Turbocharged flowbench configuration

[saltrunner]

Wow, this board it an awesome find.

I have been looking on this site and searching the web and it appears to me that all flowbenches work on vacuum. Is this an accurate means of testing intake paths if the application is for a turbocharges/supercharged vehicle? Would blowing air into the part be a better representation of the real world conditions?

I want to make a plenum with balanced flow for each runner, so I figured making a flowbench would be the best solution. I am going to test a stock plenum for comparison so I am not really concerned about the accuracy, mainly just precision. Here is a quick model of the plenum design I am looking to create

I am trying to figure out how to test each runner so that I am not adversly affecting another. Should I have a pitot tube in a short pipe connected to each runner? How would I check to see what will occur when different cylinders are open or closed?

Lastly, am I making this too complicated or missing something?

-Michael

[Tom Vaught]

I do the work for a living at Ford Research and I see a couple of issues with your basic design right off.

The runner closest to the throttle body will be starved for air.

More air will accumulate at the number 6 runner due to the velocity
of your air flow.

Look at how a Ford 5.0L GT-40 Intake is designed and study that.
Use that design if you have room as it works.

Tom V.
Turbocharger/ Supercharger engineer
Ford Motor Company.

ps as far as using a flow bench to read the different runner flows
how do you propose to take the readings. Always interested in
new ideas.

[2seater]

I'm working on the same problem. I turbocharged a Buick 3800, and the common style manifold with the air inlet on the end is pretty unbalanced. The air flow piles up on the end opposite the inlet and those cylinders tend to be lean while the ones closest to the inlet are rich. Forcing more air in makes the problem worse, no matter what the booster might be. The Ford manifold design as mentioned is a pretty good design, the flow isn't as biased to a few cylinders. A plain old carberetor manifold has better balance to the air flow.

As for flowbenches, most are designed to flow air in either direction. You can blow through them just as well. I plan to blow though from the throttle body end and use a velocity probe to check the flow at each runner. Each port would need to be checked at several points as there will be high and low flow areas. Alternatively, a long tube could be applied to the port to be tested and test the flow rate after the air has settled down (laminar flow). This is a work in progress for me as well, but some data has been collected and there is a surprising difference in the flow rates at the ports as described above.

[saltrunner]

The original factory manifold is shown here:



I am not sure what can be done to solve the unbalanced flow in the space that exists. I will see what I can figure out.

As for the flow bench I made a quick and dirty model of what I was thinking



The air will come in through the throttle body and flow through the plenum, from here the flow runs through a straight section of pipe with a pitot tube in each. I will need to calculate how long the section needs to be to be sure that the flow is fully developed. The flow then dumps into a resevoir which has a valve to allow the system flow rate/pressure to be varied along with the inlet flow rate. A manometer will then be used to get the differental pressure between the resevoir and the atmosphere to allow for repeatability regardless of the atmospheric pressure. I think if all of my assumptions are correct this system will work and will allow me to vary the flow rate of the system at different pressures.

Do you think this will work? I need to break out my fluids book and see if I am terribly butchering the theory behind this.

-Michael

[bruce]

I think you need to do your testing with the head attached. In actual use each runner is not flowing air at the same time as your test model will be doing. In order to get "good" flow numbers you need to simulate how the parts are actually working in the engine. Might even be better to mock up the whole engine minus the pistons and draw your flow through it by using the camshaft as your valve opening devise? This will let you simulate the valve open and closing sequence and see what the airflow is doing in the various runners.

More work than you probably wanted to do but if you want to do R&D and get good numbers you have to have a method to attain them. Course you will also need dyno numbers to quantify what you find on the flowbench. You can make some airflow gains on the flowbench and the dyno will show losses.

[Tom Vaught]

Your stock manifold has some features you did not show in your drawing.

1) There is a radius from the throttle body to the first runner

2) There is an expansion after the throttle body to the plenum.

3) There is a transition from the larger volume near the throttle body to the rear runner.

Each of these "small" items has a big effect on the manifold flow and distribution.

Just from a quick look without seeing the whole intake I would say the people who built the intake did a good job.

Your ideas for your flow bench are good ones.

I have played with the idea of a Intake/ Head deal like Bruce
suggests with custom sharp edged orifices in the bottom of each cylinder bore. There would be a pressure tap next to the hole on the top (head) side of each plate. The bottom side would have a a similar pressure tap.

