T04 Gas Turbine - RESURRECTED!

[finiteparts]

Hi Ash,

I was just looking at your combustor design and I commend you on running some CFD on the cold design. Often, Rolls would place an offset air hole beside the vaporizer that would strengthen the swirl in the head-end region by adding momentum in the same swirl-rotation as the vaporizer exit. Below is an image of an F402 Pegasus engine that I took at the Cherry Point MCAS airshow a few years ago. You can use your mental CFD (colorful fluid dynamics, hehee!) to visualize how each air chute between the vaporizers adds momentum to the head end swirl. Just thought I would share.



By the way, nice work on the engine! I am curious to see how the dual bearing arrangement works out...it should reduce the contact stresses at each loaded ball...have you thought about adding in a few thermocouples to measure the bearing outer race temperatures?  I feel that this might prevent any bearing issues getting too large before being noticed...the combo of vibrations (either from "feel" or accelerometers) and bearing temps would be ideal!

~ Chris

[finiteparts]

By the way, if you have access to any university that subscribes to ASME journals, here is a great paper on Rolls vaporizer development,

gasturbinespower.asmedigitalcollection.asme.org/article.aspx?articleid=1415650

The Olympus 593 T-vaporizer is discussed quite a bit...a real interesting point in the paper is that they found that the vaporizers were too long and even the Pegasus engine shown above have trimmed down the vaporizer length compared to earlier designs (Viper, Pegasus 5 & 6, etc), without any loss in combustor efficiency.


I also found that paper on elastomer o-rings in dampers from Virginia-Tech that I mentioned previously... 

proceedings.asmedigitalcollection.asme.org/proceeding.aspx?articleid=1776781

I will try to go through it and send you any relevant design points that might help....

~ Chris

[ashpowers]

Hi Chris,

Interesting about that picture you posted - I saw the inside of one of those engines several years back and that is exactly where I got the "U" tube vaporizer idea from. Unfortunately the smaller details you are pointing out did not "appear" to me back then but after all of the CFD modeling I've done on this CC, I can certainly appreciate those smaller details now. From the looks of that layout, they are trying to create opposing rotation between each vaporizer. I suppose I could take the same approach as I have an even number of vaporizers in my combustor. Good thing I haven't started assembling all of the pieces of the combustor just yet nor drilled any holes in the liners.

I do have concerns about making the vaporizers smaller in my engine though. Unfortunately not everything scales down nicely 1:1. On that same token, I dont want them too long either - should one of them burn off and send shrapnel into the turbine not a happy camper I would be. But this design has always seemed to be the most effective rather than straight vaporizers just given the locale of the vaporizers being right in the primary zone; all that surface area right where parts tend to glow. What I may do is use small wire tie to attach the head of the combustion chamber to the rest of the CC body so I can more easily remove it to tinker with it, start with long tubes and proceed through several test runs to inspect how they are holding up, where the flame center is, etc etc, and then go from there.

Adding thermocouples to the bearings is likely a pretty tall order for such a small engine but not impossible. I was thinking that I would like to at least put a thermocouple just behind the rear bearings. With all of the air and fuel/oil mix passing through the bearing tunnel I have always wondered if there is a possibility that it could hold a flame in there and torch the bearings.

EDIT: My original calculation for the required sampling rate for the DAQ device to monitor engine vibration was incorrect. I was using 120,000 RPM, which is in minutes and multiplying that forward, LOL. 120,000RPM is 2000RPs, or 2Khz at least for synchronous vibrations. So I would need a minimum of 4Khz sampling rate to detect synchronous vibrations up to this engine speed.

My DAQ device can burst sample up to 8192 Hz so I would be able to monitor synchronous vibes as well as 2nd order harmonics. To get the 3rd order I would need to go up to a faster DAQ device. I would tend to think that going with what I have would be sufficient for detecting any change in the rotorbearing group but after seeing John's videos it looks like things go really wrong really fast...

[madpatty]

The turbine has been fitted with a permanently mounted shaft rather than having a shaft extension that clamped the rear bearing when it was screwed together.

Hi Ashpowers,

Some very cool bits of the engine...

Have you welded a separate self made shaft to the turbine disc or it is just the original shaft turbine package that is commercially available...?
If it is self made shaft then how have you joined it to turbine...

