DEGREE PROJECT IN MATERIALS SCIENCE AND ENGINEERING, SECOND CYCLE, 30 CREDITS
STOCKHOLM, SWEDEN 2018
The Effect of Cooling Rate and
Solidification Time on the Ultimate
Tensile Strength of Grey cast iron
DINESH SUNDARAM
KTH ROYAL INSTITUTE OF TECHNOLOGY
SCHOOL OF INDUSTRIAL ENGINEERING AND MANAGEMENT
i ABSTRACT
Tensile strength modelling is usually done to predict the mechanical properties of lamellar graphite iron considering microstructural features. This work attempts to create a simplified model incorporating cooling rate and solidification time without considering the microstructural features. This model will save time and cost in industry with the presence of a commercially available software such as Magmasoft which simulates solidification time and cooling rate. A plate model was designed for this purpose as the test geometry to create variation in solidification time and cooling rate. By altering fraction solid, thermal conductivity, specific heat capacityin Magmasoft, a good fit was created between simulated cooling curve and experimental cooling curves. The experimental UTS data of samples from three moulds were investigated and a regression model was created using statistics tool Minitab. The effect of solidification time and alloying on the graphite length Lmax was studied for twelve samples from each mould. Quantification of the effect of cooling rate and alloying on the pearlitic properties of grey iron like matrix microhardness, pearlite inter-lamellar spacing was also investigated in this work.
The developed model has sixty three percent correlation and explains UTS well in terms of solidification time and cooling rate. Microhardness measurements show that there is an almost linear relationship between the cooling rate and microhardness of the matrix structure.
Microhardness data also provides an overview of the pearlite fineness/interlamellar spacing.
Analysis of the outliers showed that the presence of free ferrite on a fully pearlitic structure reduces the UTS significantly. Comparison of the regression model obtained from this work with previous work showed that, there is a reduction in the predicted strength with this model.
The effort to identify the reason for this reduction was not successful and needs further investigation. Pearlite inter-lamellar spacing measurement was not accurate. The relationship between pearlite interlamellar spacing and matrix microhardness needs to be investigated in the future using a better technique for pearlite spacing measurement. This will be useful to understand the effect of cooling rate on pearlite spacing and consequently on the UTS of grey cast iron.
Keywords : Lamellar graphite iron, tensile properties, pearlitic properties, cooling rate, solidification time, Magmasoft.
ii TABLE OF CONTENTS
1 INTRODUCTION ... 1
1.1 Background ... 1
1.2 Goal ... 1
1.3 Social and ethical aspects ... 1
2 LITERATURE REVIEW ... 2
2.1 Cast Iron – Structure and Constitution ... 2
2.2 Carbon equivalent ... 2
2.3 Solidification time & Cooling rate ... 4
2.4 Structure and mechanical properties ... 5
2.5 Tensile strength modelling ... 9
2.6 Microhardness of Cast iron ... 11
3 METHODS AND EXPERIMENTS ... 13
3.1 Simulation in Magmasoft ... 13
3.1.1 Parameters in Magma ... 13
3.2 Test geometry & Castings ... 14
3.3 Naming and Selection of the samples ... 17
3.4 Tensile tests ... 20
3.5 Light Optical microscopy... 20
3.5.1 Graphite morphology and Lmax ... 20
3.5.2 Pearlite spacing ... 21
3.6 Hardness measurements ... 22
4 RESULTS ... 24
4.1 Simulation and cooling curves ... 24
4.1.1 Thermocouples in the casting ... 24
4.1.2 Thermocouples in the sand ... 28
4.2 Graphite morphology and Length ... 28
4.2.1 Maximum graphite length Lmax vs ts ... 28
4.2.2 Graphite morphology ... 29
4.2.3 Pearlite inter-lamellar spacing measurement – Circular test line method ... 31
4.3 Micro-hardness ... 31
4.4 UTS modelling ... 33
4.4.1 Effect of solidification time and cooling rate on the UTS ... 33
iii
4.4.2 Effect of alloying on the UTS ... 35
4.5 Regression model ... 35
4.5.1 Tensile strength ... 35
4.5.2 Graphite length and solidification time ... 36
5 DISCUSSION ... 38
5.1 Test of the model ... 38
5.2 Comparison of the model obtained from previous work ... 38
5.3 Outlier analysis ... 39
5.3.1 Graphite length analysis and comparison ... 41
5.3.2 Microhardness comparison ... 43
5.3.3 Type D graphite and ferrite formation... 44
5.3.4 Summary – outlier analysis ... 45
6 CONCLUSIONS ... 47
7 FUTURE WORK ... 47
8 ACKNOWLEDEMENTS ... 48
9 REFERENCES ... 49 APPENDIX A : Simulation parameters
APPENDIX B : Hardness measurements
APPENDIX C : Effect of alloing elements on the UTS APPENDIX D : Fracture images
1
1 INTRODUCTION
The purpose, background and goal of this thesis project is discussed in this introductory chapter.
1.1 Background
This thesis work was performed for Volvo AB, Gothenburg within the grey iron material development project. This work focuses on creating a model for the prediction of the tensile strength of grey cast iron based on statistical data obtained from experiments. The major focus is on cooling rate and its influence on the pearlitic transformation.
The UTS was evaluated for over 100 test bars. Graphite length and micro-hardness measurements were determined for 36 samples from 3 moulds. The pearlite spacing measurement was done for 6 samples from only one mould.
1.2 Goal
The goal of this thesis work is to establish an improved model for UTS prediction where cooling rate (dT/dt) at 750 ̊C appears to be an independent factor. With such a model, we could potentially be able to calculate and predict the mechanical properties of cast components with complex structures. It could save a lot of time and cost.
A secondary goal was to quantify the effect of cooling rate and alloying on the pearlite properties like the microhardness and inter-lamellar spacing of pearlite.
