CH3NO2
Senior Member
Joined: March 2017
Posts: 455
|
Post by CH3NO2 on Oct 22, 2017 16:38:35 GMT -5
Thanks John and Chris. The design we previously discussed was entered into Solidworks. I liked the idea for it's simplicity... and I like simplicity. While the deflector plate annulus only had a small percentage of the primary flow area, a resultant recirculation zone was created but it seemed rather small and was susceptible to variations in the radial injection. So I felt like I had to try something else. So I moved on to investigating the swirler design shown here. For a swirler, it seems like a relatively easy part to have laser cut and have the blades bent to 60 degrees. While this swirler seemed easy to fabricate, when it was modeled in Solidworks the swirl performance was lousy. All it did was create a weakly rotating column of air that went straight down the axis of the flame can. No recirculation zone. It appeared there was more deflection of the flow inside the flame can due to turbulence within the plenum. The turbulence in the plenum, according to Solidworks, was so dramatic the system was remodeled to have the air brought in from the top of the plenum. It helped a bit but the swirler performance was still unsatisfactory. This swirlers performance looks more like a regular dump combustor. So after more reading of Arthur Lefebevers book it pushed me to look into possibly fabricating a real swirler. The performance would be much better with a curved vane swirler but they are much more difficult to fabricate. I spent a lot of time thinking about how a proper swirler could be made and there are many ways to make one but I could see none of them would be easy to build inside of my garage. But then I came across this video that inspired a new line of swirler fabrication thinking. youtu.be/5mW4Z4M_YFA
The breakthrough is with the Nesting of tubes cut by a 3D laser cutter. The swirler below can be made entirely from 1.25", 2.5" and 3" stainless tubing. A fabrication and welding jig wouldn't even be necessary. Piece it together on a table and kiss it with a TIG welder. Done.
The hardest part is modeling it into the computer but after that, its an easy ride from there. A vane is shown here highlighted in green. Its nothing more than a 3" diameter piece of tubing cut on a laser tube cutter. Here is the same part sectioned. The swirler can be designed with a straight vane: Or with a swept vane: These are vanes shown with the outer tube and inner hub shortened for a better view. This 2.5" OD swirler with a 1.25" hub was modeled onto a 7" diameter flame can to check its recirculation characteristics. The radial holes were left out of this example. It was just to see the effect of the swirler only. In this example it works exceptionally well. Recirculation Eureka! So now that I figured out a way to build a curved vane swirler all I need to do now is figure out the primary zone proportioning. But to do this, I must first figure out a ballpark discharge coefficient for the swirler. It's probably safe to say it probably wont be above 0.85 or below 0.4... but that still leaves a big spread to figure out. I'll see if I can build the swirler and hook it up to my high power leaf blower to take pressure drop measurements and back out a coefficient. Tony
|
|
CH3NO2
Senior Member
Joined: March 2017
Posts: 455
|
Post by CH3NO2 on Oct 22, 2017 17:50:03 GMT -5
This 2.5" diameter swirler is much too large for 33% of the Primary air flow but reducing the size will be easy to do in the computer.
|
|
|
Post by racket on Oct 22, 2017 18:31:30 GMT -5
Hi Tony
Theres a couple of points that would need to be "included" , firstly the Primary holes need to be fewer and larger to get air into the centre of the flametube so that they collide forcing some air back towards the fuel nozzle , and secondly the fuel spray from a hollow cone pattern nozzle .
There is a lotta energy being released from a high pressure fuel nozzle , I use to do testing of my nozzles by fitting them to the lid of a very large glass jar of ~6" dia by 9" high ( pickled onion bottle) , the fuel accumulating in the bottom of the jar was soon frothed by the fuel jet and the massive amounts of air swirling around in the jar .
The swirler I used in my TV84 bike engine flametube was probably (judging by another swirler I have here) 50 mm OD with vanes of ~10 mm long , the actual flow areas of the TV84 vanes were 9 X 5 mm X 12 of, the "angle" was a relatively shallow 25-30 degrees , I sorta remember having to "squash" the vanes down a tad to get the required 10% of total flametube hole area , this vane set I have here looks like a 30-35 degree angle .
I think if you add in the bigger dia holes and the hollow cone spray pattern things will change dramatically .
