Early View
Open Access

Development and experimental characterization of the thermomechanical behavior of a scaled steel ladle

Pratik N. Gajjar

Pratik N. Gajjar

Department of Civil Engineering, ISISE, ARISE, University of Minho, Guimarães, Portugal

Search for more papers by this author
Pieter Put

Pieter Put

Ceramics Research Centre, Tata Steel, Velsen Noord, The Netherlands

Search for more papers by this author
João M. Pereira

Corresponding Author

João M. Pereira

Department of Civil Engineering, ISISE, ARISE, University of Minho, Guimarães, Portugal


João M. Pereira, Department of Civil Engineering, ISISE, ARISE, University of Minho, Guimarães, Portugal.

Email: [email protected]

Search for more papers by this author
Bruno Luchini

Bruno Luchini

Ceramics Research Centre, Tata Steel, Velsen Noord, The Netherlands

Search for more papers by this author
Sido Sinnema

Sido Sinnema

Ceramics Research Centre, Tata Steel, Velsen Noord, The Netherlands

Search for more papers by this author
Paulo B. Lourenço

Paulo B. Lourenço

Department of Civil Engineering, ISISE, ARISE, University of Minho, Guimarães, Portugal

Search for more papers by this author
First published: 02 May 2024


A steel ladle usually employs mortarless refractory masonry in the innermost layer in direct contact with molten steel. The behavior of such masonry under high thermal loading is complex and essential for designing such installation. A large-scale experimental campaign was carried out focusing on a simplified, laboratory-scaled, pilot steel ladle built with multiple refractory linings as in an industrial ladle. Tests were performed under transient thermal loading up to 1400°C with new and used working linings. The results showed that the distribution of temperatures between the refractory linings was similar for all specimens and confirmed the importance of the selection of the refractory materials in the distinct linings of the ladle to limit heat losses. The heterogeneity in the distribution of dry joints and the effect of the dry joint thickness were also measured using strain gauges. The measurements indicate the effect of viscoplasticity in the working lining on the steel shell, which shows a progressive reduction in the strain starting at 1200°C. The tests performed on the working linings show lower strain built-up due to large viscoplastic deformation and increased joint thickness.


Steel ladles play a crucial role in the steelmaking industry. They are employed as industrial vessels to transport and refine molten steel and are exposed to a harsh operating environment.1, 2 Normal operating conditions of these vessels include high temperatures, high thermal stresses, slag attack, and degradation of layers in contact with liquid steel.3-7 Steel ladles are often built with multiple layers of refractory masonry linings enclosed within a thick steel shell to withstand such extreme operating conditions.8-10 Each layer comprises refractory material with different thermal, mechanical, and chemical properties. The selection of refractory materials for these layers depends on their designed purpose, such as thermomechanical behavior, thermal insulation, and chemical stability.11-13 The working lining is the most critical layer in the steel ladle. It is the innermost layer in direct contact with liquid steel therefore subjected to high operating temperatures and temperature variation. Consequently, this layer is subjected to thermal shock and a severe chemical environment, leading to material degradation.14 In short, the design and optimization of a steel ladle is a complex process influenced by material selection, lining thickness, distribution of dry joints, and process conditions.15, 16

Refractory materials exhibit complex temperature-dependent properties, from quasi-brittle behavior at ambient temperature to ductile behavior at higher temperatures.17 Therefore, understanding the failure mechanisms, as well as thermal and chemical damages of the refractories, becomes essential to optimize their consumption in extreme working environments.18, 19 The occurrence of creep in refractories and its influence on the behavior in industrial applications has been established in the literature.20-23

The working lining of a steel ladle is usually constructed with dry-stacked masonry (mortarless masonry). The joints of such masonry provide a physical break in continuous media, which can reduce the stress build-up. These dry joints show a high nonlinear behavior, which is affected by material type, contact area distribution, and the spatial and size distributions of the asperities on the surfaces.24 The experimental investigations performed for dry joints in refractory masonry under varying temperatures present a temperature-dependent normal compressive behavior25, 26 and tangential behavior10 due to thermal expansion and change in the surface asperities at higher temperatures.

Moreover, in mortarless masonry, the joint thickness is not uniform due to surface unevenness and tolerances in dimensions arising from the manufacturing process.27, 28 This nonuniform distribution can lead to local stress concentrations, affecting global behavior. Therefore, large-scale experimental campaigns are valuable to evaluate the global behavior of refractory masonry under varying thermal and mechanical boundary conditions. These campaigns also validate the material properties derived from experiments at material levels, such as Young's modulus, thermal expansion, thermal conductivity, and creep behavior, noting that limited work is available in the literature regarding the experimental evaluation of mortarless refractory masonry.

Oliveira et al.9 experimentally investigated the mechanical behavior of mortarless refractory masonry made with alumina–spinel bricks under uniaxial compression at various temperatures. They found high nonlinear mechanical behavior from the tests. The observed global stiffness and ultimate compressive strength of the mortarless refractory masonry were found to be lower compared to the alumina–spinel bricks. This work highlights the heterogeneity present in mortarless refractory due to dry joints and its influence on thermomechanical behavior. Ali et al.8 further evaluated the mechanical behavior of the mortarless refractory masonry with alumina–spinel bricks in a biaxial press under various loading at different temperatures. They found that the masonry has orthotropic mechanical behavior, and the effective stiffness in the direction normal to head joints was higher than that in the direction normal to bed joints. Tests performed at high temperatures exhibited a significant influence of viscoplasticity on global behavior. They observed permanent deformation of bricks due to viscoplastic behavior.