1.3 Social and ethical aspects
This thesis work is intended to create a model for tensile strength of cast structures. Such a model can save time and cost for the industry where actual mechanical tests could be replaced by simulated strength predictions for complex cast components. An attempt for a successful model is done in this thesis work. The social and ethical aspects of the outcome of the results cannot be predicted at this moment as the results are to be implemented by the industry. Therefore, the social and ethical aspects are considered irrelevant for this thesis work and are not discussed.
2
2 LITERATURE REVIEW
In this chapter, carbon equivalent, solidification time, cooling rate, structure and mechanical properties of grey iron, microhardness measurement and tensile strength modelling is reviewed in detail with the help of relevant literature available.
2.1 Cast Iron – Structure and Constitution
Cast iron is an iron carbon alloy with carbon≥ 2.11 wt% but also containing other important elements such as silicon, manganese, sulphur and phosphorous. The addition of these elements creates significant changes in the structure and properties. Grey cast iron has some important properties that makes it one of the most widely used cast material. It has good vibration damping because of the dispersion of graphite. It has self-lubricating qualities and hence is machinable. The price is low, and it is easy to cast even complicated structures. Its thermal conductivity is good because of the presence of graphite. Due to these reasons it is often used in the automotive industry for components like the cylinder heads, cylinder blocks, brake discs etc. It is also used for machine tool bases. Cast iron is classified into five basic types based on the graphite morphology.
• Grey cast iron (graphite in the form of flakes).
• Ductile iron/Nodular cast iron (graphite present in the form of spheres).
• Compacted graphite iron (graphite shape resembles corals).
• Malleable cast iron (graphite present in the form of nuggets).
• White iron (contains metastable cementite instead of graphite).
2.2 Carbon equivalent
The composition of cast iron is evaluated using the carbon equivalent value. Commercial irons take the carbon equivalent into consideration more than the carbon content alone. The carbon equivalent formula varies for different countries. In Sweden carbon equivalent is given by the formula shown in equation (1).
With this value we can determine whether the alloy is hypo/hyper eutectic. The carbon equivalent hence also affects the amount of dendrite framework. The dendrite framework will be smaller if the silicon and carbon contents are high. Hence the carbon equivalent is inversely proportional to the strength of the iron.
For hypoeutectic irons the liquidus temperature is affected by the addition of silicon and phosphorous. See equation (2).
𝐶𝑒𝑞 = 𝐶 % + 𝑆𝑖 %/4 + 𝑃 % /2 (1)
3
Figure 1. Iron carbon phase diagram
1669°𝐶 – 124 (𝐶𝑎𝑟𝑏𝑜𝑛 % + 𝑃ℎ𝑜𝑠𝑝ℎ𝑜𝑟𝑢𝑠%/2 + 𝑆𝑖𝑙𝑖𝑐𝑜𝑛%/4) (2) Figure 1 shows the iron carbon phase diagram. From liquidus the first phase that is formed
from the melt is austenite which is a solid solution of carbon in γ-iron and it crystallizes in the form of dendrites. The length and the pattern of the dendrites depend on the temperature gradients. Until the eutectic temperature the austenite dendrites grow. From the eutectic temperature graphite starts to form. Solidification starts from several nucleation points and deposition of graphite and austenite occurs simultaneously as eutectic cells. At 750 ̊C the eutectoid transformation begins where the austenite is then transformed into pearlite/ferrite or a mixture of both. The pearlite formation after the eutectoid transformation temperature at 750 ̊C is also very important to analyze because the matrix of the cast iron provides the strength for the material. The matrix which is made primarily of the pearlite is important to study because industrial cast iron has almost no ferrite and it is primarily pearlite. Certain alloy elements (Mn, Cr, Cu and Sn) favors the formation of pearlite instead of ferrite.
4
The form and distribution of graphite depends on the nucleation, cooling rate, melting temperature etc. The different forms of graphite according to ASTM standards are shown in Figure 2.
Figure 2. Graphite morphology (AFS-ASTM)
Depending on the type of graphite formed, the properties of the cast iron vary considerably.
Hence it is very important to study the shape, size and distribution of graphite. In Figure 2, type A graphite represents uniform distribution with random orientation. Type B shows rosette grouping of graphite and random orientation. Type C represents the hypereutectic alloy where primary graphite is often formed as superimposed flakes. Type D represents the undercooled graphite where the cooling rate is high. Type E shows a typical hypoeutectic type alloy. Both the type D & E show inter-dendritic segregation during the solidification.
2.3 Solidification time & Cooling rate
The solidification time and cooling rate are important parameters that determine the properties of grey cast iron. In general, the mold walls provide a chilling effect to the molten material and hence in complex cast structures there are regions (close to the mold walls) which have a higher cooling rate. It is known that the higher cooling rates provide greater undercooling and help in increasing the number of nuclei formed during solidification. If there are more nuclei, the structure is finer and hence possesses better properties. On the other hand, if the undercooling is lesser and time for solidification is longer, the nuclei formed will be favored for growth and hence coarser grains are formed. The solidification time influences not only the size of the graphite but also the type of graphite formed. It determines if a given iron becomes grey, mottled or white and determines the type of graphite.
5
The cooling rate which is studied with the help of cooling curves also plays an important role in the eutectoid transformation. It determines apart the mechanical properties and the amount of free ferrite formed, with decreased cooling rates we can find increased ferrite. If the cooling rate is sufficiently fast, a fully pearlitic structure can be obtained. If the cooling rate is extremely fast e.g. by oil-quenching the austenite to pearlite/ferrite transformation does not happen and martensite is formed.
2.4 Structure and mechanical properties
In a hypoeutectic cast iron melt the austenite is the first phase to nucleate. The next phase to form is graphite after reaching the eutectic composition. Graphite grows in cooperation with the austenite dendrites. Later during the solid-state transformation austenite transforms to pearlite or ferrite or a mixture of both. Higher cooling rate refines the dendritic structure of austenite and the eutectic cell size. This increases the mechanical properties. On the other hand, increasing carbon content reduces the primary phase formation which affects the mechanical properties negatively [3].
