Christmas Windmill Time

[Adriaan Kragten]

Even if energy was meant in stead of power, the battery capacity in Ah isn't a kind of energy. To calculate the energy content of a certain battery, you must know the average discharging voltage and the discharging time for which the capacity is defined. This is explained in chapter 3.1 of my public report KD 378.

[taylorp035]

For your satisfaction Adriaan, here's the performance curve of my battery.  My power meter isn't running right now on the windmill, so I have to base my power generation on the single cell li-ion cell voltage of the pack.




In other good news, it was very windy the last 3 days and the windmill was spinning for a good 24 hours of it.  It put 51 Ah into the pack and successfully triggered my homemade max voltage circuit to turn in off at 4.089v.  I'm hoping to record my first full kilowatt-hour this weekend.

Sparweb - I was thinking of a singular 1 x 2 x 1/2" magnet on each side.  I think the steel case would be just fine, especially since it's not supporting any load with my new windmill.

Have you ever considered getting FEMM to model this stuff?

- After you mentioned this, I went looking for a student version through Autodesk and didn't see one.  I probably have an ANSYS license for it if I looked elsewhere.... my brother has played with it before a few years ago.

So maybe I'm coming full circle here, but I decided to draw my own custom PM alternator that would fit in the space.  The idea would be to 3D print out of plastic the stator and rotor (yes, I know there isn't any steel...) but I'm not looking to break any records.  The main goal is to get the cut in down to 60 +/- 20 rpm at 5 volts.

Given my current setup, I'm thinking targeting 25 amps at 400 RPM and sustain 40 amps for a good wind gust while it transitions into full furling at 600 rpm.  These numbers are based on what I've seen so far the last two months.

So what I have drawn is a 8 pole rotor with 16 magnets in a Halbach array.   8 main magnets of N48 4" x 1/2" x 1/2" + 8 magnets of N52 4" x 1/2" x 1/4" in between.  4" rotor diameter.

The stator is 12 coil, 3 phase, 24 slots.   16 gauge red enamel magnet wire with 43 turns per coil.  The stator is 4 1/2" long.  It's going to take just shy of 5 lbs of wire.  I think it would be wired in star, but I was wondering if individually wiring each phase to it's own rectifier would help boost the voltage more.  Or maybe star is the best I can do...?

I searched for other people's values for RPM / # of wraps / # of coils / diameter ... .to try and roughly estimate what my new design would do.  I also bench marked a high quality hoverboard motor.   Based on all of this, I'm thinking I would be at about 68 RPM cut it, with ~0.77 ohms/phase.  At 200 RPM it would be about 10 amps and 24 amps at 400 RPM.  I was hoping my long straight wires through the stator would help keep the magnetic flux passing by the copper in the most ideal way that might be a bit better than the wedge shaped coils of a axial flux generator.  The magnets are also closer, but then again not as wide.

My hopes was that it would out perform these calculations and I could wire the 4 coils per phase in a 2 series-2 parallel fashion so it could actually do 20-30 amps continuously without getting too hot.  I figured the 16 gauge would give me more flexibility for different configurations vs. going with 14 gauge.

I think I'm looking at about $175-$200 to build this.

[SparWeb]

FEMM:
Free download.
Just work in 2D.
https://www.femm.info/wiki/HomePage

FEMM will give you the flux through the gap (it's a variable number).  You will have a better time targeting your design if you can work from a flux density and a number of turns on the rotor.  You can calculate a lot of stuff from that point.  If you can show that the voltage you need comes from "N" turns, then you can see if that many turns fit in a slot and/or pick a gauge of wire that accomplishes that.

You can draw the slot-fill that dense, but in practice you can't get it.

FEMM will also show you (in very stark terms) what the iron does for you in a motor, and why you have basically nothing if it's not there.

