Flowbench Manufacturing Australia Logo





We found that you need to flow an engine complete, head bolted on the block with the manifold carburettor, and air cleaner. Of course we flow em first separately. We let air into the air cleaner and pulled air out from the bottom of the oil pan. We also tried drilling the piston full of holes and monitoring the whole works. That didn't seem to teach us anything useful, but as we added components to the induction system, airflow slowly dropped - except when we pulled it through the exhaust pipes and measured what went into exhaust port with a good set of headers for that application, then airflow always rose! When you work out valve size differential in a cylinder head, the best always had bigger intakes than exhaust, and the intake port always flowed more air than the exhaust port. It's possible to have too much exhaust port and valve size and hurt an engine's performance. We sized the head and added all the components for the total induction system, and then did the same with exhaust side. Well, guess what? Airflow out of both sides got damn near even. Another point to consider in working out engine components: There is a big difference in acceleration characteristics of a carburetted engine vs. an injector pressurized fuel-delivery system. The carburetted engine is very sensitive to pressure changes in the main fuel-delivery area, and injected engines are much more forgiving. The injector gets fuel delivered by manufactured pressure. The carburetted one depends on the airflow and pressure of the differential at the carburettor fuel exit.


Let's get on with what we are trying to find out. How do we change it, and most Important, what is good? Let's consider a port that's cast in the cylinder head. Wait a minute. I need to back up. In the beginning I said we were gonna consider gasoline as the only fuel. Now I need to add that we are also only going to talk: pushrods, four-cycles, and V8s. There's not that much of a difference in other engines, but we can't get down to the nitty gritty with every type of engine, only those racers who already know the art would be able to follow it - and they ain't gonna read this anyhow. The intake and exhaust entrances and exits are called ports. The intake manifold is called the conduit runner.

The engine sees the intake manifold from the air cleaner to the combustion side of the intake valve. The exhaust circuit is from the combustion-chamber side of the valves to the end of exhaust system. So each is one long shape, as the engine feeds and exhausts, and each cylinder has a specific size. Let's say 45 cubic inches. Every revolution you are gonna fire four-cylinders. Every 90 degrees another four cylinder needs filling. So you can figure out how big a port or conduit has to be mathematically as long as you know what the ambient pressure is at the beginning of the system and what pressure drop there is across the cylinder. You know it as vacuum. Because that varies, you have to use straight math and check how many times you have to fill the cylinder a minute. Then figure out the size the runner has to be, to fill it 100%.

We know all kinds of things cause drag-pressure, differential changes in route that vary from cylinder to cylinder and that varies with throttle angle, RPM, torque, engine temperature, and local ambient air changes. This is way over the average cat's head. So we are reduced to sticking with a rough size, making it bigger or making it smaller in some cases by welding or using epoxies or resins. But epoxies and resins won't work with heat, at least I've never found any that can take exhaust temperature for 15 minutes. But, here's what we can do. let's talk intake conduit.

Bolt the manifold to the cylinder head, get some quick-setting rubber (head porting experts can tell you where to get it and it's not real expensive), take a port, say number one, tape the intake mouth of it closed (good high-speed safety tape will do), lube the inside of the conduit from one end to the other so when the quick-setting rubber sets (if mixed right, it is two or three part chemistry) the lube works as a release agent so the cold-set rubber can be pushed out yep, you might have to push like hell, but it will come out. Wipe the grease off, and then sit there and admire your invention. As a rule, they are l4 to 18 inches of the damnedest mess of size and shape changes you ever saw of anything.

Pure-bred overhead cam engines look like what you would expect, but these are the compromises you have to make to the cooling, location of head bolts, and pushrods. Also, to use a common intake manifold to eight cylinders, how do you keep all the runners the same considering where the valve bore centers are compared to where the bore centers of the multi-bore carburetor are? Now you have an idea of why it took 48 years of changes to make better heads, manifolds, and exhaust systems for the Ford and Chevy pushrod engines and why you can expect them to continue to get better the longer they play with them. You are like a juggler trying to control 24 balls at once.

Let's go back to our weird looking head manifold conduit sample. You'd like to have a shape that could flow the required working fluid to 100% and fill the cylinder every time from idle to 9000 RPM. Can't be done. At best, you could do good for a range of 1500 RPM, for real good only 500 RPM, but to be perfect, a range of 5 RPM is it. You are always either too big or too small except for that lousy 5 RPM, so you have to compromise.