Riveria et al [4] speculated about the nucleation event in hypoeutectic alloys. In their work they concluded that there is no separate nucleation event for the eutectic cells and these eutectic cells nucleate on or near the regions of the primary austenite dendrites. This mechanism that they propose is in accordance with the other works done by Ruff and Wallace [5] and Dioszegi et al [6]. Dawson et al [7] reported that the length of graphite which act as a stress concentration factor will be reduced if the eutectic cell count is increased. This reduction in the stress concentration factor will increase the tensile strength. L. Collini et al [8] reported the effect of graphite morphology on the tensile strength of grey cast iron.
Figure 3. UTS values for the castings with done in three different foundries [7].
6
The graphite morphology was reported for the castings in these three different foundries. They show that the tensile strength was the largest when the primary dendrites were large and the lowest tensile strength was recorded for which the dendrites were short and globular. The mechanical properties also vary considerably based on the pearlite content. A fully pearlitic structure will have improved mechanical properties. In a fully pearlitic structure the finer the pearlite the better the mechanical properties. Figure 3 shows the results of the UTS from samples of three different foundries.
Alloying elements have a remarkable effect on the UTS of grey iron. W. Xu et al [9] studied the effect of alloying elements on the mechanical properties in their experiments and reported the following results.
Figure 4. Effect of Mo on the UTS of grey iron
In this study apart from the UTS, transverse fracture stress (TFS) was also measured for the samples. Cylindrical samples of dimensions 120 mm × 30 mm were used for UTS and 350 mm
× 30 mm were used for the determination of TFS.
Figure 5. Effect of Mn on the UTS of grey iron [9]
7
Figure 4 and 5 shows the effect of Mo and Mn on the UTS. From the results it could be seen that addition of Mn upto (1 wt.%) increases the UTS and after that has a slightly decreasing trend. Mo also increases the UTS and 0.7 wt.% Mo gives the highest UTS. In this work the authors also bring up interesting results about the matrix structure. They report that matrix structure with a mixture of retained austenite and bainitic ferrite (ausferrite) shows significant increase in the mechanical properties of grey iron with 100 % ausferritic structure showed microstructure and mechanical properties like austempered grey iron. Figure 6 shows the results obtained.
Figure 6. Effect of ausferrite volume fraction on the UTS of grey iron [9].
Henrik et al [10] studied the effect of the cooling rate and alloying on the matrix structure formation in pearlitic grey iron. They conclude that Cr has higher influence than Cu on the pearlite inter lamellar spacing refinement. The results of the pearlite spacing is shown in Figure 7.
Table 1. Chemical composition of the melts analyzed [10].
Alloy C Si Mn P S Cr Cu CE
0 3.13 1.72 0.75 0.027 0.026 0.085 0.07 3.71
1 3.12 1.78 0.74 0.026 0.024 0.084 0.6 3.72
2 3.09 1.76 0.81 0.033 0.019 0.338 0.7 3.69
2(Extra melt 1) 3.06 1.84 0.72 0.026 0.025 0.332 0.75 3.68 2(Extra melt 2) 3.08 1.8 0.81 0.038 0.025 0.319 0.75 3.69
3 3.08 1.85 0.71 0.027 0.024 0.6 0.75 3.71
4 3.15 1.71 0.82 0.033 0.017 0.332 0.37 3.73
5 2.93 1.86 0.79 0.038 0.019 0.342 1.11 3.56
8
Figure 7. Pearlite lamellar spacing as a function of the simulated cooling rate.
It is also reported that the inter-lamellar distance is constant for cooling rates with higher magnitude than 0.75 °C/s. It is also important to note that the relationship between the
eutectoid cooling rates and the interlamellar spacing is not linear. Apart from the alloying and eutectoid cooling rate, the thickness of the casting plays an important role in the pearlite lamellar distance.
Figure 8. Inter-lamellar spacing and thickness of casting
The thickness of casting influences the lamellar spacing. Since pearlite formation is dependent on the diffusion of the carbon atoms, thicker alloys provide greater time for diffusion and
9
therefore longer distance for carbon atoms to travel. Hence thicker alloys have larger inter- lamellar distances as shown in Figure 8.
2.5 Tensile strength modelling
Mechanical properties of cast iron are represented by its tensile strength. Since cast iron is brittle, its yield strength could be considered equal to its ultimate tensile strength. Most of the microstructural features have an impact on the tensile strength of grey cast iron. Hence modeling of tensile strength by previous works done have all included most of the microstructural features.
Ruff and Wallace [11] produced the below equation for grey cast iron with carbon equivalent values between 3.85 – 4.45.
𝜎
𝑈𝑇𝑆 = 11.248 + 179.29𝐷𝐴𝑆 – 4949.08𝐷𝑠 + 2.74𝐿𝛾 + 90.74𝐴𝛾 +77.71𝑓𝑃𝑒𝑎𝑟𝑙𝑖𝑡𝑒 – 30.18𝑓𝑔𝑟𝐷 + 0.79𝑁𝐸 + 287.74𝑑𝑖𝑟𝛾 – 9.79 ∗ 𝐿𝑔𝑟 − 5.45𝐶𝐸 (3) Here the DAS is the dendrite arm spacing, the section size of cylindrical sample Ds, the average
dendrite length Lγ, the dendrite interaction area Aγ, pearlite content fpearlite, the fraction of D type graphite fgrD, the eutectic cell count NE, dendrite directionality dirγ, average flake length Lgr and carbon equivalence, CE.
Junming [12] suggested a model that links brinell hardness and UTS according to the following equations.
𝐻𝐵 = 421.72𝐷𝑒−0.228 + 123.03𝑓𝑐𝑎𝑟𝑏 + 2.5 𝛌 𝑝−0.5 + 90.68𝑓𝑎𝑢𝑠 – 14.74 (4)
𝜎
𝑈𝑇𝑆 = 2.273𝐻𝐵 – 322.7 (5) where De is the eutectic cell diameter, fcarb the fraction of carbide, λp the pearlite lamellar spacing and faus is the fraction of primary austenite. There are more such models created by authors considering different parameters like the stress intensity factor from Griffiths fracture theory. A review of such tensile strength models and a generic model was given by Vasilios et al. [13]. These models show that the tensile strength of cast iron can be predicted.Schmidt [14] in his work concluded that tensile strength is mainly dependent on the solidification time and the resulting graphite fineness. A linear relationship between the two parameters could be seen evident which is shown in Figure 9. Cooling rate is the main factor for the pearlitic transformation and the resulting spacing of pearlite lamellae. Hardness which depends on the pearlite lamellae fineness, is mainly dependent on the cooling rate.
