starting the intake flow: I think it depends on the exhaust pipes (timing and strength of the low-pressure wave reaching intake), exhaust valve size (strength of the low-pressure wave), exhaust valve opening time (timing and strength of the low-pressure wave), exhaust valve closing and intake valve open times (timing of the low-pressure wave), and combustion chamber volume during overlap (strength of the low-pressure wave).

The overall cam advance and retard, which is related to the piston position, doesn't really in my opinion impact the scavenging effect that much, but I might be wrong. The reason why I think so is that while the combustion chamber volume is at minimum at TDC ignoring the pin offset, it's only about 7% higher at +/- 10 degrees from the TDC for the stock S4. Things like primary diameter and the length of the valve overlap period are much more significant, by my guess anyway.

All this logic tells me that if one plans to run a lot of camshaft overlap, then one should have a relatively small exhaust valve, small exhaust port, and small exhaust primary diameter. That'll produce a strong wave, and that wave has to be timed right. A random thought: I think Greg you're working on Vanster's S3 based stroker motor, and I think that with the small exhaust valves and exhaust ports of S3 head, it might run really well with small-primary headers, very low restriction exhaust, high compression ratio, and a big cams with a lot of camshaft overlap.

Ake who's contributed a lot to this thread has tuned a very large number of motorcycle engines, and many of them make their money during the valve overlap -- see Hayabusa IVO point at the end of the post. Furthermore, the experiments with motorcycles are so much cheaper, so he's probably run every possible exhaust-camshaft combination on those. He may want to correct or clarify what I wrote above.

A similar sort of tuning effect is there at the intake side. During the IVO-BDC period, the exhaust and piston work to accelerate the intake charge to some velocity. By designing the port to be large, you'll accelerate the intake charge to a lower velocity and reduce the pumping losses. Conversely, by designing the port to be small, you'll accelerate the charge to a higher velocity and increase the pumping losses. The benefit can be reaped in the BDC-IVC period, when the work done during the IVO-BDC period can be harvested into a better cylinder filling. Higher the velocity, later you can productively make the IVC. It's like running a supercharger, it'll consume some power but pack more air mass to the cylinder. The factory isn't going to run big cams and high static compression for many reasons, including torque off idle etc. So in their calculations they like big intake ports, because the pumping losses (and thus fuel consumption) are lower, and they wouldn't really use the benefits of the small intake ports with late IVC and high static compression. So they always go for big ports for show-room stock street cars. This is why I think the 968 has those ridiculously large ports.

I do want to challenge the claim that larger valves in the 928 head don't flow more at low lifts (say 0.05") even on a 100mm bore. I think that at very low lifts, the 39mm valve is going to flow more than the 37mm intake valve, unless something is done to prevent that (such as leaving the larger valves shrouded). The curtain area is 5.4% larger, so at very low lifts if the seat is cut the same way the 39mm is going to flow more on the flow bench. So here we might have a genuine disagreement, or maybe when we're talking about low-lift flow we simply mean different valve lifts.

Now, the 39mm intake valve head may or may not make more power. If the cams have too much duration to start with and the port flows more than enough at high lifts, it's probably going to make less power, holding everything else constant. If the cams have too little duration, then it's probably going to make more power, holding everything else constant. And how large the minimum cross-sectional area is and how efficient the port is in general also matter a lot once you get in the 0.10-0.25" lift ranges, so everything else is never constant. In my opinion (not to be confused with a fact), it's not so much the high low-lift (0.05" and under) flow than the high mid-lift flow (0.10-0.25") that's beneficial for power production in a 928.

A note on opening the intake valves early on four-valve heads: The blue turbo engine has the intake valves timed to open (at @ 0.008") conservatively at 16.4 degrees BTDC. Simard's 7.0L 928 four valve engine opens the intake (at seat) at 39 degrees BTDC. Four-valve head Suzuki Hayabusa that runs well with turbos on stock cams opens the intake valve (at @ 0.012") more than 40 degrees BTDC. While these are nothing like the IVO numbers for two-valve heads, in my opinion some 4-valve engines seem to run very well with the intake valve opened quite early. Probably not 40 degrees BTDC at 0.05" for realistic 928 engines, though...

Carl Fausett

What I see missing from many conversations about larger ports and runners on 4v 928 heads is the increase in air flow by adding boost. I did not read this whole thread in depth - so it is possible that I missed an element of this discussion that may have included it.

My point is: when you add 1200 CFM of air flow above atmospheric, the calculations for optimum tube and runner size change dramatically. Add to that an icrease in CID, and you really need all the flow you can get your hands on.

What I do see mentioned is that careful back-of-valve porting and reducing the radius of any curves with an eye for NOT increasing the whole runner diameter is absolutely correct. In fact, we have learned that a round intake runner is not optimal, as the floor of the runner at the apex of the curve flows better when that section is flatter. An experienced head porter knows this.

