IN DEGREE PROJECT MATERIALS SCIENCE AND ENGINEERING,SECOND CYCLE, 30 CREDITS
, STOCKHOLM SWEDEN 2020
Inoculant measurement with thermal analysis
CHRISTOPHER ARMSTRONG
KTH ROYAL INSTITUTE OF TECHNOLOGYSCHOOL OF INDUSTRIAL ENGINEERING AND MANAGEMENT
Abstract Over time the level of inoculant will decrease due to fading and this needs to be compensated by
more additions of inoculant. When casting CGI400 at Scania a one-step method is used, and the
inoculant additions are based on the last ladle from the previous oven. Longer stops in production
will result in the previous oven not being representable and more inoculant should be added. A
method to establish the inoculation level of the base iron in the oven would make it easier to control
the process.
This study was about inoculation measurements with the help of thermo- analysis of the melt in the
oven to get a better understanding how it fluctuates between ovens and changes over time. The
study is also about how to implement this tool into production in a foundry.
With the help of thermo- analytic measurements, the inoculation level of the melt was established,
and the method was also used to see how the level of inoculant changed over. To establish the
inoculation additions needed, the saturation point of inoculant was investigated.
The results from the study shows that the starting level of inoculant is too low in the melt it also
shows that the inoculation level fluctuates over time. The conclusions that can be drawn from this
study is that thermos analysis can be used in production to establish the inoculation level, but more
measurements needs to be conducted to determine its accuracy.
Keywords:
Thermal Analysis, Inoculation, Nucleation, Compacted Graphite Iron, CGI, Cast Iron, Material
Properties
Sammanfattning Över tid så minskar ympnivån i en smälta på grund av fading och detta måste kompenseras med mer
tillsatser av ymp. Vid gjutning av CGI400 på Scania används en en-stegs metod där man baserar
ymptillsatsen i skänken på vad ympnivån var i sista skänken från den tidigare ugnen. Vid längre
produktionsstopp kommer inte ympnivån från ugnen innan vara representativ utan mer ymp måste
tillsättas. En metod för att bestämma ympnivån av basjärnet i ugnen skulle göra det lättare att styra
processen.
Denna studie handlar om ympmätning med hjälp av termisk analys i bas smälta för att få en bättre
förståelse för hur den fluktuerar mellan ugnar och ändras över tid. Studien handlar också om hur
man ska kunna applicera denna mätmetod i produktionen på ett gjuteri.
Med hjälp av termo analytiska mätningar skulle grundnivån av ymp i smältan bestämmas och även
för att se hur nivån ändrades över tiden. För att bestämma tillsatsen av ymp som borde tillföras
undersöktes mättnadspunkten av ymp och hur den kunde tillsättas i skänk.
Resultaten från studien visar att grundnivån av ymp är för låg och resultaten visar också att nivån av
ymp fluktuerar både upp och ner ökar tid vilket inte var vad man hade förväntat sig och även det
finns flera faktorer som kan påverka nivån. Slutsatsen som kan dras är att termisk analys kan
användas i produktion för att bestämma ympnivån, men mer mätningar behövs för att bestämma
dess noggrannhet.
List of abbreviations:
CGI- Compacted graphite Iron
LGI- Lamellar graphite iron
SGI- Spheroidal graphite iron
TEmin- Eutectic minimum temperature
DTEmin- Difference in eutectic minimum temperature
TL- Liquidus temperature
SEM- Scanning electron microscope
LOM- Light optical microscope
CE- Carbon equivalent
CEL- Carbon equivalent liquids
S- Saturation point
wt%- Weight percent
𝜎- Yield stress
d- Diameter
ΔG- Difference in Gibbs energy
MGM- The modification index
MGI- Inoculation index
𝑁𝑓- Grains per area
𝑁𝑤- Grains intersected in sample
𝑁𝑖- Grains inside sample
𝑅2- Coefficient of determination
Table of content
Introduction ..................................................................................................................................... 1
1.1 Ethical, social, and environmental aspect ............................................................................... 1
Theoretical background ................................................................................................................... 3
2.1 CGI ........................................................................................................................................... 4
2.2 Microstructure ......................................................................................................................... 5
2.3 Process ..................................................................................................................................... 5
2.4 Solidification ............................................................................................................................ 7
2.5 Nucleation ............................................................................................................................... 9
2.6 Defects from improper inoculation additions ....................................................................... 13
2.7 Thermal analysis .................................................................................................................... 13
2.8 Sinter cast process ................................................................................................................. 15
2.9 Thermocalc ............................................................................................................................ 16
2.10 Raw material .......................................................................................................................... 17
Experimental procedure ................................................................................................................ 17
3.1 Method .................................................................................................................................. 17
3.2 Average starting value and nucleation fading ....................................................................... 18
3.3 Saturation point ..................................................................................................................... 21
3.4 Minilab-suitcase .................................................................................................................... 21
3.5 Testing of inoculation addition in ladle ................................................................................. 23
3.6 Laboratory preparation ......................................................................................................... 24
3.7 Etching ................................................................................................................................... 25
3.8 Image J ................................................................................................................................... 27
3.9 Grain density ......................................................................................................................... 27
3.10 Grain size ............................................................................................................................... 27
3.11 Additional testing .................................................................................................................. 28
3.12 Chemical analysis................................................................................................................... 28
3.13 Thermocalc ............................................................................................................................ 29
3.14 Comparison of TL for cups ..................................................................................................... 29
Results ........................................................................................................................................... 29
4.1 Level on inoculation in oven .................................................................................................. 29
4.2 Inoculation fading over time ................................................................................................. 30
4.3 Saturation point ..................................................................................................................... 32
4.4 Testing inoculation additions to the ladle ............................................................................. 34
4.5 Grain size and density ........................................................................................................... 35
4.6 Addition of scrap iron ............................................................................................................ 36
4.7 Effects of inoculants used...................................................................................................... 37
4.8 Effects of slagging .................................................................................................................. 37
4.9 Effect of carbon additions to DTEmin ................................................................................... 37
4.10 Increased temperatures effect on DTEmin ........................................................................... 38
4.11 Comparison of nucleation level between ladle and oven ..................................................... 38
4.12 Chemical analysis................................................................................................................... 39
4.13 Charged material ................................................................................................................... 40
4.14 Thermocalc ............................................................................................................................ 40
4.15 Comparison of cups ............................................................................................................... 42
Discussion ...................................................................................................................................... 42
5.1 Average inoculation level ...................................................................................................... 42
5.2 Saturation point ..................................................................................................................... 42
5.3 Testing of inoculant additions to the ladle ............................................................................ 43
5.4 Inoculation fading over time ................................................................................................. 43
5.5 Grain size and density ........................................................................................................... 44
5.6 Addition of scrap iron ............................................................................................................ 44
5.7 Effects of inoculants used...................................................................................................... 44
5.8 Effects of carbon additions .................................................................................................... 45
5.9 Increased temperatures effect on DTEmin ........................................................................... 45
5.10 Comparison of the nucleation level between oven and ladle ............................................... 45
5.11 Chemical analysis................................................................................................................... 45
5.12 Material charged ................................................................................................................... 46
5.13 Thermocalc ............................................................................................................................ 46
5.14 Comparison of cups ............................................................................................................... 47
5.15 Overall discussion .................................................................................................................. 47
Conclusion ..................................................................................................................................... 48
Further research ............................................................................................................................ 49
Sources of error ............................................................................................................................. 49
Acknowledgment ........................................................................................................................... 50
References ................................................................................................................................. 51
1
Introduction When casting metals, the composition of the melt is very important for the finale properties of the
finished product. It is not just the alloying composition that is important also the level of inoculant
present in the melt plays a significant role. These are the starting points for the metals nucleation
and will have a big effect on material properties such as the strength, ductility, and the thermal
conductivity.