10
Figure 9. tensile strength as a function of solidification time [14].
The length of graphite apart from the solidification time is also dependent on alloying elements.
Cu has a graphitizing effect at higher concentrations and it increases the graphite length.
Addition of Cr on the other hand decreases the length of graphite [15]. It is important to note that although the length of graphite is affected by Cu and Cr content, the difference in thickness and hence difference in solidification time affects the length of graphite. Figure 10 shows how the length of graphite is affected by the composition and the thickness. The length of graphite is also affected by the pouring temperature and inoculation. In their work, these factors were kept constant by using the same inoculants and the same pouring temperature. The composition of different alloys is given in Table 1.
Figure 10. Effect of composition and thickness on the length of graphite.
Ruff’s [11] model considers most microstructural features including the length of graphite Lgr
and shows that with a linear relationship the UTS can be modelled. In the model by Junming [12], the pearlitic lamellar spacing and the eutectic cell diameter shows a power law relationship with the hardness. The hardness shows a linear relationship with the UTS.
11
Taha et al [17] published their work about the relationship between cooling rate, structure and properties of hypoeutectic cast iron. They show that low cooling rate results in decreased UTS (see Figure 11).
Figure 11. Variation of UTS with cooling rate for as-cast specimens[17].
2.6 Microhardness of Cast iron
The different phases of grey cast iron contribute to the tensile properties of grey cast iron.
Dioszegi et al [16] have reported in their research report about the microhardness measurements of the pearlitic phase of grey cast iron. Investigation of microhardness was done after colour etching so that the pearlite formed can be distinguished between the pearlite formed from primary austenite and the pearlite formed as part of the austenite-graphite eutectic.
They report a relation between the microhardness of the microstructure phases and the cooling condition in cast iron. Specifically, in the pearlitic matrix of flake iron, the hardness is higher with the primary austenite and lesser in the graphite austenite eutectic.
12
The ferrite phase is softer than the pearlitic phase. Figure 12 shows their finding about flake iron. The different cooling condition provide different results which is due to the variation of the cooling rate. Cooling condition of the insulation recorded the lowest hardness value. The chills provide rapid cooling and hence show the highest hardness value.
Figure 12. Microhardness for different pearlitic phases in flake iron [16].
13
3 METHODS AND EXPERIMENTS
This chapter will deal with the various methods and experiments that were performed for this thesis work. Details about Magmasoft, casting experiments, the nomenculature procedure of the samples and the tensile testing procedure is discussed. The graphite analysis, pearlite inter-lamellar spacing measurement, microhardness test details are also provided with illustrations.
3.1 Simulation in Magmasoft
A total of 6 pourings were done in the foundry at Skövde. In this work, three moulds from the first three pourings were analyzed, but the thermocouple measurements were obtained from pouring number 6. The cooling curves obtained were then compared with the cooling curves obtained from the software. The software has a standard material called ‘GJL-300’ for the analysis. The material data was then modified to match the experimental cooling curves obtained. The details of the values of these parameters and the values that were modified to obtain a good fit between simulation and experiments are discussed in the results section.
3.1.1 Parameters in Magma
Magma takes into consideration three important aspects of the casting.
• Geometry
• Process parameters
• Material properties
For successful casting simulations all the above three aspects are critical. The software provides options to define the geometry of the mold, the process parameters like the material definition, etc. The right geometry stands key in obtaining good simulation results. Selected phases of the casting process or all the phases of the casting process can be altered. The three phases are
• Filling stage
• Solidification stage
• Heat treatment (optional) Material definition
The material properties in Magma are present in the form of databases and one can create/modify materials. The important properties that influence the heat transfer process and the material definition aspect and their scientific implications are listed below. These properties are the ones that were modified to achieve closer fit with the experimental data.
• Specific heat capacity/Cp
The specific heat capacity is defined as the amount of heat (in Joules) required to raise one gram of the substance by one degree Celsius. Cp also influences certain other parameters that play an important role in the cooling curve.
14
• Fraction solid
The fraction solid is the ratio of the solidified portion of the total amount of alloy. Between the liquidus temperature, Tliq and the solidus temperature, Tsol, the melt contains liquid as well as solid portions. This parameter therefore plays an important role in the cooling rate and solidification time.
• Latent heat
The Latent heat is thermal energy released or absorbed, by a body or a thermodynamic system, during a constant-temperature process like solidification and solid-state phase transformation
• Thermal conductivity
The thermal conductivity of a material is the ability of a material to conduct heat. Hence it plays a vital role in the solidification and cooling rate of an alloy.
3.2 Test geometry & Castings
To start with a test geometry was designed in Magma for the experiments. The geometry was created with the goal of obtaining good variation in cooling rate and solidification time.
The model is shown in Figure 13. The model has a total of 14 plates with 7 thin and 7 thick plates. Three pin like structures were designed to obtain rapid cooling rates. The dimension of the plates and pins are provided in Table 2.
Table 2. Dimensions - Plates and Pins in mm
The castings of the test geometry were done in the foundry at Skövde. There was a total of 6 pouring for this project. In this work moulds from the first three pouring will be studied in detail. From the first three pouring that were done on 2018-01-30, 6 moulds were obtained (2 moulds for each pouring). In the first pouring one mould had thermocouples but it leaked and hence was scrapped. The other mould was chosen. In the second pouring one mould had a lot of large porosities on the outer surface and hence it was not evaluated. In the third pouring there was again one mould which leaked and was scrapped. Hence one mould from each
Plates Pins
Thin Thick X Y Z
Top Base Top base Top Base Top Base Top Base Length (x axis) 145 165 145 165 14 24 18 26 24 32
Width (y axis) 14 18 26 24 14 24 18 26 24 32
Height (z axis) 120 120 120 120 120
15
pouring was selected for evaluation. The experimental details obtained from the casting experiments are shown in Table 3.