Finite element models are often used to investigate or design industrial installations such as steel ladles with different linings of refractory masonry.2, 15 The data obtained through experimental campaigns are used to calibrate such models with varying degrees of complexity from linear thermoelastic to more complex models, including viscoplasticity and damage.29-31 However, the behavior obtained through such models for a steel ladle is not validated due to a lack of experimental investigations. Therefore, to validate numerical models, experimental work is required to be carried out on an assembly that represents a real structure.

Gasser and Poirier32 investigated the thermomechanical behavior of mortarless refractory masonry in a laboratory pilot that resembles a steel ladle. The test was conducted with only one layer of refractory lining and was performed under transient thermal loading with a maximum temperature of 1000°C. The thermal behavior was measured with thermocouples, and the mechanical response arising from thermal loading was evaluated by strain gauges and displacement sensors installed on the exterior surface of the steel shell. The elastic strains observed from the steel shell revealed the thermomechanical behavior of the refractory lining. The behavior of the lining was influenced by the dry joints present and the contact between the bricks and steel shell. Subsequently, more open joints after the test upon cooling were found, with a more uniform distribution.

Due to the complex behavior of refractories at high temperatures and the need to validate numerical models, an experimental campaign on a laboratory-scaled steel ladle is paramount. It is evident from the studies mentioned before that the experimental characterization of refractory masonry under high temperatures is limited and requires further large-scale investigations. Moreover, the tests are required for cylindrical refractory walls as in a steel ladle with multiple refractory linings under a thermal loading similar to normal operating conditions (around 1500°C).

This work aims to experimentally characterize a pilot steel ladle (with multiple refractory linings) that undergoes similar thermomechanical loadings as an industrial ladle to gather data for the validation of advanced numerical models. This work presents a novel experimental setup for large-scale refractory masonry within the framework of ATHOR (advanced thermomechanical multiscale modeling of refractory linings).33 A detailed description of the test configuration from materials, geometry, and measurement devices is given, and the results obtained from different test series are discussed along with field observations. A total of four tests were conducted, being two on new working linings (NWLs) and two on used working linings (UWLs).


There are various constraints that need to be considered during the design of a steel ladle; for example, steel ladles must be safe and durable while also being economical. The refractory linings of the steel ladle are primarily composed of three layers: the working or wear lining, the safety or permanent lining, and the insulating layers. In the case of the steel ladle used at the Tata Steel plant, the working lining has dry joints, while the safety lining is constructed using mortar joints. Figure 1A presents a steel ladle and different lining zones, while Figure 1B presents a detail of the different layers.

Details are in the caption following the image
Industrial steel ladle: (A) graphical representation of ladle with different lining zones (adapted from34); (B) different layers of refractory linings in a ladle (adapted from35).

As mentioned above, the main objective of the experiments on the pilot ladle is to characterize its thermal and mechanical behavior. This characterization performed under a controlled environment can further assist in the calibration and validation of the various numerical models with the final goal of having better models for industrial ladles. Therefore, the pilot ladle is a scaled representation that includes all the linings of an industrial ladle. The pilot ladle is tested under thermal loads of similar magnitude of an industrial ladle without the presence of molten steel, which also allows to characterize the inner surface of the linings. This experimental structure is constructed and tested at the Ceramics Research Centre (CRC) of Tata Steel in Ijmuiden.

The following sections describe the various materials used in a steel ladle (consequently in the pilot ladle) and their main thermal and mechanical properties. After that, a description is given of the design approach used for the pilot ladle's geometry and of the different measurement devices used to gather thermal and mechanical behavior.

2.1 Material properties

The final objective of a steel ladle is to have a structure that supports the harsh operating conditions imposed by molten steel while keeping heat losses to a minimum. This is achieved by employing multiple refractory linings with different materials. The working lining material presents a very high erosion and corrosion resistance but a relatively high thermal conductivity. In contrast, the safety and insulation lining aims to reduce the heat losses of the ladle, with each material operating in its optimum temperature range. The refractory materials used in the pilot steel ladle in the different lining layers are presented in Table 1.

TABLE 1. Thermal and mechanical properties of materials used in the pilot steel ladle.
Lining Working lining—WL Safety lining—SL Insulation lining—IL Insulation board—IB
Material InlineGraphics InlineGraphics InlineGraphics InlineGraphics
Fired alumina–spinel bricks (94% alumina, 5% magnesia, 1% other oxides) Bauxite bricks (90% bauxite, 10% other oxides) Alumina silicate (70% alumina, 30% silica) Microporous insulation with pyrogenic silica (fumed silica; glass cloth outer envelope)
ρ (kg/m3) 3200 2750 900 360
α (×10−6°C−1) 8.87 7.0 8.2 9.3
E20 (GPa) 21 9.5 .91 .15
λ20 (W/m °C) 6.43 1.6 .41 .026
Cp20 (J/kg °C) 805 535 827 796
  • Here, ρ is the density; α is the coefficient of thermal expansion; E20 is the Young's modulus; λ20 is the thermal conductivity; and Cp20 is the specific heat at 20°C.

