Intake flow can be hysteretic (in a baaaad way)

[Walken100]

Today I got to witness and reproduce hysteresis in the intake flow on one of my cylinders. I've been comfortable that it happens but I hadn't demonstrated it.

Some background; One of my trial ports consistently flows 232 cfm at .5 and, just as consistently flows like crap above .5. It is a pretty hard drop off. When adjusting the lift, the system becomes violently turbulent at .54. The flow drops way off, I get a headache from port noise, my string vibrates dramatically and flow path starts going across the top of the short turn (yes top). When I place a flow ball at the bottom of the entrance all is well and flow moves to 239 cfm (god if I could only keep that flow).

Now what is both interesting and a bit eye opening is that once the violent turbulence starts (at .54) it doesn't go away until I close the valve past .51.

To me, the implication is that in a running engine I'm on the cusp of creating very ugly flow that might persist for a substantial portion of the intake cycle. Or said another way, CFM taint the only thing that matters.

I've seen it written that you want flow to persist well past your target valve lift. I hadn't previously found a reason why..

Just sharing for general info, discussion, beer conversation etc.......

 

[aussie7mains]

Today I got to witness and reproduce hysteresis in the intake flow on one of my cylinders. I've been comfortable that it happens but I hadn't demonstrated it.

Some background; One of my trial ports consistently flows 232 cfm at .5 and, just as consistently flows like crap above .5. It is a pretty hard drop off. When adjusting the lift, the system becomes violently turbulent at .54. The flow drops way off, I get a headache from port noise, my string vibrates dramatically and flow path starts going across the top of the short turn (yes top). When I place a flow ball at the bottom of the entrance all is well and flow moves to 239 cfm (god if I could only keep that flow).

Now what is both interesting and a bit eye opening is that once the violent turbulence starts (at .54) it doesn't go away until I close the valve past .51.

To me, the implication is that in a running engine I'm on the cusp of creating very ugly flow that might persist for a substantial portion of the intake cycle. Or said another way, CFM taint the only thing that matters.

I've seen it written that you want flow to persist well past your target valve lift. I hadn't previously found a reason why..

Just sharing for general info, discussion, beer conversation etc.......

OK, do consider that the flow bench can only measure steady state dry flow, the airflow in reality is never like that, and flow is always turbulent, never laminar. The calculation for reynolds number (used for determining laminar or turbulent flow) in the tiny holes that ports are will give you well turbulent flow, to get laminar flow the port would need to be like an A/C duct. Also consider that at 0.5 lift you are approaching the point where the curtain area will equal the head area of the valve which is thoereticaly where maximum flow(least resistance) occurs. The instabilty you have found IMO tells you very little about what happens during the cylinder cycle. If you look at really successful engines, like the Cosworths you will find the ports are not huge and are usually round, and heavily downdrafted. If you also look at modern engines, the Godzilla is a good example, you will find the intake port is almost verticle over the back of the valve. Another good example is the intake port on the 250-2V falcon head, although this port is too big for the valve it has. So, the most valuable things on the flow bench is the low restriction of the port. Given what most inline six heads are like, bloody awful, you can only try for a reduction in restriction, and the intake manifold is almost as important in this regard. If you look to the really good inline engines, like the BMW engines, thats a good guide. The bimmer sixes easily good 1bhp per ci and totaly civilised at the same time. Even our barra is good for that, admittedly these are 4valve engines which really makes a difference. Anyway, my to bobs worth.

 

[Walken100]

OK, do consider that ..

Understand all of that. The issue is hysteresis. I realize that it may not happen in an operating engine but it may be much worse. If I can't fix the cause of that violent turbulence I'll run with a different approach on the intake ports. I have one with better average follow through out the lift curve with very little drop off until closer to .7 and no hysteretic event.

 

[Wesman07]

Very well said Aussie.

You’ll never understand what’s happening if you think the issue is related to laminar and turbulent flow. It’s not hysteresis. Who says the boundary layer has to break and reattach under the same condition?

 

[Walken100]

Very well said Aussie.

You’ll never understand what’s happening if you think the issue is related to laminar and turbulent flow.

I haven't used the term laminar..... My initial post has nothing to do with laminar flow. The term I used, "violently turbulent" wasn't an attempt to classify flow IE laminar vs turbulent rather it was a colloquialism to indicate a change in the intensity of turbulence within a flow system. Formal turbulence classification is a wee bit more complicated.

Who says the boundary layer has to break and reattach under the same condition?

This is one definition of hysteresis. If you plotted flow vs lift both up and down the scale (at .01 increments) it would form a "hysteresis loop".

Should I take it that the concept of step change in the intensity of turbulence that persists beyond its originating conditions (in both directions) isn't a concern? I understand that a flow bench is not an engine however this makes me feel that the flow in that port is going to create issues beyond it's flow numbers. In other words even if this port shape had the highest peak and average flow of my other ports I would be concerned about using it. I have another port shape that doesn't exhibit this behavior. The peak flow is lower but the average is nearly the same and the flow carries to .65 before it starts to taper off without the step function change in turbulence intensity.

If all this discussion is to be interpreted as "its fine" that would be an interesting conclusion.

 

[Wesman07]

It’s not “fine”, it’s unstable. If it’s unstable under steady state conditions, it will likely be very unsteady under dynamic conditions. Try adjusting the test pressure and see how it compares to the others. What’s tripping you up is you likely as measuring multiple variables.

 

[Walken100]

It’s not “fine”, it’s unstable.

We're aligned on that! I believe the hysteresis would have made it even worse in an engine.

If it’s unstable under steady state conditions, it will likely be very unsteady under dynamic conditions. Try adjusting the test pressure and see how it compares to the others. What’s tripping you up is you likely as measuring multiple variables.

Not tripped up yet. I was just exploring the hysteresis. I've just started troubleshooting the actual turbulence. So far I've cleaned up a couple of minor concerns which improved the peak flow, got rid of the hysteresis but not the step function turbulence. More work to do. I did vary the test pressure. Below .54 I couldn't couldn't create the turbulence. Above .54 it had little impact (I need play with this more but dinner was served)

 

[Wesman07]

The take away is boundary layer separation is not a just an isolated event. When that happens it creates room for other issues to expand on new territory.