On a short track, shoot for torque off the corner; on a long track where RPM hardly varies, go for horsepower. Everybody who runs from second on back is short of horsepower, right. Now, the above is in reference to acceleration. As you look at that rubber imprint, mark it off every 1 inch of it's length, as accurately as you can. Then measure the size of that 1 inch, and figure its area. If you can't measure it because of its complex shape, take a hot wire or rubber knife and cut the whole damn thing into 1 inch sections, marking them 1 through whatever. Put them individually in a calibrated glass of water to see how much they displace and record the measurement.

Now I've got to review the Bernoulli effect. Every change in shape or size will affect the velocity of the working fluid through that specific area. You say, I already knew that. But as you increase velocity, you have to pay for that energy some place. Every time you reduce the velocity, you waste some energy that you've already paid for. And here's how you pay for it: It takes 'bout 240 horse-power to turn a Winston Cup motor 8400 RPM. That's damn near all valve spring, compression, and expansion of the working fluid. Rings and bearings are only a small fraction of the friction. In a four-cycle race engine, the entrance and exit of a conduit is critical for maximizing flow. The space between is only one given size. For example, does a plumber use tapered or multi-size tubing to go from point A to point B? Nope. Now you can change shape and size of a given conduit and maintain a given size, but the mass flow will drop because of the energy wasted in varying the velocity.


Well, the changing world is giving us a lot of help. There are now cylinder heads and intake manifold designers who can put all of this in their computers and come up with a new port, - so far these aren't worth a damn either. But there are port designers, sharp independent porters like Mike Chapman of Salt Lake City, Robert Yates and his crew in Winston Cup, and four or five others who can get manifold and ports about right. A genius named Kenny Weld from Kansas City has designed programs and machinery that can machine ports so that they're all exactly alike. They're ready to run with no handwork except for the valve seat, and at same time, he'll machine all the combustion chambers within 0.0002 or 0.0003 cc of each other. Robert Yates and Ernie Elliott have machines that do almost all of it and do it well, but Kenny Weld was the cat who solved the one time impossible task.

Why mention it? Well, with this equipment and knowledge, the cost is gonna drop like a rock so more of you can afford it. Let's go back to our rubber part again, it's now in a bunch of pieces. Can we reduce any of the bends? Can we change the size in various places to reduce the number of velocity changes in its passage? Dealing with stock parts, as manufactured, there is untold room for improvement, but with high-buck after market parts, it don't come easy anymore.

Working with a flowbench is a cut and try deal. Half of the improvements were accidents. It's not possible to see with the naked eye dimensional changes and estimate angles, so you have to make a set of patterns for every inch and blend it in inch by inch. A good check for size is to cc the ports.

It's also important to consider that just having a cylinder head near perfect don't mean anything unless the intake manifold is matched to each port and the total flow of all eight Systems is equal. What I'm getting at is that it's like making eight exhaust pipes the exact same length, but some will have more and meaner bends. To have even length is great, but even flow is the real answer. Same with intake and exhaust flow.


In porting stock parts you gain very easily, but with many man-hours. As a rule, you will need a large collection of cutters for grinding cast iron. Buy at least carbide grade, the cheap cutters ain't worth a damn even when they're new; they don't cut and they're dull in nothing flat.

In porting stock parts you have to remove a lot of material and most times you have to add filler to bring up the lows. That only works on the intake; the exhaust side is too hot and won't stay. Watch your step going through to the water jacket. You can tell when you are getting in trouble by the sound of the cutter. The best method is to get a sample cylinder head and a sample intake manifold. Take them to a good "Do-All" saw and slice 'em up 'bout every inch and look them over before you take off with the wild-assed grinder in hand.

Before you fire up the grinder, be damn sure you've got very good eye protection, and a smart move would be to wear a good respirator and take the time to light the work area well, don't be stupid like I was. I supported an optician and eyeglass maker for 30 years. I'm lucky I ain't blind. Plus, it hurts like hell till they dig the metal or rock out of your eye.

Don't try to weld cast iron, ain't over 20 guys in the United States that know how. Aluminum is easy to weld but that takes a lot of know-how, and after you weld, you must reheat-treat a cylinder head. A welded head or manifold, iron or aluminum, jumps all over hell and has to be re-machined, and as a rule they shrink, and ports and bolt holes need work to realign. If rules permit, you can rework an iron manifold intake, iron exhaust headers, and iron cylinder heads, and easily pickup 100 horsepower on a 350-cubic-inch engine. As a matter of fact, in the last 30 years, a Chevy or Ford 350-cubic-inch engine got 300 of the 700 horsepower from cylinder heads, manifolds, combustion chamber shape, piston design, and with less camshaft. In a very stock class with iron heads, there is 'bout 30 to 40 horsepower to be had with the best valve and valve seat preparation. In this case, you deal with the seat area below differently than you do with a premium cylinder head.


