Techniques for blade tip clearance measurements with capacitive probes

Transcription

1 Meas. Sci. Technol. 11 (2000) Printed in the UK PII: S (00) Techniques for blade tip clearance measurements with capacitive probes Alexander Steiner Hytron GmbH, Georg Schröbel Weg 9, D Ober Ramstadt, Germany Received 17 November 1999, in final form 4 February 2000, accepted for publication 13 March 2000 Abstract. This article presents a proven but advantageous concept for blade tip clearance evaluation in turbomachinery. The system is based on heavy duty probes and a high frequency (HF) and amplifying electronic unit followed by a signal processing unit. Measurements are taken under high temperature and other severe conditions such as ionization. Every single blade can be observed. The signals are digitally filtered and linearized in real time. The electronic set-up is highly integrated. Miniaturized versions of the electronic units exist. The small and robust units can be used in turbo engines in flight. With several probes at different angles in one radial plane further information is available. Shaft eccentricity or blade oscillations can be calculated. Keywords: blade tip clearance, capacitive high temperature probes, real time data acquisition 1. Introduction With the increasing optimization of turbomachinery there is a growing demand for precise and easy to use measurement equipment. The minimal blade tip clearance in a gas turbine engine is one of the main parameters governing the efficiency. It is not sufficient to rely on laboratory results but to prove them under actual flight conditions or on the spot in turbo power plants. Due to the hostile turbo machinery environment mainly in the turbine the use of capacitive probes seems to be obvious. Their electrical function is described with a simple capacitor. No complicated structures, optics or active elements are involved. This is a good starting point for the fabrication of a heavy duty probe. To figure out the clearance from the small change of capacitance when a blade passes the probe, several methods were developed in the past (Chivers 1989, Killeen et al 1991). In most cases the use of these methods is restricted by some disadvantages. Concepts based on charge amplifiers or phase locked loop techniques show band-width limitations. Other concepts like pulse width modulation systems accept only a very little probe capacity (Moenich 1989). The electronic circuits in these systems have to be placed only a few centimetres away from the probe head. Therefore the electronic circuits have to be cooled even in many compressor applications. All of the mentioned systems also demand the electrical parameters of the probe to stay within a small region. An improvement was reached by using a combination of charge amplifiers, current to voltage converters and fast sample and hold stages (Bailleul et al 1993). This system is useable at blade passing frequency up to 55 khz. An autocalibration function compensates a drift of the probe capacity in the range of 10% of the initial total value. The remaining disadvantage of this system is the distance between the probe head and the first amplifier. The maximum distance is 1 m. In this article a new system will be presented. Further developments lead to the new concept based on HF signal processing and digital filter technology. Basically the probe s response to an HF signal input will be analysed. The new system allows distances up to 8 m between probe head and the electronic unit with no loss in performance. 2. Capacitive probes If any kind of probe should be used in a turbine of a turbo machine some points must be taken into consideration. The environment for the probe is very hostile. Temperatures of about 1000 C and thermal stresses must be survived. High mechanical demands exist because of vibration, pressure and different thermal expansion coefficients. Ionized and corrosive gas should not influence the properties of the probe. Because the signals to measure are extremely small, the electrical characteristics must be excellent and stable. Figure 1 shows an example of a capacitive probe head. It is a coaxial model with a centre electrode. A surrounding electrode is formed by the housing of the probe. The capacity of such a probe is about 700 pf. This includes6mofcable and a coaxial plug. If a blade passes by the capacity will change slightly due to the fringe field effect. The system described here is able to resolve changes of less than 100 ff /00/ $ IOP Publishing Ltd 865

