Introduction

Sand casting is one of the most widely used methods for manufacturing complex-shaped cast components. The metal–mold interface in sand casting in the ideal case consists of a liquid metal domain on the casting side and a porous domain on the mold side containing solid sand grains and a gaseous atmosphere between the grains. In addition to the listed solid, liquid, and gaseous phases, the mold side includes a wide variety of binders and additives, which may be organic or inorganic and form a complex environment.1,2 Metalcasters striving to protect the metal–mold interface from thermally induced interaction apply different types of coatings on the interface, decreasing the complexity of the metal–mold interaction.

When the gaseous phases are not evacuated from the metal–mold interface before the solidification of the casting skin, the gases may be incorporated into the casting domain,3 contributing to casting defect formation. To prevent gas-related defects, it is essential to study the gas formation mechanism, quantify the volume of gases evolving from different sand mixtures, and characterize the gas transport phenomena in the porous sand mixtures.

The heat from the liquid metal transported to the sand mixture is responsible for the thermal decomposition of the binders following phase transformations from the solid to a liquid and gaseous state, forming various quantities of complex gaseous phases over a wide temperature interval.4

An important parameter of the gas formation mechanisms is the content of the oxidizing agents between the sand grains.5 In the absence of an oxidizing agent or limited quantities, the binder decomposes by pyrolysis. Consequently, depending on the gaseous environment between the sand grains, various degradation processes are dominant, allowing different amounts of gases to evolve from the same binder system.

The rheology and permeability of the sand mixture play an essential role in evacuating the evolved gases from the metal–mold interface.6 Transport of evolved gases over shorter distances into low-temperature mold/core regions causes condensation of the gaseous elements on the sand grain surfaces. Different gaseous mixtures from binder decomposition are reported to be either strongly7 or less sensitive8 to the temperature variation and provoking condensation, thereby influencing the total gas volume and gas pressure between sand grains.9

Temperature distribution in the sand mixture and the consumed heat to activate the degeneration of the binder are essential parameters to characterize the gas-evolving process.10 Casting simulation models depend on the accurate definition of the thermal properties of a sand mixture, including the thermal conductivity and the heat absorption capacity.

The present work aims to summarize reported investigation methods in the literature and present an investigation methodology combining thermal analyses and gas generation investigation in sand mixtures produced under industrial conditions.

Thermal analyses focusing on the decomposition process of binders used in foundry applications are reported by Kmita et al.11 and Grabowska et al.12,13,14 These authors investigate polymerized binder particles combining thermogravimetric analyses (TGA), differential thermogravimetric analyses (DTGA), and differential scanning calorimetry (DSC) for the determination of kinetic parameters of the binder decomposition and the activation energy consumed for decomposition. Other works that have focused on gas evolution measurements from TG results have focused on converting the mass loss data to volumetric data using either the ideal gas law or by simply characterizing the different gases generated from the decomposing binder using mass spectrometry.15,16,17

Thermal analyses based on TG and DSC measurements studying the binder decomposition in real sand mixtures were reported by Bargaoui et al.18 connecting the thermal decomposition of the binders to the thermomechanical behavior of sand mixtures, including tensile and creep properties. A novel application of the Fourier thermal analyses (FTA) was reported by Svidró et al.19 based on temperature measurements in a spherical sand mixture domain. The obtained degradation characteristics by the authors were successfully verified using TGA analyses.

Besides the foundry practice, analytical organic chemistry involves various advanced methods for thermally induced gas generation analyses. Advances in this field were recently reported by Risoluti et al.20

An early attempt to measure core gas evolution from industrial-grade sand core mixtures was reported using a laboratory instrument named COGAS equipped with a melting furnace containing either aluminum or cast iron alloys.21 Evolved gas from samples immersed in the liquid metal is captured, and the water is displaced in a measuring tube to a reservoir. A balance measures the displaced water and gives input for calculating the gas volume. A disadvantage of the COGAS method is that the evolved gases pass through cold spots until collected in the measuring tube, and evolved hot gases undergo condensation. This equipment aimed to control the core gas evolution reproducibility in foundry production by introducing arbitrarily sized/shaped parts of an actual foundry core.

Another attempt to measure the gas pressure from core gas evolution was reported using a laboratory instrument, the Ridsdale gas determinator, by Zhang et al.22 including investigation of furan hot box cores, phenolic shell-molding cores, and phenolic isocyanate–urethane cold box cores with different percentages of resins.

