Sei sulla pagina 1di 4

Flow Measurement and Instrumentation 9 (1998) 7982

Vortex shedding owmeters and ultrasound detection: signal processing and inuence of bluff body geometry
V. Hans *, G. Poppen, E. von Lavante, S. Perpeet
Institute of Measurement and Control, Institute for Turbomachinery, University of Essen, Schutzenbahn 70, D-45127 Essen, Germany Received 21 March 1997; received in revised form 4 March 1998; accepted 4 March 1998

Abstract The present paper investigates vortex shedding owmeters coupled with a detection of the vortex frequency by an ultrasound barrier behind the bluff body. After a short explanation of the measurement principle a novel digital algorithm for determining the vortex frequency is presented. Another issue deals with the inuence of the bluff body geometry on the measurement process. Comparisons between numerical simulations and experiments were performed in order to analyse the relationship between the shape of the vortex body and the vortices developing behind it. 1998 Elsevier Science Ltd. All rights reserved.
Keywords: Signal processing; Ultrasound; Vortex shedding owmeters

1. Measurement principle Fig. 1 shows a pipe with a uid moving at the mean velocity . Inserting a vortex body causes vortices to separate periodically from both edges of the body. Because of the linear dependency between the velocity of the vortices and , their frequency f at a given point

behind the vortex body can be used to determine the mean ow velocity by f = S d (1)

Fig. 1. Principle of a vortex-shedding owmeter with ultrasound detection.

* Corresponding author. Tel: + 49 201 1832997; Fax: + 49 201 1832944.


0955-5986/98/$19.00 1998 Elsevier Science Ltd. All rights reserved. PII: S 0 9 5 5 - 5 9 8 6 ( 9 8 ) 0 0 0 1 4 - 4

where S is the Strouhal number and d the width of the body. Advantages of vortex shedding owmeters are the validity of Eq. (1) over a wide Reynolds number range as well as its independence of most uid properties such as uid density or pressure. Most industrial owmeters use piezoelectric pressure sensors inside the vortex body to detect the vortices and analogue signal processing hardware to calculate f from the sensor signal. As the vortices develop downstream of the body, the sensor position is unfavourable: a sensor behind the body is better suited for the task of measuring the vortex frequency. A detection sensor system which provides better results because it meets this necessity is an ultrasound barrier behind the bluff body. Since both the vortices as well as the ultrasound wave have the character of pressure uctuations, they inuence each other. This interaction becomes evident in a strong amplitude and phase modulation of the received ultrasound signal. Another advantage of using ultrasound is the high sensitivity of the system to uid pressure uctuations, which enables the use of smaller bluff bodies and less distortion of the ow prole [1].

80

V. Hans et al. / Flow Measurement and Instrumentation 9 (1998) 7982

Fig. 2.

Amplitude-modulated signal in time and frequency domain.

2. Detecting the vortex frequency The use of an ultrasound wave causes a transformation of the entire measurement process into a high frequency range: the low vortex frequency (usually less than 500 Hz) is shifted to a side-band of the ultrasound frequency (111 kHz in this case). However, the advantages of using a carrier signal are connected to a signal processing algorithm of increased complexity. The novel signal processing algorithm presented in this paper makes use of the fact that the spectrum of the original signal is narrow-banded. Fig. 2 shows an example of an amplitude-modulated signal in the time and frequency domain to demonstrate the undersampling effect. If this signal is sampled at a sampling frequency fa which exactly equals the carrier frequency, the spectrum of the original signal is shifted by fa as shown in Fig. 3. The result is a digital signal which contains only the vortex frequency [2]. In order to demonstrate the practical realizability of this algorithm at reasonable costs a measurement device was developed according to the block diagram of Fig. 4. The processing hardware of the complete device consists of no more than three components: 1. A function generator produces the sinusoidal signal to excite the transmitter and the clock signal for the analoguedigital conversion process available. This coupling between the transmitting and receiving side provides the necessary synchronization between sampling frequency and carrier frequency.

Fig. 4. Block diagram of measurement device.

2. An operational amplier circuit adjusts the receiver signal to the input range of the A/D converter. 3. The micro-controller M68HC11 forms the core of the processing system. It was chosen because it is a widespread controller which contains all the necessary processing functions. After converting the analogue receiver signal with its on-chip A/D converter, the C determines the vortex frequency by taking the maximum of the spectrum from a fast-Fourier transform. Fig. 5 shows the vortex frequency versus as measured with a system according to Fig. 4. The relationship between the Strouhal number and for two different

Fig. 3.

Effect of undersampling.

V. Hans et al. / Flow Measurement and Instrumentation 9 (1998) 7982

81

Fig. 5. Vortex frequencies measured with device according to Fig. 4.

bluff bodies is shown in Fig. 6. The results were obtained with air ow in a pipe with a diameter of 10 cm. Pitot tubes in combination with Baratron differential pressure transducers were used to determine the reference values of the mean ow velocity . A remarkable fact is the measurement range of 0.45 to 34 m/s which is reached with the simple processing circuitry of Fig. 4.

