Turbulent flow question

[Jay Kempf]

Does turbulent flow have to impinge on a surface to create an increase in parasitic drag or do vortexes in free streams also contribute? If that vortex was not associated with lift it is not part of the induced drag calculation. It would have to show up in an interference drag estimation I am guessing.

 

[Topaz]

Any energy that creates turbulence coming off an airframe is, if you follow the energy back far enough, coming from the engine and the thrust it provides. So, in the case of turbulent flow and vorticity, the energy necessary to "spin up" that air from rest is energy that's not available to push the airplane forward, whether the turbulence or vortex impinges on the airframe downstream or not. Having it impinge on the airframe just makes matters worse, as it disturbs and energizes air at that point and, if you trace it all back, that energy is coming from the engine, too. Ultimately, all that energy is dissipated as heat as the spun-up air molecules bounce into others and make them move around faster - an airplane passing raises the temperature of the air if moves through ever so slightly. In a nutshell, we're taking fusion energy from the Sun that fell on Earth millions of years ago, passing it through extinct plants and dinosaurs, and combusting reconfigured organic matter in which they stored that energy, so that we can push our airplane forward - and spin up and therefore heat the air through which it passes. Go team.

A hypothetical "no-parasite-drag" airplane would leave completely undisturbed air in its wake, and have perfect laminar flow over every square inch of the airframe. Even then, the work necessary to displace the air occupied by the airframe itself and then bring it back together again is energy that's sucked from the thrust of the powerplant, too.

As far as "bookkeeping" goes, I'd tally it with parasitic drag. Interference drag is usually meant to refer, I believe, only to drag caused by the flows around two objects interfering with each other and the drag rise that results, compared to the two objects alone - like where wings attach to fuselages, where wing struts attach to wings, etc.

You can also think of induced drag as the energy necessary to do the work of either lifting the airplane or throwing an airplane-equivalent mass of air downwards, depending on if you like Bernoulli or the downwash theory of lift.

 

[Jay Kempf]

The question is about throwing air sideways. So not induced or related to lift at all. If you have for instance two jets of air leaving the aft fuselage and converging in a horizontal plane but not crashing into any area and so not creating skin friction drag per se. Does it creat "parasitic" or "induced drag" or none at all if it isn't associated with wetted area? The only calculation has wetted area in it plus some other factors. So if you aren't throwing a mass of air downward and it isn't creating an offset and is just in the wake does it rob energy from the system that can be correlated to the drag budget sum up?Total drag is where I am going with this. If you look at a span and you take the area disturbed as a circle with the span as a diameter then you might be able to estimate all the little and big contributors related to that area I am thinking that wake turbulence and interference drag may be able to be estimated if you can tie them to the area of the larger circle as a percentage of the whole. This is just brain storming for something I am trying to figure out.

 

[Topaz]

I think where you're going awry is that the calculations we have for drag - at least, the ones we would use here as opposed to say, what NASA uses - are approximations and simulacrums of reality. Empirical representations. We don't really model the entire flow around an airframe with the parasite drag formula - it just spits out an approximation based on the variables it employs, that comes fairly close to the results of what the air is really doing.

This is getting well out of my expertise but, to do the kind of "complete flow" calculations you're describing, including turbulence, requires full solutions to the Navier-Stokes equations, something only recently becoming possible with cutting-edge supercomputers and software. In such an analysis, the distinction between "parasite" and "induced" drag pretty much disappears, since that distinction is really just an artificial labeling to support the traditional way of calculating them.

Looking back at what you're saying directly above, if it's related to lift, it's "induced" drag. If it's not, it's "parasite" drag. "Interference" drag is a subset of the latter. A vortex or bubble of turbulent, swirling air, that comes off the airplane and which isn't formed off the downwash field of the wing or tail (and so isn't related to the production of lift) is still creating parasite drag, even if it never touches the airframe again. The energy necessary to spin that air up into a vortex or tumble it into turbulence came from the motion of the airplane moving forward through the air, and what moves the airplane forward (and keeps it from slowing down) is the thrust of the propulsion system. The little portion of thrust that spun up the vortex isn't available to keep the airframe moving forward, and so that sapping-off of energy from the propulsion system is counted as a "drag".

 

[Jay Kempf]

I think where you're going awry is that the calculations we have for drag - at least, the ones we would use here as opposed to say, what NASA uses - are approximations and simulacrums of reality. Empirical representations. We don't really model the entire flow around an airframe with the parasite drag formula - it just spits out an approximation based on the variables it employs, that comes fairly close to the results of what the air is really doing.