When producing cast iron today an analysation of the nucleation level takes place in the ladle at the
foundry at Scania. Additions of inoculant and magnesium (Mg) to alter the graphite precipitation is
added to the ladle before the sample is taken. It takes several minutes for the software to analyse
the sample, valuable processing time and time where the melt in the ladle’s temperature is
constantly dropping. Therefore, the melt is cast before the results of the test is received. Depending
on the results, the base treatment will be altered for the next ladle to reach the required levels of
inoculant and Mg. The first ladles additions are based on the last ladle from the previous oven. This
could cause problems. Even though the alloying composition of the melt is within certain limits for
every compound the level of inoculants will vary depending on heating, stirring, raw material and on
how long the oven has been sitting if there is a stop in production. This could lead to undesirable
material properties and variation between the finished products.
Reduction in discarded products will lead to lower energy consumption and a reduction of material
usage. The steel and Iron industry are one of the heaviest polluters and energy demanding industries
in the world. By becoming more energy and material efficient the carbon footprint from this industry
will be reduced. [1]
Also, by improving strength of the material that they use for casting engine blocks and cylinder
heads, the diesel engine trucks will be able withstand higher combustion pressure and will be able to
get more effect out of their engines. This will lead to heavier loads being possible to haul leading to a
better CO2 to weight ratio [2] [3].
It would be very beneficial if one could know the status of the melt already in the oven in terms of
inoculant level. This would decrease discarded products not meeting the required material
properties. By analysation of the cooling curve the eutectic minimum temperature can be established
which can tell us the level of inoculation by comparing this temperature for a cup with and a without
inoculant additions. The saturation point can then be found which will tell us how much inoculant
should be added to the melt. This will be done by using Heraeus Electro-Nite testing cups and
software to analyse cooling curves of the samples. The microstructure of the samples will also be
analysed to see if there is any correlation between the grainsize and density with the results given by
the cooling curves portrait by the software.
The aim of the project is to test this method of measuring the nucleation level and to see if it will
have a positive impact when incorporated into production.
Ethical, social, and environmental aspect A producer of a product that runs on fossil fuel should try to minimize its carbon footprint, so it is
not just economically viable but also ethical and environmental. The transport sector which
utilizes these products produces a lot of carbon dioxide which contributes to global warming. This
will lead to a changing environment which will have a negative impact with rising sea levels and
drought which will lead to famine. It is in everyone’s interest to try to produce products with
2
minimum emissions and produce products that will last a very long time. Also, from an ethical and
social standpoint improved engine performance that results in vehicles travelling further with
heavier loads means communities further from civilization can be reached as well as areas were
fuel is scares. Further development will lead to an increase of trained personal needed which will
benefit the social aspect.
3
Theoretical background Cast iron is a very versatile alloy consisting mainly of iron and carbon with a carbon content
exceeding 2%. Silicon is almost always present in all cast irons due to its graphite stabilizing effect.
Other alloying elements are also added such as silica, manganese, copper to get the required
material properties. The physical and mechanical properties can also be influenced by heat
treatment and by the altering the cooling rate [4].
The iron can either solidify as grey cast iron or as white cast iron. They get their name from the
colour of the surface when the material is fractured. White cast iron is a metastable phase, here the
carbon present in the iron will precipitate as the carbon rich phase cementite (FeC3) during
solidification which reflects a white colour. Here the carbon has not had enough time to migrate to
form the more stable phase, graphite. For grey cast iron which is the more commonly used in
industry the carbon precipitates as graphite flakes which reflects a grey colour when fractured [3] [4].
There exists a variety of different grades of cast iron with different that can be used many different
applications. What gives them their different properties and differentiates them from each other is
the shape of their graphite particles and the eutectic matrix. From earlier categorizing the cast iron
depending on the colour of the fracture they are categorized depending on the shape of the
graphite. Figure 1 shows how the graphite looks in the different grades of cast iron in a SEM [3] [5].
Figure 1: Graphite structure for the three different grades of cast iron seen in SEM. From left to right: Grey-, Compacted graphite- and Ductile iron. [5]
Cast iron consists of several different classes but three of them are more common than the others.
• Lamellar graphite iron (LGI)- It is also known as grey iron or flake graphite iron. The graphite
in the stable matrix is shaped as flakes or lamellas. LGI has very high thermal conductivity,
this due to the three-dimensional graphite network within the material. This is one of the
reasons why the material is often used in the automotive industry when constructing
cylinders heads as an example. The tips of the flakes in the network have sharp points and
create notches where the graphite joins the matrix. At these points, stresses in the material
will concentrate and cracks will form when the stress is to large. This leads to a material that
is brittle. The presence of this flakes leads to the material’s good machinability [3] [4] [5] [6]
[7].
4
• Spheroidal graphite iron (SGI)- It is also called ductile or nodular cast iron. Here the graphite
is shaped as nodules/ spheroids. This type of cast iron is stronger and more ductile than LGI
because the graphite does not create sharp notches in the material where stress
concentrates. The spheroid shape of the graphite is created by usually adding magnesium to
the iron. The downside with this cast iron is that its thermal conductivity is quite poor due to
the lack of interconnecting graphite [3] [4] [5] [6] [7].
• Compacted graphite iron (CGI)- this is a cast iron that combines the high conductivity of LGI
with the ductile properties of SGI. The thermal analysis to establish the inoculation will be
done on three different grades of CGI so a more in-depth explanation of CGI will be given [3]
[4] [5] [6] [7].