[taylorp035]

I found  a page describing how to find the RPM/volt.  I'll have to work my way up to that level of complexity.  I believe I will have to send current through the correct coils.
https://things-in-motion.blogspot.com/2019/02/how-to-model-bldc-pmsm-motors-kv.html

After a few examples and videos, I was able to import my geometry and assign the appropriate materials.  I also did a test case showing a flat magnet arrangement with the Halbach array vs. no Halbach array vs. adding a steel backing.  No steel backing with the Halbach was the best result for flux at a given distance.

I might have to see what happens if I put some steel chunks outside of the coils at a later point.





Red line from the plot below for flux intensity right above the magnets

[taylorp035]

It was late and I had the magnet directions wrong.  Here's the more complete and correct study on the effects of steel and the Halbach array.

[SparWeb]

There's still some weird stuff going on in your models - it's still early days with it.
You will have to designate the steel is steel, the air is air, and the magnets are magnets.  The first model you posted has the magnets but I think the steel hasn't been defined. 
Use standard electrical iron.

For reference, the saturation field intensity of most grades of iron is about 2 T and some special (expensive) grades are much higher.  For reasonable results, aim to have a gap flux of 1.5T or more.

Your other figures are graphs seemingly exported from Excel.  If so, keep exploring the functions in FEMM - there is a tool that does that in SECONDS which I bet you would prefer to use.

[taylorp035]

I took another video of the windmill.  This time it was right on the edge of furling during the video in what I would guess to be 20-25 mph winds.

https://youtu.be/5VtbqNEjKfk


So I was able to measure the perpendicular field strength at the ID and OD of the coil edge to come up with an approximate average field strength inside the boundary of each coil.  I then used Faraday's Law to calculate the voltage and a peak AC to rectified 3 phase DC voltage to come up with what I should be expecting.  The results seemed very reasonable:

After a few hours of trying different designs, here's the results.  I'm calculating 153 RPM with my plastic design to reach 4.7 volts.  100-110 RPM would be decent for the RPM vs. current curve I'm looking for and based on the resistance of my coils.

Field Avg. Intensity          RPM                                            83.66       solid steel stator 0.1724                           153.1025   full plastic                                         268.4   no halbach magnets 0.1954                           135.1   1/8" thick steel plates as OD of coils 0.2151                           122.8   1/8" thick steel plates as OD of coils + solid steel center ID of magnets 0.23785                           111           1/8" Steel OD Case + solid steel center ID of magnets 0.21515                           122.8   1/8" Steel OD Case Only

Here's a solid steel ~1/8" casing around my plastic stator.  I'm not sure if the moving magnetic field in the outer steel casing would cause a lot of heating or eddy currents.  Maybe due to my low RPM and excess blade torque, I wouldn't care about the power loss, but I wouldn't want to melt my stator.

[SparWeb]

Yeah, heating, and yeah, probably enough to cook any plastic inside.

That halbach array isn't working for you.  Just put on pairs of magnets - they will fit in the space you have. 
That stator needs to be thin sheets of soft iron or it won't work.  The air gap field is about 1/10 as strong as it should be.
Sticking with plastic at this point just makes you like the person with a hammer, looking for nails.
Do some iterations of your design with iron in the proper place; that's your baseline.  Then try to beat that score.

[taylorp035]

Progress is being made on the design front. 

Clean slate design #2 wasn't any better by making the magnet rotor as large as possible and making all the magnets 1/2"x1/2" in cross section.  I also took my original design is tried the iron filled PLA for 3D printers... it achieved 137 RPM.

Design #3 was much different, by having 3/16" thick steel between the coils.  I also re-arranged the polarity of the magnets into groups of 2, with a spread of 20 degree between them.  The coils also now can house 45 16 gauge wires.

With no steel in the middle for the magnets to sit on, the calculations drop down to 121.4 rpm.   With a solid steel core, this goes to 90.9 rpm.    I would rather not have a steel core since that increases the cost and adds additional machining on my substandard lathe.

Looking at the plot, it seems like I need the steel to somehow extend a bit farther past the coils.  I was wondering also if I need to make the 3/16" steel out of multiple pieces of sheet steel (say 3 x 1/16" sheets).

