Exploring the Capability of Muon Scattering Tomography for Imaging the Components in the Blast Furnace

Abstract

Knowing the distribution of the materials in the blast furnace (BF) is believed to be of great interest for BF operation and process optimization. In this paper calibration samples (ferrous pellets and coke) and samples from LKAB’s experimental blast furnace (probe samples, excavation samples and core-drilling samples) were measured by the muon scattering tomography detector to explore the capability of using the muon scattering tomography to image the components in the blast furnace. The experimental results show that it is possible to use this technique to discriminate the ferrous pellets from the coke and it is also shown that the measured linear scattering densities (LSD) linearly correlate with the bulk densities of the measured materials. By applying the Stovall’s model a correlation among the LSD values, the bulk densities and the components of the materials in the probe samples and excavation samples was established. The theoretical analysis indicates that it is potential to use the present muon scattering tomography technique to image the components in various zones of the blast furnace.

1. Introduction

The blast furnace (BF) is a large industrial reactor commonly used for ironmaking. Due to the size of the BF and the harsh operation conditions (high temperature and high pressure) a direct online measurement into the BF, aiming at improving the process control and optimization, is very difficult. Nowadays the process control and optimization are implemented by different methods, which include measuring the furnace-wall temperature, measuring the off-gas pressure/temperature/composition, measuring the temperature and composition of the hot metal/slag as well as online sampling by probing into the BF, etc. The complete information regarding the materials distribution in the BF, however, is exclusively obtained by dissection and excavation of the BF after its extinguishing, which can only be carried out at an experimental BF or at an industrial BF when a re-lining work needs to be carried out (the lifespan for a normal BF is around 15–20 years).

In recent years several attempts have been made to investigate the capability of using cosmic ray muons to probe the inner structure of inaccessible dense objects including ancient pyramids,¹⁾ volcanos,^2,3) historical buildings⁴⁾ as well as industrial structures, where a long measurement time (from several days to several tens of days) is acceptable.^5,6)

When muons penetrate an object with certain density and thickness, two effects can occur:

(a) Absorption or attenuation. This happens to the low energy muons which lose all their energy and get absorbed by the matter.

(b) Coulomb scattering. This happens to the high energy muons which lose part of their energy and get scattered by the matter.

The majority of the applications, including the ones mentioned above, for online probing the inner structure of large objects are based on the muons absorption method (or muon radiography method), i.e., by counting the F (muons towards object)/B (muons towards sky) ratio in an accumulative length of time. The recent breakthrough in employing the muon absorption technique to online probe the inner structure of the BF is presented in 7–9). A summary of these works is given in Table 1. It is demonstrated in these works that the muon absorption technique can be applied to obtain the density-distribution information of the materials from certain zones in the BF. The obtained information can be further used to monitor the erosion of the refractory materials and to determine the shape and position of the cohesive zone, which are crucial for predicting the lifespan of the BF and for optimizing the BF operation.

Table 1. Available literature on online probing the inner structure of BF by employing muon absorption technique.

References	Calibration materials	Exposure time	Measured zones	Method for density calculation	Uncertainty
7,8)	–	10/20/45 days	i) Refractory brick for base plate and side wall;	Monte-Carlo simulation	3.0%a)
			ii) iron-rich part.
9)	Steel slab	–	i) Coke network;	Mathematical approximation	1.4–3.2%b)
			ii) cohesive zone.

a) For the measurement of iron-rich part within 45 days; b) for the measurement of calibration material (steel slab) within 1 hour.

As an extension of the muon radiography, the muon scattering tomography was proposed.¹⁰⁾ With muon scattering tomography the multiple Coulomb scattering of muons crossing a target volume is considered by evaluating the trajectory of each incoming and outgoing particle. The images showing the density difference of the materials within this target volume can be formed by the analysis of muon angular deflections caused by the multiple Coulomb scattering.^10,11,12)

Recently calibration samples prepared in the laboratory and excavation samples, probing samples and core-drilling samples procured from LKAB’s EBF (experimental blast furnace) have been scanned by muon scattering tomography detector. The results showed that muon scattering tomography can be used to precisely measure the properties of the materials from the EBF with an absolute precision being 7–10%.¹³⁾ The purpose of this work is to interpret the results from a metallurgical point of view and further to explore the capability of imaging the components in various parts of the blast furnace by muon scattering tomography. To the authors’ knowledge muon scattering tomographic probing of the BF has not been investigated, therefore it represents a new technique.