The mechanical properties for the different grades of cast iron can be seen in table 1. These values
are provided by Sintercast, but other sources claim that the tensile strength for LGI varies from 150-
450, CGI between 250-575 and SGI between 350-900 MPa [8].
Table 1: Mechanical properties for the three different grades of cast iron [9].
Property LGI CGI SGI
Tensile strength [MPa] 250 450 750
Elastic Modulus [GPa] 105 145 750
Elongation [%] 0 1.5 5
Thermal conductivity [W/mK] 46 37 28
Relative damping capacity 1 0.35 0.22
Hardness [HBN 10/3000] 179-202 217-241 217-255
R-B fatigue [MPa] 110 200 250
CGI Compact graphite iron is a material that combines the strength and ductility of SGI and the good
thermal conductivity of LGI. CGI possesses 70% higher tensile strength and 35% higher elastic
modulus and about double the fatigue strength compared to ordinary grey iron. This including its
good thermal conductivity is why it is a widely used product in the automotive industry especially for
engine parts. With the automotive industry being a very development intense industry, looking for a
more environmental approach, a material was needed that could make the engines lighter and with
increased mechanical properties. The engine must be able to withstand the increased combustion
pressure being generated that comes with increasing torque and horsepower [3] [4] [10] [11] [12].
According to ISO, the International Organization for Standards, to be classified as CGI with standard
16122:2006 the microstructure of a two-dimensional polished surface of a CGI sample must contain a
minimum of 80% compacted shape graphite and less than 20% nodularity and have no flake graphite.
ISO 16122 includes five grades of CGI with a tensile strength from 300 MPa to 500 MPa [5]. Zero
percent nodularity is what should be strived for when producing CGI.
5
At Scania three different grades of pearlitic CGI are produced depending the requirements of the
product. They are named in the rapport CGI400 used for product A, CGI425 used for product B and
CGI450 used for product C. The numbers at the end of the name is the tensile strength of the
material in MPa.
Microstructure The graphite particles in CGI has a vermicular shape and at a first glance they seem to be individual
particles not linked to each other. When viewing the material in the SEM as seen in figure 2, one can
see that the particles are linked to each other in an interconnecting web in the eutectic cell.
Figure 2: Microstructure of CGI seen in LOM on the left and SEM on the right [6]
This is the reason for its good thermal conductivity. The graphite particles are like that of lamellar
iron, they are randomly orientated and elongated but the ends are rounded instead of sharp. This
together with the complex graphite morphology results in a strong adhesion between the graphite
and the eutectic cell leading to its superior mechanical properties [3] [5] .
Process CGI is created by adding an element that will alter the graphite morphology to the melt, this could be
titanium, magnesium or rare earth metals such as cerium or lanthanum [12]. At Scania magnesium is
added and will alter the graphite just enough to get round edges on the fingers of the graphite
network which are usually sharp notches. This process is very precise and the window very small
where right amount of Mg additions will result in CGI, see figure 3. If too much is added SGI will be
created instead [13].
6
Figure 3: Mg vs Nodularity [14]
It has been proven that it is not the Mg itself that alters the graphite morphology, but it neutralizes
elements that that has a decreasing effect on the nodularity. Oxygen and sulphur are the two main
components that has this effect, but it is still unclear why they have this effect. Sulphur will also form
iron sulphide in the grain boundaries, which is very undesirable, but with the right amount of
magnesium manganese sulphide will be created which is harmless because it will be formed within
the grain [4] [15]. Titanium is usually also added, it has the opposite effect, it decreases nodularity
creating a larger process window. The downside of this is that titanium carbides are formed which
will lead to decreased machinability [12] [16].
Producing CGI puts a lot of demand and precision on process control. If flake graphite appears the
mechanical properties can be decreased with as much as 25-40%, and this can happen with a change
of active magnesium content of 0.001%. The aim is to get the microstructure inside the red box seen
in figure 4. With increased nodularity the strength and stiffness of the material will invertible
increase on the expense of machinability, thermal conductivity and castability [3] [4][5] [6] [17].
Figure 4: Inoculation vs Mg modification on microstructure [14].
The mechanical properties and especially the strength of the material of CGI can be altered in two
different ways. One is the ratio between perlite and ferrite in the eutectic matrix. The higher the
7
ration between perlite and ferrite the harder and stronger the material is but the toughness
decreases. Copper and tin are two alloying elements that benefits perlite formation that can be
added. Chrome and manganese can also be added but the downside of these alloying elements is
that they are carbide stabilizers and will promote the iron to solidify as white cast iron. The second
option is to alter the morphology as mentioned previously be adding for example Mg [11].
Carbon equivalent
As for all cast irons carbon has a big importance of the mechanical properties. Because of alloying
elements effect on the carbon present in the iron the concept of carbon equivalent (CE) is often
used. The carbon equivalents take into consideration other alloying elements and their effect on the
carbon in melt. When the CE is at 4.3% the composition is eutectic same as when just considering
carbon. By using CE one can treat the multicomponent cast iron as a binary Fe-C alloy instead [18].
There is a direct correlation between the phase change temperature and the composition. This
correlation has been used to map phase diagrams. The liquidus temperature (TL) will give us the %C
of the melt or the carbon equivalent (CE) which as mentioned earlier includes the influence of other
alloying elements on the carbon such as Si, Mn, P, S [18]. CE is expressed by equation 1, some will
only consider C, Si and P because the levels of Mn and S are usually quite low.
𝐶𝐸 = %𝐶 + 0.33%𝑆𝑖 + 0.33%𝑃 − 0.027%𝑀𝑛 + 0.4%𝑆 Equation 1
The solidification of a material is not a process in equilibrium, a theoretical cooling curve will look
different than the actual curve. There will be undercooling, the material will solidify below the
theoretical (equilibrium) liquidus temperature. The non- equilibrium liquids temperature will be
lower than the theoretical one [18].
But due to the difference between the non-equilibrium and equilibrium curve, the CEL carbon
equivalent liquids have been introduced instead. It is based on empirical studies and experiments
with standardized sand cups [18]. CEL is expressed by equation 2.
𝐶𝐸𝐿 = %𝐶 + 0.25%𝑆𝑖 + 0.5%𝑃 Equation 2
This improved definition should be used when calculating the carbon equivalent from the liquidus
temperature when using thermal analysis as measuring tool of the iron [19].
Solidification The right chemical composition is of great importance as just mentioned before solidification starts.
