Side pressures and attached flow
 

My Edgarwit air curtains give a measured reduction in drag (using the throttle stop method) and tuft testing shows slightly better flow attachment on the side panels behind the front wheels.

My measured pressures without, and then with, the air curtains were as follows:

https://i.postimg.cc/wj470LnP/No-edgarwits.jpg

https://i.postimg.cc/g2j0XrrK/With-Edgarwits.jpg

...giving this percentage increase in pressures:

https://i.postimg.cc/yN9scRK0/increa...-Edgarwits.jpg

But does an increase in pressure actually indicate better flow attachment? After all, more air and faster speed (eg under the car) causes a lower pressure.

I went to all my references to find out about side pressures but this data is almost impossible to find. I do have diagrams for a 1970s Mercedes sedan (and the lines of equal pressure diagram shows lots of variation on the side of the car) but the numbers are too small to read! Lots of CFD images of car pressures just show the side of the car all the same colour - either there's not enough resolution or they're not right, based on the Mercedes data and my measurements.

So, if there is improved flow attachment, do side surface pressures rise or fall?

I decided to do a test on our Mercedes. Rather than set up all the measuring pucks and the EvoScann logger, I used a single puck, static pressure ref provided by a pitot tube, and Magnehelic gauge. That is, total cost of measuring gear under US$100.

Previously with the EvoScann I measured -91Pa in the centre, lower part of the rear door. Today with the Magnehelic, I measured -73Pa (gusty wind, hotter day).

Now, what would happen if I deliberately caused separation in front of this point on the car? I attached a device in front of the area, a device which I was sure would cause separation.

https://i.postimg.cc/BQsp0ft3/IMG-0991.jpg

The measured pressure decreased to -103Pa.

So I am fairly confident in saying that when you are setting up air curtains, and possible even wheel faces, the higher the pressures you can measure on the side of the car, the better. That makes trialling different cardboard (etc) designs much more straightforward.

How much do you think (although of course there is no way of knowing without testing) of the benefit of your air curtains come from the increased surface pressures pulling back less, on the tapered sides of the car and how much do you think is thanks to better flow attachment causing better wake patterns?

I suppose my question is, all other things equal, on a non-tapered car would the resultant drag decrease be the same?

Finally summer in the southern hemisphere, so we get some more hard data posts from Julian!

Approximately what does this add up to in Cd? Any chance of a coast down test?

Julian will tell you that coastdown tests are too unreliable, he may have some cD data though.

I don't know!

As far as I can find, there is almost no coverage in the tech literature as to the mechanism by which changed side pressures/flow influence drag reduction.

Obviously early separation down the side of the car will enlarge the wake, and via tuft testing, you can see that occurring on some cars (eg from behind the rear wheels). But that doesn't answer the question about cars that have ostensibly attached flow but at lower side pressures.

Aerohead's theory that the pressures of the flow at separation imparts a pressure that's reflected in wake pressure doesn't seem to be supported, although I can see these pressures may influence wake pressure. It seems to me that wake pressures are much more influenced by vortex behaviour, which just leads us back to the question of how changed side pressures influence vortex behaviour.

Perhaps the higher pressures at side separation give less of an abrupt pressure change to the wake, resulting in weaker vortex formations? (But that seems a bit chicken and the egg.)

I've asked my experts (different thread) about the relationship between trailing panel pressures and wake pressures, but only one has got back to me - not a good time of the year to be asking.

So if you find anything in the tech lit, please let me know!

I've been doing a lot of aero testing over the last 6 months, so posting quite a lot of hard data.

I am confident that amateurs doing coastdown testing, unless perhaps done only between two very high speeds, does not give valid results.

Refer to SAE papers 950626 - ABCD –An Improved Coast Down Test and Analysis Method and 940420 - A Detailed Drag Study Using the Coastdown Method to see the level of complexity required to get valid coastdown results.

Or just do a series of tests, windows up / windows down in both directions - you should of course see a marked increase in drag with the windows down. I have never been able to detect this via coastdowns. On the other hand, throttle-stop testing sees this change every time.

The throttle-stop testing showed a change in drag with the air curtains of about 5 per cent. I never work from those figure to a proposed change in Cd, as I think such an extrapolation is too unreliable.

Do I detect a change in the weather? If a spoiler and front airdam interact, that interaction has to have some pathway.

