naca design

[tspeer]

Originally Posted by akira88

...
This is my alfa 0 for OPERi, personally I think the graph looks really weird and messy...

You probably have coordinates that are not smooth. Given the frequency of the wiggles, I'd guess you could use more coordinate pairs in your input file, too. The NACA coordinates were defined before the advent of computers, and it was a big deal to calculate more coordinates or to greater precision. Plus, there are some NACA 6-series coordinate tables that have minor errors in them, but I don't know which ones.

What I usually do for a case like this is go to MDES and use FILT to smooth out the wrinkles in the pressure distribution. Then EXEC to redesign the section and execute PANE from the main menu. That will leave you with a section virtually identical to the input shape, but with smooth coordinates and a decent number of coordinate pairs.

Edit - oops, I should have read down a bit farther in the thread to where Dr. Drela identified the real problem.

[tspeer]

Originally Posted by whoosh

Tom and Daiquirio
\often I use 2205 ss for rudder stocks on my alloy yachts, I do this for A strength and B because I have been taught thinner sections are better and the sections would be very thick if I were to go al alloy stock
what say you two?

I'd keep doing the same thing.

The thin sections will have less drag at low angles, which is where a well-balanced yacht should be sailing. You haven't said anything about the rudders tending to stall, so it doesn't sound like there would be any real benefit to going thick for thickness sake.

[Guest62110524]

Originally Posted by tspeer

I keep doing the same thing.

The thin sections will have less drag at low angles, which is where a well-balanced yacht should be sailing. You haven't said anything about the rudders tending to stall, so it doesn't sound like there would be any real benefit to going thick for thickness sake.

THANKS no they do not stall they are very good, not my design apart from the structure, but very heavy
I saw a billet in Auckland 300mm dia for a superyacht stock, sposed to be the biggest in world, , I would have gone to 2205

[tspeer]

Originally Posted by daiquiri

Hello Mr Drella,
do you perhaps know why sometimes the analysis doesn't converge at just one single alpha, while it runs smoothly for all the others in the sequence, even those very close (+\- 0.5 deg.) to the "problematic" one? What indicator should one check out, in order to understand which corrections should be done? :\

I'm sure Prof Drela can give you a much better answer, but often it seems to me to be cases where there's a strong interaction between the location of the transition point and the lift. I suspect that when the transition is a bit forward, the boundary layer is thicker at the TE, which reduces the lift. Then when it calculates at the reduced lift, the transition point moves back, thinning the boundary layer and increasing the lift. So it can't get the solution to settle down enough to meet the convergence criteria. You might try fixing the transition temporarily to see if that helps.

I usually use increments of 0.2 deg, so I often don't mind if I can just restart on the other side of the problem and miss a point or two.

[daiquiri]

Originally Posted by tspeer

If I were to actually use any of those, I'd thicken the trailing edge by maybe half a percent chord, with a squared edge and faired in over the whole section. Basically a parabolic wedge inserted into the thickness distribution. That would probably have negligible effect on the fluid dynamics and would add a lot of stiffness to the trailing edge.

Good point and good info.

Originally Posted by tspeer

I suspect that when the transition is a bit forward, the boundary layer is thicker at the TE, which reduces the lift. Then when it calculates at the reduced lift, the transition point moves back, thinning the boundary layer and increasing the lift. So it can't get the solution to settle down enough to meet the convergence criteria. You might try fixing the transition temporarily to see if that helps.

Good info again! Maybe then (if the above description of the problem is correct) it would be useful to allow the user to modify the value of relaxation parameter for the "problematic" airfoil configuration....

Thanks again.

[akira88]

Originally Posted by daiquiri

I use this this DOS software to generate coordinates file for NACA airfoils:
http://tracfoil.free.fr/airfoils/downloads/nacalte.exe
I have tested it on my PC, containes no viruses.

Once you have generated the .dat file with that software, XFOIL should work smoothly. I'm enclosing some of the results obtained. Btw, my graphics is different because I use XFLR5. It is an Opensource GUI software which uses XFOIL engine for calculations, so the results should be identical to those obtained with plain XFOIL.