If I can get to a sub-100 rpm cut in speed with 14 or even 15 gauge wire, that would be fantastic.

[SparWeb]

Try comparing with this.
Different size, different pole count, and so on...  but you want to compare your ideas with something that works.
This one gets 2T through the teeth, no flux leaking between them, and flow through the rotor is uniform.
Also going to try uploading a FEM file for you...

[SparWeb]

Try downloading these:

http://www.sparweb.ca/3_Gen_MoCo/BabyBlue/ROTOR_LAM_33a.FEM

http://www.sparweb.ca/3_Gen_MoCo/BabyBlue/ROTOR_LAM_33a.dxf

The FEM file can be parsed as plain text, but it's actually the data file for the FEMM program.

[taylorp035]

Thanks Sparweb.

I made a few more designs.  I decided to see what I could do with some easy-to-make stator laminations to get the flux up and over the coils better like a real motor would have.  First was just straight pieces of 3/16" thick steel between the coils.  This resulted in 121 RPM cut-in..  Next was a "T" shaped piece of steel (0.125" thick, 1" tall and 1" wide).  This proved to be effective, with a cut in RPM of 84.4 RPM, but I realized it took a ton of copper.... and was going to need 1321' of 14 gauge.   So back to the drawing board.

I landed on a L-shape 1/2" section shape, with the steel being closer together for better flow.  This yielded 89 RPM with 16 gauge and 0.76 Ohm's/phase (edit: 61 rpm with steel on the inside of the magnets).  This also reduced the wire down to a single 5 lb spool, but it increased the magnet count to 18.   I put 3 magnets in groups pointing the same direction.  I'll probably want to tweak the design a bit so there is more rotor clearance and hopefully integrate 14 gauge into it, as well as get the steel on the OD slightly closer.  I'm still concerned what the max. continuous current will be. 

[SparWeb]

Hi Taylor, this is creative stuff you're doing.

Can you do something for me?
In your results file, do a boundary curve, and have it compute the total flux per pole.  Put the curve just above the magnet group of one pole.
That would be "draw a red line" and then click the "integral" button, selecting "B.n" solution.
The result should pop up a value of flux per pole in Webers for you.

[taylorp035]

Here's the straight lines right on the top side of the magnets.




And here's what I've been doing to measure it.  One line on the ID of the coil and one line on the OD of the coil.  Then I average the values.   For this model, I've been doing a 45 degree arc, but it's a little off since it's not round on the OD edge of the coil and the spacing is ever so slightly off due to the "L" shapes being flipped to make it easier to contain the wires.

With no steel (shown below), I average 0.3057 Tesla.  With steel, it jumps up to 0.44535 Tesla.     Should be able to fit 39 turns of 16 gauge per coil x 4 coils x 3 phases. 




Next, I want to try and make the "L" shapes a bit closer together to improve the flow some.  I also made the magnets too close to the stator for my comfort level.

[taylorp035]

I think I've settled on a good design.  It's uses more copper than I was hoping (10 lbs of 14 gauge), but doesn't have a steel center and can be easily manufactured using some 3/4" angle iron.  The 3D printing part of the design is simpler as well.  It also shouldn't cog too bad I don't think considering the magnet layout.  It should also cool a lot better than my encapsulated plastic designs.   I figure I could fit 64 turns in it, but 10 lbs is going to only get me to 38 turns.

Cut in speed at 4.7 volts should be 84 RPM and has a total resistance of 0.67 ohms per phase.   I'm predicting 1.3 amps at 100 rpm, 9.7 amps at 200 rpm, and 18 amps at 300 rpm.  I plan on making the blades spin a bit slower, so I expect the maximum sustained speed of about 300 rpm.  Right now it's probably closer  to 350-400 I think considering how much power it makes on a really windy day (~2.5 amps average over a few hours, 2 amps over a day).

In other good news, I achieved my first kWh!  It made 71 Ah the other day over 34 hours.  I will definitely need a bigger battery once I have an alternator that can average ~10 amps+ over a day.

