Understanding Cylinder Head Flow Testing Procedures

Before getting into the subject of physically modifying cylinder heads, let’s look at the history and basic logic and techniques involved with establishing a cylinder head’s number-one criteria: airflow. A working knowledge here allows a better understanding of all the head characteristics we are attempting to improve and how they may each affect the others involved. Make no mistake; in the quest for optimal head configurations, you need to make many compromises.

This tech tip is from the full book, DAVID VIZARD’S HOW TO PORT & FLOW TEST CYLINDER HEADS.

For a comprehensive guide on this entire subject you can visit this link: LEARN MORE ABOUT THIS BOOK HERE

SHARE THIS ARTICLE: Please feel free to share this post on Facebook / Twitter / Google+ or any automotive Forums or blogs you read. You can use the social sharing buttons to the left, or copy and paste the website link: https://musclecardiy.com/cylinder-heads/understanding-cylinder-head-flow-testing-procedures-part-2

The information covered in this chapter will help you make many better decisions down the road. As you have probably guessed already, measuring cylinder heads’ capacity to flow air is our number-one priority. After that, there are a number of secondary but none-the-less vitally important factors we need to measure and manipulate in the manner most beneficial to us. Such factors include port velocity, velocity distribution, wet-flow characteristics, swirl, and the like. All of these factors tend to start at one focal point: the standard pressure drop.

Fig. 2.1. Why the crosshairs? Simple. Getting it right means making it happen here, and by “happen” I mean cylinder pressure. To produce the highest cylinder pressure means feeding the cylinder with as much air as possible. Next, the charge has to be effectively burned and then dispelled through the exhaust port as efficiently as possible. The red dot represents a prime position for the spark plug tip; the green one is poorly positioned for a high-compression engine.

The Standard Pressure Drop

Air is an elastic medium and, as such, its rate of flow has to be measured under specific conditions. An analogy of where I am going here is to imagine you have to measure the length of several different elastic bands for comparison. Just how long these elastic bands are depends on how much tension they are experiencing. To make an across-the-board comparison, we need to have a fixed amount of tension. In this case, we need to know how long these bands are when stretched by a tensile load of 2 pounds. In this instance, we call that our standard load.

By adopting a standard load we are eliminating the principal variable that changes its length, without actually involving any properties the band may have in the first place. Much the same situation applies when it comes to the measurement of airflow. The greater the suction on a given orifice, the greater the volume of airflow through that orifice. To make a comparison across the board for a number of different orifices, we have to fix the amount of suction, so that it is the same for each orifice tested. Figure 2.2 shows why we must, for comparative reasons, have a standard pressure drop or, at the very minimum, quote it on our test results. As of 2012, common practice when testing cylinder heads is to use a fixed “standard pressure drop” throughout the valve opening range tested. Although going this route may make the comparison of results fast and convenient, it can also, as we shall see, compromise port development.

Until the 1990s, when the International Automobile Federation (FIA) let the reigns go on naturally aspirated F1 engine development, it seemed the heyday of rapid piston engine development was during World War II. Back then, each engine manufacturer used a test pressure drop of its own choosing, so there really was no universally accepted pressure drop.

As postwar civilian companies, such as We slake in England, got into flow testing for engine development, nothing much changed. When it came to test pressures/vacuums, it was each to its own. But then along came the ever-innovative hot rodder Henry “Smokey” Yunick. Smokey was not only something of a rascal (and, in performance-automotive circles, a famed one at that), but a rascal who was always pushing boundaries when it came to making power from piston engines. I knew Smokey personally, but since his passing in 2001, I wish I had taken advantage of numerous opportunities to spend more time with him. Smokey was one of the great performance-engine innovators of the later part of the twentieth century. He was also very vocal about his theories and findings and any other of his opinions he thought you should know about.

Fig. 2.2. These curves show how the airflow varies with test pressure. To make viable comparisons, we need to test at or near the same test pressures. Although corrections can be made, the bigger the correction, the greater the possible error produced.

Fig. 2.3. An illustration of a simple manometer made from a hose, a tape measure, and a piece of 2×4. This shows the usual fixed pressure drop used when testing is at 28 inches, although there are good reasons for not going this route.

Fig. 2.4. In 1958 this was the combustion chamber that caused me such a dilemma. Hailed as a tech wonder by the press of the time, it proved to be in error by the amount shown in green. Weslake was a pioneer in the use of a flow bench, so why the engineers made it the way they did is still a mystery today.