2. Experimental Parts

2.1. Preparation of the Samples

To test the effectiveness of muon scattering tomography on imaging the materials in the BF, several different sets of materials were prepared. These samples are classified as calibration samples and LKAB’s EBF samples.

2.1.1. The Calibration Samples

The calibration samples include coke and ferrous pellets (hematite pellets, magnetite pellets, wustite pellets and semi-metallic iron), the latter of which represent the ferrous materials with different reduction degrees typically found at different levels in the BF. The coke (15–30 mm) used as calibration sample is exactly the same type of that used in the LKAB’s EBF in the same campaign as for producing EBF samples. All the iron-ore-pellet (10–12.5 mm) samples used for producing the calibration samples originate from the same batch of LKAB’s standard product KPBO, acquired immediately after exiting the pelletizing plant. The hematite pellets corresponds to the initial state of the iron ore pellets when charged into the EBF. The other ferrous pellets, magnetite pellets, wustite pellets and semi-metallic iron, were produced in a laboratorial vertical furnace by reducing the hematite pellets to the degrees of 11%, 30% and 80%, respectively. The different reduction degrees were obtained by using different reduction programs, i.e., each with different reduction gas composition (CO/CO₂/H₂/N₂), time and temperature, ranging from 500°C for magnetite and up to 1100°C for semi-metallic iron. The duration time for each reduction program was in the range of 3.5–4.0 hours until the reduction equilibrium was reached in the corresponding reduction state. For each of the calibration samples 4.7 dm³ materials were collected or produced and encapsulated in a cylindrical plastic bucket of the same volume (with diameter 200 mm and height 160 mm).

2.1.2. The LKAB EBF Samples

The LKAB’s EBF (the detailed description of the EBF is available in the literature¹⁴⁾) samples were procured both during the operation phase of EBF campaign 31 (autumn 2014) and after quenching the EBF at the end of the campaign. The EBF samples include probe samples, excavation samples and core-drilling samples. An illustration of procuring these samples from different positions in the EBF is shown in Fig. 1. The detailed description for procuring these samples is given as follows.

Fig. 1.

Illustration of the EBF and procurement of EBF samples from different positions.

(1) Probe Samples

The probe samples were procured during the operation phase of the EBF by three probes (upper probe, lower probe and inclined probe) installed on the furnace. The upper and lower probes were each divided in three sections corresponding to the center, the mid-radius and the wall position in the blast furnace. The inclined probe was divided in two sections, corresponding to the lower-center region and the upper-wall region. The probing was carried out one to four times per day during stable operational conditions. For each of the probe samples 4.7 dm³ materials were collected and encapsulated in the same type of cylindrical plastic bucket used for the calibration samples.

(2) Excavation Samples

In the final stage of the campaign the entire burden was quenched with nitrogen. After the furnace was completely cooled down, the materials in the burden were excavated layer by layer. The excavation samples were acquired from eight different levels (pellet layers) in the EBF at three different positions corresponding to the center, the mid-radius and the wall region, which were consistent with the probing directions. For each of the excavation samples 4.7 dm³ materials were collected and encapsulated in the same type of cylindrical plastic bucket used for the probe and calibration samples.

(3) Core-drilling Samples

After the shaft was detached, the materials in the hearth of EBF were exposed. Steel tubes (diameter 130 mm and length 500 mm) were drilled into the hearth to procure the core-drilling samples. These samples were procured from two layers and several different positions (with respect to the positions of tuyeres and taphole) in the hearth of EBF. In total 14 core-drilling samples were obtained and encapsulated in the steel tubes.

