Blast Furnace Gas Flow Strength Prediction Using FMCW Radar

Abstract

A method of predicting gas flow strength upon burden surface has been proposed in this paper. A frequency modulated continuous wave (FMCW) radar is employed to achieve the echo signal of burden surface. The radar, operating at 24–26 GHz, is mounted on top of blast furnace (BF). The burden surface scattering yields attenuation due to the burden surface dust. The propagation constant has been derived in terms of gas flow velocity, particle size, dust concentration and so on. The method has been validated in real blast furnace during production process and maintenance process, respectively. The experiment results show that the predicted value has close agreement with the measured value. The attenuation has the same periodic trend with the BF charging process.

1. Introduction

Blast furnace (BF) airflow strength has a great impact on energy efficiency. Raw material is loaded from the top of the BF. The burden surface consists of a combination of coke and ore. The harsh environment upon the burden surface becomes glutted with gas and dust. It is a dense strongly scattering medium. In the upper part of shaft, burden distribution and gas flow can be modeled using decent model and Eurgun’s equation.¹⁾ The three-dimensional gas flow distribution can be simulated using discrete element method and computational fluid dynamics (DEM-CFD) method in the entire blast furnace²⁾ and the cohesive zone.³⁾ A number of papers have been published to address the microwave propagation of sand and dust. Since wavenumber and the particle radius is small, kr << 1, the attenuation can be simplified using Rayleigh approximation.⁴⁾ Mie theory has been used to solve the scattering problem for kr~1 or kr > 1.⁵⁾ For BF application, the main particle radius varies from 0.05 mm to 15 mm. The gas flow upon the burden surface floats the small size burden surface dust. The burden surface dust and the BF gas form a gas-solid fluidized bed. The relationship between gas flow velocity and dust density was investigated in fluidized bed problem.⁶⁾ Frequency modulated continuous wave (FMCW) radar has been widely employed in burden profile monitoring.^7,8) In this paper, a method of BF gas flow prediction has been developed due to the combination of fluidized kinetic theory and attenuation in microwave propagation.

2. Method and Principle

Since the ratio of particle radius to wavelength is small, kr << 1, the propagation constant⁴⁾ is   

$$K_{V,H}(\phi )={\mathrm{k}}_0+\frac{3}{2}{k}_0\frac{{\epsilon }_m^{*}-1}{{\epsilon }_m^{*}+2}{v}_r$$(1)

Where k₀ denotes the propagation constant in free-space, ${\epsilon }_m^{*}$ is the complex dielectric constant of particle, v_r is the volume of dust particle.

The mass of dust particles M is related to particle density ρ (kg/m³), volume and visibility V_b (m).   

$$M=\rho \cdot v_r$$(2)

  

$$M=\frac{C}{V_b^{\gamma }}$$(3)

Where C = C_BF and γ = γ_BF are the experience constants for blast furnace. The volume can be derived as   

$$v_r=\frac{M}{\rho }=\frac{C_{BF}}{\rho V_b^{{\gamma }_{BF}}}$$(4)

Substitute Eq. (4) into Eq. (1), the propagation constant can be expressed as   

$$K_{V,H}(\phi )={\mathrm{k}}_0+\frac{3}{2}{k}_0\frac{{\epsilon }_m^{*}-1}{{\epsilon }_m^{*}+2}\times \frac{C_{BF}}{\rho V_b^{{\gamma }_{BF}}}$$(5)

Momentum balances has been established between gas phase and solid phase in light of kinetic theory.⁶⁾ Particles are considered smooth, spherical, inelastic, and undergoing binary collisions. The derivation is suitable for centre of blast furnace. The boundary conditions, such as blast furnace wall, are not considered. For porosities < 0.8, the particle density can be expressed as   

$$\rho =\left(\beta -150\frac{{\epsilon }_s^2{\mu }_g}{{\epsilon }_g^2{(2r)}^2}\right)\times \frac{{\epsilon }_{g}2r}{1.75{\epsilon }_s\left|v_g-v_s\right|}$$(6)

Where β represents the interface momentum transfer coefficient, v_g is the velocity of gas, v_s denotes the velocity solid dust, r denotes the particle radius, μ_g is the gas shear viscosity, ε_g and ε_s denote the concentration of gas and solid dust, respectively.