Cheers John
|
|
CH3NO2
Senior Member
Joined: March 2017
Posts: 455
|
Post by CH3NO2 on Oct 22, 2017 20:43:31 GMT -5
Hi Tony 1) Theres a couple of points that would need to be "included" , firstly the Primary holes need to be fewer and larger to get air into the centre of the flametube so that they collide forcing some air back towards the fuel nozzle , and secondly the fuel spray from a hollow cone pattern nozzle . 2) There is a lotta energy being released from a high pressure fuel nozzle , I use to do testing of my nozzles by fitting them to the lid of a very large glass jar of ~6" dia by 9" high ( pickled onion bottle) , the fuel accumulating in the bottom of the jar was soon frothed by the fuel jet and the massive amounts of air swirling around in the jar . 3) The swirler I used in my TV84 bike engine flametube was probably (judging by another swirler I have here) 50 mm OD with vanes of ~10 mm long , the actual flow areas of the TV84 vanes were 9 X 5 mm X 12 of, the "angle" was a relatively shallow 25-30 degrees , I sorta remember having to "squash" the vanes down a tad to get the required 10% of total flametube hole area , this vane set I have here looks like a 30-35 degree angle . 4) I think if you add in the bigger dia holes and the hollow cone spray pattern things will change dramatically . Cheers John Hi John, 1) Yes. I'll see if I can play with radial injection in Solidworks. Optimize jet location and penetration. I think I already have a pretty good idea where to put the jets, the next part is to figure out the optimum number and diameter. Maybe start with 8-12 jets radially? 2) I was getting a feeling that would be the case. That much fuel flow at that pressure will impart momentum into the surrounding gas. ~1/4 HP worth of momentum. I'll try to angle the hollow cone fan to the center of the recirculation zone. Alternatively, what do you think would happen if the fan was angled wider into the swirler plume? To mix fuel into the air immediately at the swirler outlet and let the mix flow/roll around the recirculation zone? Maybe it would be harder to run lean? I expect injection angle can produce distinctly different operational results. It will make good data to test a few angles to see what happens. 3) Is this your TV84 swirler? So you flattened the vane angle to optimize the flow area? i352.photobucket.com/albums/r347/Hydrofluoric_Acid/Turbine/Image_78_zpsqs4ocs9k.jpg4) " things will change dramatically " Can you elaborate here? Thanks, Tony
|
|
CH3NO2
Senior Member
Joined: March 2017
Posts: 455
|
Post by CH3NO2 on Oct 22, 2017 20:52:36 GMT -5
The swirler design I have now has an outlet angle of 60 degrees.
|
|
|
Post by racket on Oct 22, 2017 21:57:14 GMT -5
Hi Tony
Yep , thats the TV84 combustor/flametube
Your 60 degree angle might be producing too much axial velocity and not enough swirl ( coarse thread vs fine thread ), the swirl vanes I used/have are commercial oil burner ones, the 25mm dia hole in the centre was for the spray to go through , I've just measured the swirler I have, and the angle is 35 degrees ( 55 degrees from axial)
Yes, try 8 or 12 Primary holes , Jetspecs indicates lotsa small holes , but as I said in early emails to you , its a crude device, and was best suited to propane injection where the gaseous fuel is spread out close to the wall by the radial injection holes.
When guys have used a spray nozzle for kero, the rule of thumb has been to place the limited number of Primary holes at the theoretical impact point of the fuel spray against the sidewall , this invariably produces a working engine .
As for the fuel spray angle , it was hard enough for me to get a nozzle of any description , let alone an optimal one , a 60-80 degree angle would be OK in most cases , you don't want too wide an angle as full throttle running might see fuel impacting the wall and burning on it with wall overheating etc etc ......
By "dramatically " , I meant your Primary Zone airflow patterns will change , the energy in the fuel spray will force air "outwards" creating a "low pressure??" area in the hollow cone , that the large Primary hole air jets will fill once they collide in the centre , you end up with a "smoke ring" sorta effect ...............its possible to get an idea of what happens when using a garden hose if the nozzle is adjusted to a hollow cone , theres a tendency for fine droplets to move into the "hollow" .
