What to use for flow visualization of intakes?

[alriclam]

Hopefully one of you with some engine tuning experience might be able to help me out. We're trying to test some custom see through engine intakes we built for an R6 engine in order to visualize the flow through them. The caveat is that we want to test the intakes on a running engine, and not just flowbenching. Anyone know of any gasses or other chemicals that wouldn't be too harmful to introduce into the intake flows, for a short time? We don't mind if things bog a little and we aren't putting out max power. Its just the easiest way we can see how to test the real life situation. We've already done significant theoretical modelling, but now its time to test the ideas.....

I'll post pics and video if/when we accomplish this experiment.

[tzortn]

Mind if I ask why run this on a running engine? The flowbench is just as accurate. How did you make your clear intake? Make sure that it is strong enough. We imploded a PC intake that we made when we put it on the engine.

BTW I may be able to help you out with some CFD analysis if you are interested.

[alriclam]

Well, the intake team's reason is that flowbenching won't be accurate enough when you have to factor in the valves opening and closing, as well as the dynamic nature of the flow through the engine. From what we've learned so far, flowbenching will only help us so much, but the real deal would be the best simulation.

I've questioned the integrity of their plexiglass intake as well, but they assure me that it will be okay and that the intake won't see much more pressure than a vaccuum (proven by MAP data we have from the dyno). They've vaccum tested it and it looks good. Once they are done with the mockups and come down to a final decision, we hope to plastic injection mold our system with the same materials F1 intakes are made out of.

If its okay, I might PM you with any questions about CFD analysis, but for now these guys seem to have it under control. They've used Fluent (which was started by an alum of our school, so they got the chance to attend an intensive training program before school started) to model everything from the beginning of the intake to the exhaust. including all the valves. So they have the theoretical idea of what they want, now its testing time. Final task for them is to do a CFD analysis of the cooling system. Might I ask, what you do for a living tzortn? Sounds like you're an engine tuner or engineer.....

[The CyberPoet]

Hey, I just read about particle flow modeling software being shoved into the open-source realms

As for the air combination -- you should be able to use any contrived air source as long as the following conditions are met:
(A) The oxygen content has to be locked and calibrated to be the same as the air available in the atmosphere (02 = 20% to 21% by volume).
(B) The gas dye should be non-reactive, preferably a noble gas that isn't invisible (it may have to be illuminated with UV or other lighting to be made noticable -- a neon gas might be able to be raised to an excited state & dropped again, to give you a good visual). Whatever amount & form of gaseous dye you use, the rest of the balance of the mixture should be pure nitrogen (N2). The nobel gas plus the nitrogen won't interfere with the power output of the engine at all as long as the oxygen level is maintained at the right percentage (since noble gases, by definition, are inert).

This particular combination should give you the same basic burn you get from standard atmospheric air, possibly a little bit more power. Be careful though -- upping the oxygen levels by even a single percentage will radically alter the fuel-air mixture's effectiveness and increase both the power and the temperature of the burn.

If you go about introducing a dye to standard atmospheric air, do a bit of math to figure out how much oxygen is being displaced and suppliment it. One word of caution: don't use smoke bombs, as the sulfer-basis will form deposits on the valves and pistons.

I would contact a compressed gas supplier about them providing a specific blend premixed, so you don't have to deal with the hassles of controlling the blend percentages. That leaves the real work being figuring out what to use as the trace element for visualization and whether to pre-mix it, or induce it at spouts at some point.

PS - at 10k RPM, that engine will be drinking in something in the neighborhood of 50 liters of air per second -- the idea of being able to visually establish what is happening at that speed would involve taking pictures at an astronomical rate and then looking at the frames to figure out what happened. There is simply no way to get an accurate eyeball-visual of what is happening to a gas that is moving at 13 gallons per second through four venturi's without slowing the process down after the image is gathered.

Cheers
=-= The CyberPoet

[soccersteve]

someone tell me why exactly i'm trying to understand this mumbo jumbo at 2am?

[The CyberPoet]

someone tell me why exactly i'm trying to understand this mumbo jumbo at 2am?

Because:
(A) You're not a night owl;
(B) You shouldn't have opened the thread if the title didn't interest you;
(C) You just have to realize someone has to do stuff like this to get the type of performance out of bikes that you can enjoy these days.

Cheers
=-= The CyberPoet

[alriclam]

Cyberpoet, you really do have a great amount of knowledge between those two ears of yours. Thanks for the input. I'll let you know how it goes. We probably won't run the engine that fast while we do the testing. We've got a few good leads on materials that might make the required "smoke" for us, but chemicals/gasses that can be identified under UV light are also interesting to us. It might be a good idea to choose some sort of specific blend.

But we're not too worried if this doesn't work as we've hoped, we still have various intake designs that we can test on the dyno and from there we can tell what will make more power for us. But being able to see the flow might give us insights into how to make that one design even better, perhaps.

What makes the intake so important to us is because of the 20mm diameter restriction we have to run all the air we get through. Due to this restriction, a good intake design is paramount to getting a good powerband.

Thanks!