[SparWeb]

Before you go with this, how will you deal with eddy-current heating?
With (it seems) long bars of angle-iron, the eddy loops will be very large.
Could you slice the bars into thin "L" pieces before you assemble?

I think I can see a way to nest layers of smaller "L" pieces together.  Capture more flux.

Have you ever considered making a higher voltage and then bucking it?
This strikes me as easier done, with much thinner wires, maybe making a more versatile generator for you in the end.  That's a bit on the intuitive side for me - I will struggle to explain exactly why this occurs to me.  All I can say now is that...  5 Volts seems really low.  If you have also been working with packs for your car etc., then looking for at least 12 (effective charge voltage more like 20VDC) would also be useful. What's constraining your voltage below 5?

[taylorp035]

"What's constraining your voltage below 5?"

- Mostly the desire to have no BMS unit on the windmill and zero chance to overcharge a cell, though that's a moot point considering how deep I'm into this project.  It also helped when trying to hit a reasonable cut-in speed with the treadmill motor @ 35 RPM/volt.   Running at the low voltage definitely limits the power production, but I also have limited capacity in a portable battery system.   Carrying much more than 1.5-2 kWh of 18650 cells gets to be a bit bulky and less fun.  Right now my pack holds about 450 Wh.   If I make this alternator and it performs like I'm expecting, I could easily fill a 2 kWh pack in one day from empty if it's windy.

I've considered wiring the generator with thinner wire and then having multiple coils in parallel so it's re-configurable for either 5v or 12v.   I was trying to do the math on how much heat was going to be lost in the coils if say I was at 15 amps, an output of 4.7 volts and 0.67 ohms/phase and 84 RPM/volt.   But I haven't figured out how to do that math yet.   I'm assuming the alternator would be something like 25% efficient just due to resistivity in the coils..... I've played this game when I characterized the treadmill motor several times.  It was ~80% efficient at 80-100v and 10+ amps, but down at say 100 watts, it was maybe 25%.

 

Before you go with this, how will you deal with eddy-current heating?
With (it seems) long bars of angle-iron, the eddy loops will be very large.
Could you slice the bars into thin "L" pieces before you assemble?

- I can do anything at this point.  Less cutting of the angle iron would be ideal, but if it's going to help a bunch, I don't see why not. I could make slits at the "corner" of the "L" 3/4 of the way through at 1/4" or 1/2" spacing along the length if that would be dramatically better.  My quick reading on eddy currents seems to suggest this isn't going to help much considering the dimensions.

I think I can see a way to nest layers of smaller "L" pieces together.  Capture more flux.

- I could always buy thin steel sheet and layer them instead.  It would be ideal if I had a shear if I went that route.  Worst case I could cut them by hand if it was thin enough.

https://www.femm.info/wiki/SPMLoss
Looking at the FEMM website for eddy current losses and the chart they have for losses per pound, it seems like I would be at about 30 watts/lb of losses.... assuming 60 hz and a 100x multiplier for the 1/8" thickness due to thickness being squared in the equation.  I would have roughly 5 lbs of steel, so that's 150W of eddy current losses.   That sounds like roughly the upper limit as to how much additional heat I would want to get off the OD of this alternator, especially once the copper heating is added in.

[SparWeb]

Thank you for the link. I haven't plumbed the depths of the FEMM website in a long time, so I didn't know that had been posted there.

Check this out:  http://www.protolam.com/page3.html

My reading of the eddy-current loss from the chart is that you'll loose a lot more than that.  The chart is for 0.014" thick laminations.   It does NOT offer a scale factor for thicker laminations.  One of the linked papers shows that eddy current loss increases with the square of the lamination thickness.  So if your motor is 3" thick, then
(3.0/0.014 ) ^2 = 214^2 = 46,000 times more power needed.

[Mary B]

Galvanized steel flashing angles could be an option to stack. Pre bent at 90 degrees for you, just cut to size. Use the leftover bits in some other project down the road.