2 A Steiner via laser welding. Much better results in terms of durability under the harsh conditions in a turbine were found with another technique. It is possible to apply a conductive ink of platinum at the ceramic insulator. A very reliable and strong connection can be built inside an oven at temperatures above 1200 C. The mineral insulated cable and the other metallic parts of the probe head are fixed together by laser or plasma welding Probe design Figure 1. Drawing of a probe head, head diameter 6.2 mm Materials and fabrication of the probes The usual alloys for high temperature applications are applicable for the probe housing. Inconel turns out to be a good choice. Also for the seals in turbomachinery commonly used materials meet the requirements. More attention must be paid to the insulator. The use of glazed alumina is favourable because of different reasons. This material is in common use. A wide range of application exist, which are very helpful for the design of the probes (Sheard et al 1999). The thermal expansion coefficient fits the coefficient of the alloys used. At high temperature glazed alumina shows good and stable insulation properties. For applications below 600 C ceramics based on magnesium oxide are also useable. Beside the probe head the cable must be able to withstand the severe conditions. Mineral insulated cable fits perfectly to these demands. It is available with Inconel shield and nickel alloy as inner conductor. The insulation is made of silicon oxide. A major problem is to fix the centre electrode to the ceramic insulator. Probes with a thin platinum plate work. The plate is fixed to the inner conductor of the probe cable The design of the probes varies depending with the application. The diameter of the inner electrode of the probe head determines the maximum clearance which can be measured precisely. A rule of thumb says half of the diameter as the maximum clearance. In most cases the geometric form of the inner electrode is plane. In some applications the geometry of the blades have considerable changes in the region of the probe head. In such a case the electrode can be shaped to fit to the contour of the blade. 3. Signal processing The price to be paid for a heavy duty probe is the very small signal output of such a probe. The change of the probe s capacity because of the temperature, pollution, long time drift etc is much higher than the changes because of the passing blade. The diagram in figure 3 shows all functional blocks of the electronic circuits. Every probe is connected to an amplifier unit (AU). The AUs provide an analogue signal output as well as an asynchronous 2.5 MBaud serial link. The serial interface can be connected to a PC card. One PC card is able to process the data from 4 AUs. It is possible to install up to four PC cards in one PC. Figure 2. Two probe heads and two ceramic insulators. 866

3 Clearance blade tip clearance measurement Figure 3. Signal processing The amplifier units A probe is connected to an AU via a simple coaxial plug. The input stage and the HF processing stage are the heart of the analogue signal processing. An HF signal is fed to the probe. The probe s response varies due to the blade tip clearance. Further, essential for this concept, is the unit for digital filtering. At this block the information of the actual blade tip clearance is extracted from the probe signal response. This function is realized with a fast field programmable gate array (FPGA). The feed-back loop via the digital analogue converter provides wide range compensation for the disturbing changes of the probes capacity. A second FPGA manages further data conversion. At this point the value of the blade tip clearance is available as a digital number with ten bit resolution. About nine million measurements are taken per second. This results in several samples of clearance for every blade even at high blade passing frequencies. At the end of this process the data is compressed and a serial bit stream is generated. Usually fibre optic transmission is used to send this data to the following signal processing PC card. The optical transmission of digital data is very reliable. Distances up to 2000 m between an AU and a PC card can be realized The PC cards The output from the AU represents the tip blade clearance. To obtain the clearance in engineering units the data must be linearized and some calibration data must be taken into consideration. A PC card is able to process this for the data from four AUs in real time. Again fast FPGAs are used. Most numerical calculations are done by a digital signal processor (DSP) present at any PC card. Also the interface to the PC is done by the DSP. Hence the PC itself is not involved in any calculations. Only the graphical output and other kinds of user interface need the PC s capabilities. 4. Handling of the system 4.1. Calibration The characteristic of a probe output is clearly nonlinear. For the greatest possible precision the influence of the blade tip geometry must be taken into consideration. The use of a calibration wheel was found to be a good method. The calibration wheel looks like a gear wheel. The cogs are shaped like the blade tips. The wheel is mounted on 867