For a material to start to solidify the liquid phase must be undercooled, which means that the
temperature must be below the eutectic temperature [20].When a material solidifies the atoms
changes from being disorderly orientated, having an amorph structure to be rearranged into a
specific lattice structure also called crystal structure. When this occurs the atoms releases energy in
8
the form of heat to its surrounding. Less energy is needed in an ordered structure due to the
decrease of degrees of freedom of movement. Once a nuclease is formed it can continue to grow and
its speed is dependent on how fast the heat can be carried away. This is depends on the heat
generation during the phase change, flow of the melt and the mould material that transports away
the heat [20] [21].
The binary equilibrium phase-diagram Fe-C needs to be studied to get an understanding of the
phases present and the different phase changes that occurs during solidification, see figure 5. When
the iron contains approximately 4.3% C melt is said to have a eutectic composition and no primary
phase will precipitate. The melt will solidify as austenite and depending on the cooling rate the
carbon will either precipitate as cementite or as graphite. When producing CGI, the goal is to get as
close to the eutectic point as possible to get as much carbon as possible to fill out the pores without
getting any cementite or exploded graphite.
Figure 5: Fe-C phase-diagram showing both the stable and metastable phases constructed using Thermo-Calc.
9
If CE is below the value of 4.3% it is considered hypo-eutectic and above hyper-eutectic. A hypo-
eutectic composition is preferable when producing CGI. When you have hypo-eutectic composition
the iron rich phase austenite will form as the primary phase. The austenitic primary phase grows as
thin dendrites and will continue to grow until they collide with a neighbouring dendrite with a
different growth direction. When they collide two grains are created with a grain- boundary dividing
them and the dendrite will start to coarsen instead of growing larger. When the dendrites grow,
being an iron rich phase that only can dissolve approximately 2.08 % carbon, carbon will we be
rejected into the melt. If the rate of solidification is to high the carbon will not have time to diffuse
from the dendrites to the melt creating a concentration gradient in the solidified material. This is
called segregation and it is the last to solidify area that will have the highest concentration of carbon
[3] [4].
In the case of a hyper-eutectic composition the carbon rich phase graphite or cementite will form
before the eutectic transformation. Too high solidification rate which will increase the undercooling
which will lead to a metastable solidification and cementite instead of graphite will be formed which
CGI is susceptible to. This can be seen in the phase diagram seen in figure 5. When the composition
of the melt has reached eutectic, eutectic solidification will be formed. Here the remaining melt will
solidify as austenite and graphite or cementite embedding the already precipitated phase [3] [4].
What also can be seen in the phase diagram is that a solid- state phase transformation will occur at
around 738 °C. Here the austenite will transform and depending on the cooling rate, composition
and initial crystal structure either into pearlite which is a two-phased lamellar structure consisting of
88 wt% ferrite and 12 wt% cementite or ferrite with graphite will be formed. Pearlite is what
normally austenite decomposes to in conventional casting at the eutectoid temperature where the
cooling rate is not too high. Very low cooling rates will lead to a complete ferritic matrix where the
carbon will precipitate on the already precipitated graphite or cementite [3] [4] [22].
The morphology of the graphite precipitated is very much affected by the level of undercooling of
the material which is dependent on the cooling conditions during solidification. With increased
undercooling the nodularity will increase, finer and more disperse graphite particles will be formed.
Lower undercooling leads to coarser and fewer particles. By altering the cooling rate and the Mg
additions all types of morphologies can be formed [4] [15].
Nucleation Nucleation is when a small particle solidifies in the melt and there are three different types of
nucleation: homogenous, heterogeneous, and dynamic. Nucleation is a “platform” where crystal
growth is initiated during cooling. This growth is started when the temperature falls below the
liquidus line or for a eutectic composition the eutectic line. Of the three different types,
heterogenous is the most common. For nucleation sites to be able to support crystal growth it must
exceed a critical size so it will not be re-melted. This is illustrated by figure 6 which shows the Gibbs
free energy in relation to the radius of the nuclei. Once the “hump” on the green line is passed the
nuclei will start to grow spontaneously. The green line is a combination of the volume free energy
which is the driving force towards nucleation and the interfacial energy represented by the blue line
which is always positive and counteracts nucleation. To be able to reach the critical size, the
undercooling which is the driving force for the nucleation must overcome the energy barrier. There
are two different ways nuclei are formed and grows. Either they are developed continuously as the
temperature changes, or instantaneously when a specific temperature is reached [21] [23].
10
Figure 6: Critical radius for spontaneous nucleation in terms of Gibbs energy [23].
It is more energy efficient to nucleate on an already existing substrate, this is called heterogeneous
nucleation and austenite almost always nucleate this way. Surfaces support nucleation because of
wetting, angles greater than zero between the two phases will promote nucleation. Wetting is the
degree of which a liquid will spread on a solid when they come in contact with each other, one
measure of this is the contact angle as mentioned should be greater than zero. Homogeneous
nucleation, when you nucleate out of nothing is very rare and needs a lot of energy and are only
possible when the cooling rates are extremely high. Figure 7 shows the difference in energy to reach
the critical radius. The critical radius is the same for both homo- and heterogeneous nucleation but
the critical volume is usually smaller because of wetting [4] [23] [24] [25].
Figure 7: Comparison of critical radius for spontaneous nucleation for heterogeneous and homogenous nucleation in terms of Gibbs energy [23].
Inoculation
These nucleation points can be increased by inoculation additions to the melt and they have a big
effect on the mechanical properties of the material [21].
Most of the inoculants used for cast iron are based on ferrosilicon and additions from group 2 and 3
from the periodic table are most often added such as calcium, barium, strontium, titanium,
zirconium, and rare earth metals. Most ferrosilicon alloys contain between 65-75 % silica, a high
concentration of silica makes the melt exothermal and that it blends well. It is proven that pure
ferro-silica has bad inoculation effect, so that is why the alloying elements mentioned previously are
added. Because of the relative high cost of these elements there is an economical limit how much
should [26].
11
The best supported theories of additions to get heterogeneous inoculation is by adding silicon-
dioxide particles or salt-like carbides [27]. These elements will react with the oxygen and sulphur that
is dissolved in the melt to form oxy-sulphide clusters that have crystal structure that resembles that
of graphite. These clusters will benefit graphite nucleation, so it is of great importance that there is
some oxygen and sulphur dissolved in the iron. This might sound strange because it was mentioned
earlier that they had a negative effect on nodularity. The level of sulphur and oxygen in the melt will
alter the amount Mg needed to reach levels enough for the grey cast iron to turn in to CGI. Increased
or decreased levels of oxygen and sulphur will shift the Mg requirement curve left or right with
inoculation of CGI the amount of inoculation additions is critical. Too much will lead to increased
nodularity and the ratio of vermicular graphite will decrease [14] [28].
