Re: Turbines Update

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Re: Turbines Update

jon_bondy

Can I share this code with the OpenSCAD group?  Or do you wish to? 

OpenSCAD general discussion [hidden email]

I think you have to join and until then your questions get flagged in some way.  Might be easier for me to do.

OpenSCAD is not a programming language.  It is a language to define geometry.  And the geometry is static.  So all named values are in fact constants, not variables.  Thus your confusion about summation.  EVERY OpenSCAD user goes through this moment of WTF.

You are correct, of course, about using hull().  I think there is a way to use various sweep libraries to create the shape you need.  As I recall the sweep libraries call a function to generate a cross section and another function to generate a path through space.  Does the attached demo help at all?

Jon


On 6/8/2017 12:30 AM, Alexey Finkel wrote:
Blade shape is somewhat more complicated than may at first appear. If you comment out the rotor generation, and uncomment "blade_prim();" at the end, you will see how it is supposed to curve more clearly. The generic design I sent you uses a fairly narrow sliver of the whole shape, but does require that it curve in multiple directions. That is not something that hull() does well, as I learned during my last venture into SCAD. For one, it makes a straight line on the back of the cord, filling in the concavity. It will also not give a different angle of attack at the top vs. the bottom of the rotor unless I build it out of straight-line segments like before, which would again be computationally intensive and also prevent me from using "backsweep" (and backsweep is supposed to be good for both flow stability and compressor efficiency, I am told).

Attached is a version of the design with thicker, more visibly tapered (but still not buttressed) blades.

The 3Dsweep library that you suggested a while ago may be able to handle it, but I do not know how to generate the NACA airfoil profiles that it uses, or even if those can be adopted to centrifugal blade profiles. Reading the code, I also could not see right away how to set the curve between the beginning and ending profiles. Still, may be worth another look.

By the way, I had a moment of profound bewilderment when I realized that the only way OpenSCDA can do summation is through recursive functions. What sort of a respectable language would do that?!

My search for diffuser guidelines is not proving fruitful, so I am going to forge ahead with what I hope is a relatively safe, if not the most effective design. If I aim my efforts towards a test rig, then we will soon need to do some scheming to figure out how the driver turbine with its plumbing, the compressor with its, and the throttleable instrumentation pipe can all be made to "coexist peacefully together". Speaking of which: did any information on possible compressor options come to light?



On Wed, Jun 7, 2017 at 9:35 PM, jon <[hidden email]> wrote:
You use linear_extrude() to make the blade.  What if you created a profile for the top of the blade and one for the bottom, separately, and then used hull() to connect them.  Then the base could be broader than the top.  In fact, the base could be the top just scaled by 2.


On 6/7/2017 5:12 PM, Alexey Finkel wrote:
The blades actually taper veeery slightly toward the outer edge, but indeed do not include any real buttressing. Because of the complicated shape of the hub, I am really not sure how to even go about putting something like that in place. I could mess with the arc that generates the blade surface to make the taper more pronounced, but that's not quite the same thing, of course.

Excel does not, as far as I know, come with an animation package, so I think I would lose out on the visual appeal part of the market for the calculators.

On Wed, Jun 7, 2017 at 10:09 AM, jon <[hidden email]> wrote:

Preview rendered instantly here, although trying to manipulate the view screen in any way causes OpenSCAD to hang (which may be what you meant).  I worry that the blades are of uniform thickness, which means their attachment point at the base is fairly narrow.  My intuition says that they will snap off without some broadening at the base

I wrote some Stirling Engine simulators which were very naive, but I included animations.  It was to help my intuition.  I had quite a few requests for them, even though they were ... stupid.

:)


On 6/7/2017 8:08 AM, Alexey Finkel wrote:
I will ponder on the question of the Discussion group until Sunday, but as my comment on the meetup page suggests, I think putting the whole thing on hold while it is not working so great for you has a substantial risk associated with it.

Without knowing much of anything about how to use a DSP unit, I will nevertheless venture a guess that "stunning" difference may live in a very small and remote corner of available parameter space; a corner that is difficult to find unless one has either much insight knowledge, or preternatural intuitive talent. But that's just a guess.