Several authors reported studies on gas evolving from real foundry sand mixtures using specially designed measuring methods.9,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38 Druschitz et al.,32 Scraber et al.,33 and Winardi et al.9,34,35,36,37,38 used the same specially designed instrument for measuring gas generation at binder decomposition in foundry sand mixtures. The setup consisted of a furnace with molten liquid metal into which molding material samples of real sand mixtures were immersed. Aluminum, magnesium, and cast iron alloys were used to provoke various heating rates in the investigated samples. Heated samples of evolving gases due to binder decomposition were collected in a displacement container, displacing oil from the container into a beaker. Since the gases evolved to pass through a heated route, the authors report that the evolved gases do not undergo any condensation. The displaced mass of the oil was measured, and from the density of the oil, the volume of the gases evolved was calculated. The authors also tried to calculate and interpret the gas pressure in the sand grain interstices based on the estimated gas volume rate. Cylindrical samples were 28.5 mm in diameter and 50.8 mm in height. The samples used by those authors were not equipped with thermocouples in earlier reported publications but were introduced in later experiments. Gas generation due to thermal decomposition was reported from green sand samples, shell molds, cores made of phenolic urethane cold box (PUCB) binder, epoxy acrylic binder, and polyurethane cold box binders.

Samuel et al.28 reported a method developed to measure the gases evolving from foundry mold and cores. The setup was based on the observation of a direct liquid metal displacement caused by the thermal degradation of the sand mixture in a neutral gaseous environment. Polyurethane cold box and polyurethane no-bake systems were studied for different heating rates. The evolved gases were expressed as molecular weight and modeled with piecewise polynomial equations. The gas temperature reported in the investigation was considered equal to the temperature of the displaced liquid metal.

A specially designed instrument to investigate foundry sand mixtures where the heat acting on the binder decomposition generated in a furnace was reported by Mocek et al.39 The instrument was a manometer-type measurement device. The samples are placed in a tubular silite furnace and are heated. The evolved gases are transported through small tubes to pellistor sensors and lambda probes measuring hydrogen and oxygen content. Generated gas from the binder decomposition passes through small tubes, displacing the measuring liquid. Using this setup, the gas evolution rate and gas volume were measured. However, the authors do not mention the heating rate for the molding materials. The location of the thermoelement that measures the temperature of the mold is not specified. The authors studied the evolution of gas from molding sand mixed with Furan resin. An important observation was that the highest amount of gases evacuated during the first 60 seconds, where the main gaseous element is oxygen, shifting with hydrogen after 200 seconds. The authors argue that the fast consumption of oxygen in the granular interspace causes methane formation during binder decomposition, which readily decomposes to graphite and hydrogen (H2). Such graphite formed during the decomposition process was found on the instrument wall. Recently, Kirchebner et al.40 measured the gas release from cores prepared with inorganic binders using a custom-made experimental setup. Their work employed an induction analysis furnace to develop an experimental setup that could measure gas permeability and gas release. The heating rate of the sample in their measurement was 50 °C/min.

As an outcome of industrial needs concerning gas-related defects, the authors intend to study thermally induced gas generation phenomena in sand mixtures by combining a direct thermal analysis method based on Fourier thermal analyses (FTA) and gas evolution measurements in a specially designed device developed in the university laboratory. The thermal analysis algorithm based on FTA was described earlier by the authors.19 The outcome of the measurement method will be an accurate quantification of the rate and volume of the generated gases and the effect of decomposing binders on the total heat absorbed by the molds and cores.

Based on the literature reported above, the gas generation and evacuation from the intergranular space in mold mixtures is considered complex. The conducted investigation in this paper does not include the investigation of the decomposition mechanism, i.e., whether the binder decomposition was through direct gas formation (combustion) or pyrolysis. Also, an elemental investigation of the gas content was not performed.

Experimental Procedure

The experimental setup developed in the present work for the thermal analysis and measurement of the generated gas volume, called thermal analysis and gas generation (TAGG), involves several components. An illustration of the setup is shown in Figure 1.

Figure 1
figure 1

Experimental setup of the thermal analysis and gas generation (TAGG) measurements.

