In pursuit of the elusive discharge coefficient

[digger]

The valve only sees those high pressure differentials at very low lifts. So the flow bench doesn’t have to move a ton of air, just hold higher pressure.

Like warp is eluding to the pressure across a head in flow bench is across the whole induction not just the valve. the same pressure differential across running engine at high rpm for example peak hp rpm can often be upto 0.5b which is some 200" inches at peak piston speed to peak valve lift area crank angle.

Vanniks program takes the cd at flow bench pressure to populate a map so that other pressure differential can be approximated

[SchmidtMotorWorks]

How much of a power gain have you seen possible testing with TFX equipment to improve on the initial flowbench data?

A flow bench and TFX system are apples and oranges (I have both).

A flow bench is static (constant)
A TFX system is for measuring a lot of dynamic (changing) data at high speed.

Vannik has written something in EngMod4t manual about measuring CD and IIRC something about most shops not being equipped to do it properly.
Maybe he will explain here.

[GARY C]

Most amateur flow benches are limited to about 28" H2O; some only 10" H2O.

To replicate the inlet (NA) pressure ratios found in a running engine, we need to test
at 203.8" H2O; and many points in between. Only a few test benches can do this.

In addition, air is compressible. At 28" H2O, the compressible flow function is about
4%, at 204" H2O it is about 30%.

I would be interested to see a flow schematic indicating how pressure sensors could be used to
capture Cd, since it must also includes water vapor, density, viscosity and type of gas.

The closest solution I have seen is the use of in-cylinder pressure and inlet pressure measurement
using a TFX system or similar. And then use a mass flow sensor for flow measurement.

Perhaps Clint would be willing to post inlet and cylinder pressure curves.

How much of a power gain have you seen possible testing with TFX equipment to improve on the initial flowbench data?

I have never compared actual TFX data to flow bench data. I consider them to be an apples to bananas comparison.
Further, unless the tuning is optimized, no real comparison can be made; sub-optimal comparisons have limited value.
Even engine simulations can produce sub-optimal results if the inputs are not correct.

Air flow is such an input.

I was just wondering if anything had been found with TFX's equipment that might have led to runner changes that the flowbench would not show... Just a thought.

[hoffman900]

How much of a power gain have you seen possible testing with TFX equipment to improve on the initial flowbench data?

A flow bench and TFX system are apples and oranges (I have both).

A flow bench is static (constant)
A TFX system is for measuring a lot of dynamic (changing) data at high speed.

Vannik has written something in EngMod4t manual about measuring CD and IIRC something about most shops not being equipped to do it properly.
Maybe he will explain here.

The CD map is really sensitive to measurements of the valve seat / throat / valve. I had someone with a lot of experience look through my data and found some mistakes (nobody is an island of knowledge!). It certainly changed things a bit. Vannik's Help File is fantastic (I should have read it with more detail to begin with ).

Neels' software displays the pressure as a pressure ratio, which is absolute pressure divided by (the chosen) atmospheric pressure, which makes for a cleaner x-axis.

[DrillDawg]

If the discharge coefficient is so elusive and it's needed for the simulation programs, maybe someone with a little experience with it could come up with a plug in number (dare I say "rule of thumb") for different levels of performance. Like one number for 3 hp/cuin, one for 2.5 hp/cuin, so on. As I guess this number is used to predict the mach or choke numbers, if not never mind.

[hoffman900]

MAPS OF DISCHARGE COEFFICIENTS FOR VALVES, PORTS AND THROTTLES 2001-01-1798 https://www.sae.org/publications/techni ... 1-01-1798/

[DrillDawg]

So, the simulation programs need to predict the discharge coefficient numbers needed, as it seems to be dynamic numbers base on a number of variables.

[David Redszus]

If we place a typical automotive poppet valve port on flow bench and hold the pressure ratio constant, allowing the valve to increase in lift while measuring flow (or air mass), we could then graph the results.

Such a graph will show a straight line slope until some mid lift point and then form a curve with a decreasing rate of change.
But doubling the lift will not double the mass flow.