Of course real engineers have much, *much* more complete and sophisticated software packages for turbomachinery design. My spreadsheets are naive child's drivel by comparison. Their primary purpose is really to give me the illusion of knowing that I knew what I was doing when choosing some basic design parameters. It may be an intermediate stage between completely seat-of-the-pants design and real-life engineering work, but I have to wonder how big a market there is for such a beast. At any rate, if we ever built something based on those calculators, and it worked in any sense of the word, I would probably make the calculators available online for free.

It looks like I sent you the wrong version of the new blade design. This may be for the best, because I developed it into a proof-of-concept compressor rotor design by now. The result is attached. Be sure to disable preview before opening it though! The thing is very high-polygon, and pretty much kills my SCAD dead if it tries to preview it. The rotor has a radius of 50 mm, so if you wished, you could use it as a test. However, I recall you mentioned that the resin is not cheap, so I wonder if a scaled-down version would be a more practical way to go. Or, you could cut out a cylinder or a cone from the rotor hub, at least if unsolidified resin can be recovered and reused. Also, I used lower values of $fn that I might for a "final" version to save computation time. If you were interested to see how smooth we can get the surfaces, maybe increasing $fn to 100 or even 200 everywhere would help put the printer through its paces.

On Tue, Jun 6, 2017 at 8:36 AM, jon <[hidden email]> wrote:
The new/larger resin printer is assembled, but not tested, so the timing is good (I hope).

Your spreadsheets are interesting.  I wonder if they could become a commercial product (or, better yet, they could be re-coded to produce a commercial product).  I know that the Big Guys would not want/need them, but DIY folks might.

The new 5' tall speakers are interesting.  The sound is OK but not stunning in the frequency-response sense, but the "sound field" that I had assumed was just BS turns out to be palpable. You can walk behind the speakers without losing much of the sound.  It seems to just fill the room.  So, from the perspective of wanting to experience something new, it is a success.  The $700 DSP unit produces detectable differences in system response, but not stunning differences.

I seem to have run out of steam for the Discussion group.  In the past, topics just leapt to mind, but now my mind heads in other directions.  I'm not sure what to do about that.

:)



On 6/6/2017 6:36 AM, Alexey Finkel wrote:
It seems I haven't said anything on the subject in a while, but the last couple of weeks have been a relatively productive time on the turbine front, so I thought I'd give a bit of an update on the state of things.

Both the compressor and turbine calculators are coming along. Both pass the basic sanity checks I've put into them and give believable results. However, I keep adding bits to them, and finding other bits that can be done better, so I would not risk saying that they are "done" just yet (or ever, really). One conclusion is that going from 30-mm radius to 50-mm rotors makes a huge difference in power, as well as a significant, but less dramatic one in compressor pressure. Kilowatt-level output seems feasible with 50-mm rotors without going to ludicrous rpm.

I think this means that we could tentatively answer one question from the long list as follows: if the 100-mm resin printer is going to be available, we should be able to print high-precision and smooth moulds, and cast a ceramic rotor (or rotors). Whether this works in practice, and whether a stator made from a plastic printer cast can be made usable remain open questions, of course.

I've also tried to match a compressor to one possible version cold-air turbine, and found that it is sort of possible even with the very low flow rate available, if a number of adaptations and concessions are made. Still, as I think I mentioned ,if we can get higher flow rate even at the cost of pressure, it would be a lot better.

However, I have not tried to assemble a full compressor-diffuser-combustor-turbine calculator yet because my rudimentary combustor calculator is giving me unrealistically high combustion temperatures, and because useful data on even the simplest kind of radial-flow diffusers is proving tricky to track down.

On the other hand, I came up with a new way of generating blade geometries in SCAD. It uses linear extrudes with twists instead of hull, thus getting around the convexity issues I had before. A sample is attached, but be warned: it also uses differences of high-polygon objects, so it *really* doesn't like to run in preview mode. On the plus side, it seems to compile and render pretty quickly for me.

Proceeding with diffuser design may be slow for a while, as it looks like I am going to have to do some more serious reading, but I will try to keep myself entertained by modeling other parts and pondering how to fit them together.

How are your projects going by the way? Do you have more speaker cabinets than you know what to do with yet? Or did you see the light and opt for the magical "blue diamond" sound-improver?

Alexey








_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org

Sweep With Easing Demo.scad (12K) Download Attachment