Cheers!

The coordinate generator and XFLR5 are really useful, thanks!!
Regarding the Ncrit value, what would be the suitable value? My turbine is going to spin slow, probably lesser than 60rpm at about 1.5m/s, so is value 3 ok for this case?
And I also wonder is xfoil going to work well with Renold's number below 300k?

The angle of attack on a turbine is a function of how it is loaded. You have to select an operating point for the turbine and work backwards. Once you have a design for those conditions you check it across the range of conditions it will experience.

Also you will get a better result if you optimise the AoA across the span so a constant AoA is not the best.

What are your operating conditions eg The stream flow rate? The power you want to get out? What are the constraint - reaction force, swept area, efficiency and/or high energy recovery.


Something else to note is that the Re# will change over the span of the blade as well as being a function of many other parameters such as stream flow rate, blade chord and speed of rotation. Using a single polar based on a single Re# may not produce very reliable results.

Ya, my idea was to find out which AoA have the highest lift, and I will make my blade starting from zero angle slowly increase until my desired AoA (which got highest lift) at the outermost part of the blade, forming a twisted blade.

By doing so I can get an almost uniformly increasing lift force across the span, hence the velocity for the blade spinning too will increase across the span too. And according to the formula w=v/r, every section of the blade will be spinning at the same angular velocity.
Is my idea workable?

My turbine will be working in an open channel flow with constant velocity within the range of1-2m/s, and in a channel of around 0.3x0.3x5m.

Thanks again for everyone who gave me a helping hand.

[Guest625101138]

Originally Posted by akira88

..
Ya, my idea was to find out which AoA have the highest lift, and I will make my blade starting from zero angle slowly increase until my desired AoA (which got highest lift) at the outermost part of the blade, forming a twisted blade.

By doing so I can get an almost uniformly increasing lift force across the span, hence the velocity for the blade spinning too will increase across the span too. And according to the formula w=v/r, every section of the blade will be spinning at the same angular velocity.
Is my idea workable?

My turbine will be working in an open channel flow with constant velocity within the range of1-2m/s, and in a channel of around 0.3x0.3x5m.

Thanks again for everyone who gave me a helping hand.

If the turbine is not shrouded then the span efficiency will be low if you try to get high pressure at the tip. The air will just flow around the outer edge.

It seems you are making a very small turbine. Less than 0.3m in diameter. You need a foil suitable for low Re#.

Work out the Re# at 75% of the blade radius and then check the polar plot of you chosen foil at this Re#. Typically thinner foils around 8% thick work best at low Re#.

How much power were you hoping to extract? What is the intended use?

Rick

[tspeer]

Originally Posted by akira88

Regarding the Ncrit value, what would be the suitable value? My turbine is going to spin slow, probably lesser than 60rpm at about 1.5m/s, so is value 3 ok for this case?

I'd run a range of Ncrit values, including Ncrit=1 and also fix the transition very near the leading edge (but still behind the stagnation point, or the flow will be laminar) to simulate the fully turbulent case. That way you can be sure your section will work regardless of the amount of laminar flow you actually achieve.

And I also wonder is xfoil going to work well with Renold's number below 300k?

Yes, it will. XFOIL's ability to handle laminar separation bubbles makes it a good choice for low Reynolds numbers.

Ya, my idea was to find out which AoA have the highest lift, and I will make my blade starting from zero angle slowly increase until my desired AoA (which got highest lift) at the outermost part of the blade, forming a twisted blade.

By doing so I can get an almost uniformly increasing lift force across the span, hence the velocity for the blade spinning too will increase across the span too. And according to the formula w=v/r, every section of the blade will be spinning at the same angular velocity.
Is my idea workable?

Linearly increasing lift across the span is not what you want. Uniform induced velocity across the span is the most efficient.

Low angle of attack in the center doesn't make much sense to me because the rotation rate is slow there. If you have an approximately uniform inflow velocity across the span, you will need a comparatively high angle of attack there. Then the blades need to be twisted to keep the angle of attack at the tips from getting too low.