Effects on material properties
The main purpose of adding inoculation particles is to facilitate the formation of graphite so it can
occur with minimum or no undercooling. Increased formation of graphite will increase the thermal
conductivity of the material. Increased growth of graphite leads to decreased shrinkage propensity as
well, due to the growth of graphite that makes up for the solidification shrinkage in the formation of
austenite. Decrease in undercooling will also avoid cementite formation [18] [26] [29] [30].
Improved inoculation results in more eutectic grains being created during solidification to a certain
extent. Increased nucleation gives the melt more surfaces to homogenously nucleate on, but on the
same time it decreases undercooling. Increased undercooling is reported to give a finer grain
structure with increased grain density and decreased grain size. So, it is a combination of the two
that will give the optimal grain density in terms of mechanical properties [18].
More grains which will give you a harder and stronger material because the dislocations are inhibited
to move as far because the grain-boundaries. The Hall-Patch relation stats that with increased
grainsize the yield stress will be increased as seen in equation 3.
𝜎𝑦 = 𝜎0 + 𝑘𝑦𝑑−1
2⁄ Equation 3
In equation 3 the yield stress is 𝜎𝑦, 𝜎0 and 𝑘𝑦 are material dependent constants and d is the average
grainsize (diameter). This tells us that with an increased grainsize the yield stress will decrease
resulting in better mechanical properties [9].
Superheating as mentioned earlier decreases the liquidus temperature most probably because it
diminishes the nucleation potential of the melt.
Other goals for adding inoculants are to [28]:
• Reduce segregation.
• Increase ductility.
• Give a homogenous structure.
12
Fading of nucleation potential
As mentioned previously, fading is an unwanted decrease in nuclei when casting melts. It has been
observed a fading of nucleation potential when a melt has been sitting for a longer period before
being tapped, fading is time dependent. This is because the nucleation particles that differs in their
mass and density from the surrounding melt might react or dissolve into the melt or might
concentrate due to gravitational sedimentation. It has been seen that an increased temperature of
the melt decreases the nucleation potential. This reaction with the melt will lead to a decrease of the
catalytic effect of the nucleation particles and lead to inhomogeneous distribution of the particles in
the melt leading to grain-size variations in the casting. This fading of nucleation particles is also
thought of to be the result of Oswald ripening also known as coalescence. To minimize surface
energy smaller particles gets absorbed by larger particles, this gives the particles a higher volume to
surface area ratio. Particles can also be trapped in the slag or get stuck to the wall of the oven.
Decrease in holding time and a chemistry composition that is adjusted to minimize the loss of active
inoculant will decrease the fading effect [31] [32].
Method of inoculation
When adding inoculant there are three different ways, either into the ladle, additions to the melt
stream when casting or straight into the mould. The inoculant has greatest effect directly when
added and has a fading effect and most inoculants will have lost its effect 10 minutes after it has
been added. So, it is important to do it as close to the time of casting as possible [44].
Adding the inoculant straight into the ladle is the most flexible and the simplest way of adding
inoculants. For best stirring effect and to get a homogenised dispersion of the inoculant it should be
added when the ladle is 1/4 full. For this method there are generally three different ways of adding
the inoculant [28] [33].
• Additions of inoculant to the stream as the melt is poured into the ladle.
• Additions of finer grainsized inoculant to the stream by compressed air.
• Additions by inoculant wire that is added as the ladle is filled.
At Scania for their production of CGI cast irons they add the inoculant into the ladle after the ladle is
filled with a wire were the outer coating is made of a protective sheet of iron to make sure the
inoculant gets further down into the melt.
Adding the inoculant when casting is a method mostly used in foundries where they use mechanical
or automatic casting devices. The upside with this method is that the fading effect is minimized, and
less inoculant is there for needed. On the downside is that the particles might not have time to
dissolve into the melt and the inoculant will not be evenly dispersed.
13
There are as well here three different ways of adding the inoculant;
• Stream of fine inoculants particles by compressed air that is added to the melt stream as the
mould is filled. This is the method of inoculation additions for Scania grey cast iron when
producing their cylinder heads.
• Additions by wire to the melt stream.
• Additions of the inoculant straight into the mould. In this case the inoculant is placed in the
runner or the inlet to the mould. When the melt passes by the inoculant is successively
dispersed into the flowing melt.
One cannot say that one method is more superior as another, it all has to do with what type of
production you are running and at what scale. They all have pros and cons and the best way is to try
to find an option that is best suited for ones needs [28].
Material charged
The starting value of nuclei in the melt comes from the material that is being charged in the oven. Pig
and wrought iron are both claimed to be rich in nuclei compared to returns from in-house production
so a higher percentage of these materials in your charge will give you a higher starting value. This is a
known casting “hack”, to ad pig iron with rust on it to a dead melt to oxidize it and ad nucleation
points to the melt. By using thermal analysation, the level on nucleation additions are adapted to the
level present in the melt, this is called dynamic nucleation and is
Defects from improper inoculation additions If the melt is not properly inoculated could result in a variety of defects on the finished cast goods.
The ones most prominent for CGI are [34]:
• To large amount of additions with a high concentration of calcium will generate slag-like
inclusions.
• To large grainsize or inoculation at low temperatures, below 1300 °C can result in inclusions
which will cause problems when processed.
• Inoculation with high concentration of aluminium can give rise to “pinholes” especially when
casting in sand forms.
• Lack of nucleation points can promote white solidification, which will lead to porosity in the
material. The cementite will not expand as the graphite will.
Thermal analysis Thermal analysis can be used to measure the inoculant level in the melt, but a lot of other
information can be interpreted from the temperature variation over time and its derivative. By using
computer analysation, the alloy composition, grain refining, eutectic morphology, fraction solid,
amount of the phases can be given or predicted. The inoculations effect can be seen by the decrease
in undercooling and this knowledge is used to see how well a melt is inoculated. Thermal analysis is
based on the knowledge that every event that occurs in the solidification of a material leaves its
14
mark on the cooling curve. Phase changes, both solid state and liquid to solid can be seen on the
cooling curve by the latent heat of solidification realised. The cooling curve can be divided up into
four sections to better describe what occurs during solidification, seen in figure 8. In the first section
the metal is present as a melt, called the liquid area. Here stable oxides and sulphides are formed
that will be formed as the temperature decreases and will be very important as nucleation points.
The second section is where the primary phase will solidify. The eutectic temperature is the starting
point of the third section and will continue until the melt is completely solidified. The fourth and final
phase only involves decrease in temperature [18] [21] [28].