I once coasted from the top of the Siskiyou Pass to the first Ashland exit (IIRC some 30 miles) On such a course, might not elapsed time to markers give an averaged result? Downside would be the 60-mile roundtrip between passes. ...and the corners.

https://i.postimg.cc/d1pGvyy4/IMG-1004.jpg

Measured pressure at puck (just behind wheelarch spat).

Without spat: -65Pa
With spat: -53 Pa

Again, pressure rose with better flow attachment.

Now that's interesting!

I measured pressures down the side of the car and in the wake with air curtain ducts (front and rear, but a different design from Julian's, made of sheet metal). I found no change on the doors and wake (both directions showed exactly the same pressure with and without ducts), but a +10 Pa difference on the (tapered) bumper cover behind the rear wheels. Fuel economy testing suggests that overall the ducts are reducing drag slightly; I wonder if this is how?

Yes, very interesting - and thanks for reminding me.

But if we look at the angle of that rear bumper taper, and do a forward force triangle diagram from that increase in pressure, it has to be really tiny ie the taper wouldn't be more than about 10 degrees, would it?

But I can certainly see a change in pressure there having an influence on the wake pattern.

I posted my video to Linkedin, and I have some contacts there who are professional aero people - both CFD and wind tunnel. I've asked if they have any ideas about the relationship between side pressures / degree of attachment and resulting drag.

I just always took it on face value that 'improved side flow attachment = lower drag' but like all aspects of car aero, the more you think about it, the less you realise you know.

Attachment 29854

That looks to me like there is a reduction in angle of those tufts, it looks like the tufts are being sucked downwards towards the sill more without the curtain.

The curtain is blocking the downwards flow? Or there is the possibility that the wing shape is creating a vortex that acts to "seal" the underside of the car, blocking the flow. It looks like any wingtip vortex would be the correct direction to reduce airflow under the car.

As in moving outwards at the bottom and inwards at the top.

Dr Adrian Gaylard (Jaguar Land Rover) and Dr Thomas Wolf (Porsche) have now got back to me about the relationship between pressures and flows on the rear part of the car, and base pressure. Adrian also linked to an interesting paper and reference text.

Professor Joe Katz previously also got back to me on this topic.

There's lots to think about, and Adrian and Thomas take significantly different approaches in their explanations. Adrian's explanation also reminds me of something that Rob Palin (Tesla) told me, and I need to look that up again.

I'll post something when I have thought about it some more.

drag
 

1) If flow attachment can be maintained for the entire length of the body, on a body of progressive cross-sectional contraction, by default, the pressure existing at the separation line will be the highest achievable, translating to highest base pressure ( for that length ), lower pressure drag, and lower overall drag.
2) Minimizing separation is the drag reduction.
3) Un-rounding the curvature, but maintaining a streamlined profile, provided a little extra pressure recovery on the sides, with a drag reduction reward.

I have a couple of questions about a couple of possible issues with throttle stop testing. In order for this type of testing to be reasonably accurate, both the environment of the test location as well as engine power output would have to be constant over the full period of testing.

As I understand it, fuel injected cars change the air/fuel ratios pretty much constantly with changes in engine temperature, etc and the Insight, in particular, has it's lean burn mode that happens automatically. This seems to make a big difference in throttle position/speed as I have noticed and I think MetroMpg noted as well.

Are you controlling for this in your tests?

The other is atmospheric effects which can change engine power significantly, which I'm sure you are aware of. Barometric pressure can change dramatically in the space of a few hours in extreme cases, so one should probably keep tabs on that especially if tests are done on different days or times of year. Humidity also plays a role in engine power.

Have you taken these into account?

throttle stop
 

1) Technically, we'd never know if the BSFC was drifting around on the engine map.
2) And in the past, it was discovered that 30% of streamlining benefit could be lost if gear-matching wasn't accomplished to keep a constant load on the engine after aero modifications.
3) With modern, EFI, and high sampling rates, compared to carbureted vehicles, it may be that BSFC isn't as unstable as in the past. I couldn't prove it one way or another.
4) Universities competing in the mileage marathons typically have a fuel tank that can be weighed to ascertain the mass of fuel consumed over a measured distance. Thermal volumetric expansion wouldn't enter into the calculus. The only unknown you're solving for is mpg based upon the fuels mass. Easy.