You might be able to use Java Prop to design the blade shapes. The program was written for propellers, but the blade element theory for a turbine is similar.

Depending on the number of blades you have, it may be important to find an airfoil program that handles cascades. There is a lot of mutual interference between blades in a cascade that alters the section characteristics.

[akira88]

If the turbine is not shrouded then the span efficiency will be low if you try to get high pressure at the tip. The air will just flow around the outer edge.

It seems you are making a very small turbine. Less than 0.3m in diameter. You need a foil suitable for low Re#.

Work out the Re# at 75% of the blade radius and then check the polar plot of you chosen foil at this Re#. Typically thinner foils around 8% thick work best at low Re#.

How much power were you hoping to extract? What is the intended use?

Rick

I read that shrouded marine turbine will provide 3-4 times more power, but I wonder why I am not seeing many shrouded turbine used for generate power.

I think its quite hard to achieve high extraction from the stream, so maybe 1% is enough. It is for a research on tidal stream turbine.

Originally Posted by tspeer

Low angle of attack in the center doesn't make much sense to me because the rotation rate is slow there. If you have an approximately uniform inflow velocity across the span, you will need a comparatively high angle of attack there. Then the blades need to be twisted to keep the angle of attack at the tips from getting too low.

I realized my previous mistake, to get the tip move faster, the Cl need to be low according the the formula Cl= 2L/(density x A x V^2). AoA is increase proportionally to the Cl before reaching the max AoA, lets say 16degree. So my near-hub-airfoil will be having angle of 16degree from the horizontal and decrease until to the tip. This will ensure a uniform angular velocity across the span.
Something similar to this:

http://202.114.89.60/resource/pdf/681.pdf

However I am getting a little bit confuse here, I have seen some blades are having very high center angle, making the leading edge facing vertically upwards (suddenly couldnt find a good sample picture, sorry). For example the java foil suggested the center angle is 87 degree, which is facing upward, and gradually decrease while moving to the tip.



And some turbine/propeller blades are having their low centre angle (the leading edge facing forward), then the angle increase as r/R increase.

http://www.recumbents.com/wisil/hpb/prop/default.htm


http://www.recumbents.com/wisil/hpb/...on/Propfab.htm

Am I making some mistake in the understanding?
Is my suggestion of making the center angle 16 from the horizontal and slowly decrease to 5 degree at tip workable?

[daiquiri]

@Akira88:
A shroud makes sense if your diameter is limited. A limited diameter implies that the disk loading increases as the power increase, all the rest being equal. More disk loading means a higher pressure differential across the actuator disc, which increases the induced drag of blades.
If you can keep the disk loading low by increasing the diameter, then you will get better efficiencies without the shroud.

As about the question about blade angles of attack, I have like an impression that you are not considering the velocity component due to blade rotation, so you are obtaining much lower angles.
The attached pic illustrates the velocity vector diagram at three different stations along the blade. You have to imagine the turbine disc vertical, so the blue (horizontal) vector is the freestream velocity, the black (vertical) vector is velocity component due to rotation (Omega*Radius) and the red vector is the vectorial sum of the two. The picture ignores the induced velocity component, which should also be taken into account at a more detailed design stage.
You can see that the red (actually, reddish... ) vector is much more angled at the root than at the tip, and hence the blade twist must follow the same law.

I see that you have enclosed a JavaProp table, which shows an 87° angle at r/R=0. But at r/R=0 there will be no blades, just the prop shaft. So your analysis has to start at higher values of r/R (precisely, from where the hub ends).

[akira88]

Originally Posted by daiquiri

@Akira88:
A shroud makes sense if your diameter is limited. A limited diameter implies that the disk loading increases as the power increase, all the rest being equal. More disk loading means a higher pressure differential across the actuator disc, which increases the induced drag of blades.
If you can keep the disk loading low by increasing the diameter, then you will get better efficiencies without the shroud.