I'm getting the feeling you might be "over complicating" this flametube issue , I've seen lotsa very crude construction work on the Yahoo DIY site , when I made my TV84 flametube I had only a couple of text dealing with full sized aero engines to go by , and as a consequence "over complicated" it , theres nothing wrong with striving for perfection , but if years are spent on detailed research and design, the engine never gets built, a happy medium required :-)
Cheers John
|
|
|
Post by racket on Oct 23, 2017 3:22:09 GMT -5
Hi Tony On the bottom right hand side of your diagram there appears to be a bellmouthed "exit" , is that for your bleed off ?? Also I need to question why then is your air delivery tube elbow orientated "axially" , and forcing air towards the "top" of the combustor , only to have to try and bring it back down to bleed it off, the discharge from the elbow is into a somewhat restricted area judging by the colour of those velocity arrows Perhaps another option might be having the delivery tube elbow discharging in a more radially fashion like I did with my GT6041 jetandturbineowners.proboards.com/attachment/download/216 you could then have a bellmouthed bleed on the opposite side of the plenum , your turbo is basically the same configuration . Cheers John
|
|
CH3NO2
Senior Member
Joined: March 2017
Posts: 455
|
Post by CH3NO2 on Oct 23, 2017 10:26:10 GMT -5
Hi John, Yes, I am complicating it, but for me, this is a project to learn from and to get a feel for the challenges and the design optimization process. So it's all good. I can build to the jetspec design for comparison and I can also shamelessly duplicate your GT6041 flamecan which is guaranteed to work well on Kero... but I feel that particular design, with evaporators, is for Kero or similar parafins only. It wouldn't work the same on Methanol and I can't put cut nitromethane or other reactive/energetic chemicals through a red hot evaporator tube. While I love the concept of evaporators, it depends on the fuel chemistry. Some yes, some no. I'll fit the top of the dome with a 2.5" V-band flange/clamp so I can change out different swirlers easily. I selected a swirler 60' off axis angle because it was the most cited example, with the most referenced data points in Lefebvers book. Higher angles can be used but other trades creep in when going to higher angles. The Solidworks flow modeling looks great at 60 degrees so far. I'm a n00b just getting started in Lefebvers book. I still have soooo much more to learn. Yes, the 3" bellmouth exit at the bottom is to bring the bleed port as close as possible to the turbine outlet. This is to bring the jet engine thrust axis as close as possible to the rocket engine thrust axis. However this design is just preliminary for Solidworks flow modeling. It may change. For the elbow, its a curved 4" to 5" silicone coupler. www.amazon.com/gp/product/B015RAJVW0/ref=oh_aui_detailpage_o07_s00?ie=UTF8&psc=1Similar to a lobster tail diffuser in that it turns and diffuses but with flexibility for thermal expansion. If I bring in the air at the top of the plenum it will take up more space but it appears to perform better in Solidworks. It takes out a lot of the turbulent chaos in the plenum. Its a trade I havn't decided on yet but its getting closer. I will experiment with the plenum air entry and exit locations in Solidworks a little more. Tony
|
|
|
Post by racket on Oct 23, 2017 16:17:23 GMT -5
Hi Tony
LOL..........As long as the journey is enjoyable, we won't worry about the destination :-)
Yep , I'd also forget about using vapourisers for anything but kero , too many unknowns , a simple spray nozzle should reduce "variables" .
I like that 4 to 5" elbow, a nice smooth bend .
A "top end" entry is certainly "smoother" , but does take up more space and complicates the plumbing for the spray nozzle , OK for a small mobile engine or a large stationary one like yours, I think I'd go for it if I was wanting to experiment in the way you want to , make the "lid" of the entire combustor easily detached using a large diameter V band clamp , have the flametube and fueling inlet mounted in the cylindrical section of the can .................you can have the engine opened in a couple of minutes to replace whatever you want inside the combustor..........just pull it out and drop in a replacement, slip the lid on and retighten the V band and we're back in action :-)
Cheers John
|
|
|
Post by finiteparts on Oct 25, 2017 12:02:18 GMT -5
Hi Tony and John,
One of the things to consider when thinking about the fuel injection is the droplet size. You will be hard pressed to get the RANS CFD to give any relevant results on the placement of liquid fuel droplets and its impact to the flow momentum in the dome region. Why? Because to accurately capture the effects, you would need to resolve the flowfield at a scale below the geometric scale of the fuel droplets...your model would be huge and even then only give marginal information because the flow effects are highly non-steady, thus requiring some form of transient modelling, like LES (Large Eddie Simulation) or similar. If you want to estimate combustor performance (maybe a 85% correct answer), you can use a reacting flow model using methane. If you are trying to see how the injection of liquid changes your flowfield, then you have to use much more advanced modeling to capture the fluid-gas interface, momentum transfer, etc...