[taylorp035]

So if your motor is 3" thick

Oh, no... my final design doesn't have any steel on the inside.  It has 1/8" thick angle iron for the outside, which I was very roughly going off the 10x thicker calculation from 0.014" for the eddy currents.

I have some polycarbonate filament for the 3D printer, so this will greatly increase the maximum allowable temperature versus PLA.  It should be good to north of 100C before getting too soft.  The glass transition temperature is about 140-150C versus 52-57C for PLA.   This should let me dissipate a few hundred watts, especially since there will be some air gaps on the OD. 

[SparWeb]

Oh, no... my final design doesn't have any steel on the inside.

Then I still don't visualize what you're building.
Isn't it like this:  https://www.fieldlines.com/index.php?action=dlattach;topic=150017.0;attach=13888;image
But with angles inserted inside the thick part of the teeth?

Mary,
That's a really cool suggestion.
What about folding steel sheet stock into "U" shapes, arranging them side by side to make the typical toothed stator, but with the laminations turned 90 degrees.
That wouldn't be hard to test out, either...
Wheels turning in head.  Now you've done it!

[taylorp035]

Here's what I have modeled now for the 3D printed part for the stator.  The magnet rotor is pretty much a cylinder with a 1" bore through the center for the shaft and then pockets for the magnets that I was going glue in with some proven structural bonding glue.

The stator design is nice since I don't think I have to glue to angle iron in right away, so I could potentially replace them with sheet steel.  I could also put the u-shape sheet metal in afterwards to eliminate the air gap for the flux, which would be a big improvement.

Ignore the dimensions.  This was the original smaller version.  The larger diameter version is closer to 2.2 on the ID and about 2.9" radius on the OD.




I also came to the realization that my wire length calculation was off by a large amount, so now I need less wire for this design.  A 5 lbs spool of 14 gauge gets me 37 turns and a 86.4 rpm cut-in.  It also ramps quickly for amperage and will likely overpower the blades at low wind speeds, depending on my target TSR.  I played around with TSR's of 4, 5 and 6.   Even at a TSR of 4, I'm going to furl at 17-18 mph so I don't go past 30A.  20-25 amps might be a better upper limit.

Sticking with 5 lbs of wire, 16 gauge with 2 in hand is going to be about 108 RPM cut in.  Which would be ~ 7 mph at a TSR of 4.


Edit - I made the U-shape idea.  It helps a sizeable amount.   14 gauge would cut in at 71.2 RPM.  16 gauge would cut in at 44.7 RPM.  At this low of a RPM, I could wire the 16 gauge in parallel.

[ Specified attachment is not available ]

[JW]

I remember when I started messing with CAD software there was an engineering software that would interface and it cost like 10,000 dollars.

I see that in your magnet flux image.

Nice image...

[JW]

Haaaa!

I remember now- Cosmos works

http://www.solidworks.com/pages/onlinetourcosmos/COSMOSWorks.html

[Mary B]

Magnet glue reminds me of something...

I recently helped a friend do bodywork on his wife's newer car. We had to remove a rear quarter panel and replace it. It was put on with automotive panel bond... that stuff would be perfect for holding a magnet on! Very strong! Even with some heat it took a lot of effort to pop the panels apart. This was the recommended adhesive, the panel was part of the unibody structure at the roof pillar and at the top and bottom edges. https://www.amazon.com/3M-Resistant-Structural-Adhesive-Cartridge/dp/B00XC9ANTE

[taylorp035]

Mary B - That's funny you mention the automotive panel bond glue (among other things) because that's what I'm using.  It's LORD 406/19.  Excellent stuff.  LORD makes the best glue for many things.

JW - I think the software I'm using is $25k for the professional version and $2k for the light version (Autodesk Inventor).  I've been a serious user of it for about 13 years now and I find it to be easier to learn than other CAD packages and is superior for modeling shapes and many other things.  I also use Creo 4 for work and I find it to have an unbelievably steep learning curve, but it's better at finding distances with it's measuring tool.  I played with Solidworks for a few hours and found it to be similar to Inventor.    For this FEMM software, Inventor makes it super easy.... just click on any face to highlight it or sketch and it gives you an option to export that as a .dxf file.  Much easier than actually using Autocad.