Substitute Eq. (6) into Eq. (5), the propagation constant can be expressed as   

$$\begin{array}{l}{K}_{V,H}(\phi )={\mathrm{k}}_0+\frac{{\epsilon }_m^{*}-1}{{\epsilon }_m^{*}+2}\\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\times \frac{1.3125\cdot k_0{C}_{BF}{\epsilon }_s\left|v_g-v_s\right|}{V_b^{{\gamma }_{BF}}{\epsilon }_{g}r}\times {\left(\beta -37.5\frac{{\epsilon }_s^2{\mu }_g}{{\epsilon }_g^2{r}^2}\right)}^{-1}\end{array}$$(7)

The imaginary part of propagation constant Im(K_V,H(φ)) represents amplitude attenuation of the wave due to dust scattering and absorption.

3. Results and Discussion

To investigate the impact of particle size, concentration and thickness, experimental setup has been established in lab. The experimental target consists of solid bulk coke, granular coke and coke powder. The particle size is shown in Fig. 1. The granular coke and coke powder are placed upon the solid bulk coke. The solid bulk coke represents the burden surface. The granule and powder represents the BF dust. The radius of granular coke and coke powder are less than 1 cm and 1 mm, respectively. The radius of solid bulk coke is 5–10 cm. A network analyzer Agilent PNA N224A generates 24–26 GHz frequency band frequency modulation continuous wave (FMCW). An industrial radar antenna⁸⁾ has been employed to transmit and receive the FMCW. The value of kr becomes <5.2, <0.52 and 26–52 for granular coke, coke powder and solid bulk coke for frequency of 25 GHz. k is the the wavenumber of the incident wave. Consequently the powder coke satisfies the Rayleigh condition. The powder coke is the main component of BF dust. The fraction of particles with radius less than 1 mm is 95.9%. The granular coke and solid bulk coke are the main component of burden surface. The fractions of granular and solid bulk are 30.88% and 26.35%, respectively. The height between the horn antenna and solid bulk coke is 1.5 meters. The FMCW radiates the solid bulk coke through the dust. The proposed gas flow prediction method has been validated in a real BF during BF production process and maintenance process. The industrial FMCW radar⁸⁾ has been employed in the real BF measurements.

Fig. 1.

Particle size comparison among granular coke, coke powder and solid bulk coke.

Fig. 2.

Echo signal measurement using different particle size.

3.1. Influence of Particle Size on Propagation

The S₁₁ parameter, with different particle size, is shown in Fig. 3. The coke dust is parallel to the solid bulk coke surface. It is perpendicular to the radiation direction. The dust is placed on the top of foam board (the permittivity is 2 < ε < 3). The S₁₁ at the distance 1.14 meters without dust is –60 dB. The S₁₁ with coke powder is –57 dB. Consequently the coke powder generates 3 dB noise. The S₁₁ with granular coke is –50 dB. Compared with the coke powder, the granular coke generates 10 dB noise. The larger particles scatter much more strongly (σ/πr² ∝ (kr)⁴, σ denotes radar cross section) than smaller particles.⁹⁾ The S₁₁ at the distance 1.44 meters is –45 dB. The attenuation of solid bulk coke, which is caused by dust, is less than 1 dB. The signal to noise ratio (SNR) is about 10–15 dB.

Fig. 3.

Spectrum comparisons among different particle size.

Fig. 4.

Echo signal measurement with different concentration.

