Blowpipe Antenna in Blast Furnace Raceway Depth Measurement

Abstract

A high gain blowpipe antenna has been presented in this paper. The blowpipe is a pipe used to blow hot air into the Blast Furnace (BF). It has formed the antenna as the radiation unit. A multi-stages cylindrical waveguide has been designed and employed to be the feeding unit. The two units of antenna are separated outside and inside the BF. They are easy to assemble during the BF production process. The antenna operates from 24 to 26 GHz. The simulation result shows that the blowpipe antenna has 25 dBi gain, 20 dB return loss and 30° HPBW (Half Power Beam Width) at the frequency of 25 GHz. Prototype antenna has been fabricated and measured the raceway depth in a real BF. The experiment result shows that the blowpipe antenna can be used in raceway depth measurement. It has advantages of high gain and easy-to-assemble.

1. Introduction

Tuyere raceway is a kind of cavity between tuyere nose (end of blowpipe) and deadman at the bottom of BF (see Fig. 1). Hot air blows into the BF through blowpipe. Gas flow forms cavity. The length of cavity is the raceway depth. Raceway depth determines the size and shape of BF deadman. So it impacts the gas flow and heat exchange. The problem at hand is to achieve exact distance measurement. The condition is complicated due to high-temperature and the three-phase nature of raceway (gas, liquid and particles).

Fig. 1.

General view of tuyere raceway (a) and BF (b). 1. Raceway. 2. Tuyere nose. 3. Blowpipe. 4. BF deadman.

Wolfgang Birk used video monitor¹⁾ in BF tuyere to control the pulverized coal injection. Recently researchers focus on non-contact and direct measure probes, such as laser and microwave, in raceway depth detection. A intermediate catheter²⁾ achieve laser echo signal to take account of raceway depth. MEFOS developed 10 GHz pulse to measure raceway depth.³⁾ Yoshiyuki MATSU also use microwave to measure BF raceway.⁴⁾ The conventional microwave device integrated a gun (antenna) inside the blowpipe. Measurement systems, mentioned above, all permanently mounted components inside the blowpipe.

This paper proposes a blowpipe antenna for raceway depth measurement. The remainder of this paper is organized as follows: Section 2 introduces basic principle and the antenna design. The antenna performance in simulation and measurement in a real BF are illustrated in Section 3. Finally, conclusions are discussed in Section 4.

2. Method and Principle

The FMCW (Frequency Modulated Continuous Wave) principle^5,6) has been employed to detect the distance inside the BF. The measurement scenario is shown in Fig. 2. The geometry of blowpipe antenna is shown in Fig. 6. The blowpipe is similar with the horn antenna. The blowpipe antenna is composed of two parts: radiation unit and feeding unit. 24–26 GHz microwave radiates from the multi-stages cylindrical waveguide into the blowpipe through the observation port. The waveguide mounted on the port. The inner diameter of waveguide channel increases from 16 mm (observation port) to 180 mm (tuyere nose). Multi-section of the channel (see Fig. 6) enlarges gradually and discontinuously. The transmitter and receiver of continuous wave are integrated in an electric encapsulation (Fig. 2 part 3). The electric encapsulation also contains signal sampling and processing modules. The design of electric encapsulation is shown in Fig. 3.

Fig. 2.

Measurement devices. (a) The skeleton of experimental set up. (b) Devices mounted on the blowpipe. (c) Experimental procedure during BF production. 1. Tuyere nose. 2. Wave guide. 3. Electronic encapsulation. 4. Observation port.

Fig. 3.

Electronic encapsulation diagram.

Fig. 6.

Geometry of the blowpipe antenna.

2.1. Basic Principle

FFT (Fast Fourier Transform) operation of time domain signal (see Fig. 4) is shown in Fig. 5. The sampling length in Fig. 4 is N=1024. It’s different in Fig. 5, N=512. The sampling rate fs=38950 Hz. According to BF FMCW application, depth R can be calculated by frequency sampling N as:   

$$R=N\cdot \mathrm{\Delta }R=N\cdot \frac{c}{2B}$$(1)

Fig. 4.

