flowbench.txt
Time-stamp: <2004-01-09 04:32:02 lrc>
 
I've been thinking about flowbenches. At a first approximation,
anything that will improve flow, will probably improve
performance. However, in order to optimize performance, it is
necessary to optimize flow for the conditions that the engine actually
sees.

While completely accurate modeling is beyond my technical skills, it
is my belief that a flow bench, coupled with software that performs
relatively straightforward calculations, can help optimize performance
far beyond what is currently (commonly) being done.

I am not a mechanical engineer. I took thermodynamics over 20 years
ago, and the professor was surprised when I passed the class. What I
am going to do is to look at various factors that I think affect the
flow, and make some guesses as to how they can be measured, or
calculated. 

I expect that all of the factors that I will mention, will indeed
affect the performance of the engine, or the accuracy of the modeling,
however I do not know how much effect each of them may have.

It occurs to me that a common fallacy is that the goal of using a flow
bench is to maximize flow through a cylinder head. Flow is just one of
the may ways how to achieve the goal. The goal is to maximize
power. Power is approximately proportional to the amount of charge
(air-fuel mixture) in a cylinder. Too much charge under certain
conditions, or too hot of a charge, can lead to detonation or
preignition. Poorly mixed charge, can lead to not firing. I know that
turbulance and swirl affect combustion, but I don't know enough to
quantify the effects. 

If we start off with the approximation that power is proportional to
charge, our goal is to maximize the amount of charge in a
cylinder. The charge in the cylinder is the integral, over time, of
the flow into the cylinder. (For the moment, I'll ignore flow out of
the cylinder). In order to calculate the charge, you need to know the
camshaft profile as seen by the valve (i.e. multiplied by the rocker
arm ratio)
 
6000 RPM = 100 Hz = 10 mSec per revolution = 5 mSec/stroke

For example, if we assume that the engine is turning at 6000 RPM, that
means that the intake stroke will take 5 mSec. We cannot simply
measure the airflow at maximum lift and delta P (pressure
differential, aka dP), and multipy by 5mSec. We can approximate it by
splitting the intake stroke into 50, 100uSec (100 millionth of a
second, or one ten thousandth of a second) segments. We measure the
airflow with the valve at the position it is in at the 100uSec mark,
at the current dP, average that with the flow at 0 uSec from the valve
opening, multiply by 100uSec and we have an approximation of the
amount of air that flowed in those first 100uSec.

We then repeat this calculation for each condition every 100 uSec
until we have completed the intake cycle. (Note that this assumes the
intake valve opening at 0 TDC and closing at 90 TDC (90 TDC cam, 180
TDC crank). 

To perform these calculations, we would need to determine a value for
dP (pressure differential) at each of these points, and measure the
flow with the valve at that pressure and that dP, for each of those
points. Using this methodology, you could determine a cam grind that
maximizes the charge (and probably performance) for that engine. Or,
you could make modifications to the ports, valves or combustion
chamber to improve the flow at critical points of the cycle.

These calculations, ignore flex in the valve train, tappet bounce,
valves floating, and anything else that would cause the valve's
position to differ from the theoretical based upon the cam grind.

Supercharging the engine will increase the pressure on the back of the
valve. Subtract from that the pressure inside the combustion chamber
at any given moment and if the dP multiplied by the valve area, is
greater than the force of the spring, the intake valve will open.

Other than a vavle train not perfectly conforming to the cam profile,
there are a variety of other dynamic factors. The intake mainfold can
only flow so much charge at a given dP. It doesn't matter if the valve
and port will flow 2,000,000 cfm if the rest of the intake tract will
only flow 300. At some RPM an engine with 100% volumetric efficiency
will want to flow more charge than the intake tract can deliver.

The mass and velocity of the air, combined with the geometry of the
intake tract will lead to resonances and harmonic effects.

Don't need constant depression/pressure, need delta P that is
appropriate for the amount that the valve is open.

If neglect overlap, can calculate delta P at each valve position. Know
chamber volume, and with crank, rod and piston geometry can calculate
what the delta P would be. If you know the flow characteristics of the
valve/head, can calculate amount of charge in cylinder.

Software would take readings at multiple delta P for each valve
position, and would map flow in 3-d plane: delta P = X, lift=Y and
flow=Z.

With this data, engine geometry and cam profile could predict actual
effectiveness of different cam grinds etc. with a particular head.

Next stage would be to also take into account charge temperature, heat
transfer to block (etc), heat from combustion and so forth.

How does temperature, pressure, humidity and Air Fuel Ratio effect
system? What is effect of gasoline vaporizing?