2.2. Muon Scattering Tomography Detector and Detection of the Materials

The muon scattering tomography detector used in this study is located at INFN Legnaro Laboratories.¹³⁾ The detector consists of two detection planes (300×250 cm² each), which are arranged horizontally with a vertical gap of 160 cm. The working volume is located between the two planes and is around 11 m³. The upper detection plane is set to record the incoming cosmic ray muons while the lower detection plane is set to record the outgoing cosmic ray muons. During the experiment the samples were placed in the working volume, where muon multiple scattering in materials could be detected. The exposure time for the measurement of each sample is divided into subsets of 136 minutes. Several independent measurements were performed on each sample to improve the overall precision and to experimentally measure the statistical error of the measurement. The collected data from the muon detection system were used for LSD (Linear Scattering Density) reconstruction. Before the measurement of real samples, calibration of the detector was made by measuring the mock-up blocks (made of iron and plywood) with known LSD values. Countermeasures were taken to cut off the noise, to improve the algorithm convergence and to subtract the background. The detail information on experimental setup and data processing was described in 13).

2.3. Analysis of the Samples by Conventional Methods

To establish the correlations among the measured LSD values, the bulk densities of the materials and the components of the materials, the samples were analyzed by conventional methods. The bulk density is defined by Eq. (1).   

$${\rho }_{bd}=m/V$$(1)

Where ρ_bd is the bulk density; m is the mass of the sample in a specified volume; V is the control volume that the sample occupied.

(a) The calibration samples. Each of the calibration samples was weighed. The bulk densities of the calibration samples were calculated by Eq. (1), the control volume was 4.7 dm³.

(b) The probe samples and the excavation samples. Each of these samples was weighed. The bulk densities were calculated by Eq. (1), the control volume was also 4.7 dm³. Each sample was manually separated into pellets, coke and fines (< 3.5 mm); these separated fractions were also weighed. Selected pellets from the probe samples and excavation samples were analyzed by commercial Geopyc analyzer to get the apparent densities. During the analysis the pellet was placed in a cylinder filled with silica powder and the displacement volume was measured. The apparent density was taken as the weight of the pellet divided by the measured volume.

(c) The core drilling samples. After the muon scattering tomography detection, each of the core drilling tubes was opened by cutting a window, as shown in Fig. 2. The materials in the tube was divided into 5 sub-samples (10 cm in length each) and taken out one by one. Each sub-sample was weighed and screened into seven different fractions: > 16 mm, 10–16 mm, 5–10 mm, 2.8–5 mm, 1–2.8 mm, 0.5–1 mm and < 0.5 mm. The > 5 mm fractions were manually separated into the magnetic (mainly metal), coke, slag and aggregate; the 1–2.8 mm and 2.8–5 mm fractions were separated into the magnetic and non-magnetic by magnet.¹⁵⁾ The bulk density of the materials in each sub-sample was calculated according to Eq. (1).

Fig. 2.

A core drilling sample and the procedure for obtaining the sub-samples.

3. Results and Discussion

The raw muon data, which contain the information of muon track, i.e. its position and direction, can be used to compute the LSD (linear scattering density, nominated as λ) by applying the MLEM (Maximum Likelihood Expectation Maximization) algorithm.^13,16) The data from the physical analysis of the samples can be used to compute the bulk densities, ρ_bd, of the samples. According to 13) the LSD is directly proportional to the bulk density, as shown in Eq. (2).   

$$\lambda ={\rho }_{bd}R$$(2)

For materials composed by a single chemical element, the quantity R is approximately proportional to the atomic number Z of the element i (i.e., R_i ~ CZ_i, where C is a known constant). For materials composed by several elements, R is given by:   

$$R=\sum f_i\times R_i$$(3)

Where f_i is the mass fraction of element i, R_i is the value of R for element i.

3.1. The Calibration Samples

Each calibration sample consists of only one material (apart from its surrounding air) and is considered to be homogeneous in the spatial scale resolved by the muon scattering tomography detector. The measured LSD values for the calibration samples are shown in Fig. 3. It is evident that the ferrous materials have dramatically higher measured LSD values than the coke. The sharp contrast of the LSD values between the ferrous materials and the coke indicates that it is possible to discriminate the ferrous from the coke by the muon scattering tomography detector. This may imply that there is high potential for the muon scattering detector to discriminate the coke layer from the ferrous layer in the burden of the BF.