Tuyere raceway time domain AD sampling.

Fig. 5.

Frequency spectrum of echo signal.

Whereas c is the speed of light 3×10⁸ and B is bandwidth of transmit signal 1.6 GHz. 10 GHz, 26 GHz and 77 GHz FMCW radar are applicable for BF burden surface detection. The microwave transmission medium in raceway is more complicated than the medium upon burden surface. The tuyere’s structure, coal injection and fluidized feature are considered to be the main reasons. With increase in operating frequency, the measurement accuracy improves but the penetration of transmission medium decreases. Experiment results show that 24–26 GHz is suitable to be the operating frequency for raceway application. Figure 5 shows the frequency spectrum of the raceway. Campared with the frequency spectrum of the BF burden profile,⁵⁾ the raceway spectrum has lower SNR (Signal to noise ratio). The radiation energy is absorbed and reflected by the complicated transmission medium. The echo energy which distributes around the target area (frequency sampling 45 in Fig. 5) has a high level.

2.2. Design of Blowpipe Antenna

The geometry of blowpipe antenna is shown in Fig. 6. The antenna can be divided into two units: the feeding unit and the radiation unit. The feeding unit is a multi-stages cylindrical waveguide. The radiation unit is composed of observation pipe and blowpipe.

A multi-stages cylindrical waveguide is the feeding unit. The central frequency is 25 GHz. The feeding unit should be resistant to high temperature. Consequently copper has been chosen to be waveguide material. The inner diameter of observation port is D₄=8.5 mm. The inner diameter of cylindrical waveguide should equals to the inner diameter of observation port. It is difficult to achieve impedance matching between coaxial line and cylindrical waveguide with D₄. Consequently three steps of ridge have been designed to achieve better impedance matching. The whole cavity has been divided into 4 cavities by three ridges. The size of ridges have been optimized based on ridge waveguide discontinuity model.⁷⁾ The diameter of cavities are D₁=4 mm, D₂=5 mm, D₃=6.4 mm, D₄=8.5 mm. The length of cavities are L₁=7.4 mm, L₂=3 mm, L₃=5 mm and L₄=5.2 mm respectively. A transmission line radiates the wave from the cable to the first cavity. It is the same as a short dipole. The line insert into the cavity through a hole. The inner diameter of the hole is D_h=2.2 mm. The outer diameter of the transmission line is 0.5 mm. The accurate formula for determining the input impedance of a coaxial line fed linear probe in a cylindrical waveguide has been derived by Wilson.⁸⁾ Based on formula the height (L_Tr) and depth (W_Tr) of the transmission line have been optimized through the parametric study. For better matching, W_Tr=2.65 mm, L_Tr=2.35 mm. As shown in Figs. 7 and 8 the height and width of coaxial probe significantly affect S₁₁ parameter of the antenna. The height should be smaller than D₁. The depth should be smaller than L₁. With increase in L_Tr, S₁₁ improves at higher frequencies but it degrades at lower frequencies. Meanwhile S₁₁ improves at lower frequencies but it degrades at higher frequencies with W_Tr increase. The proper size (L_Tr, W_Tr and D₁) ensures that the transmit signal radiates from the transmission line to the cavity with highest radiation efficiency.

Fig. 7.

S₁₁ variation of cylindrical waveguide for different probe height (L_tr).

Fig. 8.

S₁₁ variation of cylindrical waveguide for different probe depth (W_Tr).

The observation port is the beginning of a pipe which connects to the blowpipe. The BF operator can monitor the coal injection and the production process through the observation port. The inner diameter of the observation pipe is D_ob=16 mm. The length of the observation pipe is L_ob=290 mm. The cylindrical waveguide is mounted on the observation port. A connector has been designed to convert the inner diameter of the cylindrical waveguide D₄ to the inner diameter of the observation pipe D_ob. The length of connector is L_wg2ob=15 mm. There is also a connector between the observation pipe and the blowpipe. It converts the inner diameter of the observation pipe D_ob to the inner diameter of the blowpipe D_bp. The inner diameter of the blowpipe is D_bp=122.12 mm. The length of connector is L_ob2bp=300 mm.

