Evaluation of Mixing Effects and Particle Breakage on a Cross Flow Turbine with DEM

For storage or buffering, mostly bunkers are used. Particle size segregation effects at the bunker outflow lead to fluctuations in particle size distribution. In most applications and for their following processes, an evenly distributed bunker discharge is desired in terms of particle size distribution. This especially applies to blast furnace sinter bunkers as a constant particle size distribution for blast furnace operation is needed to ensure a sufficient gas flow.

Significant segregation effects during bunker filling mainly occur due to the following two effects. Due to vibrations during the transport by conveyor belt, small particles accumulate at the bottom and large particles at the top of the bulk material heap in the conveyor belt. Thus, the large particles have a different trajectory than the small particles when discharged. Depending on the belt incline and speed, at discharge the large particles could have a higher velocity than the small particles due to the greater distance of the large particles to the center of the discharge pulley. This could lead to an accumulation of large particles in the conveying direction and an accumulation of small particles against the conveying direction of the discharging conveyor belt in the bunker [1].

The second segregation effect is noticed every time a bulk material pile is formed. As larger particles have a greater forward momentum than smaller particles, the coarse material continues moving down the side of the pile more than the fine material. The material that tumbles down the slope of a pile is called overrun. Larger particles tend to roll down the entire length of the slope and fine particles tend to settle into the side of the pile. This effect of overrun causes the outer and bottom zones of the pile to consist of coarser material, while the inner and upper zones of the pile consist of more fine material [1].

This study was conducted within the project MinSiDeg, which received funding from the European Union and was performed by the University of Leoben and several industry partners from Austria and Germany. One objective in MinSiDeg was to develop or further develop innovative conveying and storage equipment in order to reduce degradation and segregation effects during conveying and storage processes of blast furnace sinter. One of these devices is the solid state material turbine, which was originally developed and patented in [2] and was described in [3‐5] for energy recovery. The solid state material driven turbine was further developed and optimized to reduce segregation effects in [6, 7] and partly presented in [8]. This contribution is a concise version of [9, 10] with an additional focus on particle breakage. The breakage model is described in detail in [8, 11].

This study was performed on a square-shaped bunker for blast furnace sinter at a steel manufacturer with filling levels of up to 600 t. The bunker is filled by a conveyor belt with a mass flow of 300 t/h and an almost constant particle size distribution (Fig. 1). Low filling levels lead to high drop heights, which is assumed to be one of the main reasons for particle breakage in this case. As described in the introduction, the outer and bottom zones of the pile consist of large particles, which leads to an accumulation of coarse material in the outer zones in the entire bunker.

This filling process was simulated by means of DEM with the simulation software EDEM (Fig. 2). Parameters for simulations were determined in angle of repose and slip tests in [12], rebound tests in [13] and further investigations regarding contact processes in [14]. In the simulations the bunker was filled with 350 t with the original particle size distribution listed in Table 1. The small fractions 6, 10, and 16 mm were combined in the simulation (52.25% in total), and all particles were up-scaled by a factor of 3 for computational efficiency.

The bunker was filled with 350 t and then discharged. During the discharging process, the particle size distribution at the outflow was determined in the simulation. In Fig. 3 the mass fraction for each particle size in connection to the bunker filling level is shown. Small particles (6 + 10 + 16 mm) contribute with approximately 80% to the total mass flow at the beginning and decrease to around 10% at the end of the discharging process. As can be calculated from the data in Table 1, the optimum would be a constant flow of 52.25% for the small particles during the whole discharging process. In contrast to this, the mass fraction of large particles is relatively small at the beginning of the discharging process and starts to increase at approximately 50% of the filling level. This effect can be explained by the core flow effect, whereby a core flow forms during discharging. This leads to an early discharge of material in the inner zones of the bunker, which are the smaller particles. The larger particles are accumulated at the outer zones.

Various types of installations to reduce segregation effects during bunker filling were tested in simulations including cascade chutes and various types of solid state material driven turbines at different rotation speeds. The discharge from a conveyor belt was simulated with and without these installations. To determine mixing effects, the top view of the bulk material pile with only one particle size faded in was evaluated (see [7] for details). The best improvement in terms of mixing were achieved with a cross flow turbine at low rotation speeds (Fig. 4).