This is getting well out of my expertise but, to do the kind of "complete flow" calculations you're describing, including turbulence, requires full solutions to the Navier-Stokes equations, something only recently becoming possible with cutting-edge supercomputers and software. In such an analysis, the distinction between "parasite" and "induced" drag pretty much disappears, since that distinction is really just an artificial labeling to support the traditional way of calculating them.

Looking back at what you're saying directly above, if it's related to lift, it's "induced" drag. If it's not, it's "parasite" drag. "Interference" drag is a subset of the latter. A vortex or bubble of turbulent, swirling air, that comes off the airplane and which isn't formed off the downwash field of the wing or tail (and so isn't related to the production of lift) is still creating parasite drag, even if it never touches the airframe again. The energy necessary to spin that air up into a vortex or tumble it into turbulence came from the motion of the airplane moving forward through the air, and what moves the airplane forward (and keeps it from slowing down) is the thrust of the propulsion system. The little portion of thrust that spun up the vortex isn't available to keep the airframe moving forward, and so that sapping-off of energy from the propulsion system is counted as a "drag".

I am looking for a way to sum "interference" which is a subset of skin friction drag, which might have a component that isn't calculated as a function of wetted area. So looking back at that large circle I am thinking that 90% can be accounted by simple manual methods that are reliable. The ability to really look at the remaining 10% might be a path of diminishing returns for Cessna but not for Will Scheumann, right? If you have a difficult part of the design that is contributing the say half of the remaining 10% of the drag budget that would be something to really concentrate on with your full CFD analysis in detail I am thinking.

Did you make up "simulacrums?"

 

[Birdman100]

The question is about throwing air sideways. So not induced or related to lift at all. If you have for instance two jets of air leaving the aft fuselage and converging in a horizontal plane but not crashing into any area and so not creating skin friction drag per se. Does it creat "parasitic" or "induced drag" or none at all if it isn't associated with wetted area?

There will be some influences, but without knowing specific problem/setup you are talking about its hard to evaluate. Jets from jet engines do drag some air around but it is hard to tell if they make any extra drag due to this, as they generally add energy. I would say that influences are minor if they happen at aft fuselage. Problem is worse if occurs more in front introducing turbulence.

Drag, as every other aerodynamic force has to be related to wetted area after all. Force has to act on some existing object. But without adding extra wetted area you can have increased drag due to: 1) changed pressure field around the object; 2) more turbulent air.

This is getting well out of my expertise but, to do the kind of "complete flow" calculations you're describing, including turbulence, requires full solutions to the Navier-Stokes equations, something only recently becoming possible with cutting-edge supercomputers and software.

Not actually, it can be done on a powerful PC (simulation can take several days however!) with Raynolds-Average Navier Stokes equations (not the exact solutions). The accuracy depends on the specific problem and overall "quality" of simulation. It predicts and catches some interesting phenomenons.

In such an analysis, the distinction between "parasite" and "induced" drag pretty much disappears, since that distinction is really just an artificial labeling to support the traditional way of calculating them

Yes indeed. Like that.

 

[Topaz]

I am looking for a way to sum "interference" which is a subset of skin friction drag, which might have a component that isn't calculated as a function of wetted area. So looking back at that large circle I am thinking that 90% can be accounted by simple manual methods that are reliable. The ability to really look at the remaining 10% might be a path of diminishing returns for Cessna but not for Will Scheumann, right? If you have a difficult part of the design that is contributing the say half of the remaining 10% of the drag budget that would be something to really concentrate on with your full CFD analysis in detail I am thinking.

I've yet to see a really rigorous, non-CFD analysis for interference drag over arbitrary shapes. Everything I've seen is an approximation going back to... Munk, I think. Just too many variables. I know what you're working upon (haven't been ignoring your letter - just stupendously busy), and I understand why you want to go here, given what you're trying to achieve. If you're talking about the area of the airframe I think you are, remember that the prop flow field will have a positive effect as well, although that would disappear power-off. I just don't know of any non-CFD analyses for interference drag for an arbitrary shape. I wish I could be of more help. AFAIK, until very recently, even sailplanes were doing a lot of empiric testing on that area, with tuft-testing and oil-flow testing of wind-tunnel models and actual, finished aircraft. More-current designs are modeling the interference with CFD, but still assuming that the flow is largely laminar, since I don't believe that consumer- or "prosumer"-grade CFD tools are up to modeling the laminar/turbulent transition, plus fully-turbulent flow over arbitrary shapes like this. I could be wrong - I'm not exactly "current" in that area.