Full size image

The molding material sample (a) (Figure 1) has a spherical shape (Ø 40 mm) with the scope to form a one-dimensional geometrical domain when a polar spherical coordinate system is considered. The numerical considerations for heat conduction in a polar spherical coordinate system have been introduced by the authors.19 The spherical sand sample was immersed into a molten metal domain (b) (Figure 1). The geometrical centrum of the spherical sample is 90 mm below the molten bath surface. The results reported in this work were obtained by immersing the spherical sand samples in cast iron melt, held in a holding furnace at a temperature of 1315 °C.

Two thermocouples (TC) of N type are mounted in the spherical sand sample. The soldering point of TC1, where the signal for temperature variation is registered, is placed in the geometrical centrum of the sphere and TC2, 10 mm displaced laterally from TC1 (Figure 2). Thermocouples are protected from liquid metal by vertical quartz tubes (Figure 2). The tube protecting TC1 (Øouter= 9 mm, Øinner= 7 mm) has an additional role, which is to evacuate the generated gases from the sand sample. The tube protecting TC2 (Øouter =5 mm, Øinner = 3 mm) is sealed with a heat-resistant powder to ensure the gas evacuation from the sample exclusively through the central tube, protecting TC1. The thermocouples and their protecting pipes are mounted in the sample holding unit (c) (Figure 1). The thermocouples, quartz tubes, and the molding material sample are held together with a specially designed holder mechanically, including small plates and screws. Furthermore, the central tube is connected to a horizontal quartz pipe (Øouter = 7mm, Øinner = 5mm, (d) (Figure 1)) which in turn connects the sample unit to a resistance-heated pipeline (e) (Figure 1). The role of the pipeline heating unit is to avoid temperature decrease of the evacuating gases and consequent gas condensation, also reported in the literature. The collected gases are transported through the heated pipeline (170 °C) into a liquid displacement beaker (f) (Figure 1), containing preheated rapeseed oil. The oil is preheated to 140 °C before the sand sample is immersed into the molten liquid to avoid changes in the oil's viscosity and maintain a stable flowability between the liquid displacement beaker and the oil sampling beaker placed on a digital scale. The mass of the displaced oil is measured using a precision balance with a logging frequency of 5 readings per second (g) (Figure 1). Each sample measured is immersed in the melt for approximately 300 seconds. The samples measured were at room temperature during the immersion process. Signals from the thermocouples and the digital balance are collected in a data acquisition device with a logging frequency of 5 readings per second. The volume of gases evolved is calculated using the known density values of the oil. Since the maintained oil temperature during the measurement was close to 140 °C, a constant value of density 0.85 g/cm3 was used for the calculation of the volume of the gases evolved.41

Figure 2
figure 2

The test sample geometry for the TAGG experiment is19 immersed in cast iron melt. TC1 measures the temperature change in the geometrical centrum of the sample. TC2 is displaced 10 mm from TC1.

Full size image

Three sand mold samples were prepared from subangular-shaped quartz-based lake sand with an average grain size of 0,26 mm. The sand core was bound with a 1,2 wt% furan resin (furfuryl alcohol-based) and a sulfonic acid-based catalyst with 40 wt% related to the weight of the binder. The sand mixture was rammed manually and allowed to cure for 24 hours before mounting into the sample holding unit.

Results and Discussion

Measured Raw Temperature and Generated Gas

Registered raw temperature curves and calculated gas generation curves are presented in Figure 3. Three samples (A, B, and C) produced from the same batch of sand mixture were immersed in the liquid cast iron melt. The properties of the investigated samples are provided in Table 1. The variation of the sample weight and calculated density of the samples were negligible which resulted in similar heating curves registered for all the inner and outer thermocouples. Relatively high reproducibility was obtained for all samples' accumulated gas evacuated up to 40 seconds from the immersion moment. After 40 seconds, one of the samples (A) was evacuating a lower volume of gases, ending up in a lower accumulated volume of 1060 (cm3) in comparison with samples B and C. No gaseous bubbles escape from the immersed core into the liquid sample were observed other than through the described gas evacuation route.

Figure 3
figure 3

Registered heating temperatures and gas release in samples A, B, and C.

