|
Post by finiteparts on Dec 31, 2014 22:47:19 GMT -5
I was a little puzzled by some of the equations provided in Dixon and Hill's approach to recasting the diffuser choking conditions to the compressor inlet parameters, so I worked through them. It was quite interesting how they got there and I thought that I would share. Remember, this equation is for choking in the diffuser, while the previous one given was for choking in the compressor impeller. Enjoy! Chris
|
|
|
Post by racket on Jan 1, 2015 17:26:07 GMT -5
Hi Chris
Those equations are getting bigger and "scarier" :-(
Thanks for the Link to the IHI Paper , thats an interesting one , so have got it saved for further future "digestion", the first couple of reads have already uncovered some good info .
I think it might be time to put some of these "dimensionless" equations into a worked example , this is the reason why I love my Cohen and Rogers , that old text has worked examples which "fleshed out" those equation "bones" and brought them to life in a relatively simple way ................LOL, simple enough for this uneducated tinkerer to understand.
Currently I have a couple of "off the shelf" billet wheels in transit that will be used for a high pressure ratio engine ( 5 :1) that will be getting designed this year .
Comp dimensions are ..........inducer 106.56 mm , exducer 152.04 with extended tip to 156.78mm, a tip height of 8.8mm and an exducer swept back angle of 60 degrees .
I'd like to see how/what the equations would produce , compared to my already finished calcs , unfortunately my grasp of the methods required to use the complex equations is limited , schooling having finished >50 years ago and even then I don't remember doing anything this complex .
When I use the Cohen and Rogers worked examples as a guide I don't have problems as they were meant to be a "student guide" and not for highly qualified engineers to decipher , unlike the maths behind the worked examples earlier in the C and R text which go right over my head ....................heh heh , I might like "numbers" but not those sort :-)
Cheers John
|
|
|
Post by finiteparts on Jan 3, 2015 22:32:01 GMT -5
Hi John, Don't be scared! They are only numbers, they don't bite! Ha! I think once you plug in the known stuff and simplify, you will see they are easy to work with...Here is an example with the impeller choking calc.. And I went ahead and plotted it for a range of rpms and two mass flow rates, 2.0 lbm/s and 3.0 lbm/s. The red and green lines are for the larger mass flow rate case and the blue and black are for the other. I labelled them so hopefully they are clear. For the vaned diffuser choking, I assumed the rotor was 90% efficient (just the rotor, not the system) and it was experiencing a 15% slip at the discharge. These might be poor guesses, but at least you get the idea of the "ideal" behavior. And remember, this analysis assumes that the impeller blades are radial, so the backsweep that exists in reality will change things. Plus, I assumed the blade speed was defined at the tip, but I think to get a better accuarcy, we might want to use the RMS radius. So if you look at the plot you might wonder why it looks like it does. Well, think of the impeller choke like I described it before. The flow sees an approaching, fixed area (usually the inducer throat) as it rotates by...but if it rotates by quicker, the flow effectively "sees" more area kind of like an area per second thing. As for the diffuser the choke area goes down because you are putting more energy into the discharge flow from the impeller, thus increasing the exit temperature. Increased temperature means increased acoustic velocity, the flow speed required to choke is going up. But we are also getting an increase in density across the rotor, so for a fixed mass flow, the area will have to decrease....(the equation for area has the density and the square root of temperature in the denominator). This plot assumes constant mass flows (2 and 3 lbm/s) across the rpm range, which is not realistic, because you're not going to get 3 lbm/s at 30000 rpms! So this would need to be done at multiple steps to estimate the choke boundary...or, estimate the inlet mass flow as a function of rpm and then calculate the choking areas. Hopefully we are in the same ball park with each other on the numbers...due to the assumptions, we might be off by a bit from each others numbers. I totally agree with you on the worked examples in textbooks. I always hate the books that just throw the theory at you with no idea how to apply it. That goes for professors too! The good ones always show you tricks and insight gained from doing the problems or solving the equations. Dixon and Halls book is pretty good in that way too. The new version of Cohen's book still does all that too. I am still working through programming in the equations from Casey's paper, but I will keep you informed on how it predicts things. Also, I was looking through some older papers and textbooks (including Whittle's), and it seemed that a common design approach was to specify an intake velocity that was low enough so that you didn't experience large shock losses. If you finagle the variables around, that is what you are doing with the flow density approach that you are using. It would be interesting to see what flow velocity you rules of thumb give. Enjoy, Chris
|
|
|
Post by racket on Jan 4, 2015 0:30:06 GMT -5
Hi Chris
Looks like theres no choking problems with my flow :-)
Current calcs have an inlet airflow of ~630 ft/sec through a flow area ( inducer area minus 8 X 1.5mm thick X 40mm long blades and a ~26mm hub) of ~12.2 sq ins - 0.085 sq ft , I've reduced this to 0.08 sq ft for the calcs to allow for any "boundary" bits and pieces .