2.3. Noise Reduction Algorithm

DSP samples FMCW signal which is reflected by the raceway. Then DSP transmits the signal to computer. Part of signal has been refracted by blowpipe and subsequently propagated. These multiple reflected waves form a complicated frequency spectrum. The autocorrelation sequence⁹⁾ of echo signal (see Fig. 9) is defined as:   

$$R_x(\tau )=\frac{1}{N}\sum _{t=0}^{N-1}{x}_{t-\tau }{x}_t$$(2)

Fig. 9.

Autocorrelation of echo signal.

Whereas X_t is time domain echo signal, N is sampling length. The length of time lags axis is 2N–1. Echo signal FMCW¹⁰⁾ is defined as:   

$$x_t=Asin(2\pi (f_0+k\mathrm{\Delta }t)t+\phi )$$(3)

Substitute Eqs. (3) to (2), autocorrelation sequence of FMCW can be expressed by   

$$\begin{array}{l}{R}_x(\tau )=\frac{1}{N}\sum _{t=0}^{N-1}sin(2\pi ft+\phi )sin(2\pi f(t-\tau )+\phi )\\ \hspace{1em}\hspace{1em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}=cos(2\pi f\tau +\phi )\cdot \frac{1}{N}\sum _{t=0}^{N-1}\frac{1}{2}(1-cos(2*(2\pi ft+\phi )))\\ \hspace{1em}\hspace{1em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}-sin(2\pi f\tau +\phi )\cdot \frac{1}{N}\sum _{t=0}^{N-1}cos(2\pi ft+\phi )sin(2\pi ft+\phi )\end{array}$$

Trigonometric functions’ summation is zero in period interval.   

$$\begin{array}{l}cos(2*(2\pi ft+\phi ))\\ =cos(2\pi ft+\phi )sin(2\pi ft+\phi )\\ =0\end{array}$$

So the autocorrelation sequence of echo signal is   

$$\begin{array}{l}{R}_x(\tau )=\frac{1}{2}cos(2\pi f\tau +\phi )\\ \hspace{1em}\hspace{1em}\hspace{0.17em}\hspace{0.17em}\hspace{0.17em}=\frac{1}{2}cos(2\pi (f_0+k\mathrm{\Delta }t)\tau +\phi )\end{array}$$(4)

Equation (4) indicates that autocorrelation of FMCW echo signal has the same frequency spectrum. Comparison between autocorrelation sequence and original echo signal has been demonstrated below.

Non-periodic disturbance has largest correlation output at zero lag. Therefore time domain zero-padding¹¹⁾ (see Fig. 10), as Eq. (5), in disturbance area is used to improve SNR.   

$${R^{\prime }}_x(\tau )=\{\begin{array}{l}{R}_x(\tau )\hspace{1em}\hspace{1em}\hspace{1em}\epsilon \le \left|\tau \right|\le N\\ 0\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\left|\tau \right|\le \epsilon \end{array}$$(5)

Fig. 10.

Non-periodic noise interval zero-padding.

Whereas ε is zero-padding area. For the signal of the real BF, the zero-padding length is ε =50. With increase in the zero-padding length ε, the SNR improves but more signal has been filtered at low frequency. The length has been optimized for BF raceway application.

According to Wiener-Khinchin theorem,¹²⁾ FFT operation of ${R^{\prime }}_x(\tau )$ reserves periodic component.   

$$S(f)={{\int }_{-\infty }^{\infty }{R^{\prime }}_x(\tau )e}^{-j2\pi f\tau }d\tau$$(6)

3. Results and Discussion

The radiation pattern of the blowpipe antenna has been simulated in CST software. The weight and size of blowpipe lead to difficulties in measuring the pattern in the anechoic chamber. The feeding unit (Section 2.2) has been fabricated and mounted on the observation port. The performance of blowpipe antenna has been tested in a real BF. It has been evaluated by the signal to noise ratio (SNR) of echo signal. The echo signal has also been processed by the noise reduction algorithm which has been described in Section 2.3.