Fig. 2.5. The current standard for the industry is this Super Flow 600. This company makes a range of flow benches; the smallest and least expensive one is the Super Flow 110.

Fig. 2.6. An AFR small-block Chevy combustion chamber. This shape is the result of flow testing and then qualifying it on the dyno.

He was the kind of a guy who devised equipment to test theories almost at the drop of a hat, if it looked like it might bring about any kind of progress. After utilizing any possible advantages, he was ready to share what he found with the rest of the performance community. It’s worth mentioning that, apart from having a must-read autobiography, Smokey did a book for Car Tech. Here, I can say with all certainty that you may want it in your auto-tech library.

Anyway, returning to the subject of the standard pressure drop, Smokey built a number of flow benches and his final one was a monster that did a whole lot more than just deliver CFM numbers. Early on, Smokey made it one of his missions to find out which standard pressure drop value, in a meaningful manner, translated observed flow improvements to possible power improvements on the dyno. Starting at a relatively low test pressure, he ran the gamut on this and declared that tests had to be done at a minimum of 28 inches of water (H2O) depression for a flow bench improvement to consistently show up as a power improvement on the dyno.

The majority of engine builders in the United States held Smokey in high regard, and he was also widely read, so this 28-inch measurement became an accepted standard for the performance-engine building community. So when you see airflow test results in various publications, and in my books, you find that the test pressure at which the flow is quoted (as opposed to actually being measured at) in the charts and graphs is with 28 inches H2O pressure drop.

So was Smokey right in claiming that at least 28 inches was needed for that meaningful translation of results from the bench to the dyno? Well, although test results can be converted from one test pressure to another, the results, when corrected from a lower to a higher test pressure, don’t quite tally (for reasons I address shortly). Certainly, there was more than enough justification for Smokey’s claims here, but it is still a few steps short of what we may do for optimal port development. What I want to do here is to address how relevant a steady-state flow test is as far as correlating it to the pulsing flow seen in a running engine.

Let’s go back to when the other guys kept telling me a flow bench was not even worth its weight in paper because of the difference between steady and pulsing flows. The difference in flow seen on a bench versus that within a running engine is dramatic. At first sight, this makes the other guys’ argument appear very valid, but the following was my counter argument on this subject: “If we open a valve and test the flow at whatever pressure drop is chosen, we can say that in a running engine the flow will, for all practical purposes, be steady if it is considered over a short enough period of time. In other words, if we looked at the flow past the intake valve for a period of, say, one millionth of a second, the flow at the beginning of that period essentially is the same as the flow at the end of that millionth of a second. So we can say, within close limits, the flow in a running engine, when taken over a small enough time increment, is virtually steady. The implication here then is that our flow bench, on this score, does reasonably simulate what goes on in a running engine.”

In the 1980s or so, a top Japanese engineer, apparently with a large amount of funding, researched this subject right down to the bare bones. At the end of the program he proclaimed that, for all practical purposes, a steady-state flow test was a meaningful test because it substantially related to the pulsing flow seen in an engine and the dyno results produced.

So, now we have an accepted standard pressure drop from which to get fully serviceable flow figures and strong evidence that the steady flow seen with the bench still correlates well with the pulsing flow seen in a running engine.

Although we have now arrived at what are now widely accepted practices, I am far from finished on the subject. Why? I am not finished because 28 inches can present its own set of test problems.

Real-World Test Pressures

So why can 28 inches of test pressure drop be a problem? The short answer is that it is a compromise. A higher test pressure allows the development of better ports, especially the intake, but stepping it up can present real bench-motor power problems. In an effort to overcome this, I have seen flow benches that required two 50-hp electric motors to run. The space, time, money, and effort required to build a bench like this is nearly prohibitive and, as we shall see, unnecessary. The problem is: If you want to test a typical big-block Chevy head or even a really good small-block head, you need a lot of amps for the inevitably large electric motors needed to produce the required vacuums/pressures. This means buying a pretty stout commercial flow bench at something between $6,000 and $18,000, or building your own bench with as many as six high-powered vacuum motors. On top of that, you may have to have your shop or household electricity supply up-graded to deal with the amperage involved.