Figure 8: The four different regions of solidification [21].
Thermal analysation can be done in different ways but the simplest is by using a thermocouple
inserted into a test mould. The shape of the curve comes from the balance between the heat from
the material realised to the surroundings and the exothermic heat created during the phase change
when the material solidifies. The geometry of the test cups is of most importance for the results
given by the curve and should enable consistent sampling conditions. Other factors that play into
effect is the tapping temperature, level of oxidation and the pouring time. It is proven that
superheating will lower the eutectic temperature, so it is of most importance. If one wants to
examine the level of inoculation of the sample, a smaller test cup is to be preferred [18].
The analysation method provided by Electro-Nite is by using two testing cups, one with and one
without inoculant particles in it. The melt is poured in the two cups and two cooling curves are
established. The difference in the eutectic minimum temperature between the two cups will tell us
the level of inoculant [18] [19].
When carbon that is dissolved in the melt is crystalized and form graphite of different morphologies
latent heat is realised. This heat realised up on solidification is 3600 joules/ gram for graphite,
compared to approximately 210 joules/ gram for austenite. This knowledge can be used to measure
the inoculation levels in the melt. Inoculants are used as nucleation points for the crystallization of
carbon into graphite [34].
Measurements done by [26], a measuring cup that holds 370 grams of melt and the specific heat for
cast iron at the eutectic point is 0.8 J/g/°C. If it is assumed that 1 gram of carbon crystalizes to
graphite the theoretical temperature increase would be 12 °C. The accuracy of the commercial
15
thermal analytic system area gives or take 1.2 °C, which means that it would be able to detect
precipitation of 0.1 gram of graphite. The eutectic graphite on total is about 1.4 % if the cast iron has
a carbon content of 3.4%, which would equal 5.2 grams in the sample cup. The precipitation will then
be roughly two percent. This would be a sufficient mount for the measuring equipment to detect
differences in inoculation levels.
The test cups are usually made either of metal or sand. Metal can give a more precise positioning of
the thermocouples, but the sand cups are cheaper. A test cup that has Tellurium added to it
promotes the metastable white solidification. The high growth rate of the white eutectic close to
eliminates the undercooling. These cups are used to establish the liquidus temperature of the melt.
Not only the cooling curve temperature over time can give us information of the solidification event,
but also the first and second derivate can give us an estimation of the grain refinement and the
propensity to micro shrinkage. If one compares LGI with SGI for example. LGI releases very little
latent heat at the end of solidification, which will give it a large increase in cooling rate. SGI realises
more latent heat and will not have such a large increase in cooling rate [18] [29].
Sinter cast process Sinter cast is a company that provides a thermal analytic tool to measure the level of Mg and
inoculant present in the melt that is used by Scania for their CGI products. Each oven equals to five
ladles that will be casted. The melt in the oven has no additions of Mg in it. Mg has a very high
affinity to air and is therefore very reactive and should be added at the latest stage possible of the
process. Also, the inoculant additions should be done as close in time to casting due to fading. These
additions are made to the melt in ladles just before they are casted. The first ladle gets a base
treatment of Mg and inoculants based on the last ladle from the previous oven’s values. Before a
sample is taken with a sinter cast sampling cup and analysed. It measures the modification index
(MGM) which is the level of Mg, inoculation index (MGI) and the carbon equivalent for the melt
based on thermal analysis. The measured values will be shown on the screen and depending on the
values It will tell if they are acceptable or not. Depending on the values for this ladle alterations in
terms of additions of inoculants and Mg will be made for the next ladle. But the results take several
minutes to get, and the operators will not wait for the results to be able to make alterations for the
ladle are working with [14].
The previous method is used for CGI400, for CGI450 and CGI425 the addition of Mg and inoculant is a
two-step process. First a base treatment is done on the melt, here 90% of the Mg is and 10%
inoculant is added. A sample is taken, and the process is stopped until the results of MGM, MGI and
CE are received. Based on the results received from the test more additions of Mg and inoculants are
made but now usually 90% inoculant and 10% Mg.
The sampling cup is mounted with two thermocouples, one in the centre of the cup and one close to
the bottom this to allow for simulation of Mg-fading. To get optimum measurements the
temperature of the melt should be as close to 1280 °C as possible. Figure 9 shows a Sintercast
sampling cup which looks quite different then a sand cup. The shape of the sampling cup will lead to
no oxidization and fixed volume and mass to get accurate measurements [14] [22].
16
Figure 9: Sample cup for Sintercast measurements [48].
.
Thermocalc Thermocalc can be used for a variety of applications and is a thermodynamic software used for
scientific and research use. In the case for the study of thermal analysis of the inoculation level in a
melt it can be used to compare the results provided by the Electro-Nite. The software is often used in
materials engineering to calculate [35] ;
• Phase diagrams, simulation of phase transformation
• Amounts of different phases and their compositions
• Liquidus and solidus temperatures
• Stable and metastable phase equilibria
• Collect thermal chemical data such as enthalpies, heat capacities and activities to name o
few.
• With the help of the Scheil-Gulliver model solidification of a melt can be simulated.
Normally Thermocalc is used for equilibrium calculations, for these cases it is assumed that the
diffusion between the solid and the liquid is complete. These calculations are very good and useful
for many applications but is proven not to be as accurate for the solidification process, here the
diffusion in the solid state is blocked. The Scheil- Gulliver model is a non- equilibrium calculation that
has been proven to show better results for these calculations and works good for typical cooling
rates seen in foundries. This model assumes infinitely fast diffusion in the liquid phase and no or low
diffusion in the solid phase and that the solid/liquid interface is in thermodynamic equilibrium [36].
The classic model that was mentioned above works very well for many alloys but not for steel and
Iron. Carbon that dissolves interstitially is a very fast diffuser in solid state and is a very important
alloying element for steel and iron. By assuming no diffusion in the solid state will not give an
accurate result. A second form of Scheil- Gulliver model can be used that allows one to pick elements
to be “fast diffusers” in the solid phase and this addresses the problem mentioned earlier [36].
17
Raw material
Inoculant
The inoculation used by Scania for its CGI products is Odermath- wire. The composition can be seen
in table 2. The inoculation that are added to the Electro-Nite sample cups is SMW605, the
composition can be seen in table 3.
Table 2: Odermath-wire composition.
Odermath- wire (9N23) Si [%] Al [%] Ca [%] Zr [%]
Min 72 0.8 1.2 1.2
Max 78 1.8 3 1.8
Aim 72 1.04 1.43 1.53
Table 3: SMW605 composition.
