Let's talk port velocity --

[David Vizard]

I have seen a few comments stating that the average port velocity (@ used full lift) is not as important as the local velocities as measured with a pitot tube. A statement like this can be a bit confusing so let's see if I can put my point across here in a more easily understood manner.

First local port velocities as measured with a pitot tube.

What these are telling the user is where abnormally high or low velocities are occuring. These, in essence are the 'problem' areas. On an F1 head where the ports are angled up 60 degrees or more there is minimal velocity differences between the highest and lowest. Even on a now very much dated Cosworth DFV F1 head I would, as I remember, see only a 30 fps variation when the mean velocity was 300 fps.

Conversely on something like a SBC it is not out of the realms of reality to see as much as 300 fps variation between the fastest spot on the short side turn to the slowest spot on the floor of the port and the manifold face. Where the port is slow the area concerned is being underutilized, sometimes to the point of of redundancy. The areas significantly above average velocity are over utilized areas and more often than not indicating where 'overcrowding' is take place. Knowing these numbers is the first step toward understanding what a possible fix may be so they are important. The goal, though unachievable, is to have a single velocity across the port at any one position along it's length.

Mean port velocity:-

Knowing this number allows the gauging of the ports effectiveness in terms of it's ability to flow air as a result of an efficient shape rather than a large cross sectional area throughout the port. If you do not know what the mean velocity is how would you know if the port had optimal velocity for best results or not? answer - you wouldn't!

So having the mean velocity allows a porter to establish whether or not the port is too big or small.This makes it an important number as it encompasses all the problem area and allows the porter to see if those high and low speed area's are going to be a big problem or just a problem that all the types of heads similar will have.

Different types of heads have different optimal velocities and having the approximate values for each distinct type is a great help toward calculating what rpm peak power will occur at. Most conventional style pushrod engines peak at 300 to 330 fps. The last of the NA F1 engines peaked at as high as 420 fps. Last time I checked, a winning ProStocker was typically around the 330-340 fps.

DV

[ptuomov]

Here' s a novice question: In the context of this thread, is velocity measured on the flow bench? If so, at what test pressure? Or is it computed mean velocity from a running engine? Something like the formula, such as volumetric efficiency * engine speed/2 * cylinder volume / port minimum cross sectional area? Or is it an actual measurement from a running engine? Apologies in advance for asking something this basic.

[ClassAct]

I have seen a few comments stating that the average port velocity (@ used full lift) is not as important as the local velocities as measured with a pitot tube. A statement like this can be a bit confusing so let's see if I can put my point across here in a more easily understood manner.

First local port velocities as measured with a pitot tube.

What these are telling the user is where abnormally high or low velocities are occuring. These, in essence are the 'problem' areas. On an F1 head where the ports are angled up 60 degrees or more there is minimal velocity differences between the highest and lowest. Even on a now very much dated Cosworth DFV F1 head I would, as I remember, see only a 30 fps variation when the mean velocity was 300 fps.

Conversely on something like a SBC it is not out of the realms of reality to see as much as 300 fps variation between the fastest spot on the short side turn to the slowest spot on the floor of the port and the manifold face. Where the port is slow the area concerned is being underutilized, sometimes to the point of of redundancy. The areas significantly above average velocity are over utilized areas and more often than not indicating where 'overcrowding' is take place. Knowing these numbers is the first step toward understanding what a possible fix may be so they are important. The goal, though unachievable, is to have a single velocity across the port at any one position along it's length.

Mean port velocity:-

Knowing this number allows the gauging of the ports effectiveness in terms of it's ability to flow air as a result of an efficient shape rather than a large cross sectional area throughout the port. If you do not know what the mean velocity is how would you know if the port had optimal velocity for best results or not? answer - you wouldn't!

So having the mean velocity allows a porter to establish whether or not the port is too big or small.This makes it an important number as it encompasses all the problem area and allows the porter to see if those high and low speed area's are going to be a big problem or just a problem that all the types of heads similar will have.

Different types of heads have different optimal velocities and having the approximate values for each distinct type is a great help toward calculating what rpm peak power will occur at. Most conventional style pushrod engines peak at 300 to 330 fps. The last of the NA F1 engines peaked at as high as 420 fps. Last time I checked, a winning ProStocker was typically around the 330-340 fps.

DV

How does displacement enter into the picture? In terms of when the displacement exceeds the any potential cross section?

[steve316]

DV how much ABA TESTING on this have you done; to come up with these conclusions of what is the optimum air speed?

[hoffman900]

Here' s a novice question: In the context of this thread, is velocity measured on the flow bench? If so, at what test pressure? Or is it computed mean velocity from a running engine? Something like the formula, such as volumetric efficiency * engine speed/2 * cylinder volume / port minimum cross sectional area? Or is it an actual measurement from a running engine? Apologies in advance for asking something this basic.