Alumina–spinel bricks are widely used in the barrel zone of the working lining (Figure 1A). This zone will be in direct contact with molten steel and subjected to the highest temperature in the steel ladle. The primary purpose of the material used in this layer is to avoid carbon contamination as well as to provide good thermal and mechanical stability. The purpose of safety lining is primarily to prevent the steel penetration further into the ladle lining. Therefore, the material used in this lining should offer relatively good mechanical performance while offering better thermal insulation than the working lining. For this reason, bauxite bricks are usually used in the safety lining.

Insulation lining is used to reduce the heat losses in the ladle. Thus, the material used in this layer should have low thermal conductivity. The material used in this lining is aluminum silicate, which has the desired thermal properties. The insulation board is an additional lining component to further reduce heat losses. This microporous material offers excellent thermal properties while it is highly compressible. The thermal and mechanical properties of these materials are presented in Table 1. The density and thermal expansion of the materials are considered to be constant till their normal operating temperatures.2, 8 In contrast, Young's modulus (measured in this campaign), thermal conductivity,2, 36 and specific heat2 are temperature dependent. The normalized (at 20°C) values of these properties at different temperatures are shown in Figure 2.

Details are in the caption following the image
Normalized thermal and mechanical properties of lining materials: (A) Young's modulus (ET/E20); (B) thermal conductivity (λT/λ20)2, 36; (C) specific heat (CpT/Cp20).2, 36 IB, insulation board; IL, insulation lining; SL, safety lining; WL, working lining.

As mentioned earlier, the choices of materials in a steel ladle change within the industry depending on the design philosophy. As described by the material properties of different linings, it is evident that the material near the contact with molten steel requires higher mechanical performance, and subsequent lining materials require better thermal conductivity. The materials described are used for the experimental campaign on the pilot steel ladle.

2.2 Design of the pilot steel ladle

The scale of the pilot ladle should be such that it is small enough to be handled at the laboratory level and big enough that it could represent the global behavior of an industrial ladle. The other difficulties involved with the scale are the thickness of the steel shell and the size and shape of the refractory bricks. To tackle these challenges, preliminary numerical investigations were developed, assisting with the design and, ultimately, contributing to the choices made regarding the installation itself.

The objective was to assess different aspects of the pilot ladle in such a way that it would be possible to make decisions on the final configuration of the installation throughout the design phase. Different aspects were tackled, namely, the definition of the overall dimensions (scale), refractory linings' thickness, steel shell thickness, ladle bottom, and ladle height.

The preliminary numerical simulations were performed in Abaqus37 considering simplified two-dimensional (2D) partial ring models taking advantage of periodic layout of the bricks in the steel ladles. These models assumed only elastic material properties to be on a conservative side during design phase. The 2D models were then subjected to thermal loads and relevant thermal boundary conditions similar to the values used by Ali et al.2 These simulations aimed to observe the behavior of an industrial steel ladle in terms of temperature, stress, and strain distribution over the refractory linings. Similar simulations were performed at different scales for the pilot ladle, allowing to observe the influence of the thermal and mechanical behavior of the scaled ladles, the size and shapes of the refractory bricks, and so forth. An example of thermal gradient convergence between industrial ladle and pilot ladle is shown in Figure 3.

Details are in the caption following the image
An example of temperature distribution in two-dimensional (2D) partial ring model used to design pilot ladle showing working lining, equivalent safety, and insulation lining. (A) Industrial ladle; (B) pilot ladle.

Following these observations, it was possible to converge on the pilot steel ladle scale with a diameter of 1500 mm (an industrial ladle is usually 4500 mm in diameter). This pilot ladle was found to represent a mechanical behavior that is similar to an industrial ladle. Subsequently, it was decided to keep the thickness of refractory linings the same as in an industrial ladle to obtain a similar magnitude of the thermal gradient during the experimental campaign. Furthermore, steel shell thickness was optimized to 6 mm (35 mm in an industrial ladle). With this thickness, the steel shell was found to be safe to resist the thermal and mechanical loads imposed by the refractory lining without yielding.

Next, numerical simulations were performed to investigate the effect of the ladle bottom. The results found that the pilot ladle bottom imposes restraining effects on the barrel part, which also occurs for an industrial ladle. In contrast, a model without a ladle bottom will characterize the behavior of the refractory linings only in the barrel part of a steel ladle (without restraining effects). Furthermore, a pilot ladle with only the barrel part simplifies the geometry and test setup, reducing the requirement for support conditions and additional measurement devices. Therefore, it was decided to opt for the pilot ladle without a ladle bottom, which will provide insight into the behavior of the working lining and its influence on the barrel of the ladle.

From the analyses performed on models with different heights, it was observed that global boundary conditions govern the mechanical behavior of the refractory linings for the ladles with low heights (i.e., less than 300 mm). A pilot ladle with 500 mm height was deemed sufficient to neutralize these effects.

From the design optimization process discussed in this section, the final dimensions of the pilot ladle were 1500 mm in diameter and 500 mm in height. Figure 4A presents the global overview of the pilot ladle. Different refractory linings are also shown, identical in thickness to an industrial steel ladle (Figure 4B). Due to the change in geometry, the bricks in the working lining needed to be cut. The modified dimensions of the brick required are discussed later in this section. The shape of the safety and insulating linings did not need to be altered.

Details are in the caption following the image
Graphical illustration of the three-dimensional (3D) pilot steel ladle: (A) pilot ladle, top and bottom insulation; (B) detail-A indicating the configuration and thickness of the refractory linings; (C) pilot ladle and bottom and top insulations, along with placement of heating elements and a small rectangular viewing window. Dimensions are in millimeter.