Cheers
Patty

[ashpowers]

Hi Patty,

I fabricated a grinding rig for my lathe using a 12VDC electric motor that I swapped the ball bearings out for angular contact bearings and used a spring washer to preload the races. This got rid of all of the rotating slop in the motor. Then I machined a shaft extension to hold a fine white grinding stone, welded a bracket to the motor casing so I could attach it to my toolpost on the lathe. With the turbine/shaft setup between centers on the lathe I ground a 60 degree taper into the shaft at the base where it meets the inconel turbine wheel - where it is much larger in diameter than the shaft where the bearings ride. There was a short section that was also ground down forward of the taper and I cut an M8X1.0 thread pitch into the shaft, LH threads, for about 12mm length. Then I cut off the shaft forward of the threads. Then I used a tool steel rod and machined the female taper to match the turbine, bored a hole and tapped it out. I also machined a center to the front end of the shaft so I could do the final turning once it was mated. To install the turbine into the shaft I put the new shaft into my bench vise and heated the end with the taper with a blow torch - she was probably around 6-700 degrees and then I very quickly screwed the turbine into her using some welding gloves and crunk her down with everything I had. Once the parts came back down to room temp that turbine was well connected - I could not get it to loosen back up after 20 minutes of cooling. From there the shaft was turned on centers and it all came out exceptionally well.

[ashpowers]

Well, I've got the balancing rig working very nicely now and have started tinkering around with balancing the rotating group. I've been adding small bits of stacked duct tape cut into small squares to the rotors and am discovering the difficulty in balancing double overhung rotors in two planes, LOL. Adding weight to one rotor to correct its balance will induce an opposing imbalance in the other rotor... and so begins the cat and mouse chase, LOL. It took a little while to narrow down on it and the pieces of tape kept getting smaller and smaller but at a certain point the rotors start looking all boogered up with bits of tape and then trying to figure out where to remove material. I know there is a better way to do this.

I've been reading up on the web in a number of locations and have come across the traditional two-plane vector equations for balancing which I believe is the direction I need to proceed:
(listed further down in this document)
www.ni.com/white-paper/4381/en/

Even with just tinkering around with it I've been able to get the rotating group so well balanced that even turning a few thousand revs on the rig I cannot detect any imbalance at all by feel. Granted, this is nowhere close to what kind of speed it will operate at, but it is nice to see that it is't getting more out of balance. Any input on this is greatly appreciated - this is a bit new to me but perhaps a collective effort would result in something that all of us DIYers will benefit from. The hardware isn't difficult to build and the electronics aren't expensive either - would think every turbine owner out there would have one.

[madpatty]

Hi Ashpowers,
Can you upload some pics of the close view of that JOINT between the shaft and turbine wheel or provide some additional details if possible...i need some additional information on it so i can also try going with the same technique....

What was that earlier technique you used in which ""shaft extension that clamped the rear bearing when it was screwed together..?""

Cheers,
Patty

[ashpowers]

There isn't much to see that would be recognizable from what the OEM shaft looks like except for the length. Here is a quick diagram of the parts.




The old layout was kindof similar, however, there was no taper in the parts and I left the original shaft a good bit longer. Additionally, the rear end of the original shaft, around where you see the taper in the new setup, that area was machined for a 10mm i.d. bearing. The rear bearing was placed onto this part of the turbine's shaft and then the extension was screwed onto the turbine - to the point where the rear face of the extension came to compress against the front face of the inner race of the bearing.

You can kindof see the parts in this image - at least well enough to get the gist of what I'm saying..


I got that HORRIBLE idea from the design of the PhoenixMK4 engine. In the new setup the seat for the rear bearings is machined to produce a slight interference fit and they are pressed onto the shaft. The NGV has been bored out to allow the turbine/bearings to be slid into the assembly from the rear after the NGV is mounted to the bearing housing. This leaves a well-breathing port for the cooling air/lube to exit the rear bearings and dump into the area between the NGV baseplate and the turbine.

[ashpowers]

I came across this piece of software that is right up our alley. It uses the sound card of the PC as the input, which is awesome considering they typically operate at 44KHz sampling rates with 24-bit resolution and can be had for uber-cheap.