Full size image
Table 1 Weight, Volume, Apparent Density of Sand Samples, and Generated Gas Volume
Full size table

In a previous work, the authors demonstrated the role of the sand mixture density on the gas permeability when gaseous phases are transported between the sand grains.6 A variation of the local density values along the radius of the spherical sample could be caused by the hand ramming process applied during the production of the samples. In addition, the local inhomogeneity of the granular interspace can cause regional variation in access to oxidizing elements, which may delimit the binder decomposition by combustion, promoting decomposition by pyrolysis5 instead and causing the differences in the evaluated gas volumes in the samples.

Thermal Analysis

Fourier thermal analyses (FTA) is an applied analysis method initially developed for calculation of the released latent heat of solidification, based on analyses of the temperature differences registered in the solidifying cast alloy42 An extension of the FTA method was developed by the authors19 applying the technique on the heated sand mold domain for calculation of the heat absorption at decomposition of organic binders. Both thermal analysis methods are based on the numerical treatment of registered temperatures at solidification and mold heating.

The dominant heat transport mode in the sand mold, conduction is described by the heat conduction equation:

$$\rho {c}_{\text{p}}\frac{\partial T}{\partial t}=\nabla \left(k\nabla T\right)-{\dot{q}}_{\text{abs}}$$
(1)

The last term in Eqn. 1, \({\dot{q}}_{\text{abs}}\) is the volumetric time-dependent heat absorption, corresponding to the heat consumed at the binder decomposition.

Considering strictly constant material properties, division of the right side of the equation by \(\rho {c}_{\text{p}}\) , rearrangement, and substitution of \(\frac{\partial T}{\partial t}\) by \(\dot{T}\) and \({\dot{q}}_{\text{abs}}\) by \({q}_{\text{abs}}\) the heat absorption rate during the core binder decomposition process can be calculated as a function of time t.

$${q}_{\text{abs}}= {c}_{\text{V}}\alpha {\nabla }^{2}T- {c}_{\text{V}}\dot{T}$$
(2)

where \({c}_{\text{V}}\) is the volumetric heat capacity; α, the thermal diffusivity; \({\nabla }^{2}T\) is the Laplace operator characterizing the temperature differences in the domain and \(\dot{T}\) represents the local heating rate.

The product of the thermal diffusivity, α and the Laplace operator \({\nabla }^{2}T\) , in Eqn. 3, defines the zero line \({Z}_{\text{F}}\) of the temperature heating process. The zero line corresponds to the expected heating rate of the sand mixture in the hypothetical case of lacking binder decomposition.

$${Z}_{\text{F}}= \alpha {\nabla }^{2}T$$
(3)

It is important to mention that the finite difference approximation of the Laplace operator used in this calculation takes into consideration the temperature differences between the registered temperatures in the central TC1 and lateral TC2 thermocouples. In the following, the interpretation of the results of the present study is valid for the internal spherical sand domain delimited by the lateral thermocouple (TC2).

An iterative solution of the involved equations described in detail in the previous paper42 allows the calculation of the time-dependent heat absorption \({q}_{\text{abs}}\) and the total absorbed heat \(L\) , as the result of binder decomposition.

Calculated zero line \({Z}_{\text{F}}\) and the registered heating rate \(\dot{T}\) in the central thermocouple (TC1) of an experimental measurement (sample C) are shown in Figure 4. A cumulative curve of the calculated absorbed heat during the gas generation process \({f}_{\text{a}}\) is shown in the same figure where the surface area between the registered heating rate and the calculated zero curve is proportional to the time-dependent heat absorption \({q}_{\text{abs}}\). The investigated gas generation interval starts at \({T}_{\text{StartGG}}\)= 88 °C and ends at \({T}_{\text{EndGG}}\)= 735 °C. The volumetric heat capacity at the beginning \({c}_{\text{V}\_\text{StartGG}}=\) 12,24 MJ m−3 K−1. and at the end of the gas generation interval \({c}_{\text{V}\_\text{EndGG}}=\) 16,3 MJ m−3 K−1 are obtained from the literature.43

Figure 4
figure 4

Thermal analyses of heat absorption.

Full size image

Integrating the absorbed heat over the gas generation process between \({t}_{\text{b}}\) and \({t}_{\text{e}}\), the total absorbed heat \(L\) was found equal to 650 kJ kg−1.

The described procedure determines the overall absorbed heat during the gas generation process, not distinguishing the contributions from different reactions, binder decomposition or vaporization of humidity. Local minima in the heating rate curve indicate triggering points for different reactions during the decomposition.