At 630 ft/sec my static air pressure is ~11.7psia at ~270 K ( 288 K ambiant) , speed of sound ~1080 ft/sec so relative inducer tip speed is ~M 1.35 , mass flow is 3.3 lbs/sec ( 1.5 kgs/sec) equating to a flow rate of ~14.3 lbs/min/sq inch of total inducer area , similar rate to the Garrett GTX3355R 47 trim wheel at a similar max tip speed .
Inducer tip ideal angle ~25 degrees , this will hopefully be the case as the bigger X831 wheel has a ~23 degree inducer tip angle .
My exducer figures have a radial speed of ~510 ft/sec , whirl ~1480 ft/sec at the exducers ~77mm - 3.09" radius ( I averaged the extended tip) , the actual tangential velocity of ~1565 ft/sec is worth ~113 C degrees of energy .
At the diffuser leading edge at 3.7" radius , the whirl will have dropped to ~1236 ft/sec, and to produce a radial velocity of 400 ft/sec I've reduced the axial height of the passageway from the 8.8mm tip height to 7.6mm at the diffuser tip to keep the tangential angle up at ~18 degrees and velocity at ~1300 ft/sec which should equate to a M 0.95 speed to keep away from any shock problems
It looks like I'll be needing throats of ~12 -13 mm and 15 of , I've just got to measure up the patterns once I get them back to verify distances ..............LOL, my "drawing" of the FM-1 diffuser might not be what I made the pattern , ..... if I wasn't so lazy I'd remove the FM-1 engine from the test stand and pull the front off it and measure things up.
Thanks again for working those equations , they don't seem half as scary now :-)
Cheers John
|
|
|
Post by finiteparts on Jan 4, 2015 21:50:38 GMT -5
I just got done cleaning the massive amounts of cosmoline off of a compressor wheel that I added to my collection. It is an early GE type. I was hoping that I might have scored a Type B compressor wheel (the turbo on the B-17's), but unfortunately the large splined drive and lack of fuel injection holes proves that it is not. It is stamped with a GE 47962-H and F106C-015 11.0 inches tip diameter...0.9375 inch exducer height...6.5 inch inducer...2.75 inch hub diameter My guess is that it is actually a compressor wheel for a engine driven supercharger, not a turbo, because the turbos had smaller drive shafts at the compressor end. But it still looks really cool sitting with my collection of other compressors. Here is a comparison with a T700 compressor wheel. Sort of a comparison of the best compressor technology of the early 1940's and the 1970's. Enjoy! Chris
|
|
|
Post by racket on Jan 4, 2015 22:55:03 GMT -5
Hi Chris
Yep , 1830 CI radial engine , the 2600 cubic inch engines supercharger wheel has a bigger spline ............LOL, I was looking at the 1830 wheel some years ago for an engine design , I cut one down to ~8 inch exducer size and was going to power it with a 4th stage Allison turb wheel .............my couple of 2600 wheels are still intact though , just don't know what to use them for :-(
Those early supercharger wheels look pretty basic compared to modern turbine comps.
Cheers John
|
|
|
Post by finiteparts on Feb 1, 2015 0:09:32 GMT -5
I was unsatified with my previous attempts to build compressor wheels in CAD, mainly because I couldn't control the profile thicknesses and the leading edges were always just for looks and not realistic. So I spent some time working through this and got a nice rotor modeled up tonight. I was able to use a realistic elliptical leading edge and b-splines to control the suction and pressure side profiles, and also the leading edge and lower profile. I still need to figure out how to control the discharge lean angle, but I thought it turned out pretty good. It was pretty involved, so there are no splitter blades on this model...maybe next time! I might have to slap one on the Printbot to see if it will build. ~ Chris
|
|
|
Post by racket on Feb 1, 2015 0:40:31 GMT -5
Hi Chris
Thats looking pretty good :-)
Cheers John
|
|
|
Post by finiteparts on Feb 1, 2015 0:42:25 GMT -5
Some recent acquisitions to my expanding collection of turbine engine bits... This is a Honeywell GTCP36-300A turbine wheel...it is massive! I haven't been able to find the specs on this particular one, but my guess is that it is made from Mar-M-247 due to word of mouth...You can see why this one was on eBay... I also got a load compressor wheel from the same APU...nice bit of titanium with a nice sound too it when you ping it...