Has anyone used it before and if so, how well did it work? I can't seem to find any place to download it/purchase it?

users.atw.hu/aerotarget/Balancing/Balancing.html
users.atw.hu/aerotarget/Balancing/Manual.pdf

[racket]

Hi Ash

Horace has a balancing program on his Jetbeetle Site , bottom of Products page jetbeetle.com/ don't know if it would suit you, but might be worth a look :-)

Cheers
John

[ashpowers]

Email sent - not sure if it is going to work with my hardware though as the software is really specific to the hardware configuration.

I'm looking at DirectX/VB programming my own interface through the sound card like the application you see above. It really isn't that complicated of a program to write. I'll also be able to take it up to a better configuration as well since that program above only uses one input at a time and ties in through the parallel port as well. I have a really nice sound card in my PC that is capable of 24-bit resolution @ 192KHz, which is far better than what the program above was written for. It also has separate line in channels and a mic channel, all three of which will fit the bill for this setup. I also just happen to know VB6 programming really well too so I'm taking this project in that direction. Finally having a really good dynamic balancer will be great for this resurrection of the T4 engine and certainly be useful down the road as well with the TV94 engine.

Another thing I came across last night playing with the balancer is that dynamically balancing the group all as one piece can actually give you problems. I removed the compressor wheel and machined a tube to replace it. I spun up just the turbine/shaft portion and found that there is a large dynamic imbalance in the turbine wheel itself that is causing the front end of the shaft to vibrate quite a bit. With dynamic balancing in a piece like this, if you do not balance the parts independently you can still achieve good dynamic balance as an assembly, however, internal dynamic imbalances in each rotor are going to result in loads on the shaft itself which will increase as rotor speed increases. This is not good. I think the practice should be to dynamically balance the turbine/shaft, dynamically balance the compressor wheel, and then dynamically balance the entire group as one piece. In theory, you should not have to balance anything in that third step but I'm sure it is apparent that once assembled there are going to be some small variations that could stack up enough to cause an imbalance. But this approach will minimize the amount of coupling loads the shaft has to support between the two ends.

There is a slightly different approach that will have to be taken for the independent balancing since the turbine is overhung, the compressor can be centrally mounted between the transducers, and then the final assembly is the double overhung arrangement. It starts becoming a pretty complex project in and of itself but going through all the effort to build a nice engine and not properly balance it, may as well not have built the engine to start with!

[ashpowers]

OK, so have made good progress on the balancer this weekend! =)

Added a 1/2" stainless steel base, 3/4 foam vibration isolator for the main assembly, track for drive motor with adjustable height as well. The drive belt system was modified for ease of installing or removing rotor group.
I have also made progress with the programming to use the sound card as a DAQ device but I am starting to forsee a potential problem with the approach. The sound card's inputs do not like DC voltage and they have filter capacitors that will pull the signal back to a zero reference. There is a low frequency cutoff in which if you are below this frequency that "pull" of the filter cap will start skewing the result. Granted, if I am spinning the group quick enough I shouldn't have a problem with this but the other issue is that the inputs of the sound card are extremely delicate and it is very easy to damage the electronics if you send it too high of a voltage signal.

With that in mind, I am actually going back to the DAQ approach as it was working pretty darn well even though I was working with lower sampling rates. The piezo transducers are very sensitive and produce a large enough voltage signal even from the smallest of vibration so I am going to revert back to that approach and continue building the DAQFactory program I started.

I still have to build out the combustion chamber - I have already cut out the sheets to form the inner and outer liner as well as made the tubular "donut" and bisected it to form the front wall of the CC and since I have the other half of that donut to work with, I'll use the inner half of the part to form a nice radius at the back end of the inner wall coming out to that ~45 degree rear face.

[fanxing]

how to made it,I think it very hard

[ashpowers]

Not really difficult to build - just time consuming. Lots of cutting, welding, machining but it goes together like an erector set.

[madpatty]

Hi Ashpowers,

Neded some information regarding the fuel injectors in your engine...what is the ID of the needles you are using for squirting fuel into the EVAP tubes....??
I can see in the pics that they are comparatively of larger bore ...Don't the larger fuel injectors cause any problems etc.??

I am using just 0.5 mm ID syringes and they are impossible to weld except that brazing...

Moreover isn't there any difference if you use more number of smaller evap tubes than lesser number of larger evap tubes???
You are using only 6 of them and i always thought they should be atleast 12-13(though according to calculations)...

Cheers,
Patty