Fig. 2.7. Here is what happens as the test pressure and, consequently, the port velocity goes up. At low test pressures (left) the air has a better chance of staying attached to the port floor, and as a result the entire circumference of the valve is used to good effect. When the test pressure is increased to that more commonly seen in a running engine, the flow detaches from the short-side turn, and the higher-velocity air attempts to exit the long side, which is already busy passing air from the long side. Result? A drop in efficiency, which is otherwise unaccounted for during normal pressure-drop corrections.

Corrections

In the early 1970s, Neil Williams, the founder and driving force behind Super Flow, introduced the compact Super Flow 110 flow bench. I believe this was the first commercially available flow bench and it tested heads using a 10-inch pressure drop. I had the opportunity to do a comparative check on one of these benches in late 1973 at Piper Cams in England. For all its simplicity, it did produce results that compared well with the ones I saw on my monster BSI flow bench, which tested at 20 inches for small heads and 15 or so for larger heads (such as Chevy big-blocks).

Fig. 2.8. The pressure differences on a bored and stroked small-block Chevy (final displacement: 441 ci). This was a little large for the heads involved, but the trends we need to appreciate can be seen here. When the intake valve was 0.070 off the seat, the suction from the exhaust as measured in the combustion chamber (red curve) was 100 inches H2O while the suction caused by the piston never exceeded 50 inches H2O. This graph also shows that the intake port velocity (blue curve), while the piston is parked at TDC, is no less than 120 ft/sec.

As mentioned earlier, you do have an option here to correct a, say, 10-inch pressure drop to 28 inches with a mathematical correction. The problem is that this type of correction is only satisfactorily valid when corrections are small, and you are testing a simple orifice or a straight port with absolutely no bends in it. Any time a seat and valve enters the equation, errors made in corrections from a low to a high test pressure have a greater impact on flow. Equally as bad is the fact that the corrections almost always produce artificially higher flow numbers.

But there is a fix for this problem, and it’s one that has absolutely no downsides. The fix makes testing more meaningful, reduces the cost of a bench, and you are able to do it all without having to invest in any workshop electrical upgrades. To see how all this works, let’s get back to test pressures.

Floating Pressure-Drop Testing

As I just mentioned, the Super Flow 110 bench tested at 10 inches of pressure drop. When doing so, the port velocity is substantially lower than when that same port is tested at 28 inches. At 10 inches, the air could be traveling at a speed just marginally less than that needed to produce flow separation around the short side turn (Figure 2.7).

When the flow stays attached, the air makes better use of the entire valve circumference. Under these conditions, the valve is more efficient. When that same situation is tested at 28 inches, the air in the port may not make it around the short-side turn in anything like the same way and the flow past the valve adjacent to the short-side turn is consequently disrupted to a far greater extent, thus cutting the overall flow.

However, the typical mathematical adjustments applied to correct a 10-inch measurement to a 28-inch result do not work well for this. Also it is a problem that it results in a port that may work at 10 inches but be far less effective at 28. At 28 inches, the flow separation influences the development of the final seat and port used. In other words, that higher test pressure goes toward solving a problem that did not even show up at a 10-inch test pressure.

Taking stock of the situation so far, we can say that we need higher test pressures, but the downside is the requirement for increasingly larger pump motors.

The situation looks bleak until we stop to look at the cyclic pressures that exist in the intake port of a high performance engine. In Figure 2.8 we see the pressures that occur within an intake port of a street/strip 441ci (7.23-liter) small-block Chevy during the induction phase are far from being a steady 28 inches. During the overlap period, where valve lift is relatively low, we can see a depression across the intake valve as high as 100 inches, and it can be much higher in an all-out-race, two-valve engine.

This low pressure is brought about by the negative pressure wave arriving at the exhaust valve and it functions as a means of extracting the combustion residuals. It does so with sufficient amplitude to pass through the chamber and act on the intake port. This high-exhaust-generated vacuum starts the intake flow into the cylinder well before the piston even starts down the bore.

Do not underestimate the value of this effect. In practice, it can make or break the degree of success seen on the development of a high performance engine. Although at high valve lift, the pressure drop across the intake valve for our example is some 50 inches we find that for a highly developed engine with decent seat and port design, depression is more in the 15to 20-inch range. If in-cylinder pressure measurements show anything much more than that, it’s a sign of inadequate heads for the displacement/RPM involved.

