Benjamin Bizjan (born 1988) obtained a master degree in 2012 and a PhD in 2014 at the University of Ljubljana, Faculty of Mechanical Engineering. In 2018, he was promoted from teaching assistant to associate professor at the Faculty of Mechanical Engineering in Ljubljana. He is also a postdoc researcher and developer at the high-tech Slovenian company Abelium. He is a member of an interdisciplinary team involved in challenging R&D projects solving complex problems from the industry: mineral wool technology development, flow velocimetry and thermometry solutions, cavitation avoidance and applications, precision farming, seaport environmental studies and big data analytics.
Experimental Investigation of Mineral Wool Fibre Formation by High-speed Imaging
Benjamin Bizjan* & Brane Širok
University of Ljubljana, Faculty of Mechanical Engineering, Aškerčeva 6, 1000 Ljubljana, Slovenia
Mineral wool is a fibrous insulation material produced by fiberization of silicate melts on spinning machines. The most commonly used types of mineral wool are glass wool and rock wool. Especially in the case of rock wool, the fibre formation phase of the manufacturing process is particularly complex due to the design of spinning machines. Typically, rock wool spinning machines comprise two to four fast spinning wheels in an asymmetric arrangement, over which the mineral melt is poured from a furnace, cascading between the wheels and forming a thin film on the surface of all wheels. Through the interaction of inertial, surface tension and aerodynamic forces, unstable flow disturbances occur on the melt film surface and grow into ligaments, which then rapidly solidify to mineral fibres.
Despite the fact that presented rock wool production technology has been used for more than 70 years, the manufacturers still encounter major operational challenges related to the fiberization process, namely poor melt film adhesion and formation of unfiberized material in form of pearls and shots. These phenomena significantly reduce the rock wool production capacity and negatively affect the quality of final products, but cannot be easily solved by a mere adjustment of integral production line parameters. Nevertheless, in the last decade, local measurements by high-speed video cameras have become a viable experimental method due to improved performance and greater affordability of such imaging equipment. Consequently, fiberization phenomena can now be studied on much smaller length and time scales, allowing for more accurate modelling.
Figure 1. A concept of high-speed imaging of the fiberization process of mineral melt
Our study of the mineral melt fiberization process was performed in two phases. The first phase was represented by cold experiments on a single wheel spinning machine. Instead of melt, liquids at ambient temperature (different water-glycerol mixtures) were used as working media to simulate the fiberization process by the hydrodinamically similar process of ligament-type liquid disintegration. Disintegration of the radial liquid film to ligaments was recorded by a high-speed camera, using diffuse background lighting to detect liquid structures which appeared relatively darker due to the light absorption. Image acquisition was followed by analysis of characteristic dimensions of observed liquid flow. The ligament foot spacing, s, and the mean diameter of ligament head droplets (dHD) were chosen as two most representative parameters for characterization of ligament formation hydrodynamics. Both s and dHD were determined to drop with the Weber number (We = ρω2R3/σ) of the liquid film (Figure 4). This implies that the total number of ligaments attached to the wheel perimeter increases with rotational speed, while the diameter and mass fraction of head droplets decreases. Head droplets are known to solidify to pearls in the melt fiberization process, consequently increasing the bulk density of produced mineral wool and reducing its mechanical strength (especially when used in composite materials). With that said, other fiberization parameters can also be obtained from visualization images (e.g. fibre diameter and length, melt film structure etc.).
Figure 2. Formation of glycerol ligaments on the surface of a spinning wheel (R = 45 mm, 1200 rpm)
To validate cold experiment results, further experiments were performed using typical silicate melts from the rock wool production process. Melt was initially heated to 1500°C and then poured onto the spinning wheel, forming a thin radial film. Formation of melt ligaments was visualized by a high-speed camera at different rotational speeds of the wheel (1800-3000 rpm). Then, parameters s and dHD were determined by a methodology similar to cold experiments. Experimental results suggest a good hydrodynamic similarity between cold and melt experiments (fig. 4), since cold experiment values of s and dHD extrapolate well into the higher Weber number range of silicate melt experiments. Therefore, a conclusion can be made that models of the fiberization process obtained from cold experiments are, to a large degree, also valid in an actual industrial fiberization process.
Figure 3. Examples of a silicate melt film on a spinning wheel (R = 95 mm, rotational speed 1800 rpm in the upper panel and 2400 rpm in the lower panel), and related melt structures in the fiberization process
Figure 4. Dependence of ligament foot spacing and head droplet diameter on the Weber number, and the similarity between cold experiments (water-glycerol mixtures as working media) and melt experiments
 B. Širok, B. Blagojević and P. Bullen, Mineral Wool: Production and Properties; Woodhead Publishing Limited, Cambridge, 2008.
 J. Vad, B. Morlin, Fluid Mechanical Model for Formation of Mineral Wool Fibers Applied in Polymer Composites. Materials Science Forum, 537-538, 269-276, 2007.
 J. Liu, Q. Yu, Duan W., and Q. Qin, Experimental investigation on ligament formation for molten slag
Granulation. Applied Thermal Engineering, 73, 888–893, 2014.
 E. Villermaux, The formation of filamentary structures from molten silicates: Pele’s hair, angel hair, and blown clinker. Comptes Rendus Mecanique, 340, 555-564, 2012.