Fig. 3.

Contrast of the measured LSD values for the calibration samples (the ferrous and the non-ferrous) by muon scattering tomography detector.

The reduction degrees of the ferrous material increase as it descends in the BF. It would be of great interest to know the exact reduction degrees at different positions in the BF since this would allow a more precise control, and hence a more energy efficient production. However, it seems that it is difficult to discriminate the ferrous materials with different reduction degrees by using the present muon scattering tomography technique. Indeed the measured LSD values for the ferrous materials with different reduction degrees are quite similar if the measurement errors are also considered as shown in Fig. 3. These results could be explained by considering that (a) the shapes (shown in Fig. 4) as well as the bulk densities of the ferrous pellets (shown later in Fig. 6) only slightly get altered due to the subtle swelling of the pellets during the reduction; (b) the removal of the oxygen atoms from the ferrous pellet due to the reduction has little influence on the factor R in Eq. (3), as shown in 13).

Fig. 4.

Appearance of the laboratory-prepared ferrous materials with different reduction degrees.

Fig. 6.

Correlation between the measured LSD values by muon scattering tomography detector and the measured bulk densities of the calibration, probe and excavation samples.

3.2. The Probe Samples and the Excavation Samples

3.2.1. Correlation between the LSD Values and the Bulk Densities

The probe samples and the excavation samples represent the real materials which can be typically found in the EBF as well as in the industrial BF. Without prior knowledge of the components, the excavation samples were measured by the muon scattering tomography detector, assuming that: (a) each of the procured materials encapsulated in the plastic bucket is homogenous and therefore (b) each sample can only produce one signal by the muon scattering tomography detector. The results in terms of measured LSD values of these samples are shown in Fig. 5. It is seen that muon scattering tomography technique is capable to discriminate the materials procured from different positions of the EBF’s shaft, since different LSD values have been produced by the muon scattering tomography image reconstruction analysis. The results from the bulk-density measurement as well as the correlation with the measured LSD values are shown in Fig. 6. It is seen that the higher the bulk density of the materials is, the larger the LSD value would be. A positive linear correlation between the bulk densities and the LSD values with correlation coefficient being 0.9762 is evidenced. In this sense the LSD values marked in Fig. 5 represents the bulk densities of the materials and the LSD values can be used to derive the bulk densities of corresponding materials by applying the linear regression equation shown in Fig. 6.

Fig. 5.

Illustration of measured LSD values by muon scattering tomography detector for the probe samples and the excavation samples procured from different positions of LKAB’s EBF (The differences of the measured LSD values between the excavation samples and the probe samples procured at almost the same levels are attributed to the differences in components in different samples, as to be shown in Fig. 7).

3.2.2. Components of the Samples and their Correlation to the Bulk Densities

The manual screening and separation of the probe samples as well as the excavation samples show the characteristic of these materials. As shown in Fig. 7 the probe samples and the excavation samples consist of ferrous pellets, coke and small fractions of fines, the latter of which thus has little contribution to the bulk densities of the samples. It is also found that in general the samples collected from the wall positions contain larger fractions of the ferrous pellets and smaller fractions of coke while the ones from the center contain smaller fractions of the ferrous pellets and larger fractions of coke. These have been controlled by the charging system of the EBF and are also common in the industrial BFs to promote the gas flow and the reduction process.

Fig. 7.

Mass fractions of coke, ferrous pellets and fines found in the probe samples and excavation samples (Fig. 1. can be referred to for sampling positions).

