Will Blade Check Airfoil Per Pwa 330

[scott bean]

Does anybody know if blade will check an airfoil per PWA330 airfoil specification?

Thanks
Scott Bean
[email protected]

[Eric Hollenbeck]

Hi Scott,
The answer is yes. The best fit method used in Pratt's PWA330 spec is the same as "2 point" best fit used in Blade.

Eric

[Pat Ahern]

PWA 330 defines both Airfoil SETUP and analysis as "best-fit". In order to create the stacking axis (cmm coordinate system) you must equalize material condition at the Airfoil setup sections defined on the B/P. How can you inspect the setup sections with a "coordinate" measuring machine, if you don't have coordinates to start with? How do you know the cmm is on the correct A/F section line when you collect the data to use for equalizing the setup?

Anyone care to comment?

[Pat Ahern]

录音卡技术参数：　　接口方式：录音管理系统 模拟中继　　录音输入阻抗录音管理系统 直流>= 470K欧姆?交流（1000HZ时）>=10K欧姆　　隔离阻抗脚直流均：>= 470K 欧姆　　录音管理系统 录音模块直流阻抗： >1M　　录音管理系统 隔离耐压：>=500VDC　　录音格式：LAW、ADPCM，GSM格式的WINDOWS WAV文件　　数据速率：64 KBPS /32 KBPS /13.6 KBPS　　录音管理系统 全通道信噪比：>= 42 dB　　频响：300~3400HZ　　放音输出功率 >= 0.5 W　　全通道失真：< 5%　　音量调整：-20db?~?+20db　　硬盘容量 (40G最多可存放7000小时)

wtf?

[Brian Gudauskas]

I know this is old, but for reference: You scan the airfoil and then the software best fits and balances the sections automatically. It uses correlatable code to perform it, just like P&W does internally with their blade software (PWAirfoil).

PWA 330 defines both Airfoil SETUP and analysis as "best-fit". In order to create the stacking axis (cmm coordinate system) you must equalize material condition at the Airfoil setup sections defined on the B/P. How can you inspect the setup sections with a "coordinate" measuring machine, if you don't have coordinates to start with? How do you know the cmm is on the correct A/F section line when you collect the data to use for equalizing the setup?

Anyone care to comment?

[Pat Ahern]

But how do you know that the physical scanning is at the correct location (on the section line) all the way around the part, when the part is not set up yet?

[Brian Gudauskas]

Pratt is very specific when it comes to datums and "setup". They will have specified on the print or in a spec how to bestfit for setup. Read carefully and you will find it. Once you do the setup you are then on section line.

[Pat Ahern]

Yes, the setup and analysis spec is PWA 330. I've read and used the turbine portion many times. Set up is accomplished using a guillotine gage and an optical comparator.

Catch 22 this. Before a bestfit airfoil setup is complete, there is no stacking axis. It hasn't been created yet. The stacking axis is your zero for both X and Y directions, and is used to create the radial direction datum point. The cmm is a "coordinate" measuring machine. It needs coordinates to be able to move. Coordinates come from a point, that you set to be 0,0,0. But that point, in a bestfit setup, hasn't been created yet. So how does the cmm even move? What coordinate zero point are you using to do the initial scanning of the airfoil? Certainly not the part's stacking axis.

The part is physically sitting on the cmm, but it's not properly setup yet. And since the part is not setup yet, the cmm is not parallel and perp to the part yet. So while the z height of the cmm may be exactly the same on both the concave and convex sides of the scan, relative to some coordinates, but not the castings' bestfit coordinates, the scan was NOT actually at the proper section all the way around the part. So now Blade comes in and uses this data that's not from the correct section, to do it's bestfit. If the physical scan is not the correct data, than the results of the bestfit will not be correct either. It can't get a proper analysis/bestfit, when it starts with incorrect data.

You said it yourself, "once you do the setup you are then on section line". But the cmm needs to be on the section line BEFORE the software does the bestfitting.

Lots of luck.

[JEC_31]

Pat,

So are you bringing up the possible catch-22 from the concern that ScanBlade jumps onto a part with a rough alignment (or none at all?) and therefore is producing suspect data?

We have a seperate PCDMIS program that first finds the chunk of metal with a rough 6-point iterative DCC alignment before scanning the prescribed datums and setting-up using the prescribed Best Fit method. It finishes with a clearance move up in the air and then saves the alignment externally. The ScanBlade program then starts, recalls that alignment, and transforms the directions of this coordinate system to suit it's methods so that it can work it's magic.

Even if if only a rough aligment were used and Blade used it's Best Fits to "align" the raw data with nominal, would not the fact that ScanBlade runs without probe compensation greatly reduce cosine error?

Just my 0.00000000002 cents.

- Josh

[Pat Ahern]

Hey Jec_31,

Do you physically scan the airfoil set up sections multiple times (at least twice)? Something like this:

Align it to the 6 point nest, scan the set up sections, best fit the setup sections, create a new alignment to the bestfit setup sections (can Blade even do this?), save this new alignment, then physically scan the airfoil again to the new bestfit alignment. Re-bestfit the new scan data and report it.