An assessment of the electrical power requirement is crucial for the experiments on the pilot steel ladle. From this calculation, the number of heating elements required can be defined. Analyses considering different parameters found that the steady-state power consumption is 21.5 kW for all the cases to achieve a surface temperature of 1400°C, which is slightly lower compared to molten steel's temperature. However, this target temperature was deemed sufficient due to the limitations of the power supply unit and the fact that a significant amount of damage and viscoplasticity of working lining bricks occur from 1300°C, which could be observed with this target temperature. Nevertheless, this is a considerable leap forward in terms of large-scale experiments on refractory masonry, where most of the tests are conducted at or below 1000°C.9, 32 These results determined the number of the heating element and their spacing. Eighteen silicon carbide heating elements (Kanthal Globar SR) were placed 138 mm away from the surface of the working lining. The spacing between the elements is also 138 mm.

Afterward, the design of the top and bottom insulation was addressed. The top insulation was designed to reduce heat losses and to support the heating elements. Figure 4C shows the graphical representation of the top lid. The thickness of the steel shell is 6 mm with a diameter of 1612 mm. Three stiffeners were considered to increase the lid's stiffness and reduce the vertical displacement during service conditions. Additionally, holes were made to place the heating elements, and a rectangular window was considered to allow monitoring (discussed in Section 2.3). Bottom insulation (shown in Figure 4C) was designed to limit heat losses, support the gravity load of the whole assembly and protect the ground floor.

2.3 Measurement devices

Experiments on the pilot model were performed to observe the behavior of refractory linings at elevated temperatures. Thus, the expectation is to measure the temperature distribution, strains, and dry joint opening and closing. These measurements are essential for the calibration and validation of the developed numerical models. The temperature distribution will assist in identifying critical thermal boundary parameters, such as heat transfer coefficient and emissivity at the exposed surfaces of the linings. Additionally, the temperature measurement acquired within the surface will help to reconfirm the thermal properties of the material (i.e., thermal conductivity and specific heat). Another key aspect in the validation of the numerical models is the definition of the local mechanical boundary conditions (i.e., the interaction between the bricks and linings). The measurement of the mechanical-related fields (strain and displacement) can assist in confirming the thermomechanical properties of the material (thermal elongation, elastic–viscoplastic behavior, and damage). Therefore, suitable measurement devices were necessary so that reliable data could be used to calibrate numerical models.

Thermocouples are widely used for the instrumentation of industrial devices to observe temperatures,9, 38 and several thermocouples were used at different locations. An assembly of thermocouple placement was installed at three different locations and layers of the pilot ladle (spaced at about 120° in the plan view), as shown in Figure 5A (total of 18 thermocouples). As shown in Figure 5B, six type K thermocouples (with a range of up to 1260°C) were installed in different refractory linings. Two thermocouples were installed in the working linings at the midthickness and 3/4th thickness by cutting grooves at the bottom face of the bricks. Three thermocouples were installed between the interface of the different linings (working–safety linings, safety–insulation linings, insulation lining–insulation board). The remaining thermocouple was installed at the outer side of the steel shell. One type S thermocouple (with a range of up to 1600°C) was placed near the innermost part of the working lining. Additional thermocouples were placed inside the bottom insulation layers to monitor the heat losses.

Details are in the caption following the image
Installation and placement of measurement devices; (A) plan view showing the location of thermocouples; (B) detail-A indicating placement of six thermocouples in different linings; (C) digital image correlation (DIC) setup to measure displacement fields in the top refractory linings and the steel shell.

Measurement of the mechanical fields is a challenging task in a high-temperature environment. Measurement devices require stable operation conditions to obtain reliable data. Digital image correlation (DIC) is an optical full-field measurement technique, which allows the determination of an entire experimental field of strains and displacements based on the gray-level conservation principle.39, 40

The use of DIC is proven for high-temperature applications41, 42 and can be used for continuous measuring. Therefore, DIC was employed to monitor displacement and strains in the working lining through a cut window in the top lid (see also Figure 5C) and on the exterior section of a steel shell, as shown in Figure 5C. Additionally, three high-temperature spot welding foil strain gauges from Kyowa (self-temperature compensating) were placed on the structure's cold face, close to the thermocouples shown in Figure 5A. The shell (cold face) temperature reaches 250°C during the experiment. The use of strain gauges on the cold face to evaluate the behavior of refractory lining has been reported in previous high-temperature experiments.32 Moreover, extensive manual field measurements to observe changes in shape, joint thickness, and damage are planned before and after testing the ladle.

2.4 Construction of refractory linings

Once the geometry, composition, and choice of measurement devices were finalized, the construction of the pilot steel ladle was carried out. The steel shell was placed on the finished bottom insulation layer to construct the pilot ladle linings. Subsequently, the insulation boards of 5 × 300 × 300 mm3 (length × width × height) were placed by applying an air-hardening mortar, as shown in Figure 6A. Next, the insulation lining was constructed by placing bricks of 32 × 114 × 230 mm3 dimensions. Afterward, the safety lining was built by using bricks of 40 × 114 × 230 mm3 dimensions (Figure 6B). For these layers, the air-hardening mortar was used to close the joints. Finally, the bricks in the working lining were placed, dry-stacked, as shown in Figure 6C.

Details are in the caption following the image
Construction of pilot ladle linings: (A) placement of insulation board; (B) building insulation and safety linings; (C) building working lining.