Table 3. Casting - Parameters
The summary of the pouring and the mould details are presented in Table 4. The composition of the alloy after chemical analysis is shown in Table 5.
Table 4. Summary of the pourings and moulds
Mold Name Pouring no Alloy Note
Plate mold 1 1 VIG275/190 (0519) Evaluated
Plate mold 2 1 VIG275/190 (0519) Mold leaked, scrapped
Plate mold 3 2 VIG290/205 (0520) Large porosities on outer surfaces
Plate mold 4 2 VIG290/205 (0520) Evaluated
Plate mold 5 3 VIG250/190B (0419) Evaluated
Plate mold 6 3 VIG250/190B (0419) Mold leaked, scrapped
Table 5. Alloy composition of molds evaluated
Pouring No. 1 2 3
Gerometry obtained 2 Plate 2 Plate 2 Plate
Alloy VIG275/190 (0519) VIG290/205 (0520) VIG250/190B (0419)
Furnace no. LFR4 LFR3 LFR6
Melt quantity (kg) 785 830 867
Pouring temperature (° C) 1410 1400 1400
Pouring time plates (s) NA 10-12 10
Ladle inoculation 1,6 kg Superseed 1,6 kg Superseed 1,6 kg Superseed
Alloying No No 1,5 kg Cu
Thermo-couples Yes No No
Alloy C Si Mn P S Cr Ni Mo Cu Sn Ti V Nb CE
Mould 1
3,27 2,00 0,58 0,025 0,074 0,12 0,04 0,20 0,92 0,050 0,016 0,014 0,014 3,78
Mould 4
3,08 1,86 0,58 0,025 0,076 0,12 0 0,2 0,9 0,05 0,015 0,01 0,012 3,56
Mould 5
3,31 1,98 0,6 0,025 0,074 0,18 0 0,1 0,5 0,03 0,015 0,01 0,013 3,81
16
Figure 13(a). Test geometry showing the plate positions from a-g in thick and thin plates. The name of the samples are done based on this designation.
Figure.13(b) – Test geometry that shows the inlet, plates and pin
17 3.3 Naming and Selection of the samples
The samples for tensile tests and the micro-structure analysis were selected based on the cooling rates and solidification times. The thin plates were designated as A and the thick ones as B. The full designations were done based on the mould number, plate number and sample position.
Example:
1A-2-d
Mould Number – A(Thin Plates)/B(Thick Plates) – Plate number- Sample position.
.
Figure 14. The solidification time for one of the plates to show the gradient and position selection. Here the gradient is symmetrical.
Figure 15. Thick plate dimensions and position designations
Figure 16. Thin plate dimensions and position designations
18
Figure 17. Pins X,Y,Z. with the location of the microsample position
24 samples for each mould were selected for tensile tests. Apart from the tensile tests, samples for microstructure were also selected for image analysis. Another important aspect while selecting samples was to choose plates that had the least gradient within a single plate so that there are no experimental artefacts. Therefore, samples from plate numbers 1 and 3 that had an asymmetric gradient were not chosen for analysis. Plate numbers 1 and 3 in both A an B plates showed asymmetric gradients. The selected samples were cut, and the top of the samples were marked. The sample fracture position and its height from the top of the sample were noted.
Then the cooling rates and solidification times were noted from Magma for that fracture position.
Figure 17 shows the pin like structures. The micro-samples were cut at the same position for all the samples of the pin type irrespective of the moulds. The micro-samples for the plate structures were selected based on their location in the simulation. Since the solidification time varies according to the height of the sample. The sample location was selected for each sample such that the position where there is least/no gradient was selected. For mould 1 the same location for all the samples was chosen. Figure 18 shows the position for the micro-samples.
From the second mould a variable called ‘h’ value was introduced.
The variable ‘h’ is the value at which the sample is cut. ‘h’ is then formulated to calculate the position (based on the coordinates) in the simulation to identify the corresponding solidification and cooling rates. The selected samples had cooling rates and solidification time as seen in Figure 19, 20. Samples were selected to get a large variation in the cooling rates and solidification times.
19
Figure 18. (a) microsample position for mould.1 (b) microsample positions for moulds 4&5.
Figure 19. The tensile samples selected and their corresponding parameters that show variation in the selected samples. Figure 20. The samples selected for microstructure analysis and their parameters.
0 200 400 600 800 1000 1200
0 0.5 1 1.5
Solididification time(s)
Cooling rate (°C/s) Tensile samples
The height ‘h’ at which the samples are cut
80 180 280 380 480 580 680 780
0 0.5 1
Solidification time
Cooling Rate (°C/s) Microstructure samples
20 3.4 Tensile tests
The plates were cut in the Volvo workshop, Gothenburg. They were marked with a red dot on top. (see Figure 21) and sent to Swecast, Jönköping for machining and tensile tests. This was done so that the sample position after the fracture could be identified. With the help of the fracture position and the total height, the Z-axis position in the simulation could be formulated.
The samples were of the type 7C35 (7mm-gauge diameter, C- cylindrical and 35 mm - gauge length).
Figure 21. The tensile sample with marking on top.
3.5 Light optical microscopy
3.5.1 Graphite morphology and Lmax
Graphite length, distribution and size was analyzed for 12 samples per mould. The graphite morphology was analyzed using light optical microscopy using the Leica Q win software. The software allows one to process images obtained from the LOM. Fields can be selected based on the statistical requirements. Per sample 8 × 8 fields with 64 images in total were selected for analysis. The software also allows one to omit particles detected less than 10µm to avoid inclusion of porosities, potential dirt/dust, oxides and other such misleading elements.
The obtained data was exported to the excel and analysis of the graphite length was performed for the samples. Here Lmax is defined as the maximum ferret of the graphite particle. See Figure 22 that shows the maximum and minimum ferret.