Those really big heads of ours - the ones that are CNC machined - are meant for highly modified engines with boost. 6.5 or 7 liters, plus boost. They can carry those runners sizes well, while I would expect those ports to perform poorly on a 5.0L NA motor.

I have pics of the ovate-type port and the large CNC port to compare against it, and as soon as I have some keyboard time I will post them.

Carl Fausett

One thought I just had before I disappear into the shop... as luck would have it, I should have two 6.54L 928 engines going to dyno this month. They are identical in bore and stroke.

One is built for NA, and the other is built for boost. To that end, the differences between these two motors is the CR, the cams, and the type of porting that was done on them. Where the boosted engine has the big CNC porting job, the NA motor has a much more refined porting job done to them specifically in an attempt to not over-port them and keep fluid velocities high.

It will be interesting to see the two compared. Unfortunately, one will be measured on a chassis dyno, and the other on an engine dyno, but I cant help that.

V2Rocket

how does this apply to the 928S3 (85/86 32v) heads?

Carl Fausett

That's a great question. I have long been of the opinion that the S3 intake is the best flowing intake manifold system Porsche ever made for the 928. Were it not for the difficulty in servicing the ignition wires and spark plugs it might have survived longer. To me, it looks like that pipe-organ manifold was a solid attempt at best flow possible no matter what.

That said, it may have just been a stop-gap solution until they could get the complex casting completed for the "compact" intake was see on the 87 and-up motors.

Anyway, back yo your question. Do you know that the ovoid cross-section of the 85-86 intake runners in the heads is 90 degrees off when compared to the ovoid shape of the 87-95 heads? This is why a 85-86 owner cannot just bolt on a late-model intake - the port will not line up and they are of a different shape.

I wish I could tell you I have done some flow-bench and porting work on the S3 heads, but I have not had any reason to as yet. It would be interesting to see how they perform compared to the S4-type heads.

V2Rocket

i've got a pair of S3 heads...i only need one for my project.
ill send you one to practice on, if you apply the lessons learned to my other head and send it back at no charge...


i've seen it tossed out around here that the narrow/tall S3 port might aid in low-rpm power production and improved mixture vs the S4+ style.
though when looking at ALL OTHER DOHC engines since the 80s they all use short/wide ports like S4+ rather than "round" S3 types.
my thinking it it was a sort of "in between" stop-gap while they sorted out how to really build a DOHC motor.

ptuomov

Whether big intake ports are good or not for supercharged engines is one question, and opinions on that vary.

However, I think the logic of ignoring the density of the charge is erroneous. If the port is straight, it doesn't _need_ to be any larger under 2 bar boost than normally aspirated. It'll flow about the same number of CFM -- and 3x SCFM and mass -- just fine. Upstream of the compressor you need about 3x cross-sectional areas, but not downstream of it. That's how the world works, and if this is not the starting point of the logic then the results of the applied logic are right only coincidentally.

It's a separate question whether there's any penalty from too large intake ports with supercharged engines. If it's a positive displacement belt driven compressor, I'd say not much. Maybe the shortest runner and biggest port work just fine in those applications, because they minimize pumping losses. Why run a "wave supercharger" on the intake side when you are already mass flow limited by the actual positive displacement supercharger?

With a centrifugal belt-driven supercharger, there's more logic to running smaller intake ports, and with a turbo even more, as long as the compressor isn't at the sonic choke that limits the mass flow. Still, a large intake port is probably not imposing a big penalty to filling at high rpms and there's some less pumping losses so who knows how it nets out. With a turbo, the main reason why factory street cars run smaller, not larger, intake ports and valves on turbocharged version of the same engine is probably that the engine needs to produce torque before the turbine has spooled.

On the exhaust side, with a turbo the dumbest thing you would do is making the exhaust ports too large. You need fast exhaust ports, and correctly sized turbo. Too big exhaust ports or primary runners are just going to make everything a little bit worse, or a lot worse, and a short cam can only compensate to some extent for a too large exhaust port. For belt-driven supercharger cars I'd guess this depends a lot on the camshaft specification, but I'm not sure.

For supercharged high boost engines and turbo engines with significantly higher intake pressure than exhaust pressure, the porting needs to take into account prevention of crossflow from intake valve right out to exhaust during the overlap. Looking at various 39mm intake valve heads that have been posted here, clearly some of them have made efforts to reduce crossflow (and reversion) and other have not. I think that this should be one of the main areas of focus for supercharged heads, because it allows one to run more overlap and to get the associate benefits. The worst case scenario is that the fresh intake charge flies straight out the exhaust port while the residual exhaust gas stays in the cylinder. The best case scenario is that the fresh intake charge pushes the exhaust gas out the exhaust valve.