From Figs. 5, 6 and 7 it is easy to deduce that the sample that contains a larger fraction of ferrous pellets will have a higher bulk density, which thus has a higher measured LSD value by the muon scattering tomography. Nevertheless, it is of great interest to resolve the materials in the samples by establishing the correlations among the components of the materials, the bulk densities and the LSD values. In this paper this is realized by applying the recognized linear packing density model presented by Stovall.¹⁷⁾ By (a) assuming that the electrostatic, Van der Waals and all other cohesive or repulsive interactions among the packing components are negligible and (b) considering that the packing system has at least one component of identical particle size being fully packed, the packing density of a given system, η_i, can be expressed as Eq. (4).   

$${\eta }_i={\alpha }_i+\left(1-{\alpha }_i\right){\sum }_{j=1}^{i-1}g\left(i,j\right){\mathrm{\Phi }}_j+{\sum }_{j=i+1}^{n}f\left(i,j\right){\mathrm{\Phi }}_j$$(4)

Where α_i is the packing density of the component in size group i; Φ_j is the partial volume of the component in size group j; g(i, j) and f(i, j) are the interaction factors which respectively take into account the wall effect and the loosing effect due to the packing of components in size group i and that in size group j. According to Stovall, g(i, j) is a function of the effective radii (r_i and r_j) of components in size group i and j, f(i, j) is a function of the effective radii of components in size group i and j as well as α_i and α_j; the mathematical expressions of the two functions, g(i, j) and f(i, j), are tentatively given as Eqs. (5) and (6).   

$$g\left(i,j\right)=1-\frac{r_i}{r_j}$$(5)

  

$$\begin{array}{l}f\left(i,j\right)=\\ 1-\frac{{\alpha }_j}{\raisebox{1ex}{${\alpha }_i$}\!\left/ \!\raisebox{-1ex}{$r_j$}\right.\left[\left(1-{\mu }^3\right)\left(1-{\alpha }_i\right)-3{\mu }^3\right]\left(r_j-r_i\right)+{\alpha }_j\left(1-{\mu }^3\right)}\\ \left(1-\frac{{\mu }^3{r}_j^3}{r_i^3}\right),\mathrm{\hspace{0.17em} }{r}_i\ge \mu r_j\mathrm{\hspace{0.17em} }and\mathrm{\hspace{0.17em} }\mu \approx 0.2\end{array}$$(6)

Considering a packing mixture with components of n group size, there will be n bulk density values and the true packing density, η, is the minimum value among them, as expressed by Eq. (7).   

$$\eta =\text{the}\mathrm{\hspace{0.17em} }\text{minimum}\mathrm{\hspace{0.17em} }\text{of}\{{\eta }_1,{\eta }_2,{\eta }_3,\dots ,{\eta }_n\}$$(7)

Obvious the packing conditions of the pellets and the coke in the plastic buckets and in the BF satisfy the assumptions shown earlier; therefore it is believed that the Stovall’s model obeys. Assume that (a) the probe samples and the excavation samples consist of only ferrous pellets and coke (as shown in Fig. 7 the fractions of fines are small) and (b) the ferrous pellets have effective radius r_pellet and the coke particles have effective radius r_coke, then the bulk density, ρ_bd, can be expressed in the term of packing density, η, as shown in Eq. (8).   

$$\eta =\frac{\left(m_{pellet}/{\rho }_{ad.pellet}+m_{coke}/{\rho }_{ad.coke}\right)}{\left(m_{pellet}+m_{coke}\right)/{\rho }_{bd}}$$(8)

Where m_pellet and m_coke are the mass fractions of the pellets and coke, respectively; ρ_ad.pellet and ρ_ad.coke are the apparent densities of the pellets and the coke, respectively.

Define X as the mass ratio of the pellets to the coke in the samples:   

$$X=\frac{m_{pellet}}{m_{coke}}$$(9)

Considering Eq. (10), the bulk density, ρ_bd, could be transformed as Eq. (11).   

$$m_{pellet}+m_{coke}\approx 1$$(10)

  

$${\rho }_{bd}=\frac{\eta }{\frac{X/\left(1+X\right)}{{\rho }_{ad.pellet}}+\frac{1/\left(1+X\right)}{{\rho }_{ad.coke}}}$$(11)

Inset the experimentally determined parameters (as listed in the Fig. 8 caption) to Eq. (11), a figure which illustrates the correlation between the bulk density, ρ_bd, and the mass ratio of pellets to coke is drawn in Fig. 8, in which the experimental data are also plotted. It is seen that the experimental data correlate fairly well with the calculation, which may justify the application of Stovall’s model in the present case. The scattering of the experimental data is attributed to the changes in particle sizes, shapes and apparent densities of the materials during the reduction and descending of the burden materials in the EBF and the relevant parameters due to these changes were otherwise considered as constants in the model calculation.