 

[Dana]

You can also think of induced drag as the energy necessary to do the work of either lifting the airplane or throwing an airplane-equivalent mass of air downwards, depending on if you like Bernoulli or the downwash theory of lift.

Bernoulli and downwash (aka Newton) aren't competing theories; they're two different ways of looking at the same thing, two different but complementary ways of mathematically describing it.

Depending on what we're trying to calculate and the tools we have available, it makes sense to break it up in various ways. Lift and drag aren't really two separate forces; we just break it up into two orthogonal vector components to make the math easier. Similar with induced and parasite drag, then we break parasite up into form drag (pressure effects) and skin friction (viscosity effects).

But when you're talking about separated flow (turbulent flow is usually taken to mean a boundary layer condition) the math gets really hairy, to the point that traditionally all data was empirical rather than calculated; to calculate it requires some serious computing power that has only recently become available.

Dana

 

[autoreply]

I've yet to see a really rigorous, non-CFD analysis for interference drag over arbitrary shapes. Everything I've seen is an approximation going back to... Munk, I think. Just too many variables. I know what you're working upon (haven't been ignoring your letter - just stupendously busy), and I understand why you want to go here, given what you're trying to achieve. If you're talking about the area of the airframe I think you are, remember that the prop flow field will have a positive effect as well, although that would disappear power-off. I just don't know of any non-CFD analyses for interference drag for an arbitrary shape. I wish I could be of more help. AFAIK, until very recently, even sailplanes were doing a lot of empiric testing on that area, with tuft-testing and oil-flow testing of wind-tunnel models and actual, finished aircraft. More-current designs are modeling the interference with CFD, but still assuming that the flow is largely laminar, since I don't believe that consumer- or "prosumer"-grade CFD tools are up to modeling the laminar/turbulent transition, plus fully-turbulent flow over arbitrary shapes like this. I could be wrong - I'm not exactly "current" in that area.

Fuselage/wing design is notoriously hard. A lot of data has been gotten from prototypes (the Akaflieg Braunschweig B14 and the Flugtag Esslingen E14 come to mind), wing tunnels etc, but even with a shoulder wing it's surprisingly hard to model it well.

The Mü31 is the future IMHO, it's one of the first designs where several configurations have seriously been compared w.r.t. interference drag.
http://www.akaflieg.vo.tu-muenchen.de/index.php/mue-31

For boom/wing intersections, empirical data is fairly accurate. For fuselage/wing intersections, it's often almost impossible to estimate. Low wings are almost impossible to get right. You can only "fix" things, i.e. resolve detached flow, but you'll still end up with considerable intersection/interference drag. No surprise that virtually all fast, well developed low-winged planes still have detached flow aft of the wing.

Interference/intersection drag at itself is actually a nonsensical concept, especially once we're talking about laminar flow, since you're altering the flow pattern on both bodies (fuselage/wing) considerably, using the aero data for both original bodies is in fact virtually useless.

This thread covers the subject in a lot more depth:
http://www.homebuiltairplanes.com/f...ference-drag-laminar-airframe-pylon-wing.html

 

[Jay Kempf]

Interference/intersection drag at itself is actually a nonsensical concept, especially once we're talking about laminar flow, since you're altering the flow pattern on both bodies (fuselage/wing) considerably, using the aero data for both original bodies is in fact virtually useless.

This part is interesting. There are so many factors and people tend to compare apples and orangutans when trying to estimate things. Almost every example both pusher and tractor of wings and fuselages started with some design and when they figured out they had the fattest part of the fuselage exactly lined up with the fattest part of the top of the wing they wonder why they find massive areas of air being torn apart. Not suggesting area ruling but pressure gradient separation and then pressure recovery areas. For the same reason that it is hard to design a winglet, you are trying to visualize a flow in 3D and get the desired result in the free stream by meddling in the wetted are that directly contributes to the flow in the free stream. Haven't run CFD yet but I am trying to come up with some sort of general assumption of cause so that I can find a way to change the assumptions to get rid of the affect. If the assumption of a low wing is that adding large areas of fast flowing low pressure areas together causes the problem then certainly there should be a way to rearrange the components to not do that and get a gain. My premise is staring with some simple assumptions like controlling laminar flow all the way to the max thickness of the wing and forward fuselage and taking into account the circulation in front of the wing and it's slope you should be able to back calculate. the transition area around the lowest pressure on top of the wing and somehow put geometry in place to keep it from creating downstream vortex effects at two critical angles of attack. Velocity should not matter if the streamlines stay behaved. Power on and low AOA shouldn't be a problem because CL is so low. The streamlines can be modeled but you have to have some sort of prediction based on assumptions before creating the geometry to test. I have a PHD aerodynamist with the right experience and the right tools to help at that point as long as I do the actual work and run the problem it will be overseen and my inexperience won't be the issue of getting the correct results and then analyzing them.