The cross flow turbine consists of 10 segmented blades, which form an ideal curvature for material flow. In this case the turbine diameter is 1.6 m. The center of the cross flow turbine is hollow, which allows the material to flow through the turbine. The material flow through the turbine is shown in Fig. 11, which was simulated with the particle size distribution listed in Table 1. The color scale represents the translational velocity of the sinter particles.

Simulations with different rotation speeds were performed and an optimum in terms of mixing was found at 5 rpm. Depending on the mass flow, this low rotation speed can be achieved by regenerative braking or by an electric drive at low mass flows.

The cross flow turbine was developed and optimized for mixing. Energy recovery is a secondary application. The power output of the cross flow turbine was investigated in [9, 10], whereby an efficiency of η = 39.5% at 35 rpm is determined. Other types of solid state material turbines [3‐5] are optimized for energy recovery to achieve a higher power output.

The bunker filling process was simulated again with the cross flow turbine (Fig. 5). The distance between conveyor belt discharge and turbine center was 2556 mm. The 100 mm particles were represented by 50 mm particles, otherwise they would not pass the turbine due to the upscaling by a factor of 3. A significantly more evenly distributed particle size distribution is noticed at bunker discharge when the bunker is filled with the cross flow turbine (Fig. 6). An almost constant particle size distribution till a filling level of 100t is noticed, which is a great improvement compared to Fig. 3.

A recently developed breakage model for DEM was used [8, 11, 15‐17] using an API (Application Programming Interface) in the simulation software ThreeParticle from Becker3D. The novel breakage model is based on a probabilistic particle replacement with voronoi tessellated fragments. Depending on the stress on the individual particle, the initial particle is probabilistically replaced by different breakage patterns (Fig. 7). Depending on the resulting fragment sizes, the breakage pattern is generated by means of the voronoi algorithm [18‐20] with a certain amount of seeds. In contrast to [21] and other particle replacement models [22‐26], where the particle is replaced by a number of spheres, in this model the initial particle is replaced by an exact copy of the initial particle, which has been previously tessellated. This ensures a mass and volume consistency. Polyhedral particles of any shape can be used. As the model is based on probabilities, a high amount of particles is necessary, but then the model delivers high accuracy in terms of fragment size distribution.

In Fig. 8 the breakage model is demonstrated with the impact of sinter particles on a steel plate. As the model is based on probabilities, a number of particles are necessary. The impact of 25 identical initial particles (40 mm in diameter, 47 g) is simulated at different velocities. Figure 8a shows the particles before the impact, moving towards the steel plate. Figure 8b–d shows the rebound of particles after the impact with different velocities, moving away from the steel plate. The color scale represents the particle or fragment mass (0 to 47 g). Shortly after the impact process is completed, the initial particles are replaced by one of the breakage patterns from Fig. 7, following experimentally determined probabilities [11]. The initial 40 mm particles appear red, the 16–25 mm fragments appear green and all smaller fragments (10–16, 6–10, < 6 mm) appear blue in this case.

The probability for each breakage pattern is dependent on the mechanical stress and is experimentally determined by single particle impact tests in [27‐29]. The impact tests were performed with a specially developed automated single particle impact tester (Fig. 9), which is described in detail in [27] and allows rapid analysis of breakage behavior of bulk materials. The model was verified and validated with a trial of shatter tests and trials with two different transfer systems in [8, 15] using different batches of sinter from two different manufacturers.

Further breakage of fragments can also be simulated, which allows a simulation of long and complex conveying systems with multiple breakage (Fig. 10). How often a particle and its resulting fragments can be further broken is described with the breakage level L in Fig. 10. For example, a breakage level of L = 2 means that the initial particle for this breakage process is a fragment of a previous breakage process at L = 1.

It is considered that the cross flow turbine could also cause damaging effects on particles. On the one hand, the turbine splits and reduces the free fall height by its diameter, but on the other hand, it causes additional impacts against the turbine and among the particles. To quantify particle breakage caused by the cross flow turbine, a comparison of the drop with and without the turbine was performed with DEM in ThreeParticle (Fig. 11) with the previously described breakage model. For computational efficiency, the simulation was simplified in order to simulate fewer particles in the bunker. To simulate the drop without the turbine, only the pile peak was simulated as a drop into a bulk material bed has a damping effect. The bulk material bed was placed at the same height as the bottom of the turbine, which would allow a direct comparison. The turbine reduced the drop height for the discharged material from h = 3.356 m to hT = 1.756 m but caused additional impacts with the turbine and within the material. For the evaluation, a fragment collecting box was implemented, otherwise, the resulting fragments would have been deleted outside the simulation domain. Very soft material properties were assigned to the collecting box so that no breakage would occur due to impacts on the collecting box. A time step of 5 × 10−6 and the Hooke contact model were used. A detailed description of this simulation and parameters are described in [8].