In airflow and a poppet-valved four-cycle engine, I can't describe how much that damn valve butts airflow and the problems it causes trying to get the proper amount of working fluid evenly to all cylinders. New high-buck engines have variable-length runners, and varying numbers of valves for various RPM or power modes, and variable cam action in lift and timing. Well, you know the air can't go through the valve, it has to go around it. It's amazing how many experts worked on airflow for 20 years before they included that in their plan of attack.

To see how air flows around a valve, get a ruined good flowing aluminum cylinder head. Cut a 3 5/8 inch circle off the center of the intake valve, clean through from the head surface to the top of the head. Machine it down to 3 1/2 inches. Epoxy it into a 3 1/2 inch ID clear plastic tube 'bout 1 foot long. Now cut a hole through the plastic tube where the intake runner adjoins the tube. Add a manual screw-type lever to open and close the intake valve with a weak valve spring, 30 pounds on the seat is plenty. OK, when the valve is open to say, 0.5OO inch, you got a circuit. Intake port (what's left of it) to and around the intake valve ending at the bottom of the combustion chamber and into the remaining 6 inches of length of the 3 ½ -inch plastic tube.

Now let's epoxy a cover over the port orifice. Cut into it and epoxy a garden hose fitting to it. We want to run water pressure through the port and valve and exit into the combustion chamber to see what happens at 0.050 to 0.600 of valve lift. Most water pressure is 30 to 60 PSI, so we need to have a pressure regulator and cut it back to 'bout 5 PSI so we can vary velocity to look for changes. Water will do exactly what air does. You are gonna notice the water comes around the valve from the bottom of intake port where guide ends, to some distance after the valve head in combustion chamber. The shorter the cone the better the flow; the higher the pressure, the more defined the cone is. A perfect 360-degree even cone is affected by shrouding. Anything closer than a 1/2 inch to the edge of the valve slows flow down. And it's possible in heavy shrouding to only flow about 300 degrees around the valve and the other 60 degrees just bubbles, rolls, and twists. You'll also notice, that not a drop of water hits the lower inner area around the valve stem, or for that matter, over a 1/4 inch (at worst) inward of the valve, and you'll notice on the bottom side of the valve combustion-chamber side that it isn't touched by water either.

So you see, swirl in the port and whatever has been done to the valve or valve exit in the combustion chamber is overridden by the huge pressure drop at the valve seat caused by the tremendous increase in velocity. Remember, before that valve cracks open, the air column's velocity was at zero and pressure was at the maximum. That tremendous flow acceleration, accompanied by the sudden steep pressure drop, overrules all other physics. As the valve opens farther, velocity starts to drop and pressure starts to rise, but the kinetic energy induced by the start of the event remains in total control.

There is more to it. The intake and exhaust are open together quite awhile and the temperature and kinetic energy of the exhaust gas cycle play a part in the total result, and so does the pressure in the combustion chamber. I agree, it's getting deep now, but again, what has to be considered is whether you want to port knowledgeably or blindly. We've played the game a "blind hog gets an acorn every once in a while" for too long already.

What's the best combustion chamber? A flat-bottom head with no combustion cavity, no shrouding, and valves inline with pistons. What's the best port angle? Shallow as you can get-the top of the port up against the valve spring seat. Take any port shape, straighten it out all that you can, and for every degree you straighten it out, with no other changes and for every degree you straighten it out, the more you flow. That's great for airflow, but the compression target has to be addressed.

Piston shape has to be considered. The best piston theory-wise is flat with a minimum surface area. In real life, the best is 0.5-inch around the bore, 0.030 to 0.040 inch clearance to the flat head area, and a concave inner shape.

But how about valve clearance? Also, it's best to have straighter ports and raise the top of the engine. Can you close the hood? What's best valve seat angle? Probably 45 degrees, and remember, you do better with sharp, short angle changes than with pretty radii, but only when turning over 10 degrees. If the run is essentially straight, radiuses that are aerodynamically correct will increase velocity and drop pressure economically enough to affect a net gain in airflow.

In production engines there are two concepts argued all the time: Is it better to straighten the bore to valve angle and increase the port angle, or to decrease port angle and spread the valve to piston angle wider? The first is Ford, the second is Chevy. I'll tell you this: The combustion chamber probably still has 50 horsepower left in it when the piston and combustion chamber are maximized. Consider this: When you react the working fluid, all entrances and exits of the chamber are closed. If you start with a 14:1 air/fuel mixture, when you have consumed half of the charge, you have polluted the remaining half of the air/fuel charge with reacted gasses and the mixture is now 7:1 too rich to support combustion. It turns out, reacted working fluid weighs less than it did unreacted. So if the original charge is spun (as in a centrifuge), the reacted and unreacted would never mix, Therefore, you can react the total volume of the working fluid and pick up another 50 horses. I have a patent on this. I have run it, but I'm sure we'll hear more about this later because there is always more than one way to skin a cat.