SMW605 %Si %Al %Ca %Bi %SE
Min 62 - 1.8 0.8 0.8
Max 68 1 2.4 1.2 1.2
Charge
For the three different CGI compositions different recipes are used when charging the oven. These
recipes are altered in terms of weight of ingoing material depending on what melt was run in the
oven previously and how much that is left the oven from previous run. If the melt run previously has
to high levels of certain alloying elements that should not be in the following the remaining melt will
be discarded into small ladle to be re-melted later. The charge is made up of four different
components;
1. CGI- chips
2. CGI- returns
3. Pig Iron- low phosphorus
4. Wrought iron- scrap iron
Experimental procedure
Method For each oven of melt which has a weight of approximately ten metric tonne equals five ladles that
will be cast into engine blocks or cylinder heads. The main goal is to establish the level of inoculant in
the oven, so the first ladle will get proper inoculation additions. The second ladle will get inoculation
additions by the Sintercast measurement tool based on the levels from the first ladle.
The level inoculation level will be established using thermal analysis, for these two different sampling
cups are used. One empty sampling cup with a fixed volume and an identical cup but with 0.05 wt%
inoculant added to it. The difference in the eutectic minimum temperature (DTEmin), will tell us the
level of inoculation.
18
All the samples taken during this investigation will be taken on from untreated melts, no magnesium
has been added to the melt to create CGI. The melt will have the morphology and structure of grey
cast iron.
The tests were conducted on all grades of CGI, though most test were done on CGI400 due to its
production volume. The grade of CGI will be mentioned when presenting the results, but due to the
very close composition between the three grades of CGI they are expected to behave the same way
and will be the same type of alloy when analysed unless big differences between them are noted.
Average starting value and nucleation fading First, the average starting level of inoculant should be established for the melt in the oven by taking
several measurements. Simultaneously Inoculation fading will be investigated by conducting tests
over time on ovens that are sitting due to stop in production.
For each measurement taken, three different sampling cups was used. The first cup has an addition
of tellurium in it. This measurement is taken on a different mount connected to a different computer
and software. The volume of the cups contains tellurium is different compared to the other cups. The
cooling curve is calculated by the software and is based on a fixed volume, so changing the cup size
will alter the result. This is used to establish the liquidus temperature for the melt that correlates to
the carbon equivalent, which is an important process parameter.
It is important so that we can tell that we have the right chemistry for the melt. It does not matter
what order you use the cups, but a good idea is to use the cup with tellurium first to see that the
melt have the right chemistry, so time and cups are not wasted.
The chemistry of the melt is checked with a spectrometer, this gives the amount of each component
in the melt. The carbon content observed by the spectrometer is not reliable because some carbon
will be burnt off during the procedure by the spectrometer. Therefore, the tellurium cup is used to
establish the CE and TL. The samples to establish the chemistry and the TL is taken after the melt has
been heated to 1400 °C. The results for the TL just take a few minutes to receive and depending on
the results additions of carbon will be given. The results from the spectrometer takes longer time and
the melt is heated up to 1460 °C and de-slagged while waiting for the results. Depending on the
results from the spectrometer, the chemistry might be adjusted though this is seldom needed after
years of fine tuning.
The second sampling cup which has no additions is taken after the melt has been de-slagged and the
chemistry and TL checks out. Heating of the melt will lead to stirring of the melt due to convection
which will lead to a homogenised melt. The samples are taken a few minutes after to ensure the
movement of the melt has stopped. The first samples taken from an oven after it is de-slagged is
considered as t=0, sample taken at time zero. This is the starting value of the melt.
19
Figure 10: Picture of the set up by the oven.
The testing cups with the melt are left on the mount until the instrument has taken out the eutectic
min and max temperature or when they can be seen on the curve if the instrument as failed to point
them out. The sample is removed from the mount, having a temperature of around 1000-1100°C and
is left to cool down for further investigations. The sample should not be left on the mount longer
then needed so the connectors do not get heated up before and can cool down before the next
measurement is taken.
The third cup with inoculation additions to it is mounted and a new sample is taken and the same
procedure as for the cup with no inoculation is conducted. A measurement takes approximately
maximum 5 minutes for the three cups. The set up by the oven can be seen in figure 10 and picture
of the two cups used for DTEmin measurements can be seen in figure 11.
To get proper measurements it is important that the tapping temperature does not exceed 1370 °C
and that it is at least 20 °C above the TL. The upper limit is not to exceed the range of the
thermocouple, so samples are most often cooled for a few seconds before being poured into the
sampling cup. This goes for all the measurements for all the different cups. The sample is taken from
the top of the melt in the oven using a ceramic dipping cup. When the sample is taken from the oven,
any slag at the surface of the oven is removed, the cup is dipped in the melt once to heat it up a rinse
out any impurities. An additional scoop is taken, some of the melt is poured back to remove any
unwanted slag that might float on the surface.
The cups are mounted on a heat resistant contact block that is connected by a type K extension wire.
The wire conducts the electrical signal from the thermocouple of type K (NiCr-Ni) with a range of 400
- 1370 °C to the Heraeus Electro-Nite Quick-lab-E measurement instrument connected to a computer
where the software Meltcontrol 2020 is installed. The software will display a cooling curve and from
analysing the curve and knowing how different properties of the melt would affect the curve and
what type of sampling cups is being used information can be gathered.
20
For ovens that has been sitting and cooled down, they are heated again to the correct measuring
temperature which is set to between 1400-1460 °C. These ovens are used to take measurements
over time to investigate nucleation fading. This can be done when there is a stop in the production
line.
Figure 11: Picture of the two cups used to measure the DTEmin. The one on the left is the cup with inoculant additions and the cup to the right is empty.
The difference in eutectic minimum temperature (TEmin) for the cup with and without inoculation is
calculated and this determines how well the melt is inoculated. Increased inoculation as mentioned
previously will increase the TEmin temperature, the undercooling will decrease. In figure 12 the two
solidification curves for the two different cups can be seen. It is the temperature difference between
the two when it is at its lowest before the temperature increases due to the exothermal reaction
occurring. These points are marked with a black dot in the image, and it is the temperature
difference between these two points that is donated as DTEmin. Several tests are done, and an
average starting value is calculated.
Figure 12: Cooling curve for cup containing inoculant and cup not containing inoculant. The black points show the eutectic minimum temperature points.