Another assumption that I am sort of working towards is that there should be nothing in that critical zone behind the max thickness of the wing to perturbed separation. By carefully shaping what comes after the max thickness and playing with pressure gradients and infill streamline shapes one should be able to find reductions in separated flows. Even just a thin boundary layer would be adequate that isn't laminar but is somewhat behaved, right? Will Schuemann did quite a bit of this sort of work a long time ago when it came to arranging planforms for pressure gradients on changing geometry. Get the negative and positive pressures to form a sort of constructive interference for some net gain and you get efficiency gains in 3D flows.

My original question is based on breaking up the large circle into segments that need attention. So if 80-90% of the exposed span is flowing as you would like it to and you are only working with the root 5-10% and 60% of the forward wetted area of the fuselage is flowing perfectly then you are down to a small area of affected wake in the overall drag budget. How much can that area contribute if it was fully separated and turbulent? How much if you cleaned that flow up to have 50% less wake disturbance?

 

[Himat]

Drag, as every other aerodynamic force has to be related to wetted area after all.

I do not agree on that one. To make aerodynamic drag a force must act over time to transfer energy, but this force doesn't have to be related to a wetted area. Given a certain force, mass and time the energy transfer can be calculated and hence the drag. An electromagnetic water jet work without having any "area", there might be possible ways to move air without being in contact with it too.

 

[Himat]

...

My original question is based on breaking up the large circle into segments that need attention. So if 80-90% of the exposed span is flowing as you would like it to and you are only working with the root 5-10% and 60% of the forward wetted area of the fuselage is flowing perfectly then you are down to a small area of affected wake in the overall drag budget. How much can that area contribute if it was fully separated and turbulent? How much if you cleaned that flow up to have 50% less wake disturbance?

Sidestepping a bit.

To find the total drag of an vehicle moving in air all energy that is transferred to the air must be accounted for.

For bookkeeping terms like induced, parasite, interference, skin friction, wave and others are used, but in the there is two forms of aerodynamic drag on an aircraft, drag necessary to keep the airplane flying and drag unnecessary to keep the airplane flying.To minimize drag the part that is not necessary to keep the airplane flying must be identified and minimized.

Back to your first question, it might not matter if the vortices act on a surface or not as long as the energy spent to create the vortices is accounted for. As I understood some of Synergy's (John's) posts, he looked at aerodynamic drag as a matter of energy transfer in a flow field.

 

[Birdman100]

I do not agree on that one. To make aerodynamic drag a force must act over time to transfer energy, but this force doesn't have to be related to a wetted area. Given a certain force, mass and time the energy transfer can be calculated and hence the drag. An electromagnetic water jet work without having any "area", there might be possible ways to move air without being in contact with it too.

I think you misunderstood me. It may be that I wasnt clear however. What I meant was that all phenomenon in airfield (turbulence, separation (read turbulence again) and so on) has to transfer their effect/influence to the surface of the body to manifest as a force (drag). I really cant see how could air show its influence without interaction with body walls (am focusing on classic aerodynamics and mechanical concept here). That doesnt mean that some distant disturbance doesn have its effect on a body, it has, but in a way that the effect propagate through the fluid, combines with other effects and ends up acting on the surface of the body in some form at last. Without reaching the surface in some way, there is no way to measure its influence.

An electromagnetic water jet work without having any "area", there might be possible ways to move air without being in contact with it too

Im not sure how it works exactly but i see your point. There are many examples of "invisible" forces, but the origin of these forces are somewhat different as to aerodynamic forces so I wouldnt compare. However, no mater what kind of device you have, the force it produces acting on something (to water in case of water jet), is equal to reactive force that acts from fluid to device (3. Newtons Law). That reactive force has, after all to act on some material object. The same with airplanes, with the exception that the airplanes body is the only object the force can transfer to.

 

[don january]

I know so little about this question, mainly the only Turbulent flow I have is after my wife feeds me burrito's

Does turbulent flow have to impinge on a surface to create an increase in parasitic drag or do vortexes in free streams also contribute? If that vortex was not associated with lift it is not part of the induced drag calculation. It would have to show up in an interference drag estimation I am guessing.

 

[RonL]

Does turbulent flow have to impinge on a surface to create an increase in parasitic drag or do vortexes in free streams also contribute? If that vortex was not associated with lift it is not part of the induced drag calculation. It would have to show up in an interference drag estimation I am guessing.