3. Inuence of different vortex body geometries The bluff body is the part of the measurement system that is responsible for the production of the vortices. Its geometry should be optimized for producing regular vortices to simplify the detection of a dened vortex frequency. Fig. 7 summarizes the results of this work for two different vortex body geometries. The left column shows the simulated and measured values for a triangular bluff body. The second body, which is described in the right column, differs from the rst one only by its truncated downstream tip. It will be shown that this slight geometrical difference produces a signicant difference in the vortex detection process. The top diagram is the result of the numerical simul-

Fig. 6.

Strouhal number versus mean ow velocity .

ation for the particular vortex generating body. The numerical results are achieved using a NavierStokes solver, capable of handling unsteady, compressible and viscous ows in two-dimensional geometries. The numerical scheme is based on Roes Flux Difference Splitting in the nite volume form [3]. The computational domain is divided into several blocks to simplify the grid generation about the bluff body as well as the formulation of the boundary conditions. Starting initially from a uniform ow data, the governing equations are integrated in time using an explicit RungeKutta method [4,5]. After approximately 0.3 s, the typical periodic asymmetrical vortex shedding is established, forming the usual vortex street. The diagrams at the top row of Fig. 7 show the mean value of the simulated pressure 4 cm behind the bluff body calculated along a path from the transmitter position to the receiver position. The index n is the number of the simulated time step. The distance of 4 cm was chosen because this is the distance between the vortex body and the ultrasound wave (see Fig. 1). The ultrasound wave that reaches the receiving transducer contains the integral of the pressure uctuations between the transmitter and the receiver. As vortices are locations of low pressure, the minima in the top diagram are the points where a vortex crosses the ultrasound wave. The diagrams presented in the middle part of Fig. 7 show the measured receiver signals after amplitude demodulation with the undersampling algorithm that was described in Section 2. Vortices passing the ultrasound wave are again the cause of low signal amplitudes. The main difference between the two vortex body geometries are the local minima in the simulated as well as the measured time signals for the truncated bluff body (marked with S). These minima do not appear in the signals of the triangular body. A minimum value is obviously caused by a vortex, so these minima result from secondary vortices which detach from the edge of the truncated bluff body. Fig. 7 shows secondary vortices and one example of the connection to a local minimum in the measured signal by dashed lines. Since a backward edge is missing in the case of the triangular bluff body there is no production of secondary vortices and a more regular pressure signal is obtained. The secondary vortices are disadvantageous because they become visible in the spectrum of the receiver signal at twice the vortex frequency f . In the case where the amplitude of this component is higher than the amplitude at f , a correct determination of the ow velocity becomes impossible. It is important to note the remarkable agreement between simulation and measurement, which both lead to the conclusion that the triangular vortex generating body is better suited for the present application due to its more regular vortex generation.

82

V. Hans et al. / Flow Measurement and Instrumentation 9 (1998) 7982

Fig. 7.

Simulations and experiments for two different vortex bodies.

4. Conclusion Flow measurement using ultrasound to detect the vortices behind a bluff body is an interesting alternative to other methods. It was demonstrated that handling the measurement process is easy if combined with advanced signal processing methods such as the one that was presented in this paper. Investigations of the geometry of the vortex generating body and its inuence on the form of the recorded signals resulted in an excellent agreement between the experiments and simulations for the two vortex bodies analysed. Due to the success of this method, research work will continue aiming at new vortex body shapes giving better, clearer signals.

References

[1] Miau JJ, Li YP, Chou JH, Huang YC, Yang CC, Shaw JH. Integration of vortex shedder and sensors for a vortex owmeter. In: Proceedings of FLOMEKO 96. Standard Press of China, Beijing, 1996;8994. [2] Hans V, Poppen G. Measuring vortex frequencies using undersampled ultrasound signals. In: Proceedings of FLOMEKO 96. Standard Press of China, Beijing, 1996;72428. [3] Lavante EV, Yao J. Three dimensional numerical simulation of the ow in a four-stroke i.c. engine. In: Proceedings of 6th ICFD Conference. Butterworth-Heinemann Ltd, London 1995. [4] Johnson MW. Computation of ow in a vortex shedding owmeter. Flow measurement and instrumentation 1990;1:2018. [5] Vatsa VN, Thomas JL, Wedan BW. NavierStokes computations of prolate spheroids at angle of attack. AIAA Paper, NASA Langley Research Center, Hampton, Virginia 87-2627 1987.

Acknowledgements This project is supported by the Deutsche Forschungsgemeinschaft (DFG), ME 484/29-1.

Potrebbero piacerti anche