3.2. Influence of Dust Concentration on Propagation

The distance between antenna and coke powder is 1.1 m. S₁₁ parameter with different dust concentrations is shown in Fig. 5. Different weight ((a) 5 g, (b) 10 g, (c) 15 g, (d) 20 g, (e) 25 g, (f) 30 g) is placed in a circle area. The radius of the circle is 15 cm. As shown in Fig. 5, the S₁₁ parameter of solid bulk coke surface, located 1.44 meters, is –50 dB. The variation of coke surface’s S₁₁ with increase of dust concentration is less than 5 dB. Meanwhile the phase shift variation of coke surface is less than 0.1 meters. On the other hand, the dust concentration has a considerable impact on the coke dust’s spectrum. The amplitude of coke dust increases 1 dB·g^–1 in average with increase of dust concentration. Figure 6 shows that the noise signal intensity is proportional to dust concentration. Compared with coke surface signal, the SNR increases 1 dB·g^–1 with increase of the dust concentration. According to Eq. (5), the attenuation is proportional to the concentration ρ and the dielectric constant ${\epsilon }_m^{*}$ . The measured results in Fig. 6 seem to correspond to volcanic ash results.⁹⁾ This is partly because the concentration has impact on dielectric constant. For gas flow prediction the dielectric constant of BF dust is assumed to be constant.

Fig. 5.

Spectrum comparisons among different concentration.

Fig. 6.

S₁₁ variation of dust for different concentration.

Fig. 7.

SNR variations for different concentration.

Fig. 8.

Echo signal measurement among different thickness of coke dust.

3.3. Influence of Dust Thickness on Propagation

The stronger the BF gas flow is, the thicker the coke dust is. The scenario is simulated in the lab using multilayer coke dust. The coke powder has been placed on the top of foam board. The height of foam board is 10 cm. Each layer has the same radius (15 cm) and weight (5 g). The S₁₁ parameter of multilayer is shown in Fig. 9. The coke dust layer has noise signal respectively in the distance of 1.34, 1.24, 1.14, 1.04 and 0.94 m. The distance is correlated with layer height. The noise intensity is respectively –52.5, –52.2, –53.9, –64.3 and –64.9 dB. However the coke surface has an intensity of –44 dB. The difference of coke surface signal is lower than 2 dB. With a constant target signal the SNR represents the noise signal intensity. The SNR decreases 0.5 dB·cm^–1 with increase of dust thickness. According to Eq. (5), the relation between the attenuation and visibility is shown in Fig. 11. The parameters are as follows:⁴⁾ the frequency f = 25 GHz, complex dielectric constant ${\epsilon }_m^{*}=4.0-1.325j$ , the dust concentration ρ = 2.44 × 10³, the constant C_BF = 6.9 × 10^–5 and γ_BF = 1.07. It is shown that the simulation has 0.38–0.59 dB·cm^–1 attenuation for dust height between 0.94 and 1.34 meters. It is consistent with the experimental result (0.5 dB·cm^–1). It also shows that the attenuation decreases with the increase of distance. The distance between radar and burden surface is about 4–8 metres. The attenuation within the range of 1–10 meters is shown in Fig. 12.

Fig. 9.

Spectrum comparisons among different thickness of coke dust.

Fig. 10.

SNR variation for different thickness of dust.

Fig. 11.

Attenuation simulation in experimental distance range.

Fig. 12.

Attenuation simulation using BF parameters.

3.4. Influence of Gas Flow Strength on Propagation

Comparison of radar SNR, BF charging signal and burden surface height is shown in Fig. 13. The radar is mounted on top of 10th BF 1080 m³ within Yonglian Iron and Steel Corporation. The radar radiates the centre of the BF. The centre has stronger gas flow than the edge. The SNR is calculated by the amplitude of burden surface spectrum and the amplitude of the dust spectrum upon the burden surface. The dotted line represents the BF charging signal. The value 1 and 2 represents charging coke and ore, respectively. At the end of charging, the raw materials reduce the gas flow strength upon the burden surface. Consequently the SNR has peak value at the end of charging. Due to the increase of gas flow strength the SNR decreases during the non-charging process. It shows that the SNR has the same periodic trend as the burden surface height. The charging mechanism changes the airflow distribution. The airflow strength has a great impact on the amplitude. The amplitude of the radar signal has metallurgy characteristics.