Now here is where we get a break. If we hook up a typical vacuum cleaner to a cylinder head and progressively open the intake valve, we find the pressure drop at low lift is high and at high lift is low. That is exactly what we are looking for. Accepting a floating pressure drop is the key that immediately unlocks the door leading to the building of a very simple yet extremely effective flow bench. The construction of such a bench is explained in Chapter 3, but I first want to address the relative importance of the intake versus the exhaust.

This tech tip is from the full book, DAVID VIZARD’S HOW TO PORT & FLOW TEST CYLINDER HEADS.

For a comprehensive guide on this entire subject you can visit this link: LEARN MORE ABOUT THIS BOOK HERE

Intake Fixation?

It is patently obvious that any cylinder head has two distinctly different ports that need to be developed to produce high-flow efficiencies for maximum power: intake and exhaust. The question being posed here is: Which of these two is the most influential toward the production of power? At first sight, it may seem to be the exhaust because, after the charge has been burned, the volume is so much greater. Let’s assume for the moment we are dealing with a naturally aspirated (non-supercharged) engine. At the point of exhaust valve opening, there is between 70 and 120 psi waiting to escape from the cylinder. When the exhaust valve opens, the flow velocity between the seats in the head and those on the valve momentarily exceeds the speed of sound.

Fig. 2.9. The green curve is the intake flow of a big-block Chevy 049 casting in stock form but with a 2.250 valve. The blue curve is for the same head ported. The solid red curve is for a head having a flow efficiency of 100 percent based on the valve O.D. In practice, it is helpful to look at the flow efficiency based on the throat diameter of the port (red dotted curve). In either case, the flow efficiency of most heads is a long way from the best that can theoretically be obtained.

For the intake valve, the situation is far different. The greatest pressure available to drive the charge into the cylinder is that of atmospheric pressure (i.e., 14.7 psi). Worse yet, of that 14.7 psi, we can’t really use more than 1 or 2 psi at most. Therefore, we need to see the minimum drop across the intake valve for best breathing. From this, we can see it is going to be a lot harder to fill the cylinders than it is to empty them. In addition, if the cylinder fails to adequately fill due to poor induction efficiency, the engine’s power potential is reduced no matter how good the exhaust may be. So, initially at least, we need to focus on finding intake flow from a performance cylinder head. It is not until average flow efficiencies of the intake throughout the intended valve-lift envelope have exceeded about 60 percent that we need to start considering intake-to-exhaust flow ratios and the valve sizes involved.

Flow Efficiency

For a given pressure drop, there is a relatively well-defined limit for how much air can pass through a given area available for that flow. At 28 inches pressure drop, an opening like the size of the one created between a valve and the seat in the head flows 146 cfm for every square inch of available opening area. At this level, the aperture created is 100-percent efficient. Of course, the area available to flow in or out of the cylinder varies with lift. As the aperture geometry varies, so does the flow efficiency. At very low lift (0.0 to 0.050 inch), the gap at the valve seat resembles a venturi.

At this size of gap, there can be some pressure recovery, so at these low lifts the valve seat configuration acts as a nozzle. Then the flow efficiencies appear very high and can often exceed 100 percent. However, that is more a function of how we determine what the 100-percentile figure is. In practice, a nozzle requires a different equation to compute its flow, hence the apparent anomaly here. Figure 2.9 shows the difference between a stock big-block Chevy head, a ported one, and the 100-percent mark that we should strive for.

A similar high-efficiency situation at low intake lift can exist for a well designed exhaust port. Not only can very high efficiencies be realized at low lift, but also at lift values exceeding about 0.28 inch of the valve diameter. At these lift values, a steeply up-drafted port can start to take on the properties of a nozzle. As a result, the flow figures can appear extremely high and in certain instances exceed the 100-percent mark.

I address flow efficiency more in Chapter 10. But it may help here to show some typical efficiency figures versus the figures seen if the valve and port were 100-percent efficient. Figure 2.9 shows this. Note how a typical stock port is usually barely more than about 50-percent efficient. If porting time is put into production two-valve heads, efficiency can be raised to 65 percent or more, depending on the casting involved. That, without resorting to bigger valves, is a 30-percent increase in airflow. What that can do for power output, especially if it is combined with a compression ratio (CR) increase, is very gratifying.

Written by David Vizard and posted with permission of CarTech Books