For the comparison of particle damaging effects caused by the turbine, the same process was simulated with the cross flow turbine at different rotational speeds of 5, 10, 20, 30, 40, and 50 rpm. 5 rpm is the optimum for mixing and 30–40 rpm for energy recovery. Simulations in [8] show that at 5 rpm the material flows through the turbine, which results in a mixing effect and reduces segregation. This is not the case at 30 rpm, where the material remains on the same turbine blade and causes a higher torque on the turbine. The material forms a bulk material bed inside the turbine, which is assumed to have a significant damping effect. As the damping effect depends on the PSD in the material bed and smaller particles lead to higher damping, the small particles were also simulated in this case. The PSD for the simulation equaled the original PSD from the bulk sample of the sinter plant in Table 1. It has to be stated that only the size fractions 10–16, 16–25, and 25–40 mm were breakable in this case, which were represented by polyhedral particles of spherical shape with diameters of 16, 25, and 40 mm in the simulation. No experimental data for the fractions 6–10, 40–50, and 50–100 mm are available. The fines and the non-breakable particles were represented by spheres with a diameter of 6, 10, 50, and 100 mm in the simulations. A total mass of 122.9 kg sinter was simulated, which was equivalent to 160,512 particles with this PSD.

As a minimum grain size of 6.3 mm is required for a blast furnace operation, fines generation is especially critical for blast furnace sinter and is one main focus in conveying and storage processes in this case, apart from equipment wear [30]. These fines have to be screened out before reaching the blast furnace and have to be re-sintered (return fines), which is quite an energy consuming process causing high emissions and costs [30]. According to [31], 6% of the total mass flow of sinter provided at the blast furnace are fines due to conveying and storage processes in the EU average.

The results of the particle breakage evaluation for 5 rpm (mixing) and 30 rpm (energy recovery) are shown in Fig. 12, which shows the increase in mass fractions for each particle size due to the bunker filling process. The diagram does not include the size fractions 40–50 and 50–100 mm as 50 and 100 mm particles were not breakable in this simulation. More breakage occurs at the drop with the cross flow turbine at 5 rpm and about 1% more fines are produced in this case, compared to the case without turbine. It is assumed that volume breakage occurs mainly due to impacts on the blade edges, where also the most wear occurs [9, 10], and the generation of fines is mainly due to abrasion among sinter particles inside the turbine. Significantly less particle breakage occurs if the turbine is operated with a higher rotation speed of 30 rpm for energy recovery. 0.5% less fines are produced at 30 rpm compared to the case without turbine. It was assumed that the turbine has an even more particle-preserving effect at higher rotation speeds, but is limited to a certain rotation speed as the particle preserving effect of a damping material bed on the blades also decreases with higher rotation speeds. For this purpose, the fines production in dependence on the rotation speed was investigated for this case (Fig. 13). A minimum for the fines production is noticed at 30 rpm, which confirms the assumption.

Various studies confirm that replacing a drop by several smaller drops leads to less particle breakage [32]. Thus, particle breakage is not only dependent on the rotation speed, but also on the mounting height of the turbine and bunker filling levels. The reduction of particle breakage due to splitting the drop height into two smaller drops requires additional simulations. Overall, it is assumed that the turbine leads to less particle breakage during a whole bunker filling process but this needs to be confirmed by further investigations.

The cross flow turbine has a significant mixing effect. This was proven at the example of a bunker used for blast furnace sinter. A more evenly distributed particle size distribution at the bunker outflow is noticed when the bunker is filled with the cross flow turbine. The optimum rotation speed for mixing is 5 rpm and is 35 rpm for energy recovery for this turbine. Particle breakage with a focus on fines generation due to the turbine was evaluated using a novel breakage model for DEM. At 5 rpm slightly more fines are produced due to the turbine in this case. At higher rotation speeds (30–40 rpm) significantly less fines are produced due to the turbine. With this turbine a simple device to increase mixing effects or for energy recovery is provided, which is easily installable in existing plants. A trial with a prototype of the cross flow turbine is currently being performed at a steel manufacturing plant (Fig. 14).