BLEMAB European project: muon imaging technique applied to blast furnaces

Like electrons, muons are charged elementary particles belonging to the lepton family, a type of particle that can interact through all types of forces except the hadronic force. Muons are continuously produced in the upper layer of the Earth’s atmosphere due to the collisions of very high-energy cosmic rays with atmospheric gas. Due to their high energies and their large mass, about 200 times that of the electron, atmospheric muons can reach the earth’s surface and penetrate up to hundreds of meters of rock, depending on their energy.

The basic idea of muon radiography is to exploit the natural atmospheric muon flux to perform radiographic studies of large massive structures, thus reproducing in a very similar way what is usually done with x-rays in the usual radiographs. The most direct result of a muon radiograph is the measurement of the directional transmission of the muons, that is, the fraction of incoming muons that is able to cross the thickness of the material encountered along one direction. Transmission can be experimentally estimated by comparing the muon flux measured downstream of the target under examination with the flux measured upstream in the same direction. The interpretation of the measured transmission requires a comparison with simulations that takes into account the geometry of the volume under examination, the characteristics of the muon flux and the other information available on the material structures placed in the acceptance of the experimental apparatus used for the measurement. Many simulations, algebraic formulas and experimental measurements can be found in the literature describing muon flux at ground level, but often these are not sufficient to fully describe the energy and angular ranges required for muon radiography applications. Atmospheric muon flux has important dependencies both on energy and direction. The INFN-UNIFI team in Florence measured muon energy and angular spectra at ground level in the momentum range from 0.1 to 130 GeV/$c$ and in the zenith angle range from $0^{\circ}$ up to 80^∘ using the ADAMO detector [2], a magnetic spectrometer made with a permanent magnet and a microstrip silicon tracker, with a Maximum Detectable Rigidity of about 250 GeV/$c$. Based on this measurement, a preliminary version of a muon software generator was implemented and successfully used to compare different muon radiographs with dedicated simulations [3, 4, 5, 6, 7]. ADAMO results have been also used recently for the implementation of a realistic atmospheric muon generator [8] integrated with the GEANT4 software package [9].

The comparison of the measured two-dimensional angular distribution of muon transmission with realistic software simulations allows reconstructing the two-dimensional average angular density distribution of the target, as resulting from the measuring position. Combining multiple measurements performed from different viewpoints, information on the three-dimensional density distribution can be derived.

Detailed reviews about muon imaging and the relative instrumentation can be found in [10, 11].

The main aim of the BLEMAB project is to test the capability of muon transmission radiography technique in internal volume imaging of blast furnaces and to establish a non-invasive investigation methodology for online monitoring of internal density variations of such structures.

An important test of the performance of this imaging technique concerns the ability to measure the shape, position and thickness of the so-called "cohesive zone", the place in the blast furnace where the iron-containing raw material begins to melt. The knowledge of the processes that take place inside the blast furnaces and the maximization of the extension of the cohesive zone are crucial points for improving the productivity of these plants and therefore of great interest for steel companies. Until now, this type of study has been based only on direct measurements, such as the tracking of radioisotopes injected from the top of the blast furnace or the use of vertical probes with conductive cables connected to conventional instrumentation, such as vertical multi-point probes (MPVP), which provide essential information about the state of the internal volume [12, 13], or other very expensive and invasive [14] techniques. Unlike these techniques, muon imaging is a safe non-invasive radiographic methodology based on a completely natural and freely usable radiation, which therefore does not even cause any radiological issue.

Muon imaging has been proposed in the past for blast furnace imaging, as described in [15, 16]. These works demonstrated the feasibility of measuring the density distribution of the lower part of a blast furnace using the muon radiography technique. However, a direct measurement of the height of the casting surface had not yet been hypothesized. A simple muon detection system has already been employed by AM to verify the possibility of detecting density variations inside a BF’s inner volume by studying the corresponding variation in the flux of atmospheric muons crossing it. The study of the average density distribution inside the BF would add important hints about the position and extension of the cohesive zone as a function of the BF’s running parameters. According to the BF’s experts the measurement of average density variations in some regions of the BF’s inner volume in a timescale of some days after changing the running parameters would be extremely useful for improving the BF’s steel production process.

This is the main objective of the BLEMAB proposal, for which a muon tracking system of large acceptance and appropriate angular resolution is being developed. This equipment will be installed at the blast furnaces hosted in the ArcelorMittal (AM) site in Bremen (Germany) for many months, in order to allow long-term data collection and the development of a monitoring methodology based on the online analysis of muon trajectories and on the comparison with realistic simulations. Muographic measurements will also be compared with measurements obtained using an enhanced multipoint vertical probe (E-MPVP) and standard blast furnace models for validation.

Section 2 introduces the software simulation framework used for a preliminary simulation and a new framework based on the GEANT4 package, already available but still under development for further improvements, that will be used to realize the full simulations using the most realistic models of the AM’s blast furnaces.

Some details of the prototype detector under development for the BLEMAB project are presented in section 3, together with preliminary results obtained from a test of a small prototype tracking plane implementing the BLEMAB concepts.