You would make a multiple vertical manometer and hook up the pressure side to the wells and the vacuum side to the
upper leg of each manometer.

When you did your flow test the flow "trends would be shown
on the manometers.

This is similar to graphing a "duct" or "pipe" with multiple manometers.

Hope everyone can comment on this idea.

Tom V.

[2seater]

The stock manifold looks pretty good. Decreasing size toward the far end will help minimize the "dead end" effect. Increasing flow rate with a turbo motor is a lot easier than an n/a one, just turn up the boost. Of course the downside is more heat. That stock manifold looks like a good place to start and work to optimize the basic design. I think your basic design for the testing will work, but it will be critical to get all the "sensors" to read the same. Orifice plates could be used too but I have a feeling the restriction of the plates might affect the actual flow characteristics. In other words, the process of testing changes the results. Just my speculation.

Regarding the question of flowing all ports at the same time or combinations following the firing order will likely drive you nuts. I have thought about this problem at length and talked with the people that make the Power Plate for the Buick GN guys. They flow all ports at once and try to achieve a balance for all of them. In simple terms they restrict the flow to the "far" cylinders to achieve this balance. The result is better air/fuel ratios for each cylinder reducing knock problems and the bad things that happen from there. I do believe flowing all ports at the same time is valid. The amount of time for cylinder filling is extremely short (thousandths of a second), and even though air is light, it still has mass and doesn't want to change direction. The air flowing to each cylinder is essentially constant, with small breaks as the valves open and close. Without sophisticated computer modeling or test equipment to measure the shock waves from this stuttering flow, assuming constant flow is much easier to deal with. The downside is it will requires hundreds of cfm.

[bruce]

A good read on this is:

Scientific Design of Exhaust & Intake Systems, Smith & Morrison, ISBN 0-8376-0309-9

[saltrunner]

Where can I find pitot tubes suitable for this application, all the ones I am finding are for 10" ducts?

So if I redesign my model of the plenum to better distibute the flow and I make a flow bench, is it just a trial and error process to figure out how to make small balance changes between runners?

Thanks for making this such an informative discussion.

-Michael

[84-1074663779]

I doubt if you are going to find a solution to these air distribution problems on a flowbench. The problem is that it is not just a question of steady state flow down each runner. Firing order, and inlet cam duration are going to set up a confused pulsing flow that is just not possible to duplicate on a flowbench.

Your best bet is to first get it running on a dyno under load, and measure individual exhaust gas temperatures after you have fitted a set of carefully matched injectors. Air distribution problems will really stand out. Only when you understand the problem can changes be made to try to fix it.

Invariably the cylinders nearest the throttle body run rich through air starvation, and the most remote cylinders run lean. You might be able to fit ridges or deflectors in the plenum volume to even things up at very high air velocity. There is really no easy way around these problems.

[Tom Vaught]

I really like Tony's idea of using the dyno for getting a handle on the air
flow vs exhaust temp vs the fuel flow.

Having matched injectors could be possible if you made a simple tester
and used a pulse device to cycle the injectors.
You would have to ASSUME that the fuel delivery to your test stand did not have the same issues as your intake manifold (one injector getting more fuel than another.) The way around this would be to test one injector at a time for flow into a burette over a specific time period. Applying a fixed 12 volts just causes the the injector to go "Static" and will give you a false reading of flow. You need to test at a given duty cycle like 85 percent of max flow.

After you get the fuel side in order you then will have to test several Type
K thermocouples to find ones that are close on the exhaust tems over the
speed range.

Now you will be close on the fuel and close on the Thermo-couples.

If you run the engine on the dyno the last deal is the mapping of the engine as you change the timing, fuel, and the load on the engine.
Each time the values can change the overall temperature numbers so you
will need to look at trends.

It typically takes months of dyno time to accurately map an engine for fuel
spark and airflow so I hope you have a good friendly dyno shop.

Mind you this is a carb statement now but I had a guy call me one time
and he said "Tom, I have perfect distribution on all 8 cylinders" (on his NASCAR engine). I told him to change the rpm by 500 rpm and then tell
me what happened.
He was not happy, his fuel distribution had changed (cam related) and his
"perfect distribution" was now history.

This is why the OEM companies use dual Heated Exhaust Oxygen sensors in the exhaust, to correct for this issue.

Tom V.