Sign In
Navigate Up
Project

Why the REFRASORT project ?

​​​​​​The refractory production industry in Europe is vital for all high temperature processes, such as the production of metals, cement, ceramics and many others. Raw materials account for about 40 to 50% of the refractory costs and have a major influence on the quality of the finished products. The European industry is largely dependent on import, mainly from China, for high quality raw materials. Graphite and magnesite have been classified as critical raw materials by the European Commission. Refractory grade bauxite is for 95% produced in China. The prices of refractory raw materials have risen significantly in the last decade, by 30 to over 300 percent depending on the raw material.

One way to reduce the supply risk and raw material costs is recycling of spent refractories. However, for high value recycling of refractories in refractory production, these secondary raw materials must meet stringent quality criteria, as the raw material properties have a direct and indirect effect on important refractory properties. The spent refractories must be separated from impurities such as metal and slag, and classified into high purity fractions according to their chemical composition. For industrial application, such a cleaning and sorting step must be performed in an automated way. The ongoing FP7 project REFRASORT, funded by the European Commission, is developing just such an automated sorting system.​ 

​Objectives​

The REFRASORT project aims at developing a new separation technology that enables automated sorting of coarse refractory material into pure material fractions under industrial conditions, using non-destructive technologies.

By increasing the purity of the refractory waste streams and thus the availability of high-grade raw materials, reuse should increase from 5% to 20% of raw material demand

Specifically, the project will:

  • Develop a reliable sensor for inline identification of refractory material, suitable for use in an industrial environment.
  • Develop an automated sorting equipment for coarse, heavy particles that is able to sort up to 8 material types.
  • Integrate the sensor and sorting equipment into a functioning demonstrator
  • Validate the technology by the production and assessment of refractory material with a significant content of secondary raw material.

Challenges

Automated sorting of refractories has to face a number of challenges due to the specific nature of the material. First of all, an enormous variety of refractory materials is in use today. The type of refractory used depends on the application, and often multiple types can be present together in one vessel. The REFRASORT project focuses on 8 types of refractories that represent 95% of all refractories used in the steel industry, which is in itself responsible for 60% of all refractory use worldwide. The 8 types can be grouped into three main classes: magnesia-based, doloma-based and alumina-based.

Group​​ Type​ Composition​
MgO-based Fired MgO High MgO, no C, low CaO 
MgO-C with antioxidant High MgO, 5 – 15 wt-% C, lo​w CaO, antioxidant ~3%
MgO-C without antioxidant High MgO, 5 – 15 wt-% C, low CaO, no antioxidant
Doloma-based Fired d​​oloma​ High MgO, high CaO, no C
Doloma carbon​ High MgO, high CaO, 5-15 wt-% C​
Alumina-based Fired bauxite High Al, Al/Si ~8/1, low CaO/MgO/C
Fired andalusite High Al, Al/Si ~2/1, low CaO/MgO/C
Fired chamotte High Al, Al/Si ~1/1, low CaO/MgO/C
 

To determine the boundary conditions for the automated sorting installation, a thorough characterization of fresh and spent refractories was performed. During use, refractories are subjected to extreme conditions due to the interaction with hot steel and slag. At end of life, refractories are broken out of steel vessels mostly without separation. As a result of demolition and usage, dust and reaction layers are formed at the outer surface of the refractories. These dust, slag and carbonation layers (arrows point out carbonation layers in the figure below) are on average 100 – 200 µm thick, but may be as much as one to several cm thick on the hot face (in contact with slag). Such contamination layers pose a significant challenge to many commonly used sensors based on surface properties (e.g. colour sensors, XRF, FTIR). 



Additionally, the spent refractory bricks are heavy, coarse objects that vary significantly in shape, size and weight. Within the particle size range considered in the REFRASORT project (60 – 300 mm), object weight ranges from several hundred grams up to 20 kg. While the majority of objects are brick-shaped, the remainder consists of broken pieces and unshaped material. This significantly complicates the design of a suitable handling system.

Results

The REFRASORT system combines a novel identification technique based on LIBS (Laser induced breakdown spectroscopy) with a mechanical handling system able to deal with large and heavy objects.