[The CyberPoet]

We probably won't run the engine that fast while we do the testing.

I'm assuming that the engine will be raced after it's built, right? Then you really need to run it at 10k (or around where-ever torque peaks) to get an accurate idea of the air modeling for the intakes. The problem is that the valve overlap and the start of the piston compression upstroke don't become idealized until around the point that your torque values start really ramping up (to within about 75% of peak), so testing at much lower RPM rates ends up giving false conceptions of what happens closer to peak output.

What makes the intake so important to us is because of the 20mm diameter restriction we have to run all the air we get through. Due to this restriction, a good intake design is paramount to getting a good powerband.

Is that 20mm per cylinder, or 20mm for the whole airbox?
Is the system permitted to use RAM-air, turbo- or supercharged-compression before the restrictor plate (to load the passage at higher-than-vacuum pressures for faster through-put)?
Is there a mandate on how thick/thin the restrictor plate has to be? I ask because a well designed venturi with the restrictor plate in the center can help overcome the physical limitations of the edges of the restrictor plate (such as induced turbulance at the leading edge of the restrictor plate opening). Looking at the image below that I just generated (airflow right to left), the vortices at the lower air restrictor (item B) would be much more degenerate on the intake stream than the one for item A, where you gain two benefits -- the immediate decompression after the restrictor with limited vortice capture, and the venturi ramp-up just before the restrictor.

How big is the stock airbox opening when in the original OEM configuration?
Is it a 4-valve per cylinder or 5-valve per cylinder engine, and are you shutting down any of the valves to compensate at any RPM level? Are the valves still cam-controlled, or have you gone to servo-electric valves?

Curious...
=-= The CyberPoet

[alriclam]

Hey Cyber,

First on the intake issue, its 20mm diameter for all of the cylinders total, not each one individually.

We can't cool the air, ram the air, or otherwise mess with the air before the restrictor, though you can have a filter and ducting before the restrictor. Ours is right at the top of the diffuser, which then goes into the airbox, and then into the runners. This year, after some CFD analysis, the team wants to go with a diffuser/airbox hybrid, basically a big long diffuser that will go into the runners.

You can run turbos and the such, but only after the restrictor.

The engine is from a 2001 R6, so its 4 valves per cylinder, and the valves are still cam controlled, we haven't yet messed with variable timing (usually a little too complicated for a one year project, though some teams do run variable runner length intakes.) Airbox of the original R6 config is maybe half the size of the katana tank, its rather bulbous.

We did place in new, sharper cams into the engine this weekend, and managed to dyno it for an average 35 ft-lb of torque pretty flat all the way from 2k to 10k rpm. It was alot nicer than the stock cams, which do have significant ramp up of the torque. Unfortunately, that run couldn't be replicated, and we were turning out worse numbers after that. The engine team pulled the head and think the springs are shot, are you generally supposed to put stiffer/softer springs in if you change cam shapes?

Back on the topic, someone on KP who has done work on flow visualization for an auto company mentioned using incense in their tests, and we like that because its cheap and easy. However, we wont be able to get the 10k fps they were getting with their cameras. But we figure it would be worth trying out.

We might be able to do the 10k rpm test with the incense, though it might not make much sense with a slower speed camera than what we should have. We'll see.

[The CyberPoet]

Hey Cyber,

First on the intake issue, its 20mm diameter for all of the cylinders total, not each one individually.

We can't cool the air, ram the air, or otherwise mess with the air before the restrictor, though you can have a filter and ducting before the restrictor.

Then the filter should be before it, and ensure that the filter has a much larger surface area, so that it has minimal drag on the airflow.

Ours is right at the top of the diffuser, which then goes into the airbox, and then into the runners. This year, after some CFD analysis, the team wants to go with a diffuser/airbox hybrid, basically a big long diffuser that will go into the runners.

You can run turbos and the such, but only after the restrictor.

That would be feasible -- if you can increase the vacuum level at the backside of the restrictor plate enough, you can effectively make the system act as if the restrictor plate isn't there, especially with the appropriate venturi design.

The engine is from a 2001 R6, so its 4 valves per cylinder, and the valves are still cam controlled, we haven't yet messed with variable timing (usually a little too complicated for a one year project, though some teams do run variable runner length intakes.) Airbox of the original R6 config is maybe half the size of the katana tank, its rather bulbous.

We did place in new, sharper cams into the engine this weekend, and managed to dyno it for an average 35 ft-lb of torque pretty flat all the way from 2k to 10k rpm. It was alot nicer than the stock cams, which do have significant ramp up of the torque. Unfortunately, that run couldn't be replicated, and we were turning out worse numbers after that. The engine team pulled the head and think the springs are shot, are you generally supposed to put stiffer/softer springs in if you change cam shapes?

valve springs are a specialized science that's above my head in everything except the theoretical sense. Generally, you need to ensure that the springs aren't (weren't) compressing beyond the deformation limit by the new cams, and that they are just strong enough to return the valves to rest when the time comes... Imagine in your head what happens if the springs are too strong and if they are too weak. Also have the team look at the contact point between the valves and the cams -- you may have worn away some of the material with the sharper grind, which would contribute to not being able to replicate the dyno run.