I don't see how an accurate measurement of port velocity can be obtained? It seems anything that is used to probe velocity is going to either disrupt the airflow, or change the csa because of something (a probe) has to occupy the space. So then really port velocity is a good guesstimate?????

Formula One (and other high end programs) will measure particle flow with optical sensors, but yes, a probe will influence it.

Larry Meaux has had good posts on velocity and port angle over the last decade and a half on this site.

[David Vizard]

I am somewhat surprised by the vigor of the responses here. All good to great questions but it seems that a lot of what is being asked is actually history not DV related. Don't have time as of now to do a two page response but that is what it will take.

I will see if I can put together something by Friday but no promises.

DV

[David Vizard]

Well here is my first opportunity to answer some of your questions.

Firstly the subject of peak power occurring at 300 ft/sec for mean port velocity and peak torque at 240 ft/sec. Fortunately I do not have to take a single step here to justify the claim that the peak power/mean port velocity numbers I quoted need to be backed up with my data.

I don't know the exact date but, apparently, the port velocity/peak power link was an observation made by the genius mind of Sir Harry Ricardo (1885 - 1974) toward the end of WW1. I really don't have time to relate just how innovative and pioneering this gentleman was (if you want to know more then google his name)

During WW1 he teamed up with Shell as a consultant. The result was not only a better understanding of detonation but also the birth of the 'octane' numbers assigned to fuel to put a value on it's resistance to detonation. The variable compression octane test engine, which is about a world standard for such tests, was largely his efforts. Also, I think he had a hand in the design of the Farnborough Indicator, an all mechanical device for measuring instantaneous cylinder pressures on a running engine. Now think about that for a moment. what if somebody asked you to design such a piece of equipment that, without the aid of any electronics, could do the job on an engine running at 5000 rpm.

In 1922 he designed the 4 valve per cylinder engine that powered the GP cars entered by Vauxhall (now part of GM UK). I don't think this guy slept at night. Not only did he develop and design many heads and engines but he also developed test equipment to investigate related issues. It is said that his work on cutting engine oil consumption saved the lives of hundreds of WW1 tank crews. Before so doing the engine-smoke heralded their impending arrival long before they were heard. That smoke made them an easy target acquisition.

Archive film of his work for Ford on side valve engine combustion with transparent heads is still out there. Indeed, some of it may have found it's way to YouTube. (anyone know if this is so?)

Among his writings is a book entitled The High-Speed Internal-Combustion Engine. If you have not read it then at least check it out from the library.

But Sir Harry is not the only one who has adopted the mean port speed Vs peak power philosophy. There are other well-known engine development greats who also used this method to determine where peak power would occur. However, over the years since the WW1 Ricardo tests technology has moved on and we can now assign slightly different velocity numbers to engines of various types and design sophistication for more accurate answers. This I will get to down the road awhile.

Meanwhile here are some answers to comments so far made.

Ptuomov-
In the context of this thread, is velocity measured on the flow bench? If so, at what test pressure? Or is it computed mean velocity from a running engine? Something like the formula, such as volumetric
efficiency * engine speed/2 * cylinder volume / port minimum cross sectional area? Or is it an actual measurement from a running engine?

Answer-
In this case both. The original figures are computed, in the same way Ricardo did, from running engines. From this Stan and I worked our flow bench program so it would duplicate the running engine situation. The numbers given on the IOP flow program are the numbers that would exist in a running engine. With this port velocity # you can see right off the bat if your port is too small or big. Flow bench numbers relate to a 28 inch depression.
I will put post formulas from my ‘How to Build HP’ three-day seminar in the next few days.

Classact-
How does displacement enter into the picture? In terms of when the displacement exceeds the any potential cross section?

Answer-
Displacement, RPM and port cross section are the prime factors toward calculating the velocity.

Exhaustgases-
I don't see how an accurate measurement of port velocity can be obtained? It seems anything that is used to probe velocity is going to either disrupt the airflow, or change the csa because of something (a probe) has to occupy the space. So then really port velocity is a good guesstimate?????

Answer-
You are right here but don’t forget we are calculating the mean velocity here not a local velocity. The mean velocity an be calculated very accurately.

Steve 316
DV how much ABA TESTING on this have you done; to come up with these conclusions of what is the optimum air speed?
Answer-
Ricardo, Taylor, Hayward, Westlake, and many others did the work on this before I was even born!

More on this to come!
DV

[David Vizard]

Hoffman Ref:-
Formula One (and other high end programs) will measure particle flow with optical sensors, but yes, a probe will influence it.


Got a good story to tell here. There I was working away in the back of my shop in California when my daughter comes in and tells me there is a salesman with some interesting hardware I should see.