Fig. 13.

Trend comparison among charging signal, burden surface height and SNR during BF production process.

3.5. Comparison of Gas Flow Prediction and Measured Value

The echo signal attenuation can be expressed in terms of dust concentration, gas flow velocity, particle size and so on. The BF gas flow can be predicted in light of attenuation derivation. The parameters are as follows: the frequency f = 25 GHz, complex dielectric constant ${\epsilon }_m^{*}=4.0-1.325j$ , the particle radius r = 14.8 mm, the gas viscosity μ_g = 1.78 m²·s^–1, the gas concentration ε_g = 1.34 kg·m^–3, the solid concentration ε_s = 1950 kg·m^–3, the constant C_BF = 2.3 × 10^–5 and γ_BF = 1.07. Figure 14 gives the comparison of airflow velocity measured value and predicted value. The data is achieved during BF maintenance process. The strength of airflow has discrete increasing during BF maintenance process. The results show that the prediction is close agreement to the measured value.

Fig. 14.

Comparison of airflow volume measured value and predicted value under BF maintenance process.

4. Conclusions

A method of predicting gas flow strength in blast furnace has been proposed. The gas flow refers to the flow upon the burden surface. The method presents the possibility to get more information from the echo signal of burden surface. It can be employed to small particle radius dust and small transmit wavenumber, kr << 1. The relationship between gas flow strength and propagation constant has been developed. The formulas have been derived according to Rayleigh approximation and gas-solid fluidized bed theory. The method has been validated in a real BF. The results show that the predicted value has close agreement with the measured value. The amplitude of noise signal, generated by dust, increases 1 dB·g^–1 with increase of dust concentration. The SNR decreases 0.5 dB·cm^–1 with increase of dust thickness. It also shows that the radar signal attenuation has metallurgical characteristics. The charging mechanism changes the gas flow strength. The attenuation has the same periodic trend with the charging process.

References

• 1)   J.-I.  Park,  U.-H.  Baek,  K.-S.  Jang,  H.-S.  Oh and  J.-W.  Han: ISIJ Int., 51 (2011), 1617.
• 2)   S.  Natsui,  H.  Nogami,  S.  Ueda,  J.  Kano,  R.  Inoue and  T.  Ariyama: ISIJ Int., 51 (2011), 41.
• 3)   H.  Kurosawa,  S.  Matsuhashi,  S.  Natsui,  T.  Kon,  S.  Ueda,  R.  Inoue and  T.  Ariyama: ISIJ Int., 52 (2012), 1010.
• 4)   Q.-f.  Dong,  Y.-L.  Li,  J.-d.  Xu,  H.  Zhang and  M.-j.  Wang: IEEE Trans. Antenn. Propag., 61 (2013), 910.
• 5)   Y.  Kuga,  D.  Rice and  R. D.  West: IEEE Trans. Antenn. Propag., 44 (1996), 326.
• 6)   L.  Huilin,  D.  Gidaspow,  J.  Bouillard and  L.  Wentie: Chem. Eng. J., 95 (2003), 1.
• 7)   J.  Liu,  X.  Chen and  Z.  Zhang: Meas. Sci. Technol., 17 (2006), 135.
• 8)   X.  Chen,  J.  Wei,  D.  Xu,  Q.  Hou and  Z.  Bai: ISIJ Int., 52 (2012), 2048.
• 9)   J. W.  Bredow,  R.  Porco,  M. S.  Dawson,  C. L.  Betty,  S.  Self and  T.  Thordarson: IEEE Trans. Geosci. Remote, 33 (1995), 1071.