Fig. 8.

Correlation between the bulk density and the mass ratio of pellet to coke in the probe samples and the excavation samples (The determined parameters used for Stovall’s model¹⁷⁾ calculation are shown as follows: r_coke=11.8 mm; r_pellet=5.7 mm; ρ_ad.coke=1.00 g/cm³; ρ_ad.pellet=3.20 g/cm³; ρ_bd.coke=0.52 g/cm³; ρ_bd.pellet=2.00 g/cm³).

Figure 8 quantitatively establishes a correlation between the bulk density and the mass ratio of pellets to coke in the probe and excavation samples. By considering the results shown in Fig. 6, a correlation between the LSD value and the pellets to coke mass ratio in the probe and excavation samples can be indirectly established as well. This implies that the muon scattering tomography detector may have the capability to resolve the components in the burden of BF. It should be noted that Stovall’s model is not decisive, as, for example, CFD modeling may provide more comprehensive results in resolving different materials including pellets, coke as well as fines, nut coke and limestone, which may exist in the BF burden.

3.3. The Core-drilling Samples

The core-drilling samples represent the real materials in the hearth of the EBF as well as in the industrial BF. The results in terms of measured LSD values of these samples by muon scattering tomography are shown in Fig. 9. It is evident that the divided sub-sample fractions were ‘seen’ as different by the muon scattering tomography, as the results show clear contrast in LSD values. Similar as the probe samples and the excavation samples, the bulk densities of the sub-samples were measured after the muon scattering tomography detection. The bulk-density results could thereafter be compared with the LSD values, as illustrated in Fig. 10. It is also seen that the sub-sample which has higher bulk density could generate a larger LSD value by the muon scattering tomography detector and there is likely a positive linear correlation between the bulk densities and the LSD values. However, it seems that the experimental results are quite scattered. The reason for this could be due to the complexity of the materials in the drilling cores.

Fig. 9.

Illustration of measured LSD values by muon scattering tomography detector for the core-drilling samples procured from the hearth of LKAB’s EBF.

Fig. 10.

Correlation between the measured LSD values by muon scattering tomography detector and the measured bulk densities of the core-drilling samples.

The complexity of the materials in the drilling cores is confirmed by the physical analyses, which show that the materials in the drilling cores are a mixture of coke, slag and the magnetic, etc. in big variations of particle sizes and with different sizes of voids filled in between these materials. In general it is found that: (a) the materials become coarser further down in the EBF hearth; (b) there are more fines generated in the drilling cores, which are procured beneath the three tuyeres. One may expect the continuity of the measured bulk densities and as well the LSD values for the first-layer core-drilling samples and the second-layer core-drilling samples. However, this cannot be confirmed by the measured results. This may be due to several reasons including: (a) the compression of the materials as well as the movement of the fine materials from the top to the bottom in the steel tube during the drilling; (b) the movement of fine materials in the drilling tube during the long distance transportation (from LKAB, Sweden to INFN, Italy).

One may also expect to establish the correlation between measured LSD values and the components of the materials in the sample by, for example, applying the Stovall’s model. However, this is believed to be very difficult at the moment, as the materials in the core-drilling samples are so complex and the parameters that would be needed in the Stovall’s model are not available. A computerized modelling in line with applying the muon scattering tomography detection as well as a comprehensive understanding of the materials distribution in the BF’s hearth may provide the capability in resolving the materials in such a complex case, and this could be interest of the study in the future.

4. Further Exploitation of the Results

4.1. Correlation among the Measured LSD, Bulk Density and Components of the Materials

Figures 6 and 10 confirm that the LSD values have positive correlations with the bulk densities. However, the slopes of the regressed straight lines don’t correspond to the R values in Eq. (2). This is due to that R is always changing with the components in the materials. In the provided samples C and Fe are the prevailing elements in the samples. Considering the R values for Fe and C (7.2 rad²m²/ton vs. 2.3 rad²m²/ton) as well as the fact that the materials containing larger fractions of Fe (or ferrous) will in general have higher bulk density, it can be therefore deduced from Eq. (2) that the LSD values will be higher when the bulk density is higher. The regressed straight lines in Figs. 6 and 10 could therefore be used as the first approximation to correlate the measured LSD values with the bulk densities for the samples procured from the EBF shaft and hearth. By applying the mathematical models or the computerized simulation it is likely that the corrections among LSD, bulk density and components of the materials could be established.