The main evaluations concerning the feasibility of the experiment have been finalized thanks to a simulation tool already available to the team and successfully used for previous studies related to measurements carried out at the Temperino mine in Campiglia Marittima, near Livorno (Italy) [4, 5, 6], at the Bourbon Gallery in Naples [3] (Italy) and at river embankments near Florence and Pistoia [7] (Italy). In general all simulations for muographic applications require the knowledge of the target material thickness $l(\theta,\phi)$ and its average density $\rho(\theta,\phi)$ as seen from the hypothetical detector’s installation position along all directions in its field of view, defined by the zenith ($\theta$) and azimuth ($\phi$) angles. The product of these two quantities defines the target material opacity along each direction, $X(\theta,\phi)=l(\theta,\phi)\times\rho(\theta,\phi)$, which in turn determines the minimum momentum $p_{min}(\theta,\phi)$ that a muon coming from that direction must have to cross both the target material and the detector tracking planes, to be detected. In this simulation $p_{min}(\theta,\phi)$ is determined by using tabulated muon range values available in the literature. The corresponding muon transmission $t(\theta,\phi)$ is finally calculated as the ratio between the integral of the muon momentum spectrum $\Phi(p,\theta,\phi)$ (for which analytical expressions have been fitted to the ADAMO data) in the momentum range beyond $p_{min}(\theta,\phi)$ (simulating a measurement performed downstream the target volume) and the integral of the same spectrum in the whole momentum range down to the minimum value allowing muons to cross the detector modules (simulating a free-sky reference measurement).

For the BLEMAB application a 45^∘ tilt of the detector pointing direction with respect to the vertical direction is assumed for pointing the blast furnace’s cohesive zone; in this configuration a total muon rate of approximately 40 Hz is expected with a single BLEMAB detector.

The sketch on the left in figure 1 shows the geometry considered for this simulation, implementing a simplified model of the blast furnace located in the Bremen site of ArcelorMittal.The two dimensional angular distribution of muon events expected downstream the blast furnace is show at center, considering 10 h muon events approximately. This distribution is based on a real free sky measurement performed with the MIMA detector and on the simulated muon transmission distribution, reported in the plot on the right side.

In this simulation a 2^∘ bin size is chosen both for elevation and zenith angles. The presence of the blast furnace in front of the muon detection system can be identified as an elongated angular region in the central part of the distributions, where the number of detected muons is significantly suppressed.

In each bin two to three hundred muon tracks are expected after a data taking of about 24 h in the angular region where the cohesive zone is located, allowing a measurement of muon distribution with statistical fluctuations below 10 % with such an angular resolution.

Yet more precise information can be obtained integrating muon events over larger angular regions. Statistical fluctuations at percent level in half day are expected when observing a $10^{\circ}\,\times\,10^{\circ}$ region in the direction of the cohesive zone.

Figure 2 shows the main results of this simulation. In particular, the expected two dimensional angular distribution of the average density has been evaluated for the whole blast furnace structure and after subtracting the contribution due to the toroidal structure surrounding the furnace and to the outer metal shell. The angular average density distribution implemented in the simulation is shown on the right side plot for comparison. This plots show that some details of the internal material distribution can be argued already after a 10 h data taking, but is must be observed that in reality the material layers inside a blast furnace are constantly falling down to the melting zone, taking approximately 8 h to complete a full cycle. Therefore an average over time of the average material density will be measured in this time interval. In a more realistic scheme, the difference in material density in the lower and upper parts of the blast furnace’s volume is expected to be larger (0.7 g cm^-3 versus 1.5 g cm^-3) and the interface between the two regions should be yet more evident.

A more detailed simulation has been developed, which implements an accurate representation of the geometry and material composition of the blast furnace where the measurement will be carried out. The new simulation is based on the GEANT4 software package and on a recently developed atmospheric muon software generator [8]. The basic structure of this simulation was derived from the one created previously within the European MU-BLAST project [15].

Figure 3 shows a sketch of the geometry considered for the full simulation of the BLEMAB experimental configuration.

In this simulation 24 identical detectors, 80 cm $\times$ 80 cm $\times$ 80 cm are placed all around the blast furnace, 8 m far from the wall and some meters below the tuyere, the large toroidal pipe surrounding the structure, used to blow hot air inside the blast furnace. The multiple detection systems allow increasing the ratio between the number of events entering the detector’s acceptance and the total number of simulated muon events, which can be generated from a cylindrical surface surrounding the blast furnace.

Results from this complete simulation, which takes into account a detailed description of physics interactions of muons inside materials, will be used for more careful evaluations and for comparisons with the experimental measured distributions.

A first partial prototype of the tracking plane has been produced to estimate the detection efficiency and spatial resolution that is possible to obtain in the BLEMAB detector configuration. This prototype is limited to 5 scintillator bars, assembled as shown in figure 5 (left).

For this test we have used a LeCroy HDO6054 digital oscilloscope as DAQ system. The scintillator assembly has been enclosed in a protective aluminum box, used also as a darkening cover. Each bar is read by two 4 mm $\times$ 4 mm SiPM optical sensors whose signals are summed together. A study of the dependence of detection efficiency on the impact coordinate of muons along the bars (80 cm length) has been carried out by using an external trigger system made of two auxiliary plastic scintillators read by means of standard PMTs. This scintillators have a small surface and are suitable to select a small region of the system under test. The complete setup can be seen in figure 5 (center). The first full BLEMAB tracking module under construction is shown at the right side in the same figure.

Data acquisition with the small prototype was repeated for seven different positions of the trigger system, with steps of 10 cm, in order to cover the whole prototype’s length. For each position a total of approximately 1000 muon events have been collected.

Figure 6 (left) shows the histogram obtained for the total signal released by cosmic ray muons and integrated with the DAQ oscilloscope. The histogram is peaked for a value of almost 5 V$\cdot$ns and has approximately the shape of a Landau distribution. In this measurement for 2 events of a total of 864 a null signal was registered, resulting in an efficiency of 99.8%. Figure 6 (right) shows the variation of the detection efficiency for the different impact point positions. Even if some tendency could be argued, the values are all compatible with a constant value of 99.7% within the statistical uncertainties. All measured values are beyond 99.6%.