Blade Tip Timing Frequently asked Questions Dr Pete Russhard Rolls-Royce plc 2012 The information in this document is the property of Rolls-Royce plc and may not be copied or communicated to a third party, or used for any purpose other than that for which it is supplied without the express written consent of Rolls-Royce plc. This information is given in good faith based upon the latest information available to Rolls-Royce plc, no warranty or representation is given concerning such information, which must not be taken as establishing any contractual or other commitment binding upon Rolls-Royce plc or any of its subsidiary or associated companies.
Introduction 2 At Rolls-Royce we perform around 10 large BTT tests each year (40 probes+) and a number of question arise several times. This presentation picks on some of a few of the more in depth questions and attempts to answer them from a Rolls-Royce perspective. Each industry sector may have unique conditions and so these answer may not be necessarily applicable to everybody.
The probe 3 Q. Why do I need an optical probe? Q. Why can t I use a capacitance or eddy current sensor? This question is usually based upon the operational requirements, often it is not possible to get to the probe for maintenance and non-optical sensors have higher reliability and lifetimes in most applications. At RR we use exclusively optical sensors for development work. Q. Why? In the next slide we will discuss the probe resolution and relate it to the physical measurement. The probe resolution on its own does not define the need to end system resolution.
Optical probe typical performance 4 If we assume that we take a simple trigger point at 70% of the amplitude and the trigger uncertainty is ±10% (typically much better) then we can estimate the positional uncertainty as 0.1mm. This changes to 0.2mm as the probe to blade clearance changes from 1mm to 3mm. Typical noise at the timing generator is 40mV. (max output 5V) Laser power is either closed loop based upon probe amplitude or constant power maximum 80mW at 655nm. (visible for H&S reasons) For high temperature applications 80mW at 405nm. Optical receiver bandwidth is 3.5MHz.
Capacitance probe typical performance 5 Trigger points for slow rise time waveforms require more complexity. Typically the uncertainty is driven by signal noise. It is possible to achieve a similar resolution of 0.1 to 0.2mm as seen in the optical system. Typical noise at the timing generator is 60mV. (max output 5V) @1mm and 125mV @2mm Conditioning bandwidth is 150KHz. Probe sense diameter is 7.5mm
Optical v s Capacitance 6 Q. So which is best? To answer this we need to consider a few points. Blade vibration can be characterised by the shape, amplitude and frequency of the defection at the tip. The measurement point can be difficult to define for BTT. 120 m pk-pk 195 m pk-pk For a change in axial position of 0.3mm we get approximate 40% change in response amplitude. Note that the deflection at each point may have a different phase 300 m
Field of View 7 Optical sensor field of view Capacitance sensor field of view. A. If you can relate the time of arrival from the probe to blade behaviour then there is no problem. you can use capacitance or optical. Typically Large blade low deflection gradient either technology Small blade high deflection gradient - optical
Probe radial position 8 Q. What are the probe radial positions? A. Probes are positioned to best satisfy the mathematical requirements for the BTT algorithms. Most of which use a matrix to solve for amplitude. The matrix is designed to best solve for a particular engine order Probe positioned are designed to best satisfy many engine orders They may also be acting simultaneously The condition number is a mathematical property of the matrix and is the ratio of the maximum to minimum singular values for the matrix for each of the engine order models. There are many probe positions that can satisfy a set of requirements. A theoretical uncertainty value is associated with the value calculated based upon the BTT data processing
Condition number 9 Q. How do I calculate condition number? The following illustration shows how condition number can be calculated for a simple sinusoid For a square real, matrix, the square roots of the eigenvalues of A T A where A T is the transpose, are called singular values. Substitute = 0.0698, = 0.209, = 0.646 and EO = 3 Calculate the eigenvalues of A T A to be 4.734, 0.025 and 1.25 The singular values are the square root of the eigenvalues 2.175, 0.158 and 1.118
Condition number 10 Q. And what does it really mean? A large condition number (greater than ten??) means that the solution to the equation is sensitive to small changes (i.e. noise in the measurement) and may tend to give numerical problems when calculating a solution. A first approximation shows the range of errors have been created for a number of simulation files to illustrate this.
Condition number 11 Q. Is condition number the best way to define probe position Currently it s the most used method although as the number of probes increases it effect on the final result uncertainty is reduced.
Probe axial position 12 Q. I understand radial position but what about the axial position on a blade. In order to interpret the blade deflections it is necessary to know the sense position on a blade. Initial methods for achieving this were by static measurement and predictions of relative movements. (depicted by the dashed lines) If the stress gradient at the between the dashed lines at the measurement point is high then we need to add further compensation to the measured displacement value. At Rolls-Royce we use additional probes to allow us to extract the change in axial position during the test. typically we can resolve axial shifts of 0.1mm (these may be caused by lean, untwist or axial position change)
Mode Shape 13 Q. My blade modes are close in frequency and may cross how do I identify them? A. This can be achieved by using both leading and trailing edge probes. This type of configuration also allows axial shifts to be determined. This gives confidence that the LE and TE derived displacement's have some correlation. We can obtain the blade shape at speed to compare with our blade FEM Note the ratio change for when the response ND changes. As with any BTT interpretation you need a validated FEM
Frequency Results 14 Q. The blade would have failed within seconds if the amplitudes reported by BTT were true what s happening? A. Most BTT systems report peak amplitudes over a give analysis range. In order to create more realistic values rules can be defined to filter out obvious noise spikes. These are usually based upon continuous coherence against the model for a number of revolutions. In order to give a better perspective on response amplitude then we can also report the distribution of amplitude. 120 100 80 Histogram of Blade 02 (LE mm) Normal Mean 3.423 StDev 0.3313 N 806 Max amplitude = 2 revolutions @ 4.15 4.25mm. By plotting the number of occurrences of amplitude 60 alongside the maximum value a better 40 20 understanding of the response can be gained. 0 1.6 2.0 2.4 2.8 3.2 Blade 02 (LE mm) 3.6 4.0
Resolution 15 Q. Do I really need 10µm resolution? Q. Can you actually achieve 10µm resolution? Depends on the application. Here we show the results for a mid stage HP compressor blade. Here the analysed displacement resolution is approximately 50µm which represents around 2% blade endurance. Currently acquisition to analysis BTT resolutions at Rolls-Royce for a large civil engine (1.3m diameter) are around 25µm Its very easy to get analysis paralysis checking things that don t really matter just takes time, effort and money!