3.1. Radiation Pattern of the Blowpipe Antenna

The blowpipe antenna operated as a conical horn antenna. The conventional horn antennas have a directional radiation pattern with a high antenna gain. The main difference between blowpipe antenna and conventional horn antennas is the multi-stages aperture enlargement. As shown in Fig. 6, there are two enlargement stages. One is L_wg2ob, the other is L_ob2bp. The radiation pattern of different stages has been simulated in CST software. The simulation results are shown in Fig. 11. There are three lines in Fig. 11. The dotted line denotes the cylindrical waveguide whereas the length of cylindrical waveguide is L_cw. The dash line denotes the cylindrical waveguide and the first stage aperture enlargement (L_cw+L_wg2ob, see Fig. 6). The solid line denotes the cylindrical waveguide, the first stage aperture enlargement and the observation pipe (L_cw+L_wg2ob +L_ob, see Fig. 6). The S₁₁ comparison results in Fig. 11(a) illustrates that the return lose of the complex antenna structure (L_cw+L_wg2ob) is 3–5 dB higher than cylindrical waveguide (L_cw) over the operating frequency band 24–26 GHz. The performance of the S₁₁ parameter becomes worse according to the assembly of observation pipe (L_cw+L_wg2ob +L_ob). However the gain in Fig. 11(b) has been increased 4 dB by the first stage aperture enlargement (L_wg2ob). The comparison between Fig. 11(b) and Fig. 12(b) shows that the total gain has been increased by 16 dB with the two stages aperture enlargement (L_wg2ob and L_ob2bp). Meanwhile the S₁₁ parameter changes a little with and without these two enlargement structures. The physical aperture of cylindrical waveguide is D₄. The assembly of cylindrical waveguide and the first enlargement stage (L_wg2ob) enlarges the physical aperture to D_ob. The total blowpipe antenna has the physical aperture D_bp. The comparison of gain illustrates that multi-stages enlargement of aperture can increase the gain.

Fig. 11.

Antenna comparison between different parts assembly. (a) S parameter. (b) Gain.

Fig. 12.

The blowpipe antenna simulation. (a) S parameter. (b) Gain. (c) Radiation pattern at 25 GHz.

The radiation pattern of blowpipe antenna has been simulated in CST software. The results (see Fig. 12) indicate that the antenna has good impedance matching and radiation performance. The S-parameter results are shown in Fig. 12(a). Over the operating frequency bands, insertion loss S₂₁ nearly equals to 0, return loss S₁₁ is less than –18 dB. Figure 12(b) shows VSWR. Almost all VSWR are less than 2 in frequency range 24–26 GHz. Over the operating frequencies, the gain of antenna is shown in Fig. 12(c). The gain is more than 24.5 dB over 23–28 frequency band. In addition, the antenna radiation pattern at 25 GHz has been simulated. Figure 12(d) shows that the gain increases to 25.4 dB, the side-lobe decreases to –11.2 dB, and the main lobe of radiation pattern maintains the direction of pointing forward, which is suitable for raceway measurement system.

3.2. Antenna Performance in a Real BF

Frequency spectrum (see Fig. 13) shows that the raceway signal has low SNR (signal noise ratio). The spectrum includes a large number of interfering signals. The coal injection inside the tuyere blowpipe generates near-field signal interference. The fixed noises are introduced into the spectrum by the bulk inside the blowpipe channel. Compared with frequency spectrum (dash line in Fig. 13), 10logS(f) (solid line in Fig. 13) has almost 10 dB higher SNR. Two spectrum peaks are the end of the blowpipe and the raceway depth respectively. The depth of raceway is 1.35 meters according to spectrum after autocorrelation filtering (Fig. 13) and Eq. (1). From the graph in Fig. 14 it can be seen that the measured depth varies between 1.05 and 1.65 during 400 seconds operation. The inner diameter of tuyere was 180 mm. The blow pipe flow rate was 6600 m³/min with 32 tuyeres. The flow rate of coal injection was 160 kg/t. The dips in the graph resulted from the coal injection. The raceway depth variation shows smooth and well-correlated with expert experience.