4.2. Possibilities and Challenges

In this paper it is demonstrated that the muon scattering tomography technique is capable to discriminate the coke, the ferrous pellets, the materials in the EBF’s shaft and hearth. It is therefore expected that the muon scattering tomography detector can be used to image the burden layers in the shaft and the components of the materials in the hearth. Besides these, it is also expected that the detector can be used to image the cohesive zone (its position, thickness and shape), the levels of the molten slag and hot metal, the shape of the raceway, the corrosion of the refractory linings, the scaffolding, etc., as the bulk densities of these materials are known to be different from their surrounding materials and thus the materials in these zones could generate different LSD values by the muon scattering tomography system. Imaging the components of the materials in these areas is of great interest for the BF operation.

In this study the measurement was obtained with isolated samples at an extended exposure time. Whereas, in the operational BF the materials are dynamic; although some phenomena (e.g. the cohesive zone position) can persist quite a long time at stable operation conditions, some other phenomena (e.g. descending of the burden) may change with time very fast and this would require the detector to have a short response time. Further, the industrial environment is very harsh and the materials in the BF are surrounded by an enormous amount of additional matter making the discrimination of different components by the detector much more complex. At the moment requesting a reasonable response time and installing the detector with suitable size around the operational BF to monitor different phenomena are still the challenges.

5. Conclusions

In this paper the capability of using a non-invasive detection technique, known as muon scattering tomography, to image the components in the blast furnace is explored. The exploration is implemented by using the muon scattering tomography detection system to (a) scan a set of materials, which includes the calibration samples (ferrous pellets and coke) and EBF samples (probe samples and excavation samples from the EBF shaft, core-drilling samples from the EBF hearth); and thus to (b) produce a group of data, in terms of linear scattering density (LSD), of the measured samples. Imaging the components of the materials in the BF can be potentially realized by establishing the correlations among the measured LSD values, the bulk densities and the components of the materials in various zones of BF. According to the experimental results and the theoretical analysis, the conclusions can be drawn as follows:

(1) The sharp contrast in the measured LSD values between the ferrous materials and the coke indicates that it is potential to use the muon scattering tomography to discriminate the coke layer from the ferrous layer in the BF;

(2) The measured LSD values have positive linear correlation with the bulk densities of the measured materials, i.e., a higher measured LSD value of the material indicates a higher bulk density of the same materials;

(3) By applying Stovall’s model, a correlation among the measured LSD values, the bulk densities and the components of the materials in the probe samples and excavation samples is established;

(4) Muon scattering tomography technique can be potentially used to image the components of the materials in different zones in the BF, where the materials can form a distinguishable contrast in the measured LSD values.

Acknowledgement

This work was funded by the European Commission via the Research Fund for Coal and Steel for the Mu-Blast project (RFSR-CT-2014-00027).