Secondly, the spray from the fuel injector should really not impact the flow field as much as one might think. If the atomizer is operating properly, the fuel droplets will be very small, which means that their drag vs momentum is large. They should follow the flow much like a fog. The large amount of energy that you are putting in is primarily getting used to form more surface area (increased surface tension required for smaller droplets). It's like spitting out the window of a moving car...the drag of the liquid going out of the car is so high that if you didn't spit hard enough, it can turn back and hit the car...thus it doesn't really impact the flowfield around the car.
As for your swirler discharge coefficient...you have a CFD model, just calculate it from the data in the model....that will be way more accurate than trying to guess or use a 1-d correlation. You can even use your CFD model to back-calculate the flow splits for the cold flow model.
I don't think that you are over complicating things. I think that there is a good balance between design and testing. But without the grunt work required to get a good estimate of the basics, how can you trouble shoot the combustor when things don't go as planned. Simple 1-d correlations can get you in the ball park, but when things go wrong, you are just guessing as to what the problem is. Plus, some of us enjoy the design work as much as the building...probably not as much as the running of the engine though. Everyone likes that the best!
Chris
|
|
CH3NO2
Senior Member
Joined: March 2017
Posts: 455
|
Post by CH3NO2 on Oct 25, 2017 17:06:59 GMT -5
Hi Guys, Here is the Solidworks flow modeling with 40% flow area dedicated to the swirler. I sized it up a bit because no discharge coefficient has been applied by me. If Solidworks can automatically apply it, I'm not sure. In the image below I added red lines to illustrate the lines intersecting near the center of the recirculation bubble and the 12 primary zone radial jets. It looks like it would work well with a 80 degree hollow cone fuel nozzle. What are your thoughts? Thank you, Tony
|
|
|
Post by racket on Oct 26, 2017 3:10:47 GMT -5
Hi Tony
Thats starting to look more like it should :-)
Cheers John
|
|
CH3NO2
Senior Member
Joined: March 2017
Posts: 455
|
Post by CH3NO2 on Oct 26, 2017 17:44:42 GMT -5
Looking good so far. The swirl number is 1.46, from the Leferbvre equation =(2/3)*((1-(Dhub/Dsw)^3)/(1-(Dhub/Dsw)^2))*TAN(theta) I think the Primary Zone design is good enough to proceed to Secondary Zone optimization. The image below is "Flow Pipes" in Solidworks and shows the flow lines. The holes in the Secondary Zone were sized according to Jetspecs, which is a good first pass, but it looks like the holes are too numerous and small. The resulting jets are laying down too quickly and not penetrating very well. It looks like the Secondary Zone could use fewer and larger holes. I'll experiment with Secondary hole sizing and report back. The Tertiary zone looks good from what I can see. Tony
|
|
CH3NO2
Senior Member
Joined: March 2017
Posts: 455
|
Post by CH3NO2 on Oct 26, 2017 18:07:36 GMT -5
Gotta be careful in how air is introduced into the plenum. All kinds of unwanted recirculation and turbulence can be introduced into the plenum even with a 5" diameter inlet. I may implement a longer entry tube with flow straighteners. This is one of the nicer images of the chaotic plenum turbulence.
|
|
|
Post by racket on Oct 26, 2017 18:47:58 GMT -5
Hi Tony
Yep, you'll need a longer axial section at the delivery tube, probably at least a couple of "diameters" long , thats why I mentioned earlier about the top entry producing a bulkier engine , a bundle of tubes in that axial section can allow a reduction in length .
With my TV84 flametube I had 12 Primary zone wall holes, 15 Secondary and 10 Tertiary , I think theres a mention in Lefebvre that the Secondaries don't need to penetrate as deep " since most of the gases emanating from the primary zone are fairly close to the wall " page 140 , mid way down the page .
Cheers John
|
|