LIBS (Laser induced breakdown spectroscopy)​​

The basic principle of the LIBS sensor for sorting is shown in the figure above. By focusing the beam of a pulsed laser on the refractory, power densities in the range of GW/cm² are reached locally for a short time. These are sufficient to vaporise the material, dissolve the chemical bonds and heat the material to temperatures above 10 000 °C. In this state, the material emits light in specific spectral lines, which is detected with a spectrometer and allows the qualitative and quantitative determination of the composition of the object within a few microseconds.

The LIBS system is also uniquely suited to deal with the abovementioned surface contaminations as it uses a laser burst consisting of two components co-located on the moving object. A first burst is used for cleaning and penetrates the contamination layer to a depth in the order of 100 µm. The second burst can then take analytical measurements with significantly reduced interference from the contamination layer. By repeated application, analytical information can be derived even from deeper regions. The data evaluation takes into account the emissions from the major matrix elements Ca, Mg, Si, and Al, as well as from minor elements of interest such as C, Cr, and Fe. Using a pre-defined classification scheme, each piece of refractory is assigned to a sorting fraction in such a way as to keep the output within the adjustable recycling specifications. To establish the classification scheme, LIBS measurements were carried out on a set of used and freshly produced refractory materials of known composition. 

 

Laser measurements on fresh samples of the 8 types of refractories have demonstrated that the three main types of refractories (Type A = MgO based, Type B = Doloma based and Type C = Alumina based) can be identified based on the spectroscopic signals of selected elements (Mg, Ca, Si). The statistical significance of the separation is indicated by the 1-sigma error bars of repeated measurements each taking few milliseconds. Incorporating further elements (e.g. Al, C) in the data evaluation is expected to allow identifying also the composition of the sub-classes and providing a reliable basis to sort refractory material with high purity and enable improved high-grade recycling.​ First sorting trials on spent refractories have shown that, with the current settings, 29 out of 30 bricks could be correctly identified into the three main classes. Further optimization is needed however to improve the cleaning of the surface to reduce the influence of surface contamination, and adapt the LIBS thresholds to meet recycling specifications.

During the measurements, it was noted that the spectroscopic signal of C is lower than expected in the spent carbon-bonded refractories. This can be explained by the oxidation of C during refractory use, which leads to a decarbonized layer at the surface of the brick that can be as much as 1 cm thick (Figure 6) while the LIBS measurement typically measures at some 100 µm depth. Complementary identification techniques are therefore investigated to aid in the reliable identification of C containing bricks.  

Mechanical handling system​​​

The mechanical handling system has to ensure that each ​particle is presented to the sensor separately and each object ends up in the correct fraction according to the sensors measurement. Due to the nature of the sensor, the objects have to be presented with a defined minimum distance to each other. The mechanical handling consists of four sub-processes, namely lining up, isolation, spacing and separation.

The selected approach starts with a vibratory feeder with a v-shaped metal trough, which forces the objects to line up behind each other (Lining up). Behind the vibratory feeder is a plate conveyer, which runs at higher velocity than the vibratory feeder. This way the objects are accelerated and isolated from each other (Isolation). From the first plate conveyor, the objects are pushed into a parking position by set of pneumatically driven pushers. Objects that are classified as too big or too small for the subsequent handling remain on the plate conveyor and end up in a residual fraction. A second plate conveyor transports the objects to the identification after they are released from the parking position (see figure). The distance between the objects on the second conveyor can be adjusted (Spacing) by changing the timing of releasing the objects from the parking position. After identification the bricks have to be separated into different fractions according to the sorting decision, which is again achieved using pneumatically driven pushers (Separation). 


The first tests with the mechanical handling demonstrator, performed on objects in the size fraction of 60-160 mm showed that the vast majority of objects can be handled and presented to the sensor. The number of sorted fractions is currently limited to three but can easily be increased by extending the length of the second conveyor and increasing the number of pneumatic pushers.​ 

Facts and figures​

  • Start date: 1-11-2013
  • Duration: 36 months
  • Total cost: 2,38 Million EUR
  • EC contribution: 1,75 Million EUR
  • Instrument: FP7 – ENV.2013.6.3-1
  • Grant Agreement no° 603809