Fig. 13.

Frequency spectrum of echo signal and signal after autocorrelation filtering.

Fig. 14.

Distance variation of the raceway from a real BF.

4. Conclusions

A blow pipe antenna for raceway depth measurement in the BF has been presented. The antenna includes two separated parts inside and outside the BF: the feeding unit and the radiation unit. The two units’ design of blowpipe antenna has practical value. It has the advantage of easy to assemble. The simulation results show that the gain of the antenna with the radiation unit is 16 dB higher than without the radiation unit. Meanwhile the S₁₁ parameter is below –18 dB over the operating frequency band. Experiments at 3# BF of Baosteel Co. Ltd. (Shanghai, China) have been reported. Firsthand echo signal with tuyere antenna has been achieved. The Echo signal, obtained during BF production, has metallurgy characteristics. The noise reduction algorithm has been applied to the echo signal. The Echo signal, processed by the noise reduction algorithm, has 10 dB higher SNR than without processing by the algorithm. The results indicate that the signal of blowpipe and raceway mix in frequency spectrum. The depth of measurement was 1.35 m measured from tuyere nose to raceway. By using of blowpipe antenna and noise reduction algorithm the raceway depth can be identified from mixed spectrum exactly. The result coincides with expert experience. The transmission medium of the raceway is complicated. The raceway has fluidized characteristic. The electromagnetic scattering characteristics of the fluidized bed need to be investigated in further research.

References

• 1)   W.  Birk,  O.  Marklund and  A.  Medvedev: IEEE T. Ind. Appl., 38 (2002), 571.
• 2)   W.  Xue and  R.  Zhen: 2nd Int. Conf. on Digital Manufacturing and Automation ICDMA, IEEE, Piscataway, NJ, (2011), 889.
• 3)   D.  Malmberg,  P.  Hahlin and  E.  Nilsson: ISIJ Int., 47 (2007), 533.
• 4)   Y.  Matsui,  Y.  Yamaguchi,  M.  Sawayama,  S.  Kitano,  N.  Nagai and  T.  Imai: ISIJ Int., 45 (2005), 1432.
• 5)   X.  Chen,  J.  Wei,  D.  Xu,  Q.  Hou and  Z.  Bai: ISIJ Int., 52 (2012), 2048.
• 6)   X.  Chen,  F.  Liu,  Q.  Hou and  Y.  Lu: 9th Int. Conf. on Electronic Measurement & Instruments (ICEMI ’09), IEEE, Piscataway, NJ, (2009), 599.
• 7)   H.-W.  Yao,  A.  Abdelmonem,  J.-F.  Liang and  K. A.  Zaki: IEEE T. Microw. Theory, 42 (1994), 2371.
• 8)   W. W. S.  Lee and  E. K. N.  Yung: IEEE T. Microw. Theory, 42 (1994), 1468.
• 9)   H.  Chen,  Y.  Li and  X.-M.  Wang: IEEE Int. Conf on Control and Automation (ICCA 2007), IEEE, Piscataway, NJ, (2007), 2843.
• 10)   J.  Liu,  X.  Chen and  Z.  Zhang: Meas. Sci. Technol., 17 (2006), 135.
• 11)   G.  Qi: Proc. 7th Int. Conf. on Signal Processing (ICSP ’04), IEEE, Piscataway, NJ, (2004), 2078.
• 12)   L.  Cohen: IEEE Signal Proc. Lett., 5 (1998), 292.