References

• 1)   R.  Alfaro,  E.  Belmont,  V.  Grabski,  L.  Manzanilla,  A.  Martinez-Davalos,  A.  Menchaca-Rocha,  M.  Moreno and  A.  Sandoval: Proc. 30th Int. Cosmic Ray Conf., ed. by R. Caballero, et al., Universidad Nacional Autonoma De Mexico, Mexico, (2008), 1265.
• 2)   H. K. M.  Tanaka,  T.  Uchida,  M.  Tanaka,  M.  Takeo,  J.  Oikawa,  T.  Ohminato,  Y.  Aoki,  E.  Koyama and  H.  Tsuji: Geophys. Res. Lett., 36 (2009), L17302.
• 3)   H. K. M.  Tanaka,  T.  Kusagaya and  H.  Shinohara: Nat. Commun., 5 (2014), 3381.
• 4)   A.  Zenoni: Proc. 4th Int. Conf. on Advancements in Nuclear Instrumentation Measurement Methods and their Applications (ANIMMA), IEEE, New York, NY, (2015), 1.
• 5)   W. B.  Gilboy,  P. M.  Jenneson,  S. J. R.  Simons,  S. J.  Stanley and  D.  Rhodes: Nucl. Instrum. Methods Phys. Res. Sect. A, 580 (2007), 785.
• 6)   W. B.  Gilboy,  P. M.  Jenneson,  S. J. R.  Simons,  S. J.  Stanley and  D.  Rhodes: Nucl. Instrum. Methods Phys. Res. Sect. B, 263 (2007), 317.
• 7)   A.  Shinotake,  S.  Matsuzaki,  K.  Kunitomo,  M.  Naito,  M.  Hashimoto,  A.  Hatanaka,  T.  Nagane,  K.  Nagamine and  H.  Tanaka: Tetsu-to-Hagané, 95 (2009), 665.
• 8)   K.  Nagamine,  H. K. M.  Tanaka,  S. N.  Nakamura,  K.  Ishida,  M.  Hashimoto,  A.  Shinotake,  M.  Naito and  A.  Hatanaka: Proc. Jpn. Acad. Ser. B, 81 (2005), 257.
• 9)   J.  Sauerwald,  A.  Franzen,  A.  Fischer,  T.  Buhles and  J.  Janz: Proc. 6th Int. Cong. on the Science and Technology of Ironmaking (ICSTI), Curran, Norwich, (2012), 1371.
• 10)   K. N.  Borozdin,  G. E.  Hogan,  C.  Morris,  W. C.  Priedhorsky,  A.  Saunders,  L. J.  Schultz and  M. E.  Teasdale: Nature, 422 (2003), 277.
• 11)   K.  Borozdin,  T.  Asaki,  R.  Chartrand,  M.  Galassi,  A.  Greene,  N.  Hengartner,  G.  Hogan,  A.  Klimenko,  C.  Morris,  W.  Priedhorsky,  A.  Saunders,  R.  Schirato,  L.  Schultz and  M.  Sottile: Proc. Institute of Nuclear Materials Management (INMM) 46th Annual Meeting, Institute of Nuclear Materials Management, Illinois, (2005), 1.
• 12)   J.  Burns,  S.  Quillin,  M.  Stapleton,  C.  Steer and  S.  Snow: J. Instrum., 10 (2015), 10041.
• 13)   E.  Åström,  G.  Bonomi,  I.  Calliari,  P.  Calvini,  P.  Checchia,  A.  Donzella,  E.  Faraci,  F.  Forsberg,  F.  Gonella,  X.  Hu,  J.  Klinger,  L.  Sundqvist Ökvist,  D.  Pagano,  A.  Rigoni,  E.  Ramous,  M.  Urbani,  S.  Vanini,  A.  Zenoni and  G.  Zumerle: J. Instrum., 11 (2016), 07010.
• 14)   L.  Hooey,  J.  Sterneland and  M.  Hallin: 60th Ironmaking Conf. Proc., Iron & Steel Society, London, (2001), 197.
• 15)   L.  Sundqvist Ökvist,  N.  Eklund and  J.-O.  Wikström: The Iron & Steel Technology Conf. and Expositon (AISTech 2010) Proc., AIST, Warredale, PA, (2010), 701.
• 16)   M.  Benettoni,  G.  Bettella,  G.  Bonomi,  G.  Calvagno,  P.  Calvini,  P.  Checchia,  G.  Cortelazzo,  L.  Cossutta,  A.  Donzella,  M.  Furlan,  F.  Gonella,  M.  Pegoraro,  A.  Rigoni Garola,  P.  Ronchese,  S.  Squarcia,  M.  Subieta,  S.  Vanini,  G.  Viesti,  P.  Zanuttigh,  A.  Zenoni and  G.  Zumerle: J. Instrum., 8 (2013), 12007.
• 17)   T.  Stovall,  F.  de Larrard and  M.  Buil: Powder Technol., 48 (1986), 1.