 

[MattM]

Rhino 3D had dxf support and was free.  I doubt it compares favorably to professional grade CAD.  But it's a cheap option for home users.

[JW]

Hi taylorp035 thanks for the reply.

That sounds about right. I have had a seat on solidworks for over 20 years now. Ive done some freelance engineering and I tell them I use my own PC and will transfer files as needed. Im also weird about NDA's once someone grants you access to a software they hold your seat for you and own everything you do on that software.

I have several patents and it educated me alot.   

[SparWeb]

Rhino looks like an easy way to work with complex solids.
It wouldn't be so much fun for assemblies of parts, by the looks of it.
I've been considering trying it out...  haven't had the chance yet.

I try out new CAD software fairly often
Creo - the free version is just a toy, but I'd love to try out the commercial version
CATIA - very complex.  Over my head until I can take a week-long course or something to learn how to use it properly.
Solidworks - easy to start using, lots of features, too many overlapping workflows, hard to become efficient
Fusion 360 - meant to be easy to learn for non-CAD users, but since I am one, it's just frustratingly limited.

Strongly agree with Taylor's opinion of Inventor.  Some designers who grew up on CATIA or Solidworks will knock it down, but I think it's good value.  CATIA users usually aren't confronting the huge licensing cost personally.  Solidworks users usually aren't doing design work in complex assemblies of components.  If anyone ever gets a chance to work with Inventor, try to take a course to get over the learning curve, and then so many design possibilities will open up.  I pushed the abilities of Inventor on a really fast workstation for several years and it was usually enjoyable to use.  The worst drag was opening parts and projects done by incompetent users!

I'm still a bit of a shil for Autodesk, it seems.  Bad time for that, because Autodesk has just booted a huge following of users OUT of Fusion 360 and cranked the price up.  I don't know what they were thinking.

[JW]

For me using CAD is great since it can be transitioned to CAM software. Just reminded me of that. I will work with a good CNC operator to make the parts. I remember working with this really knowledgeable beyond engineer guy. Were sitting there using a 5axis Mill to test cut a piece of Styrofoam and I was watching the monitor and address's were scrolling down. He was making molds for aluminum alloy car wheels out of a super Inconel alloy as the mold. I wasn't allowed to understand the pressurized flow process of the liquid aluminum in the mold.       

[taylorp035]

I have a mini CNC 4-axis mill at home, so I need the drawing and CAM software.  Luckily I still have access to the licenses so I can have the latest and greatest for my home projects.  At work, I probably use a few hundred thousand or millions of dollars of ANSYS licenses at a time (120 solving cores at a time for static structural simulations.... my company is one of very few that has unlimited licenses through them).

I finished up the assembly and worked out how I'm going to hold the stator while still wrapping it.  I don't have the sheet metal U-shapes in it yet.  The bearing holders are held in place with some dowel pins so I can slide the magnet rotor inside.

[Mary B]

Rhino 3D had dxf support and was free.  I doubt it compares favorably to professional grade CAD.  But it's a cheap option for home users.

key word... WAS free... I am on the hunt for 3d modeling software to go with my new Creality Ender 5 Plus... Fusion 360 is an option but it is a resource hog an won't run well on my older computer, tried Bender and it sucks, tried the cloud software and they really sucked/lacked features...

[MattM]

Wings3D has dxf support, too.

It's also open source and located on sourceforge.net

An Introduction To Wings3D – Day 1 (by Jordy Lakiere)
https://www.youtube.com/channel/UCbZOxqC1_Y1KM9MchuBApIA
FINGER EXERCISES TO GET YOU STARTED
http://www.wings3d.com/?page_id=679
WINGS 3D USER MANUAL
http://www.wings3d.com/?page_id=252
The Wings3D Handbook
http://www.wings3d.com/forum/showthread.php?tid=73