Lots of prints inspect the airfoil to a 6 point nest (which is not a bestfit setup), and I have no doubt Blade can handle that. And if the airfoil shape is constant in the radial direction, then scanning points that are not on the actual section isn't that big a deal. But I've also seen airfoils that had such a large bow in them, that it almost created a 45 degree angle. It doesn't matter whether you're using probe comp or not on a part like that. For every .001 (in the radial direction) that the scan isn't on the proper section line (because you scanned it to a 6 point nest alignment instead of the print's bestfit alignment), the data being collected is inaccurate by .001 too. And when then profile tolerance is +/- .003, even just .001 is significant.

Sometimes close is good enough (using a 6 point nest when the print has a bestfit setup).
Sometimes it's not. Is Blade just being close? Or can it actually scan the airfoil, bestfit it, re-align to the bestfit, scan it again using the new alignment, re-bestfit it, and then report it?

[JEC_31]

Hey Jec_31,

Do you physically scan the airfoil set up sections multiple times (at least twice)? Something like this:

Align it to the 6 point nest, scan the set up sections, best fit the setup sections, create a new alignment to the bestfit setup sections (can Blade even do this?), save this new alignment, then physically scan the airfoil again to the new bestfit alignment. Re-bestfit the new scan data and report it.

Here's our system:
1) PCDMIS runs a 6-point iterative alignment, target 0.002", measure all always, limit 99, runs twice. I tried 0.0015 and it ran 10 times, so 2 thou' is the way it is.
2) PCDMIS scans datums, which in our case are tenons on both ends of the blade (a gas turbine vane) and take points on Z datums in packing areas inside the tenons.
3) PCDMIS 3D Best Fit the scans, vector fit. There's a lot of movement here as it adjusts from the rough 6-point, 0.030" to 0.040" in X and Y.
4) PCDMIS 3D best Fit, constrain to Z only the Z datum points.
5) Save alignment externally.
6) PCDMIS ScanBlade program loads, recalls saved alignment, transforms it, and begins scanning.
7) Raw Data is dumped to Blade. From this point on I do not know the inner workings of Blade.

Due to the nature of the datums, especially having the Z location solidly defined, we are confident that we are scanning on nominal section well within reason. The amount of movement from the 6-point to the 3D Best Fit initially had me worried but I got over it, it's mostly the third primary point not ecompassing the tenon it's on.

Lots of prints inspect the airfoil to a 6 point nest (which is not a bestfit setup), and I have no doubt Blade can handle that. And if the airfoil shape is constant in the radial direction, then scanning points that are not on the actual section isn't that big a deal. But I've also seen airfoils that had such a large bow in them, that it almost created a 45 degree angle. It doesn't matter whether you're using probe comp or not on a part like that. For every .001 (in the radial direction) that the scan isn't on the proper section line (because you scanned it to a 6 point nest alignment instead of the print's bestfit alignment), the data being collected is inaccurate by .001 too. And when then profile tolerance is +/- .003, even just .001 is significant.

Aha! With a bow that serious now I understand and can see how concerned you are on section-line setup. I learned much of my CMM craft working in the plastics field where it's common to set-up and measure large parts warped literally inches away from nominal - I had to teach myself variables just to be able to find the part! My mantra has always been, "The CMM probe is blind. It doesn't know if it's probing in the wrong spot." Once the tolerances shrink this does become critical.

Sometimes close is good enough (using a 6 point nest when the print has a bestfit setup).

Sometimes it's not. Is Blade just being close? Or can it actually scan the airfoil, bestfit it, re-align to the bestfit, scan it again using the new alignment, re-bestfit it, and then report it?

IMO, the scan-align-rescan-realign-final scan scenario sounds like a job for the programmer to accomplish with PCDMIS first before handing the airfoil over to ScanBlade for the ultimate-final scan and Blade to crunch the numbers thereof. That way we don't have to worry about relying on Blade's inner math workings to iron-out probing errors.

- Josh

[Pat Ahern]

Hey Josh,

My original involvement in this thread was to remind people that PWA330 is about setup as well as inspection. Your scanning system sounds good. And it sounds like you are doing 2 sets of scans. Using PCDMIS to 3d best fit, and creating a new alignment is a good idea. I didn't see how "Blade" was going to create an alignment to re-set up to, for the re-scanning.

I've worked on a few vanes with tenons, and I believe the shape was constant in the radial direction. That approximate 45 degree surface I mentioned was not a deviated part. That was the nominal shape (crazy Pratt part). A constant radial shape really does give you a lot more room for error. And an actual part that is close to nominal really helps too. You know, if the part was actually nominal, then the 6 point nest setup and the airfoil bestfit setup would be exactly the same. The further away from nominal that the part is, the more critical the setup technique becomes. But you knew that already

Good luck.

[JEC_31]

Thanks Pat - I was hoping we were on the right track.

Blade was first described to me as a reporting software, it doesn't do the real CMM work. I think that this thread would've been better titled, "Will The PCDIMS-Bladerunner-Blade Turnkey Package check Airfoil per Pwa 330".