Back on the topic, someone on KP who has done work on flow visualization for an auto company mentioned using incense in their tests, and we like that because its cheap and easy. However, we wont be able to get the 10k fps they were getting with their cameras. But we figure it would be worth trying out.

We might be able to do the 10k rpm test with the incense, though it might not make much sense with a slower speed camera than what we should have. We'll see.

Actually, even with a slower speed camera, it can be worthwhile, if you make sure the camera is not in sync with some multiple of the RPM's. You wouldn't get every action in a single row, but by taking frames from disseperate time slices, you should be able to build up frames that represent the actions in order. To put that another way, if you get one frame with the valve starting to open at minute 1 second 42 slice 18 and another frame with the valve slightly further open at minute 2 second 18 slice 31, you can rearrange them (manually) to give you a good approximation of what it would have looked like in linear order.

Good Luck!
=-= The CyberPoet

[tzortn]

Might I ask, what you do for a living tzortn? Sounds like you're an engine tuner or engineer.....

Yes, I am a mechanical engineer (BTW you may remember me as 600newbie from KP). I graduated from RPI where is was a member of the FSAE team. After that I worked for Roush industries. There I worked on production engines as well as race engines (IRL, Petit LeMans and NASCAR). Currently I work for a company whose core business is NBC (Nuclear, Biological, Chemical) protection. So I do a lot of FEA and CFD analysis along with my design work.

What are you trying to achieve with your testing? If you are just looking for an A to B comparison (ie this intake is 20% better than that intake) the bench is fine as long as you have the same error in every run.

What are you doing for your restrictor? Like CP mentions a simple plate with a hole will not cut it. Actually a taper helps, but you get better results with higher continuity. For our LeMans engine we originally used a simple radius tangent to the cylinder where the restriction is measured. We saw a notable difference when we started using conic sections before and after the cylinder. Of course cutting the restrictor is more of a problem.

[Meikol2]

This is the most intersting thread I have read all day.

Sorry, Al, I don't have anything to add other than that.

[alriclam]

No problem Meikol2. My head is spinning after reading the responses too, this stuff is pretty new to me and i'm probably only absorbing 50% of whats being said. Its kinda weird, cause as a captain I'm more of a jack of all trades, but a master of none. Though the "experts" on the each of the individual teams (ie suspension, diff/brake, intake and fluids, etc) should more than make up for it. I figure I'm better off even if I can only understand a portion of whats going on than not learning about this stuff at all.

tzortn, sounds like you took your FSAE experience and continued it as a career path, thats really cool. In terms of your comments, yeah the intake team is trying to get A to B comparisons (which can be decided by our dyno), but also try to improve on individual designs by use of modelling clay and small changes to the clear intakes they've built. This is where it would be nice to have the flow visualization, but its icing on the cake, I'd be happy if they came close to a perfect intake with the A to B comparison. The past few years worth of intakes have been more about space constraints than of theoretical design, so any intake from them will probably be an improvement.

On last year's car, our diffuser serves as the choke point. Basically looks like an hourglass shape but the neck is closer to the beginning of the intake. This year, we hope to have a retangular cross section that tapers gradually from runners to the restrictor.

CP, good call on the rearranging of frames from a slow camera, that might be feasible. I'll have to look more into that.

Oh, and as an update on the problem we were having with our new cams and our thought that we had screwed up the stock springs, we pulled the head today and noticed scorching of both intake valves of two cylinders (3 and 4). We think that the problem had to do with overheating, but not related to our new cams. I'm still worried about why and how it overheated. We are cutting the valves and the seats and hopefully we'll have it fixed......

Thanks guys!

[alriclam]

Here's another technical question: We need to get wheel speed sensors (we're going to use 1 for each wheel on the car, and data log with our new data acq system). Anyone have preferences as to which ones to use, ie hall effect, variable reluctance, or electromagnoresistive (sp?) I'm leaning to hall effect, but the variable reluctance is nice because it doesn't have to be powered.......

[The CyberPoet]

Here's another technical question: We need to get wheel speed sensors (we're going to use 1 for each wheel on the car, and data log with our new data acq system). Anyone have preferences as to which ones to use, ie hall effect, variable reluctance, or electromagnoresistive (sp?) I'm leaning to hall effect, but the variable reluctance is nice because it doesn't have to be powered.......

I don't know what the budget is here, but you might be able to simply adapt the sensor off an BMW motorcycle with ABS braking (or from a similarly equipped Honda VFR, Honda ST1300, etc). The other option is to pull an automotive system directly from an early 90's Audi as a good example (depends on the wheel size/shape contraints -- Audi's solution is quite small). Do you need to take the wheelspeed at the wheel, or will pulling much further inboard (at the axle) serve your unsprung weight equation better? Do you run rotors inboard or outboard? Placing a magnet or two into the rotor material close to the axis of rotation and then using a hall sensor there will give you what you are looking for as well.
Finally, an optical sensor at the axles would also work, if you're not overly concerned with contamination from the road surface...

Cheers
=-= The CyberPoet