The graphite length is highly dependent on the solidification time. The length and type of graphite for various cooling rates and solidification times will give a clear idea about the possible mechanical properties for the corresponding graphite morphology.
21
Figure 22. Maximum and minimum ferret of graphite particle
3.5.2 Pearlite spacing
Pearlite spacing for grey cast iron is a challenging task since the lamellar distance is highly distributed within the samples with same solidification time and cooling rate. Nevertheless, one could expect a relationship between the pearlite spacing and the cooling rate since the pearlite would be finely spaced if the cooling rate is high enough. Since pearlite measurements are time consuming, measurements perform measurements only for 6 samples from mould 1.
These samples were etched moderately using Nital. The pearlite measurements were done using light optical microscope. Images such as the one shown in Figure 24 were taken with random field selection. 15 such images were taken for analysis for one sample. The images were analyzed using photoshop where grids were placed in the images. Inside each grid a circle of known diameter was placed on the pearlite colonies.
The number of cementite plates that hit the edge of the circle were counted. This was done for each of the grids present. The grids which have graphite or grids in which the lamellae is very small and is not resolvable in the optical microscope were not measured. A total of approximately 570 such measurements were performed for each sample. The formula for the mean true, mean directed and mean random lamellar spacing is shown in figure 23. Since the un-resolvable lamellae are not measured, this technique might slightly exaggerate the size of the lamellae for each sample and hence the test accuracy.
22
Figure 23. Pearlite measurement methodology adopted from various methods suggested in the review by G.F. Vander voort et al [18]. σd – mean directed spacing. σr – mean random
spacing σt – mean true spacing.
. Figure 24. Measurement methodology for pearlite spacing – circular test line method
3.6 Hardness measurements
Hardness of grey cast iron is usually measured using brinell hardness technique because of the presence of graphite. Brinell hardness measurements results in the overall hardness value of the sample. In general, the presence of graphite reduces the hardness value substantially and since cast iron comprises of graphite and pearlite the brinell method is usually implemented.
In this work Vicker’s microhardness was measured for two reasons. Firstly, the matrix of the grey cast iron provides the bulk of the strength to the material. Secondly this work focuses on investigating the pearlitic properties of grey iron. This measurement gives one an overview of the matrix hardness and hence a relationship between the pearlitic microhardness and the
23
tensile strength could be achieved. Struers microhardness equipment was used to measure the Vickers microhardness of pearlite. Several measurements for the same sample with load of 50 g and 200 g were done and since a lot of scatter in the results were observed, 100 grams load was chosen for the measurements.
100 g load resulted in a diagonal size of 25 ± 3 µm. The scatter was high but relatively lower than the 50 and 200 g load.
Factors that were considered
• Load value should result in least scatter of measurements.
• Load value must be moderate such that the indent does not hit on underlying graphite and affect measurements.
• Load value should result in the recommended (by Struers) diagonal size.
Figure 25. Plots showing the scatter in HV for measurements with different load and their corresponding diagonal values for a test sample.
250 300 350 400 450
0 5 10 15 20
Microhardness (HV)
No. of measurements
HV 0.05
200 250 300 350 400 450
0 5 10 15 20
Microhardness (HV)
No. of measurements
HV 0.2
250 270 290 310 330 350 370 390 410 430 450
0 5 10 15 20
Microhardness (HV)
No. of measurements
HV 0.1
Diagonal - 25 µm
Diagonal – 35 µm Diagonal 17 µm
24
4 RESULTS
This chapter will deal with the results obtained from experiments and analysis. The results for each experiment are discussed with plots and illustrations.
4.1 Simulation and cooling curves
The cooling curves were obtained from the thermocouple data. The thermocouples were positioned in the mould as shown below in Figure 26. TC 1-4 are in the casting. TC 5 and 6 are in the sand.
Figure 26. Thermocouple positions
4.1.1 Thermocouples in the casting
The experimental cooling curves obtained from different thermocouple measurements and the respective simulated curves are shown in Figures 27, 28. The TC 3 during experiments had a measurement error which was evident from the curves. Simulation cooling curve results show that the rate at which the liquid melt solidifies is higher in the case of the simulation and that it needs to be reduced by modifying the fraction solid (fs) curve. It was also found that the solidus temperature and the liquidus temperature was slightly higher.
The latent heat of solidification was also slightly lower than the experimental curves. The specific heat capacity (Cp) was altered in the database at different temperatures. These changes were done in Magma material database so that the solidification time has a good match between actual cooling curves and simulation for eutectic and eutectoid transformation.
25
Figure 27. Cooling curves experimental and simulation before the modification of data in Magma – Eutectic solidification
Figure 28. Cooling curves experimental and simulation before the modification of data in Magma – Pearlite transformation
26
Figure 29. Changes done in the fraction solid curve.
Figure 30. Changes done in the specific heat capacity (only the lower scale is shown to highlight the changes in lower temperature ranges)
Figures 29, 30 show the changes that were made in the material database. The fraction solid amount was changed so that the amount of solid from the melt is gradually increasing. The specific heat capacity Cp of the alloy was also modified slightly in the lower scale regions. It is seen from Figure 31 that the changes done in the material properties has given a good match between simulation and actual cooling curves which give us the right solidification time and cooling rate. In case of TC 3 as mentioned earlier there was an error in the thermocouple measurement, which is the cause for the mismatch, nevertheless the solidification time has a good match. The exact values of the parameters modified in the database are presented in Appendix A.
0 0.2 0.4 0.6 0.8 1 1.2
1150 1160 1170 1180 1190 1200 1210 1220 1230
Fs
Temperature (°C)
Fraction solid (Fs) curve
Original Modified
400 600 800 1000 1200
0 500 1000 1500 2000
Cp Specific heat (J/kgK)
Temperature
Specific heat capacity curve
modifications
Modified Original
27
Figure 31 (a). Cooling curves experimental and simulation after the modification of data in Magma – Eutectic transformation
Figure 31 (b). Cooling curves experimental and simulation after the modification of data in Magma – pearlite transformation
28 4.1.2 Thermocouples in the sand
The cooling curves of the thermocouples placed in the sand are shown below. They are compared with the simulated sand curves. There were no changes made in the sand properties in Magmasoft database. The cooling curves are shown in Figure 32.