Yes of course.

I see a display of air/fuel ratio on my dash the whole time.

My car's lean cruise mode doesn't change during the test runs (if it did change, I'd simple reprogram the MoTeC ECU so it didn't).

I do test runs back to back, with and without the aero changes - normally over a period of 15 minutes or so.

The test runs in the one aero configuration are extremely consistent (typically under 1 per cent variation in top speed).

BSFC, fuel consumption and fuel density have nothing to do with throttle-stop testing.

testing
 

1) altering the road load at any given throttle setting can cause the engine to perform at a less efficient manner.
2) Unless the engine is loaded ( through gear-matching), it will operate at this diminished efficiency.
3) You could lower drag by 30% and you'd never know it.
4) It's the thermodynamic efficiency of the engine which is called into question.
Just sayin'

I don't think you understand very well what is being described by throttle stop testing. We're talking changes in engine rpm of 40 or 50 rpm - basically inconsequential in terms of torque change. (And it's torque we're working with in this case, not power.)

But hey, don't take my word for it. Go and do some testing yourself, eg windows up / windows down and see if the measured drag changes matches what you'd expect in the two different configurations. Practical testing, not just theorising.

not power
 

Torque and rpm is power.
You're not accounting for the actual net amount of power derived from the fuel charge. Sure the rpm can remain steady, but that says nothing about how efficiently, or not, the engine is converting the chemo-thermo energy into mechanical energy.
A course in internal combustion and air pollution was required in my course work. We had an engine test cell with a General Electric engine dynamometer at Texas Tech. BSFC is probably the most important thing to know about an engine.
I've already told you that I did coastdown testing to SAE protocols, and top speed at the Chrysler Proving Grounds. Top speed testing at Bonneville. And General Motors Aerodynamics Laboratory helped with the data reduction.
No theorizing necessary.

Ok, you definitely haven't read how throttle stop testing works.

The engine is at a constant throttle. RPM varies inconsequentially The air/fuel ratio is constant. Ignition timing varies inconsequentially.

But look, don't worry about it. The technique works brilliantly - if you can't get your head around it, that's OK.

constant throttle
 

1) If you've altered the drag you've altered the road load horsepower requirement of the vehicle.
2) If the engine is not kept at a constant load, its BSFC will wander to a less efficient 'topographical region' of the engine's 'map.'
3 It doesn't matter if the throttle position, RPM, and stochiometric ratio are constant, you've altered the thermodynamic properties of the engine and they haven't been accounted for.

All right, it's obvious you don't know much about engines. For example, for a given engine and fuel, the stoichiometric ratio is a constant! (I think you mean air/fuel ratio.)

Don't worry about it. Maybe try doing some mapping of an engine from scratch, have your own chassis dyno, your own wideband air/fuel ratio meter, etc, etc, and then come back to me.

You know, experience and testing versus just theory (yet again!).

For anyone wondering....

In top gear, engine rpm varies little for speed changes. So for example, in the Insight, a few km/h difference in speed may be only 40-50 rpm.

In throttle-stop testing, the engine is at a constant throttle. The air/fuel ratio is constant. Ignition timing varies little - and in fact in the Insight with this variation in engine speed and manifold vacuum, is constant. Volumetric efficiency varies inconsequentially. Internal frictional losses in the engine vary inconsequentially.

As is then obvious, the torque output of the engine is very nearly constant over this very small change in rpm. This means the push backwards on the road by the wheels is very nearly constant.

The power output of the engine varies a little (very little, but a little) but that's why when we do the maths to calculate change in drag, we use the square rule (for force) not the cube rule (for power).

It's not like this was all just dreamed up in some random way: it was thought-through very carefully, discussed with some top experts for their thoughts, and then tested to see if a deliberate change in drag gave the expected test results. (Which it did.)

As with any car testing, you can do it really badly and get completely unreliable results (eg testing in low gear, on a peaky turbo engine at revs where it comes onto boost, on two days with completely different weather, and with different run-up speeds to the test section) but you can also do it and get excellent, repeatable results - and vastly better than normal coastdowns.

My new way of measuring..........................
 

1) I located the videos back on page 5 of the forum.
2) I wanted to revisit them before responding.
3) The video portion plays, but there's no audio.
Any idea on how to proceed?