The bricks used in the working lining are trapezoidal in shape. They are 140 mm in thickness and 100 mm in height, with the width at the shorter face equal to 116 mm and 148 mm at the other face. This lining is made with dry joints; therefore, no mortar was used. Additionally, these bricks are cut from manufactured bricks with larger dimensions to account for the reduced diameter of the ladle. As in the manufactured bricks, these cut bricks had a dimensional tolerance of ±1 mm, which led to uneven distribution of the dry joints with varying thicknesses. As shown in Figure 7, the thickness of these dry joints varies due to dimensional tolerance and the placement of bricks in the linings.

Details are in the caption following the image
Nonuniform dry joint thickness caused by the dimensional tolerance and placement of the bricks.

As shown in Figure 7, the joints in the working lining were classified as head joints and bed joints. Field measurements for the thickness of joints were carried out for all the joints. Table 2 presents the mean values of joint thickness observed for the two specimens (NWL-01 and NWL-02) considered for this experimental campaign. The scatter is quite large, with coefficients of variation (ratio between standard deviation and mean) about 100%. It can be observed that the average thickness of the bed joint is similar for both specimens. The thickness of head joints is larger compared to bed joints, and it is different between both specimens as well. This difference is due to the dimensional tolerance of the bricks, as well as their placement within the linings. The largest joint is also shown in Table 2, whereas the smallest joint is defined as a joint with no observable gap between the bricks (0 mm thickness). It is noted that those joints cannot be described as closed joints due to the surface roughness of the bricks, which induces a small amount of joint thickness25 and joint stiffness.

TABLE 2. Joint thicknesses in the working lining of the pilot ladle before the test (two specimens are shown).
NWL-01 NWL-02
Head joint Bed joint Head joint Bed joint
Mean thickness of joints (mm) .32 .20 .23 .18
Standard deviation (mm) .30 .22 .27 .21
Largest joint (mm) 1.68 1.45 1.77 1.30


After all the linings were built, the top insulation lid with heating elements was placed on top of the pilot ladle. Above the top lid, a protective cover was placed to isolate heating element connections. Figure 8 shows the finished experimental setup of the pilot ladle. Thermocouples and strain gauges were placed and connected to a data acquisition system. The figure also shows the DIC setup used for the steel shell and working lining. Additionally, a fan was used to mitigate the hot air draft arising from the working lining, which can influence the DIC.

Details are in the caption following the image
Experimental setup of the pilot ladle. DIC, digital image correlation.

As shown in Table 3, a total of four tests were conducted on the pilot ladle in this experimental campaign. The first two tests were performed on a pilot with NWLs, and the other two tests were performed on the UWLs of NWL-02. Therefore, UWL tests signify subsequent thermal loading cycles, where an initial thermal load of 1400°C was applied with NWL-02, the second thermal cycle was applied with UWL-01, and the third thermal cycle was applied with UWL-02. Once the target temperature was achieved, an 8-h dwell time was used for all the tests. This duration was selected considering the calibration of numerical models for constant high-temperature applications. The test specimen NWL-01 was tested till 1250°C as a precaution and to check the structural safety of the pilot ladle components, while the other tests reached 1400°C. The results of the tests are presented in the following sections.

TABLE 3. Summary of tests performed on the pilot ladle.
Test series Specimen Target temperature (°C)
NWL NWL-01 1250
NWL-02 1400
UWL UWL-01 1400
UWL-02 1400
  • Abbreviations: NWL, new working lining; UWL, used working lining.

3.1 Test series—NWL

Figure 9A,B shows the temperature evolution observed in the installed thermocouples at the 120° locations in the ladle linings. At the beginning of the thermal loading through heating elements, the whole assembly of the pilot ladle was at room temperature of around 16°C. The thermal loading was applied at a rate of 4°C/min. It is possible to note that the applied rate of thermal loading was sustained till 500°C of the furnace temperature. Afterward, the rate of loading decreased due to increased power requirement. However, at the 20th hour, the furnace temperature reached 1250°C for both tests. For the test specimen NWL-01, the applied temperature was kept constant for the next 8 h, and for the other test specimen, the applied temperature was increased to 1400°C. Once the 1400°C desired temperature was achieved at the 36th hour, the same was kept constant for the next 8 h.

Details are in the caption following the image
Test series—new working lining (NWL): (A) average temperature evolution observed during experiments (three locations around the circumference); (B) location and legend for installed thermocouples. Average circumferential strain evolution: (C) with respect to time; (D) with respect to hot face temperature.

For both specimens, temperature evolution observed in the linings is very similar. Moreover, irregular temperature evolution can be observed at the different linings at 100°C, which can be attributed to the evaporation of free water. During the dwell time of 8 h, a constant temperature at the hot face (HF) of the working lining can be observed. However, temperature rise in subsequent linings of the ladle can be identified for both specimens as the system moves to a steady state.

At the end of the experiment of NWL-02 (at 44 h), the furnace temperature was 1400°C. The temperature observed at the HF of the working linings was 1383°C. The temperature at the thermocouples, Working-1 (installed 65 mm from HF) and Working-2 (100 mm from HF), was 1219°C and 1105°C, respectively. The temperature at the safety lining (140 mm from HF) was 1000°C. Therefore, a drop of approximately 383°C can be observed within the working lining.