Figure 32. Cooling curves for Thermocouple 5 and 6 in sand
4.2 Graphite morphology and Length
4.2.1 Maximum graphite length Lmax vs ts
The image analysis results for the moulds show that as expected there is a steady increase in the graphite length as the solidification time increases. The maximum graphite length as a function of the simulated solidification time is plotted and shown in Figure 33.
The plots show a strong relationship between the two parameters for solidification times (< 500 s) and the length of graphite converges to a constant value as the curves seems to flatten out for solidification times (> 500 s). This flattening effect might be due to impingement of the graphite cells. In this work since the change in composition is low for the moulds there is no significant effect noticed regarding the graphite length. The effect of Cu and Cr on graphite length is therefore not given importance. Primary importance was given to the effect of the solidification time and its influence on the graphite length as that is most influencing parameter.
0 100 200 300 400 500 600 700 800
0 2000 4000 6000 8000 10000 12000
Temperature (°C)
Time (s)
Cooling curve - Sand
TC5
TC6
TC_05(simulate d)
29
Figure 33. Maximum graphite length as a function of solidification time.
It is also important to note that the area chosen for the analysis in Qwin software might vary slightly and that variation in the area affects the measurement as the solidification time varies in the same sample.
The samples A2-a-1, A6-b-3, A4-e-3 and A7-f-1 have lesser variation in the solidification times, and hence are slightly indistinguishable. The percentile of maximum length chosen for the plots is 99 percentile. Increasing the percentile to a higher value gives us more distinguishable results. Since higher percentile might include errors in the image analysis such as scratches or two graphite particles clinging together (which might be counted as one), 99 percentile was chosen as the optimum value and was considered.
4.2.2 Graphite morphology
The type of graphite could be seen in the micrographs shown in Figure 34 (a,b). The samples with their solidification times and their micrograph is shown. It could be seen that the 1A-2-a- 1 is mainly comprised of type A with smaller graphite. 1B-5-e-3 also is of type A graphite but with longer graphite length. 1B-2-c-3 shows type A graphite with the longest length. 1B-2-c-3 is the sample with highest solidification time (ts-961.7). The samples x, y, z has the shortest solidification times with x being the shortest. 1-x (ts- 83.5) is made of type E graphite and also the shortest graphite length. It also shows a very small percentage of type D. 1-z shows similar type E graphite with slightly longer graphite with a higher percentage of type D graphite. 1-z shows type E with the longest graphite length and also type D graphite mixed. In the z samples type A graphite is also present. The amount of type D graphite measured for a number of samples with low solidification times are discussed in the discussion section.
4A-6-b-3 4A-7-f-1
4B-2-c-1 4B-2-c-3
50 100 150 200 250
0 500 1000
Lmax at 99th percentile (µm)
Solidification time (s)
Mould 4
5A-2-a-1
5A-4-e-3 5A-6-b-3
50 100 150 200 250 300
0 500 1000
Lmax at 99th percentile(µm)
Solidification time (s)
Mould 5
30
Figure 34 (a). Micrograph of samples showing type A, D and E according to their respective solidification times.
Figure 34 (b). Micrograph of samples showing type A, D and E according to their solidification times.
31
4.2.3 Pearlite inter-lamellar spacing measurement – Circular test line method
The results of the pearlite measurement using a circular test line method are shown below.
Figure 35. Lamellar spacing as a function of cooling rate (circular test line).
Table 6. Summary of pearlite measurement
The measurements show a relationship between the lamellar spacing and the cooling rate, but as mentioned earlier the methodology used exaggerated the values of the lamellar spacing due to the inability of LOM to resolve the most finely spaced lamellae. The scatter of pearlite measurement in the same sample was high. The plot with the average spacing did not provide good result. The spacing which had the highest frequency on the other hand gives an overview.
4.3 Micro-hardness
The results of the micro-hardness measurements with 100 g load for samples from different moulds are shown in Figure 36. The measurements were performed without etching to avoid any potential bias during measurements.
Since the samples measured were un-etched, there are chances that the indent might have hit underlying ferrite or graphite to record low values or carbides to record high values. This affects the measurements and gives a scatter. A histogram was prepared for each sample for this purpose. For example, sample 4B-2-c-1 has the values as shown in Figure 35 The HV value with highest frequency is in the range of 260-280. This range is specified for all the samples in Figure 36. Figure 37 shows HV as a function of cooling rate considering the average HV.
Sample Highest frequency Average Cooling rate
1A-6-b-3 0,4 0,50 0,346
1A-7-f-1 0,35 0,45 0,343
1B-2-c-3 0,8 0,64 0,056
1A-2-a-1 0,45 0,44 0,086
1B-7-a-3 0,35 0,44 0,217
0.4 0.45 0.5 0.55 0.6 0.65
0 0.1 0.2 0.3 0.4
Lamellar spacing (µm)
Cooling rate (°C/s)
Average
0.3 0.4 0.5 0.6 0.7 0.8 0.9
0 0.1 0.2 0.3 0.4
Lamellar spacing (µm)
Cooling rate (°C/s)
Highest frequency
32
1-x, the sample with highest cooling rate had the highest average of 341, while 1B-2-c-3, 4B- 2-c-3 and 5B-2-c-3, the slowest cooled samples recorded 268, 268 and 261 respectively. The summary of the measurements is presented in Table 7. As a general trend the microhardness of samples (from the same position) of mould 5 were lower than the other two moulds. Mould 1 and 4 did not change in the average values for the same locations. This could be because of the decreased Cu content in mould 5. Cu and Cr both have a refining effect on the pearlite spacing.
In mould 5 although the Cr levels have increased from 0.12 % to 0.18 , there was no refining effect seen. This could be because of decrease in the Cu (from 0.8 to 0.5 %) is relatively higher than the increase of Cr (0.06 %).