Interesting how different the turbine design is from my GTCP36-150 turbine rotor... Then the final goody that I have recently picked up is a rotor assembly from a Rotax\Lucas CT2023 maybe? I am going on the looks of it and some other stuff in the literature. Well, I had more images, but the uploads are not working...I will try again in another post...
|
|
|
Post by finiteparts on Feb 1, 2015 0:49:26 GMT -5
So here are the rest of the pictures... It was a nice surprise because I thought that I was only getting the compressor and diffuser, but the turbine was included too. I doubt that it is good enough shape to run, since it has some odd repairs were they "blended" out the impeller section just under the inducer (the compressor is a two piece construction). But I will try to take some measurements and post them in case anyone wants to use its dimensions as guidelines for their design...since it is roughly turbocharger size. Enjoy! Chris
|
|
|
Post by racket on Feb 2, 2015 0:44:11 GMT -5
Hi Chris
I'd be very interested in some numbers from the compressor stage .
Diffuser vane height and how it compares with wheel tip height, and diffuser passageway divergence angle .
Cheers John
|
|
turbotom
Junior Member
Joined: June 2011
Posts: 58
|
Post by turbotom on Feb 2, 2015 7:33:51 GMT -5
I was unsatified with my previous attempts to build compressor wheels in CAD, mainly because I couldn't control the profile thicknesses and the leading edges were always just for looks and not realistic. So I spent some time working through this and got a nice rotor modeled up tonight. I was able to use a realistic elliptical leading edge and b-splines to control the suction and pressure side profiles, and also the leading edge and lower profile. I still need to figure out how to control the discharge lean angle, but I thought it turned out pretty good. ... It was pretty involved, so there are no splitter blades on this model...maybe next time! I might have to slap one on the Printbot to see if it will build. ~ Chris Chris and all - There's a free tool available that's a perfect choice for blade design, they even include a fully parametrized impeller design sample that can be easily adapted to individual applications. Moreover, export functions to all major 3D CAD exchange formats are provided. The user interface is somewhat peculiar but since it's meant (in its commercial version) for automatized optimization of CFD models, it's not been primarily designed as a CAD tool. There are tutorials included that help a lot understanding the "way of thinking" behind the system. And best of all, it's free! You may download the package from this page: www.caeses.com/products/caeses-free/Modifications to the impeller design take their time to process depending on the functions enabled. The package requires a powerful PC and I found it to be not too stable once in a while, but this may also be the result of my rather packed and quite peculiar overall software configuration. After playing around a little with this package I really appreciate the fast and easy way to model a complete impeller, especially having done this several times before "manually"... Have fun and all the best, Thomas P.S. I have been lazy posting for a while since I had to change work (...) but there have been some questions regarding "our" own engine design - so far we got it running quite well up to approx. 70% (70krpm) with temperatures continously decreasing as we go faster (reached approx. 340°C already). Due to a lot of smoke from the exhaust, we switched from our own disc type combustor to one borrowed from a T312 APU and since this wouldn't make no difference, to LPG fuel and still found the machine to smoke like initially. So probably our disc/air blast combustor is performing well but we've got a problem with the oil seals. On LPG fuel we're not able to accelerate any further due to pressure/flow limitations so the next step will be returning to kerosene. We've got some difficulties with the torch ignitor -- we've been using one salvaged from a Walter M-601 turboprop but apparently the fuel flow through this unit is way too much so we may have to make our own design. Currently we will ignore all the current shortcomings of the design and try to accelerate to 100% and then put a load on the engine to find out if our turbomachinery is performing well. If this works, the required modifications can be considered just a question of convenince, especially since we already know what we've got to address.
|
|
|
Post by racket on Feb 2, 2015 19:07:19 GMT -5
Hi Thomas
Thats one fine looking engine hooked up to the dyno , thanks for the update , its always interesting to hear about development issues and how they're overcome :-)
I'm certain all of us hear would love to hear regular updates on the engines progress .
Cheers John
|
|
|
Post by finiteparts on Feb 2, 2015 23:27:05 GMT -5
Hey John,
I will get those numbers next weekend when I have some free time...
Thomas, I tried that program a while ago and found it so frustrating that I just gave up on using it. I am glad to hear that you got it to work for you...I might have to revisit it someday. Since I use CAD a lot at work, I need to get better at surface modeling and this is a nice reason to work on that. But, thanks for sharing it, maybe other readers will have better luck than I did.
Your engine is looking sweet! I agree with John, keep the updates coming. Any videos?
Thanks,
Chris
|
|
|
Post by finiteparts on Feb 5, 2015 23:53:56 GMT -5
It has been told that Frank Whittle pitched his idea for the jet engine to the chief engineer of the Roll Royce Merlin engine program, by saying:
"It is so simple. Unlike your reciprocating piston engine it has parts that only turn, whereas your engine has parts that turn, go up and down and in and out and from side to side. My engine is simple."
The chief engineer replied, "Simple, eh? Never mind lad, we'll design the simplicity out of it."
No truer words were ever spoken!!!
|
|