Gas Generation Rate Analysis

The accumulated gas evacuated from the test sample and the calculated rate of the gas evolution over the experimental time interval is shown in Figure 5.

Figure 5
figure 5

Measured rate and volume of gas generation as a function of time.

Full size image

The thermal analysis investigation indicates a complex origin of the evacuated gases. Gas expansion of the intergranular atmosphere and the decomposition of the water and the core binders are supposed to contribute to the total evacuated gas volume. A significant rate of gas release is observed at the beginning of the measurement process, in agreement with previous literature reports. More than 80% of the total gas volume is evacuated during the first 65 seconds of the measurement process.

In an attempt to characterize the gas evolution in the analyzed sample, the start and the end time of the heat absorption process are indicated separately for the outer domain of the sample (between the outer sand surface and the position of TC2) and the inner domain of the sample (between TC 1 and TC2).

An immediate start of the gas transport from the sample surface towards the centrum of the sample is observed at time 0. The peak of the released gas at the measurement's beginning is attributed to the fast expansion of the intergranular gaseous environment and the instant vaporization of the moisture present in the sample, driven by the rapid heating of the outer sand mixture layer. The transition from the expansion of the intergranular gaseous environment to the gas release due to the binder decomposition cannot be identified accurately since the surface temperature of the sample is not measured. A raw approximation of the intergranular volume in the spherical sample considering a 40% porosity would result in a volume equals 13,4 cm3 of void/air. Heating the gaseous environment between the sand grains from ambient temperature to 88 °C, the expanded gas volume is negligible (16 cm3) compared to the total evacuated gas volume during the measurement (1126 cm3). Gas generation in the outer domain ends at t = 65 seconds, when \({T}_{\text{EndGG}}\)= 735 °C is measured on TC2.

Gas generation in the inner domain of the sample starts at t = 23 seconds, when at \({T}_{\text{StartGG}}\)= 88 °C is measured on TC2 and is completed at t = 114 seconds, when \({T}_{\text{EndGG}}\)= 735 °C is measured on TC1.

The gas generation in the outer and internal domains overlap, making the interpretation of the gas evaluation as a function of temperature difficult, since the measured heating rate and the calculated zero line become equal at t = 114 seconds. The evacuated gases after this moment are not supplied by any decomposition process in the whole sand domain. Thanks to the temperature observation in the geometrical centrum of the sample, the last 72 cm3 gas of the total 1126 cm3 evacuated amount is attributed exclusively to the volume expansion of the gaseous environment between the sand grains due to the forward increase of the sand mixture temperature until the end of the measurement.

Conclusion

The experimental method described in this paper helps estimate the amount of gases a particular type of foundry mixture could produce during the casting process. During the experimental procedure, the cores undergo a thermal shock in a process very similar to the actual foundry conditions. The rate of gas evolution as a function of time and the rate of heat absorption as a function of temperature is computed from the measured data. The combination of temperature measurement and the gas release calculation allow to approximate separately the volume of gases released by the binder decomposition and the gas release generated due to the thermally induced gas expansion. An important outcome of the performed experiment is the observation of the multiple sources of the gas generation in the intergranular sand domain, including the decomposition of various components such as binders and thermally driven volume expansion of the gases present between the sand grains. They overlap with each other during the measurement process indicating the complexity of the gas generation measurement.

Although the decomposition process as a function of temperature is approximate in nature (due to the lack of temperature measurement at the point of contact between the liquid metal and the sand sample), the total gas evolution is accurate. This type of information is critical for foundries that determine the type and amount of binder system to use in their production processes. Apart from the results of gas generation, the method aids in performing thermal analysis to estimate thermophysical properties of foundry mixtures, such as the absorbed heat at gas generation, including the binder decomposition and mobilization of the gaseous environment between the sand grains. The results obtained from this experimental setup could be used directly to create models and simulate the gas evolution process.

The results obtained from the method are consistent with very little scatter; hence, the experimental procedure is repeatable. Since all the relevant parameters are well controlled in the experimental setup, the reliability of the results is also high. This setup can be used to study different foundry mixtures with varying aggregate types and binder systems and to serve as reference measurements for numerical simulation of thermally driven gas generation and gas transport phenomena.