4 A Steiner an electric motor. For a calibration the probe is mounted on a linear traversing system. A calibration starts with a clearance of zero: the probe is moved towards the rotating wheel until the probe touches the wheel slightly. This is no risk for the heavy duty probe. From this position the probe is moved back from the wheel in discrete steps up to the maximum clearance of interest. With the data from the positioning system and the measured data from the probe the PC calculates the calibration parameters. In actual operation these parameters are used by the DSP to perform a real time correction of the data. This correction is based on spline approximation. An AU provides a certain gain of the probe s signal input. This gain is chosen according to the probe used and the clearance to be measured. For some applications the gain can be switched in operation. In this way coarse and fine measurements are possible with traversed probes. In a wide range the gain does not affect the linearization. An example for such a set-up is blade grinding machines. Changes of the probe s electrical characteristics because of changing temperature or pollution etc usually do not affect the calibration. Effects from these influences are compensated by the electronic units Visualization and PC software It is possible to use the system without the PC cards. If only a coarse value of the clearance is wanted, the analogue output of an AU can be used. The monitoring for tolerance limit violation in gas turbines is one example. Also customized stand alone units with a FPGA and maybe a DSP are possible. These units are connected to the serial link of the AUs. The utilization of a PC and PC card(s) offers the most flexibility. The graphical display provides an overall view of a large set of data. Minimum, maximum and average clearances of a stage are also possible as single blade trend analyses. Several screens can be selected: an overall view of all applied probe data or zoomed display of specific data. All data are available from hard disk for later offline analysis. With specific set-ups advanced measurements can be achieved. One example is the use of three probes in one stage. These probes are mounted in radial angles of 120. The PC is now able to calculate the shaft eccentricity from the probes data. Another example is the calculation of blade oscillations from a set of data from two probes. 5. System performance As mentioned before, the range of the clearance measurement depends on the probes and the gain of the AUs. With increasing distance between the probe head and the blade tip the signal to noise ratio decreases. If the signal to noise ratio reaches a significant value for the desired accuracy the end of the measurement range is reached. Also for the minimal clearance a limitation exists. With a given gain the amplifiers will run into saturation at a certain clearance. For example, a probe with 6 mm head diameter is used. With a specific gain the amplifiers saturate at 0.3 mm clearance. For this set-up single blade clearance tracking will be possible up to 1.8 mm. Even higher clearances can be measured, but the increasing influence of noise demands averaging over several blades. Because of the ten bit data processing in the AUs the theoretical resolution is about the 1/1000 part of the difference between the maximum and the minimum clearance from the calibration. The useable resolution is reduced because of some factors. First the influence of noise must be mentioned. Less important effects arise from residual errors of the digital filters or the linearization. To use the above given example again, an accuracy of at least mm can be expected in the range from 0.3 to 1.8 mm. In many cases this is less than the uncertainty of the probe mounting in the turbo machine. Practically probes for the evaluation of clearances up to 5 mm are available. Probes for clearances up to 8 mm are in development. Due to the high data processing rate blade tip clearance can be measured at blade passing frequencies in the range from 0.3 to 45 khz. This standard range can be shifted to lower or higher frequencies. At present the absolute maximum blade passing frequency is 55 khz. The absolute lower limit is about 0.1 khz due to AC coupled signal paths. Probes were successfully tested at several turbines. This includes more than 50 hours of operation at temperatures of 800 C and thermal stresses up to 1000 C. 6. Further developments Although the system performance has been proved for many applications the further development of the system is in process. One point of interest deals with the easier handling of the system. One example is a more automated calibration procedure. Another point is the utilization of an additional microcontroller chip in the AUs. An AU has to be manually adjusted to a specific probe in the present system. The microcontroller will exempt the user from any adjustments. The aim of the second group of developments is to expand the practical useability of the system. Hence efforts are being spent to develop a special probe for blade oscillation measurements. For these measurements not the tip clearance but the exact position of the tip at a certain time must be evaluated. A prototype of this probe uses a differentialcapacity principle. Other objectives of the developments are set-ups for the monitoring of centrifuges, slide bearings and high speed shafts. 7. Conclusion The presented system provides a powerful and reliable way for blade tip clearance measurements even under extremely harsh conditions. The heavy duty probes are built of special materials. They are able to withstand the severe environment conditions in a turbine for continuous operation. Sophisticated electronic circuits deliver exact results from the very small output of a probe. High speed data processing allows single blade tracking and a large variety of possibilities for data presentation and advanced data calculation. The previous use of the system shows its potential for use in further applications. 868

5 Clearance blade tip clearance measurement References Bailleul G and Moenich M 1993 New digital capacitive measurement system for blade clearances Proc. 39th ISA Int. Instrumentation Symp. ChiversJWH1989 A technique for the measurement of blade tip clearance in a gas turbine AIAA Forster R L 1989 Linear capacitive reactance sensor for industrial applications SAE Killeen B, Sheard A G and Westerman G C 1991 Blade by blade tip clearance measurement in aero and industrial turbomachinery Proc. of the 37th ISA Int. Instrumentation Symp. Knoell H, Schedl K and Kappler G Two advanced measuring techniques for the determination of rotor tip clearance 5th Int. Symp. on Airbreathing Engines (Bangalore) Moenich M 1989 Konstruktion eines kapazitiven Elongationsmessystems zur Betriebsüberwachung eines Verdichterprüfstandes TH Darmstadt Mueller D, Sheard A G, Mozumdar S and Johann E 1997 Capacitive measurement of compressor and turbine blade tip to casing running clearance ASME J. Eng. Gas Turbines Power Sheard A G, O Donnell S G and Stringfellow J F 1999 High temperature proximity measurement in aero and industrial turbomachinery ASME J. Eng. Gas Turbines Power