Here's a good experiment for a first grade lesson in airflow: Go to a clear creek, stream, or river if you can find an unpolluted one that has a fair flow speed with rocks on the bottom, Notice the rocks are worn in various smooth shapes. Now reach down to the river bed and rotate one of the rocks 90 degrees. Notice how turbulent the water is now, compared to the way it was before you disturbed it? This will show you what happens with clear air, it's the same thing. Therefore, if you experimented with see through parts and added colored smoke to the air when doing flow tests, you could see what's happening as well as you can see what's happening in the water reaction. This would also show you that a better plan for doing aerodynamic studies on a car body would be with a water tunnel rather than a wind tunnel, and that colored smoke is much better than clear air.

For those of you who know enough about airflow, I'd recommend that you buy a flowbench if you can afford it. I think there are newer ones coming that should put a lot of the present and used equipment on the market, inexpensively. There is a mystery we've been aware of for about 15 years: Even in a very good running engine the cylinder firing pressure varies, even at steady state, RPM, and load. That may or may not turn out to be an airflow problem, but more likely it has to do with cylinder scavenging and the state of homogenization of the working fluid or the ignition system. If you have, make, or buy a flowbench, do your testing at a lift close to the maximum of the cam you intend to use.

The newest airflow deals that apply to even running engines (such as those I have) give you the opportunity to use gasoline, so there are no compromises in wet flow for safety (in a running engine). I haven't brought this up yet, so let's cover the fact that multi-cylinder engines don't operate with all the fuel vaporized. It turns out that in a V8 manifold of clear plastic, along with the vapor, there are almost always eight little rivers of liquid fuel going into the combustion chambers why? Because vaporization is caused by pressure and temperature drop. There are no means to attain super vaporization (as in an air conditioner), so some vaporized fuel goes back to liquid before it is reacted. Why? Velocity, temperature, and pressure changes caused by the shape, temperature, pressure differences, and unlaminar flow in the length of the conduit.

To think out a proper starting point for manifold design, do the math. If RPM multiplied by the number of cylinders, multiplied by the cylinder size says you need 810 cubic feet of air per minute in order to fill the cylinders 100% at 7500 RPM, you have 15 psI ambient pressure, and you average 2 inches of manifold depression each, the conduit has to flow approximately 100 cubic feet per minute. Figure

Remember, you have a valve that's not wide open all the time. For example, in my flowbench we have to get around 280 cubic feet of air at 0.600-inch lift to be in the ballpark for 8000 RPM for a 45 cubic inch cylinder. I'd advise you not to try to design a head or manifold from a clean piece of paper unless you've spent at least 10 or more years of your life working and studying engine airflow.

Because most of you are interested in carbureted engines (two or four barrel) as you work in the plenum of the manifold or the port entrances, consider that all two or four air columns must be bent or twisted to fit each individual cylinder entrance. That's how it works. That's the biggest reason for carb spacers. Also, carb bore spacing is critical to how well this can be done. Wide bore spacing carbs are much worse to deal with because of increased turning angles and the long distance from the source of individual columns to serve a given intake port. There is a possibility that chemical milling, such as Extrude Bone, may be very beneficial, particularly in low-buck porting. It's not as good as Kenny Weld's port milling, but it allows you to get into areas where no amount of work can be accomplished by hand. I used a homemade deal on the order of Extrude Hone 40 to 45 years ago and it helped like hell. Extrude Hone has controls I didn't have. I had to really watch that we didn't eat through the water jacket, and sometimes we did. I was involved with the Mexican road race in 1952, and at a jungle town at the race start, I watched an 18-year-old Mexican boy start to straighten a Cadillac front bumperette that had smacked into a mountain. He cut the damn thing in four pieces straightened each out, welded them together again, and then ground, dressed, and plated them. You couldn't tell it from new. He taught me, without saying a word, how to cut a cast-iron intake manifold apart and oven-braze it back together again. After I did the trick to the inside, I could then reach it.

Well, I'm sure you have questions. Some of you got lost, some of you have specific questions 'cause there was something I didn't cover, or I was unclear. Pick up your pen, or better yet, sit down at your typewriter or word processor and ask your questions. Maybe we can fill in all the blanks. Most of all this is a big topic. Nobody is gonna get it all in one shot. From time to time, CIRCLE TRACK can get other input from guys who are hitting the home runs today.


fma logo grey

Copyright © 2020 Flowbench Manufacturing Australia. All rights reserved.