The temperature at the insulation lining was 852°C (40 mm from the safety lining). Thus, a reduction of 148°C can be attributed to the safety lining, which has more insulating thermal properties compared to the working lining. Furthermore, a decrease of 367°C can be noticed within the insulation lining, from 852°C to 485°C within 32 mm. In the end, the temperature at the exterior of the steel shell was 258°C. This is a reduction of 227°C within the insulation board (5 mm thickness) and steel shell (6 mm). Since steel has very high thermal conductivity, the reduction must be attributed to the insulation board.

The temperature evolution and distribution in the ladle linings exhibit the expected thermal behavior of a steel ladle. Temperature measurement data provide insight into the behavior of different layers in the lining during the application of thermal loads. Considering the selection of materials used in the linings and their different thermal properties, a complex distribution of temperature from the HF to the exterior of the steel shell can be observed. These data can be used to evaluate parameters that are crucial for numerical modeling (such as heat transfer coefficients at the interior and exterior and gap thermal conductance between the linings). Furthermore, these data can be used to validate the thermal properties of the materials.

The applied thermal loads and the temperature distribution within the linings dictate the mechanical behavior of the pilot ladle. Given the local and global boundary conditions, thermal expansion of the linings will exhort pressure within and between the linings. The highest pressure will be at the HF of the working lining due to higher temperature (thus higher thermal expansion). This pressure will be transferred to the subsequent linings and ultimately to the steel shell. As the working lining is made with dry joints, a part of the thermal expansion in this lining will close these joints without exhorting pressure. Additionally, viscoplastic behavior (or creep) will also be present at very high temperatures.

As mentioned, the performance of the steel shell is a critical issue in the behavior of the refractory linings subjected to thermal loads. Figure 9C shows the average mechanical strain observed in the circumferential direction through installed strain gauges on the exterior surface of the steel shell. The strain evolution confirms a highly nonlinear behavior during thermal loading. During the early stage of the test (till 2 h), a slight increase in the strain can be observed. This increase can be attributed to the closing of the head joints in the working lining as well as the closing of the gap between the linings (i.e., working–safety lining). Once these joints are closed, a steady growth in the strain can be observed till about the 20th hour. The effect of the joint sizes can also be noticed from the maximum strain observed. Specimen NWL-02 exhibits stiffer behavior compared to NWL-01. This difference was expected as the average head joint thickness in the working lining is smaller for NWL-02 (.23 mm) compared to NWL-01 (.32 mm), as shown in Table 2.

After the 20th hour, strain reduction can be observed till the end of the test. In NWL-01, the maximum applied temperature was lower (1250°C) compared to NWL-02 (1400°C). Therefore, during the dwelling period for NWL-01, no significant drop in strain was observed. However, for NWL-02, despite a rise in applied temperature after the 20th hour (consequently an increase in thermal expansion), a gradual reduction in the strain was observed (from .57 to .48 mm/m). This reduction is due to the viscoplastic behavior of the alumina–spinel bricks in the working lining. The viscoplastic behavior of this material becomes prominent after 1200°C. Figure 9D shows the strain evolution with respect to the HF temperature of the working lining. From the figure, strain reduction can be observed to be starting from 1200°C. Therefore, the strain measurement from the steel shell of the ladle reveals the thermomechanical behavior of the refractory linings and the effect of joints. These data are relevant for the validation of numerical models.

After the experiment, the pilot ladle was subjected to natural cooling with the top lid in place. Once the ladle reached the ambient temperature, the top lid was removed. Measurement of head and bed joints after the experiment was carried out to observe any change in the joint thickness. No significant change in the bed joint thickness was observed. Table 4 presents the measurement of head joints taken before and after the experiments. It can be observed that, for test specimen NWL-02, average joint thickness increased by .24 mm (100%) compared to an increase of .08 mm (25%) for NWL-01. This significant increment in head joint thickness was expected as the test specimen NWL-02 was subjected to higher temperatures compared to NWL-01. This increase in the joint thickness can be attributed to the plastic deformation of the bricks in the working lining, primarily due to viscoplasticity. The results show that the .44 mm mean thickness of the joints became much closer for the two specimens (difference is now less than 20%). In contrast, the 1.36 mm maximum head joint opening was reduced, and the coefficient of variation also reduced to 65%. The latter value still indicates a large scatter in the head joint opening.

TABLE 4. Change in the thickness of head joints.
NWL-01 NWL-02
Before After Before After
Mean thickness of joints (mm) .32 .40 .23 .47
Standard deviation (mm) .30 .26 .27 .31
Largest joint (mm) 1.68 1.46 1.77 1.25

As discussed earlier, DIC was employed to measure displacements and strains at the steel shell and top layer of the working lining. Figure 10 shows images taken from the camera for both DIC setups and the marked area used to perform analysis. For the steel shell, the measured area was 50 mm wide and 300 mm in height from the bottom. The measured area for the top layer of the working lining was 90 mm wide and 120 mm in depth with a dry joint. The DIC analysis was done only for test specimen NWL-02. After the test, the open-source software Ncorr was used for the DIC analysis.43

Details are in the caption following the image
Digital image correlation: (A) steel shell; (B) top layer of the working lining.

DIC analysis measures the total strain on an inspected surface. Under thermomechanical loading, the total strain comprises elastic, thermal, and inelastic strain. Figure 11A shows the evolution of circumferential strain observed from DIC analysis on the steel shell. The figure also shows the expected thermal strain due to temperature rise in the steel shell (12 × 10−6°C−1) and total strain (calculated thermal strain plus measured average strain from strain gauge). The total strain observed from the DIC shows similar behavior as the calculated strain. The differences observed between the observed and calculated strain may be partly due to the hot air draft that affects the image captured by the camera, particularly when DIC values are lower.