I ask her to send him in. A few moments later in comes this very expensively dressed guy looking all business. Almost the first words out of his mouth were along the lines of - 'I have some test equipment here that you will need if you intend to do any more consultancy for Cosworth or any of the other engine manufactures you might have worked for'.

As an attention getter that was a great opening line. What he had was a laser doppler anemometer and he had just sold one to Cosworth. From the interference pattern of two micro laser beams this device could measure the speed and direction of a patch of air just 50 millionths across. Well we played with this on the flow bench for a while and it really was pretty trick. I asked 'how much'? Answer $400,000 and that is in 1985 dollars not current ones. I told him that was way out of my league thanked him for his time and sent him on his way.

A few days later I had a call from one of the then top guys at Cosworth and (wouldn't you know) the subject got around to flow testing. After going through the usual spiel about not putting much faith in flow benches I asked why then did they spend 400 grand on a laser doppler anemometer? The response went something like this - 'What - how the heck did you know about that - we only got it last week!'

Just to confuse him I told him Harvey Crane told me!

DV

[CFM-Z440]

Is there anything that we can use to estimate ideal mean air speed for different head types? I'm assuming the 420ft/sec F1 port is very straight, with little turn into the bowl, and has perfectly even flow all the way around the valve.

I work with 4 valve single's in motorcycles and atv's with peak hp in the 9000-10,000 rpm range. The ports are not inclined like I assume the NA F1 ports to be, but they are pretty straight with most of the turn at the bowl, and they have very little SSR. They seem happy right around 285-300'/sec mean velocity, between the guides and divider, and I even up the local velocity as much as I can, but they are faster on the floor. I'm a long ways from 420ft/sec, but I feel like the design isn't worlds apart. Could I get it to perform with higher mean velocity if I could get my local velocities more even?

[David Vizard]

Is there anything that we can use to estimate ideal mean air speed for different head types? I'm assuming the 420ft/sec F1 port is very straight, with little turn into the bowl, and has perfectly even flow all the way around the valve.

I work with 4 valve single's in motorcycles and atv's with peak hp in the 9000-10,000 rpm range. The ports are not inclined like I assume the NA F1 ports to be, but they are pretty straight with most of the turn at the bowl, and they have very little SSR. They seem happy right around 285-300'/sec mean velocity, between the guides and divider, and I even up the local velocity as much as I can, but they are faster on the floor. I'm a long ways from 420ft/sec, but I feel like the design isn't worlds apart. Could I get it to perform with higher mean velocity if I could get my local velocities more even?

What will eventually stop you achieving the 420 ft/sec F1 mark will be the lack of down draft angle and the tight SS turn.But 285 to 300 is a little on the low side for a 4 valve head.
As soon as I can I will post ft/sec figs for a variety of head types so you can better judge where you stand on this.
DV

[Multiplex]

David ,

Did the IOP software have a possibility to output velocity data for the multi-valve engines ?

Klass

[Stan Weiss]

3 heads same lift and flow, port area and length.
Head #1 has 1 2.02" intake
Head #2 has 2 1.01" intakes. Same curtain area as head #1 different valve area
Head #3 has 2 1.428413" intakes. Different curtain area, same valve area head #1

Stan

Flow.pdf

[Multiplex]

Stan Weiss wrote:
> 3 heads same lift and flow, port area and length.
> Head #1 has 1 2.02" intake
> Head #2 has 2 1.01" intakes. Same curtain area as head #1 different
> valve area
> Head #3 has 2 1.428413" intakes. Different curtain area, same valve
> area head #1
>
> Stan
>
> Flow.pdf


Thanks, that's how i used it so far.
it only requires a little more calculation :-)

Klass

[David Vizard]

Well I finally got around to it. here are the port velocity numbers at which peak power occurs for various styles on cylinder heads.

These are from my power point seminar that i am going to present at PRI on the Friday afternoon from 2 pm to 5 pm with a question and answer session going on from there.

DV

[ptuomov]

Interesting.

If you go to 1980's and 1990's four valve heads from Germany or Japan, I think you're going to find very low port velocities with this formula.

Take for example 1987 Porsche 928 S4. The port minimum cross-sectional area is about 2.7 square inches. The engine displacement is about 302 sqin with eight cylinders, so 37.75 CID per cylinder. The peak power from factory is at 6000 rpm, making 316 hp. Unless I made a mistake, the formula says 6000*(302/8)/(2.7*360) = 233 ft/sec.

The head itself flows decently considering the port minimum area. It flows 290 CFM with 2.7 sqin MCSA at 28" of water. It's just very big and very slow compared to these numbers listed on the sheet.

Cams are also very short duration, in case that's relevant.