Find an old "Lava Lamp" and watch it for awhile, you will see a lot of "cause and affect", that might give a few hints of how things in motion affect other surfaces in the different amounts of closeness.

 

[autoreply]

This part is interesting. There are so many factors and people tend to compare apples and orangutans when trying to estimate things. Almost every example both pusher and tractor of wings and fuselages started with some design and when they figured out they had the fattest part of the fuselage exactly lined up with the fattest part of the top of the wing they wonder why they find massive areas of air being torn apart. Not suggesting area ruling but pressure gradient separation and then pressure recovery areas. For the same reason that it is hard to design a winglet, you are trying to visualize a flow in 3D and get the desired result in the free stream by meddling in the wetted are that directly contributes to the flow in the free stream. Haven't run CFD yet but I am trying to come up with some sort of general assumption of cause so that I can find a way to change the assumptions to get rid of the affect. If the assumption of a low wing is that adding large areas of fast flowing low pressure areas together causes the problem then certainly there should be a way to rearrange the components to not do that and get a gain. My premise is staring with some simple assumptions like controlling laminar flow all the way to the max thickness of the wing and forward fuselage and taking into account the circulation in front of the wing and it's slope you should be able to back calculate. the transition area around the lowest pressure on top of the wing and somehow put geometry in place to keep it from creating downstream vortex effects at two critical angles of attack. Velocity should not matter if the streamlines stay behaved. Power on and low AOA shouldn't be a problem because CL is so low. The streamlines can be modeled but you have to have some sort of prediction based on assumptions before creating the geometry to test. I have a PHD aerodynamist with the right experience and the right tools to help at that point as long as I do the actual work and run the problem it will be overseen and my inexperience won't be the issue of getting the correct results and then analyzing them.

If that PHD is worth his salt, he'll point out that there are not that many general rules. We simply don't have the power yet to directly design complex flow sections for this kind of work and as such are (and will be) stuck with trial and "fixing" things for the near future.

Well, there's one, low wings suck w.r.t. interference drag and are rather draggy in general. Other reasons usually prevail despite this.

But please note that all the aircraft in competitive sports (Reno certainly isn't even close) have mid to high wings for exactly this reason.

As long as a low wing is still on, I wouldn't worry too much and simply patch up flow to a decent level once it's flying. No need to go for small improvements in a fundamentally draggy concept

My original question is based on breaking up the large circle into segments that need attention. So if 80-90% of the exposed span is flowing as you would like it to and you are only working with the root 5-10% and 60% of the forward wetted area of the fuselage is flowing perfectly then you are down to a small area of affected wake in the overall drag budget. How much can that area contribute if it was fully separated and turbulent? How much if you cleaned that flow up to have 50% less wake disturbance?

Having a 6" wide separated section at both roots of a bad sailplane junction (a manned one that is..) easily doubles drag.

 

[mcrae0104]

...all the aircraft in competitive sports (Reno certainly isn't even close)...

Huh?

 

[autoreply]

Huh?

Competitive in the sense that the designs are competitive and thus small changes are significant. If you look at the Reno racers, the difference between sports gold winners and 5th place is like a 40-60% higher power/drag ratio. For what I consider competitive aircraft racing (sailplanes and the red bull air races), we're talking about 1-2% in drag and sometimes way less than that between the number 1 and 5.

 

[TFF]

I would not really call Red Bull stuff racing; more like falling with style.

 

[Jay Kempf]

If that PHD is worth his salt, he'll point out that there are not that many general rules.

He's worth his salt. Flow patterns are not drag predictions just visualization techniques to get from one condition to another. Try modelling a compressor stall sometime. Tough stuff.

Generally I feel that almost every attempt at doing a low wing has been done by designing the structure first and then trying to figure out the aerodynamics after. Most times in small planes all the fattest parts line up and create a sort of mess where the bulk of the low pressure peaks are. Sacrificing or deleting pressure gradients in one area to pull hard to evacuate another area longitudinally has not ever really been examined. I think there is room to follow behind Mike Arnold and to try to understand such phenomena. And so it goes. There are only two choices of shapes to look at and move around.... convex and concave. Most simple concave blends have not worked. So that leaves the other or combinations of the two changing along a longitudinal line. The other one is to shed off into the free flow. If separated flow starts but doesn't have any real wetted area to follow then it is minimized. How much of the large tailboom of some of the competition sailplanes are in the turbulent zone? For what distance after separation? 3-4 meters?