Frequency range 16 Q. What is the highest frequency you can measure? A. Don t know. This is probably the most difficult to answer with any certainty. BTT measures displacement. If there is displacement at a high frequency that is above the resolution of the system then BTT data sets will contain it However, For higher frequencies that are often short lived it may be difficult to find it in the data set, So employ an expert to dig it out for you!!
Component certification 17 Q. Can I use BTT alone to certify my component? A. We would never use a single measurement system alone to certify components, and never have. BTT is a tool that can be used alongside modelling, fatigue testing, and many other standard techniques to gather sufficient certification evidence. Each company must make its own defence of techniques used to the certification authorities.
BTT analysis tools 18 Q. What analysis tools are available and do I have to be an expert to use them? A. Analysis now takes it most of our time in a BTT experiment We no longer regard the conversion of TOA data to displacements and frequency as the hard part. Interpretation is where most effort is now put. The skill of conversion resides in the software suite and, at RR, has to be automated to cope with the volume of data now being generated. The output of the BTT system is displacement data in a format that is compatible with interpretation tools. A standard file format for data interpretation tools would allow the use of third-party software. ISA107.1 please note!!
BTT analysis tools 19 Q. What standard input/output formats are there? A. We have a mixture of formats. Where we read data from our FEM analysis we are tied to a proprietary format however we have produced Excel tools that can mimic this for other users. The majority of our output files are created as.csv and can be viewed in Excel or Minitab. Data from all of internal processing steps are available to encourage thirdparty development of algorithms and methods. these formats are released to registered Users of Rolls-Royce analysis software. These contain similar data to the.ind format but may also include some additional pre processing of the data such as zeroing and filtering
BTT weakest link 20 Q. What the hardest part of a BTT test? A. This differs for each industry sector but for Rolls-Royce it tends to be the application of significant numbers of probes to a vehicle. There are numerous failure points during engine build. Repeat running may be necessary to capture the required data once the engine leaves the test bed there is no way to collect more data. At RR the OPR systems are notoriously unreliable. There are methods available both internally and in the supply chain to remove the need for the OPR which are currently being validated.
Additional data 21 Q. I don t have access to blade endurance data or mode-shapes can I still use BTT? A. We have provided assistance to a number of third party s who are trying to use BTT for the first time. Where no blade endurance data is available then BTT may be no better than a random number generator. You can measure displacements and frequencies but have no way to interpret it. An increase in blade displacement is not always a bad thing. You can trend data over time and relate it to performance or efficiency. As with many other measurements you need supporting information to use the BTT data effectively. BTT is a measurement technology and not just a measurement system.
I want one I think! 22 Q. What do I get when I buy a BTT system? A. A small part of a very complicated measurement process. (sometimes its just a big box of bits) Not only do you need hardware but also an understanding of what you are doing and a knowledge of how to apply it and the resulting data Q. So should I buy a turn-key system? Manuals not included A. If you are new to BTT then we would recommend that you employ experts in the beginning. If you wish to develop your own capability then ensure that you capture any knowledge from third party support.
Vendors 23 Q. I am a vendor, How do I get you to buy my products? A. In the past we have had great difficulties in using off-the shelf parts for our BTT system. Even today, although commercially sourced most of the design are based upon our own designs. We do allow access to our hardware interfaces to allow third-party designs to be integrated. Currently we are looking for new lasers systems, probe maintenance tools, IEEE1588 Ethernet compliant acquisition equipment. This is changing and we hope ISA107.1 will move this forward more quickly. Already from this group we have identified companies to work with to further our capabilities. We would like to leverage our $$$/ by working towards common goals. We will continue to publicise what we do and we would expect vendors to bring validated products to our door. We will invest R&D funding where standards do not exist and there is a clear capability improvement for us.
Vendors 24 Q. So what sort of products are you looking for? A. At Rolls-Royce we operate our systems in as a repeater technology. i.e. we cannot afford to have experts allocated to individual tests. Probes. Ruggedised lasers. Acquisition system. System operators for test support. Open source interfaces. Analysis algorithms. Installation design. Blade calibration services. Analysis off-load. Maintenance tools We want to buy standards parts but not at a premium cost!!
Vendors 25 Q. What facilities can you provide to vendors to assist in developing products? A. We have a number of spin test and environmental test rigs, however we need vendors to meet an initial set of requirements before we commit our resources System pass off rigs High temperature spinning rig Calibration facilities
BTT system development facilities 26 Probe resolution rig System pass-off rig Derby BTT supplier capability. Blade vibration lab University of Manchester.
We buy all of this in reasonable quantities 27 Application Acquisition Development Where do you fit in?? Training Operation Common Tools and Techniques shared across a wide range of people and functions