Figure 36. Histogram for 4B-2-c-1
. Figure 37. Average HV as a function of cooling rates for samples in all the moulds.
It is evident from the results that the average HV value for each sample show that the microhardness of the matrix has a clear correlation with the cooling rate and the relationship is almost linear. It is safe to conclude that the values from the average show clearly the difference in the pearlite spacing of these samples. It is already known that for higher cooling rate the pearlite fineness increases. Thus the microhardness measurements provide an estimate of the matrix structure and pearlite fineness. From literature it is known that the pearlite formed from the primary austenite is harder than the pearlite formed from the graphite austenite eutectic.
Measurements in this work did not consider this differentiation because the goal for the microhardness measurement was to find an overview of the matrix structure which includes
0 5 10 15 20
230 250 270 290 310 330 350 370 390 410 430 450 More
Frequency
Bin
4B-2-c-1
250 260 270 280 290 300 310 320 330 340 350
0 0.2 0.4 0.6 0.8 1 1.2
Microhardness HV
Cooling rate (°C/s)
Microhardness HV
33
both the types of pearlite. For detailed measurement values of each sample please see Appendix B.
Table 7. Hardness (HV) and their ts and (dT/dt)
4.4 UTS modelling
4.4.1 Effect of solidification time and cooling rate on the UTS
The UTS as a function of cooling rate is shown in Figure 38. The results show a clear correlation that with increasing cooling rates the UTS is increasing. There are a few samples (marked red) which did not record values as expected. These samples were analyzed further and are discussed in detail in the discussion section. The fastest cooled samples (0.956 °C/s) were the 1-x and 4-x samples which show the highest tensile strength of 324 and 329 respectively. The slowest cooled sample 4B-2-d (0.057 °C/s) recorded the lowest UTS of 223.
It is important to note that the results for all the samples of three different moulds have also other parameters such as solidification time and composition varying. Figure 39 shows how the solidification time influences the UTS
Sample Average Cooling rate Solidification time
1-x 341 1,124 83,5
1B-7-a-3 299 0,217 393,5
1B-5-e-3 283 0,134 481,5
1B-2-c-3 268 0,056 961,7
1A-7-f-1 314 0,343 278,8
1A-6-b-3 315 0,346 294,7
4A-4-e-3 280 0,088 318,5
4A-6-b-3 310 0,347 301,6
4A-7-f-1 309 0,344 293,4
4B-2-c-1 280 0,056 798,1
4B-2-c-3 268 0,056 965,2
4B-7-a-3 301 0,217 388,2
4-y 331 0,914 104,2
5-z 295 0,595 149,1
5B-7-a-3 278 0,217 388,2
5B-2-c-3 261 0,056 965,2
5B-2-c-1 264 0,056 798,1
5A-7-f-1 298 0,344 293,4
5A-4-e-3 262 0,088 318,5
34
Figure 38. UTS as a function of cooling rate for all three moulds (Outliers are marked red).
Figure 39. UTS as a function of solidification time for all three moulds (Outliers marked red).
It could be seen from Figure 39 that increase in solidification time reduces the UTS. This is due to the increase in graphite length as the solidification time increases. Again, some of the outliers are marked with red.
1A-2-d
1A-6-d
1-Z 4A-2-d
4A-4-d
4A-4-g
4B-2-d
4-y
4-Z
5B-6-b
5-x
5-Z
200 220 240 260 280 300 320 340
0 0.2 0.4 0.6 0.8 1
UTS (MPa)
Cooling rate (°C/s)
1A-2-d
1A-6-d
4A-2-d
4A-4-g
4B-2-d 5B-6-b
200 220 240 260 280 300 320 340
0 200 400 600 800 1000
UTS (MPa)
SolidificationTime (s)
35 4.4.2 Effect of alloying on the UTS
The effect of carbon equivalence on the UTS is shown in Figure 40. Since there are other varying parameters, the results don’t show any conclusive effect, it is well known that increase in CE is detrimental for the UTS. It is seen in most cases that mould 1 (CE = 3.78) shows the highest UTS for the samples. The effect of Mo, Cu and Cr have all been plotted, but as mentioned earlier, since the other factors vary considerably such plots become less effective, the values and these plots are provided in the Appendix C.
Figure 40. UTS with varying CE for three moulds
4.5 Regression model
4.5.1 Tensile strength
A regression model which includes the solidification time, cooling rate, Cu, Cr, Mo and CE was developed (Courtesy of Fethi Zerhab) using statistical software Minitab. This regression model showed that there is good correlation between the predictor variables and the response UTS showing 63.4 % correlation. However the CE, Cu, Cr and Mo content did not have a significant impact. The summary of the regression model is given in Figure 41. Hence with only solidification time and cooling rate the following equation was achieved. For achieving this model two samples from the results with very high variance 1A-2-d and 1A-6-d was neglected.
UTS = 263.18 – 0.0437 * ts + 126.2 * (dT/dt) – 65.6 * (dT/dt)2 (6)
200 220 240 260 280 300 320 340
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
UTS
Samples
Effect of Carbon equivalent (CE) on UTS
3.78 3.87 3.85
36
This equation (6) can be used to predict UTS for specific values of ts and (dT/dt) or set the parameters of ts and (dT/dt) for the desired UTS. Here ts is the solidification time and (dT/dt) is the cooling rate. The correlation coefficient will be even higher if other outliers are analyzed and if they could be removed. All these outliers are analyzed in detail in the discussion section.
Figure 41. Regression model for UTS
4.5.2 Graphite length and solidification time
The relationship between the solidification time and maximum graphite length shows high correlation with a logarithmic model. The plot with the regression equation is given below.
There is 91 % correlation between the two variables. Figure 42 shows the fitted line plot and the equation.
The relationship between the two parameters is given mathematically according to equation 7.
𝐿𝑚𝑎𝑥 = 84.43 ∗ 𝑙𝑛(𝑡𝑠) – 297.4 (7) From literature [19] we have a relationship between UTS and Lmax
𝜎𝑈𝑇𝑆 = 𝐾𝑡 / √𝐿𝑚𝑎𝑥 (8)