Why should the air flow graph be curved when the pressure ratio and lift (actually valve curtain area) are linear?
If we repeat the process using different pressure ratios, we will find other dissimilar curves.

Since a real engine incorporates ever changing pressure ratios and cam lift curves with each crank angle, we will need to create a three dimensional look-up table that would indicate the Cd at any operating condition.

Blair displays such a three dimensional Cd graph, (Four stroke engine simulation, chapt 3).

Somewhere in my archives resides a paper by GM that displays flow curves for various GM engines. Each is different which indicates that each engine needs its own specific Cd lookup table.

While the correct and accurate air mass flow values entered into a simulation program is essential, it does not end there.
Test bench measurement cannot be considered accurate unless we accurately consider discharge coefficient values.

[ptuomov]

The valve only sees those high pressure differentials at very low lifts. So the flow bench doesn’t have to move a ton of air, just hold higher pressure.

Like warp is eluding to the pressure across a head in flow bench is across the whole induction not just the valve. the same pressure differential across running engine at high rpm for example peak hp rpm can often be upto 0.5b which is some 200" inches at peak piston speed to peak valve lift area crank angle.

Vanniks program takes the cd at flow bench pressure to populate a map so that other pressure differential can be approximated

I'm not an expert here, but the pressure differentials that my engine is seeing with the valve at mid and high lifts is nothing like what's been discussed in this thread. There are high intake-valve pressure differentials only at low valve lifts. Furthermore, at mid and high lifts of my head, the flow bench CFM scales pretty close to the incompressible square root formula whether the the head is flowed at 10" or 28". Yes, most benches can't pull more than 28" (and some not even that) with a four-valve head and high lift, but it's not like the engine sees high intake valve pressure differentials either at high valve lifts. At low lifts, the Cd in my heads does vary with test pressure, but since the mass flow is low it's more feasible to get a flow bench to pull a large pressure differential at low lifts. At low lifts, the ambient pressure and the pressure in the port upstream of the valve are similar, because mass flow is low. It's also the case that when the engine runs the mass flow at very low lifts is not very high, so it's probably not the area of the flow map that is the most critical to the simulation results.

I'd be interested in two things. First, some indication of where we need high resolution in the Cd(lift, pressure ratio) map. Like in that SAE paper that was linked earlier. Second, a simple and accurate-enough procedure for measuring the Cd(lift, pressure ratio) map on the flow bench that focuses on those areas of the map that actually matter. It makes sense to me that every head needs it's own table, so getting a measurement recipe that is practical and economical would be valuable.

[ptuomov]

If the discharge coefficient is so elusive and it's needed for the simulation programs, maybe someone with a little experience with it could come up with a plug in number (dare I say "rule of thumb") for different levels of performance. Like one number for 3 hp/cuin, one for 2.5 hp/cuin, so on. As I guess this number is used to predict the mach or choke numbers, if not never mind.

If you have the head flow curve at one pressure ratio across all lifts, that's in my opinion a really good start. Good news is that a lot of people have that. I think there's value in not letting the best be the worst enemy of good. Most people have access to the flow bench flow curve at one pressure differential across all lifts but didn't try all the different pressure differentials. I think it's more productive to talk about how to get most out of those data rather than just say "you're doing it wrong" (comment not aimed at any person specifically).

From that curve, one can assume the same Cd per lift regardless of the pressure differential -- If someone has a better procedure, I'm all ears.

In my opinion and in my very limited experience, the mid and high lift Cd's will be close enough for government work by that method. I think (but do not know) that the low lift flow that's less well behaved doesn't matter as much, because mass flow is much lower at low lifts.

[David Redszus]

The pressure ratio for an inlet valve is indeed highest at low lift. But it is reversionary flow; it's trying to go backwards and adds little (or detracts little) from overall air mass flow. The meaningful pressure ratio must be positive, not negative.

I am curious how one would measure pressure ratio in a running engine. Inlet pipe to ambient is not useful except to determine bellmouth shape and inlet runner length.

Depending on the PAD (displacement and rpm) and the valve curtain area curve, the pressure ratio can easily exceed 1.4.
If resonant tuned inlet pressure waves are present, a PR of 1.8 is not unreasonable.