Details are in the caption following the image
Digital image correlation (DIC) results. On steel shell: (A) circumferential strain evaluation at 250 mm height of the shell; (B) circumferential strain distribution, ε𝜃 (mm/m); (C) circumferential displacement, 𝜃 (mm). On working lining: (D) circumferential strain evaluation at 25 and 130 mm from the hot face on working lining; (E) circumferential displacement distribution (mm). r is the radial and 𝜃 is the circumferential direction.

During the test, a change in the color of the white paint was observed due to high temperatures, but the quality of the readings was validated with the calculated thermal and strain gauge strains. The observed strains and displacements (Figure 11B,C) show nonuniform distribution, with higher strains at the midheight of the shell (250 mm from the bottom) and relatively lower strains at the bottom. This difference is likely due to the boundary conditions.

For the DIC on working linings, a black metallic paint that can withstand high temperatures was used. Additionally, to minimize the effects of material glow due to high thermal radiation, two bright white lights were used in addition with the fan to reduce hot air interference. However, background noise was observed during the DIC analysis. This noise primarily affects the local strain calculation, as the light passing through hot air (near the working lining) to relatively cold air (near the camera) bends away, which creates an artificial displacement of a pixel during DIC analysis. Therefore, the strains were calculated by assuming virtual extensometers, where each end represents the average displacement of 25 × 25 pixels. Figure 11D shows the evolution of the strains in the circumferential direction, measured at 25 mm and 130 mm from the HF of the working lining brick. The values presented are the average strains observed from virtual extensometers at both bricks, as shown in Figure 11E. The strain evolution shows an increase in strain at both locations as the temperature increases due to thermal expansion. However, the strain level observed near the HF (i.e., at 25 mm from HF) is low compared to thermal strain. The thermal strain at that location should be 12 mm/m (calculated by multiplying thermal expansion with the temperature at that location), which compares to about 9 mm/m. The difference is due to the combination of thermal expansion, onset of viscoplasticity, and joint closing. However, the strain observed at 130 mm from HF is comparable with the thermal strain due to expansion at that location, 8.8 mm/m expected versus 7 mm/mm measured. This lower difference suggests less relevance of joint closing and no viscoplasticity at that location.

As mentioned earlier, the strains observed by DIC are total strains that comprise thermal, elastic, and inelastic strains. The decomposition of such strains can be found by numerical simulations using the results obtained by DIC.

3.2 Test series—UWL

In an industrial ladle, the working lining is subjected to cyclic thermal loading due to molten steel. Therefore, two additional tests were performed on the pilot ladle with the working lining used in the test specimen NWL-02 to understand the behavior of a past UWL. The reheating tests were performed when the specimen NWL-02 cooled down to ambient temperature and are denoted as UWL. In between the tests, all the linings were not disturbed. As shown in Figure 12A,B, the specimen was subjected to a similar thermal load profile as in the earlier series. The observed temperature distribution for both tests is identical. Test UWL-02 was loaded only till 1360°C due to a problem with the power supply unit. The temperature distribution at different linings is similar to the observed values in NWL-02. Moreover, in this test series, no irregular temperature distribution at 100°C due to water evaporation was observed, as all the linings had been subjected to high temperatures during the previous test.

Details are in the caption following the image
Test series—used working lining (UWL): (A) average temperature evolution observed during experiments (three locations around the circumference); (B) location and legend for installed thermocouples. Average circumferential strain evolution: (C) with respect to time; (D) with respect to hot face temperature.

Figure 12C,D shows the average mechanical strain observed in the circumferential direction through installed strain gauges on the exterior surface of the steel shell. The observed strain evolution presents a similar behavior for both test specimens. As in the case of test series NWL, similar behavior can be observed during the beginning stage of thermal loading. This is due to the closing of the joints between the bricks and linings. As the joint closes, a steady growth in strain can be observed till the dwelling period. It is interesting to observe that for this test series, the maximum strain observed (.44 mm/m) is lower compared to the strain observed in NWL-02 (.57 mm/m). This difference is due to the change in the head joint sizes that was increased after the first thermal loading (NWL-02), as shown in Table 4. Moreover, the rate of strain increase is lower compared to the NWL-02.

Figure 12C,D shows that for both test specimens in this series, no decrease in strain was observed starting at 1200°C as observed in the test series NWL. This difference in behavior between NWL and UWL can be due to an increase in head joint thickness, due to which joints are not fully closed at a higher temperature. Between the test specimens UWL-01 and UWL-02, no significant difference was observed, which suggests that after the first thermal loading (NWL-02), the refractory linings are not subjected to significant additional viscoplasticity when exposed to a similar magnitude of thermal loading. The behavior observed in this series will also help calibrate and validate numerical models. In this test series, DIC measurements were not taken.


Additional field observations were made during and after the experiments. Measurements were taken before and after the experiments to observe any change in the dimensions of the steel shell. No change in the dimension of the steel shell was observed, which indicates that the steel shell was subjected to loading in the elastic range. As shown in Figure 13, the lifting of the steel shell was observed during the experiment. For the example shown in the figure (NWL-02), the steel shell was raised by 2.5 mm. This behavior can be due to the closing of gaps between the refractory linings and the steel shell. Once the gaps are closed, due to the difference in thermal expansion between refractory linings (inside, high temperature) and steel shell (outside, low temperature), and frictional interaction between linings will lift the steel shell across the ladle cross-section.