As about the question about blade angles of attack, I have like an impression that you are not considering the velocity component due to blade rotation, so you are obtaining much lower angles.
The attached pic illustrates the velocity vector diagram at three different stations along the blade. You have to imagine the turbine disc vertical, so the blue (horizontal) vector is the freestream velocity, the black (vertical) vector is velocity component due to rotation (Omega*Radius) and the red vector is the vectorial sum of the two. The picture ignores the induced velocity component, which should also be taken into account at a more detailed design stage.
You can see that the red (actually, reddish... ) vector is much more angled at the root than at the tip, and hence the blade twist must follow the same law.

I see that you have enclosed a JavaProp table, which shows an 87° angle at r/R=0. But at r/R=0 there will be no blades, just the prop shaft. So your analysis has to start at higher values of r/R (precisely, from where the hub ends).

Ahh, my concept was totally wrong Thank you for spending time explaining it to me.

By the way I found a site about the basic blade design, http://www.windmission.dk/workshop/b...gn/angles.html
It is helpful, but there is a graph I dont get it, http://www.windmission.dk/workshop/b...ign/polar.html

It is stated that this is the profile curve, however what is the x-axis? cause I would like to get it for other airfoils section using xlfr5 too.

Thanks again!!

[tspeer]

Originally Posted by akira88

...there is a graph I dont get it, http://www.windmission.dk/workshop/b...ign/polar.html

It is stated that this is the profile curve, however what is the x-axis?

The X axis is the drag coefficient. Drag polars are often plotted with drag on the X axis because for a flying vehicle it is the horizontal force. With lift being the vertical force, it makes sense to plot it on the Y axis, even though one normally thinks of lift as being the independent variable and drag as the dependent variable in the graph.

The straight line drawn from the origin tangent to the curve locates the point of maximum lift/drag ratio. If the blade operates at a single operating condition, this would be the best one to pick, because the section is most efficient there.

You will probably also be looking at the lift curve, http://www.windmission.dk/workshop/B...esign/cla.html. The lift coefficient shown in this plot is the same as the Y axis on the drag polar plot.

Both of these curves will be included in the section data for any airfoil shape. However, they are only good for two-dimensional flow. There is an induced velocity due to the lift on the blades that will slow the incoming stream. So while the vector diagrams of http://www.windmission.dk/workshop/B...ign/angles.htm are correct, the incoming flow is not the free-stream velocity. That may be why they use 2/3*v as the incoming flow in the diagram, with the 2/3 being an empirical factor.

If you decide to go with a shroud, you will want to have the exit area greater than the inlet area. That will increase the velocity going into your turbine and allow you to get more power from the same size rotor. However, if the total drag is important, there will be extra drag from the shroud and you will have to trade off that drag vs a larger rotor without a shroud to get the solution with the lowest total drag.

[Guest625101138]

Originally Posted by akira88

I read that shrouded marine turbine will provide 3-4 times more power, but I wonder why I am not seeing many shrouded turbine used for generate power.

I think its quite hard to achieve high extraction from the stream, so maybe 1% is enough. It is for a research on tidal stream turbine.



......

Getting into the foil design without understanding the basics of a turbine is putting the cart before the horse.

This link might help you with the fundamental relationships for a turbine:
http://en.wikipedia.org/wiki/Betz'_law

If you use a venturi shroud then the basic relationships stay the same but you can exceed the Betz limit for the stream that you are operating in.

So the first step is to understand the gross flow conditions. Then you need to understand the velocity profile over the blades. Finally you get into optimising the foil.

If you describe in more detail what your stream looks like I might be able to help more with the design. If you only have a 0.3m square flow then there will be local effects if you use a turbine that spans the entire flow. It would be best to fully shroud the turbine blades to reduce the complexity.

I have attached photos of different types of tidal turbines to give you an idea of what is being done.

[Guest625101138]

If you are using a tidal stream then you need foils that will work equally well in either direction of flow. The foils will be symmetrical unless you have the ability to re-orientate the turbine with respect to the flow.