Does any one have actual flow curves that could be posted for discussion?
Or TFX inlet pressure curves?

While we can measure (but not easily) the inlet duct pressure, obtaining the cylinder pressure is another matter.

[hoffman900]

Sir Yun in here wrote a very good peice in his blog:

https://www.google.com/amp/s/aseriesmod ... mbers/amp/

From Vannik’s help file:
http://vannik.co.za/download/EngMod4T/PortFlow4T.pdf

[digger]

The valve only sees those high pressure differentials at very low lifts. So the flow bench doesn’t have to move a ton of air, just hold higher pressure.

Like warp is eluding to the pressure across a head in flow bench is across the whole induction not just the valve. the same pressure differential across running engine at high rpm for example peak hp rpm can often be upto 0.5b which is some 200" inches at peak piston speed to peak valve lift area crank angle.

Vanniks program takes the cd at flow bench pressure to populate a map so that other pressure differential can be approximated

I'm not an expert here, but the pressure differentials that my engine is seeing with the valve at mid and high lifts is nothing like what's been discussed in this thread. There are high intake-valve pressure differentials only at low valve lifts. Furthermore, at mid and high lifts of my head, the flow bench CFM scales pretty close to the incompressible square root formula whether the the head is flowed at 10" or 28". Yes, most benches can't pull more than 28" (and some not even that) with a four-valve head and high lift, but it's not like the engine sees high intake valve pressure differentials either at high valve lifts. At low lifts, the Cd in my heads does vary with test pressure, but since the mass flow is low it's more feasible to get a flow bench to pull a large pressure differential at low lifts. At low lifts, the ambient pressure and the pressure in the port upstream of the valve are similar, because mass flow is low. It's also the case that when the engine runs the mass flow at very low lifts is not very high, so it's probably not the area of the flow map that is the most critical to the simulation results.

I'd be interested in two things. First, some indication of where we need high resolution in the Cd(lift, pressure ratio) map. Like in that SAE paper that was linked earlier. Second, a simple and accurate-enough procedure for measuring the Cd(lift, pressure ratio) map on the flow bench that focuses on those areas of the map that actually matter. It makes sense to me that every head needs it's own table, so getting a measurement recipe that is practical and economical would be valuable.

What's the cylinder pressure at peak hp with respect to ambinent/ manifold ? What's the peak port velocity and how does that compare with the velocity on the flow bench.

[DrillDawg]

For those with the means, rig your engine with heads, intake and carb up on a spintron with no rod or pistons and apply your suction through a modified oil pan, rig up a couple of transducers, vary the rpm (you will need a method of tracking the valve lift) and test pressures and see what it tells you. But that still won't take into account all the variables, so seeing the numbers are only needed to get better results from the program, why not just use the program to get the numbers for you? For someone with a lot of dyno information, back into the numbers by adjusting the DC numbers in the program until the program numbers match your dyno numbers. After about a couple of dozen you should start to see a pattern of the numbers needed for whatever level your engines are.

[ptuomov]

For those with the means, rig your engine with heads, intake and carb up on a spintron with no rod or pistons and apply your suction through a modified oil pan, rig up a couple of transducers, vary the rpm (you will need a method of tracking the valve lift) and test pressures and see what it tells you. But that still won't take into account all the variables, so seeing the numbers are only needed to get better results from the program, why not just use the program to get the numbers for you? For someone with a lot of dyno information, back into the numbers by adjusting the DC numbers in the program until the program numbers match your dyno numbers. After about a couple of dozen you should start to see a pattern of the numbers needed for whatever level your engines are.

Based on the Mahle piston book, the motored vs. fired operation is so different that at least for almost all piston design questions, "spintron" type testing is not useful.

The problem with calibrating the Cd to match dyno numbers is that there are so many similarly unknown quantities that could also be calibrated in this way, and there's no statistical identification of all those parameters.

I'm not saying that the flow bench is "be all" or anything, but one thing that the flow bench should be good at is measuring the intake valve Cd at different valve lifts and pressure ratios. Exhaust is a bit harder, I believe.