Details are in the caption following the image
Lifting of the steel shell of pilot ladle: (A) elevation view of the steel shell indicating observed area; (B) before experiment; (C) during peak thermal load.

After the experiment, once the pilot ladle reached ambient temperature, additional observations are possible. Apart from the increase in the joint thickness (shown in Table 4), some damage was observed on the HF for the working lining. Figure 14A shows crushing of corners of the working lining bricks. As mentioned earlier, the distribution of joints is nonuniform. Therefore, such crushing was observed at locations where the joints were closed before the experiment. This damage is due to thermal expansion of the bricks and stress concentrations before viscoplastic effects can develop. Apart from such local damage, no spalling was observed near the HF.

Details are in the caption following the image
Field observations of the working lining bricks: (A) crushing of the corner near the hot face; (B) cracks near the hot face; (C) cracks in the centerline of the hot face and cold face.

Furthermore, cracks were also observed on the bricks of the working linings. Figure 14B presents an example of cracks observed in 42 bricks near the corner of the HF surface. This crack is again likely due to the concentration of compressive stresses at the corner due to higher thermal expansion, which ranged from 10 to 60 mm. Apart from such cracks, cracks in the centerline of the HF and cold face were observed in 29 bricks out of 140 working linings bricks in total. Figure 14C shows an example of such cracks, which ranged from 25 mm to 140 mm in length. This crack is likely due to stress concentrations generated by different joint closing behavior at the head joints of the bricks due to the staggering of the units. The different joint closing behavior is expected due to nonuniform joint thickness. Such cracks in the working linings bricks reduce the thermomechanical integrity of the material, which can induce further damage in the bricks when subjected to longer exposure under sustained or cyclic thermal loading. In the case of an industrial ladle with molten steel present, such crack paves way for molten steel/slag infiltration which causes further material degradation and increased corrosion. Apart from the damage observed in the working lining, no damage was observed in other refractory linings of the pilot ladle.

To confirm the viscoplastic deformation on the HF of the working lining bricks, the variation of the dimensions of selected bricks was calculated for the test specimen NWL-02. An average reduction of .28 mm in the horizontal dimension of the HF was observed. This reduction in size translates to an average viscoplastic strain of 2 mm/m on the HF of the working lining bricks. This amount of strain relates to the pressure induced due to the thermal expansion of the material and restraints imposed by joints, linings, and shell. The comparison of the observed viscoplastic strain with numerical results will be made in a following publication.


Four tests were performed under transient thermal loading, two with NWL and two with UWL of the pilot ladle. All tests were subjected to a similar rate of thermal loading. The thermal behavior observed through various thermocouples presents a similar distribution of temperatures between the refractory linings for all tests. Tests performed on specimens with NWLs show the influence of evaporation of free water around 100°C, whereas such influence was not observed for the tests with UWL.

The working lining of the pilot ladle presents heterogeneity in the distribution of dry joints primarily due to the placement of bricks in the lining. Their effect can be observed by measuring mechanical strain in the circumferential direction on the exterior side of the steel shell. The working lining with a larger mean thickness of joints demonstrates lower strain on the steel shell than the lining with a small joint thickness. The effect of viscoplasticity that becomes prominent around 1200°C in working lining bricks was also observed through strain gauges on the steel shell. DIC was used to monitor the behavior of the steel shell and working lining during the thermal loading. The results obtained show a similar response. Moreover, DIC measures the total displacement/strain on the surface. Hence, the decomposition of the thermal, elastic, or viscoplastic response of material with this measurement is complex, requiring analytical or numerical modeling.

During the thermal loading, the bricks in the working lining are subjected to thermal expansion. The joint closing leads to stress concentration due to the thermal expansion of the bricks and undergoes viscoplasticity at higher temperatures, which induces plastic deformation in bricks. Due to this effect, the dry joint thickness increases after the test. The tests performed on the UWL (consequently, working lining with plastic deformation and larger joint thickness) show different mechanical behavior compared to the NWL. The results show a lower rate of strain increase during thermal loading, and the effect of viscoplasticity in the working lining bricks was not observed for the dwell testing time.

This experimental campaign describes a novel simplified approach for the thermomechanical characterization of a steel ladle's cylindrical shell without the constraints of ladle bottom and molten steel. The results gathered from this campaign can be used for the calibration and validation of numerical modeling approaches. These experimental data give essential data regarding the influence of different material and geometric parameters on the global behavior of the pilot ladle, which is essential for the calibration of numerical models.


This work was supported by the funding scheme of the European Commission, Marie Skłodowska-Curie Actions Innovative Training Networks in the frame of the project ATHOR—Advanced THermomechanical multiscale mOdelling of Refractory linings 764987 Grant. The first, third, and sixth authors acknowledge the partial funding by FCT/MCTES through national funds (PIDDAC) under the R&D Unit Institute for Sustainability and Innovation in Structural Engineering (ISISE), under reference UIDB/04029/2020 (doi.org/10.54499/UIDB/04029/2020), and under the Associate Laboratory Advanced Production and Intelligent Systems ARISE under reference LA/P/0112/2020. This work is financed by national funds through FCT—Foundation for Science and Technology, under grant agreement 2021.05961. BD attributed to the first author.


    The authors declare no conflicts of interest.