1050
1100
1150
1200
1250
1300
1
16
31
46
61
76
91
10
6
12
1
13
6
15
1
16
6
18
1
19
6
21
1
22
6
24
1
25
6
27
1
28
6
30
1
31
6
33
1
34
6
36
1
37
6
39
1
40
6
42
1
43
6
45
1
46
6
48
1
49
6
Tem
per
atu
re [
°C]
Time [s/10]
without inoculation Inoculated
21
Saturation point A melt has a saturation point for inoculation, after this point the TEmin will not be as large as
previously with additions of inoculants. This point can be seen in figure 13 where S is the saturation
point. After this point the temperature difference between the two cups with different amount of
inoculation will be small, this point is the goal to reach. If you have less, you do not have enough
inoculant additions and past you are spending money on excess additions of inoculation which is not
economical. According to Electro-Nite, the supplier of the measuring equipment the DTEmin should
be less than 2 °C for it to be considered saturated and if the difference is larger than 3 °C more
additions of inoculant is required. The yellow curve in the effect of increased silicon in the melt, the
green the effect of the inoculant which is FeSi and the blue curve is a composition of them both, the
total effect. This will be established by using the Mini-lab suitcase provided by Electro- Nite.
Figure 13: TEmin vs amount of nucleation. Showing the saturation point marked S.
Minilab-suitcase
Procedure
To find the saturation point of the melt, the Minilab- suitcase provided by Electro-Nite was used seen
in figure 14. Each melt will always vary a bit in composition, heating practices, raw material etc. and
this will affect the level on inoculation of the melt. The needed addition of inoculant will be different
to reach the saturation point for each melt. By using additional cups that have stepwise more
additions of inoculation; 0.05%, 0.10% and 0.15% one can find between which cups the difference in
eutectic minimum temperature is below the reference value for the saturation point. When that is
22
found the cup with the lower additions of the inoculant of the two cups compared will tell you how
much inoculant should be added to the melt.
Figure 14: The minilab- suitcase used to fill the testing cups with increasing amount of inoculant.
Equipment
The suitcase contains the tools to be able to prepare empty test cups with inoculants to be able to
conduct the tests described. It has a calibrated spoon that holds 65 milligrams of inoculant, this
equals to 0.05 weight percent of the sample that has a fixed volume and weighs 130 grams. For these
test Scania’s own inoculant 9N23 will be used instead of SMW605 that the pre-filed cups are filed
with. A piece of the inoculant wire used in production was cut off and the inoculants inside the wire
was removed. The size of the inoculant is desired to be in the range between 125 to 355 μm. To get
this size a sieve is used with a top filter with the mesh size of 355 μm and a bottom with a mesh size
of 125 μm. The inoculant is collected from in between these two filters to get desired size.
The test cups to be used for the minilab are the cups called QC4000Cov2T. These cups have two
pieces of tape that covers the inlet hole from both sides, and it is between these two pieces of tape
that the inoculant is contained. This is to simulates the stream inoculation and should give the
inoculation a homogenous dispersion in the melt. The cups that was at disposal for the test was the
QC4000Cov, which is the same type of cup except that it lacks the two, cover tapes. The first test was
done by pouring the inoculant straight in the hole, so it was at the bottom of the cup. To get a set up
closer to the cups with two pieces of tape, for the rest of the tests the inoculant was trapped
between two pieces of tape just above the inlet hole, see figure 15. This also to get a better
dispersion of the particles.
23
Figure 15: Preparation of the cups with increasing inoculation additions.
The spoon is used to scope up the right amount of inoculant and is poured in the funnel to make
filling easy and correct. A metal clip is put on the sample with the sample id and the amount of
inoculant added written on it.
Testing of inoculation addition in ladle When the amount of inoculation is determined with the help of the mini-lab suitcase this will be
tested for inoculation additions to the ladle. Inoculant wire used in the production will be cut in the
required length to get enough inoculant, this will be placed in the ladle. By placing the wire in the
bottom of the ladle will ensure good mixing when the melt is poured on top of it. The samples will
later be taken on the ladle’s way to the casting floor and the same method as previously will be used
to determine the DTEmin, see figure 16.
Figure 16: Pictures how samples are taken from the ladle
24
Laboratory preparation The samples chosen for further investigation of the microstructure must be prepared. Samples with
values ranging from high to low where chosen and it is the samples with no inoculation additions
that are to be analysed because they represent the actual melt in the oven. Some inoculated samples
will also be prepared but just for comparison. The sample is cut just below the thermocouple as seen
in figure 17. The sample is mounted into a Bakelite puck to make handling of the sample easier using
a Struers CitoPress-30. The sample is then grinded using Struers AbraPol-20 and then polished using a
using four different polishing surfaces going from coarse to fine with three different grades of
diamond paste, 6, 3 and 1 my meter. When the samples are finished being polished, they are rinsed
with ethanol, wiped with cotton soaked in ethanol and the rinsed with ethanol again before being
blow-dried with a hairdryer. After this a clean scratch free surface is achieved ready for further
investigation.
Figure 17: Picture of sample with a line demonstrating where the sample is cut for further investigation.
The samples are first investigated using a light optical microscope of type Zeiss Axio Imager.M2m to
look at the microstructure of the sample. When conducting further analysation on the samples only
the samples with no additions will be analysed because these samples represent the melt in the
oven. The programme axiovision is used to be able to analyse the pictures using the microscope on a
HP Z440 compute. With this imaging programme a mosaic picture, an image that is several pictures
put together to create on large image of the surface was used. The software also has a feature that
can determine the area of graphite present in the sample and the nodularity of the graphite.
Nodularity is not of interest, but an increase of inoculation points could lead to an increase in
graphite because it promotes graphite precipitation. The microstructure of the analysed sample is
different than from the finished CGI product. There has been no addition of Mg yet to the melt, so
the sample we look at are not CGI but grey iron. Therefore, the samples contain flake graphite some
nodularity with a ferritic background. The microstructure is analysed to see a difference from the
different samples in terms of primary precipitation of dendrites and graphite formation of the
microstructure, see figure 18. To be able to get a cell count of the image which is how the inoculation
level can be measured the sample has to be etched.
Thermocouple Sample cut
25
Figure 18: Microstructure of sample seen with LOM. The left with 2.5x zoom and the right with 10x.
Etching
Nital- etching
The first sample tested was etched with 2% Nital. The etchant was dropped on the sample and was
left to react with the substrate for 10 seconds. After that, the sample was rinsed with water and then
ethanol before blow-dried. The etchant removed the graphite in the sample and darkened the white
ferrite, but no cells could be detected. The same procedure was done again for an additional 10
seconds. The only difference was that the ferrite got darker, figure 19 shows the result after 20
seconds of Nital etching. You were able to detect small outlines of grain boundaries if you zoomed in
close enough, but not enough to get a good cell count. After some research, the best approach to be
able to see and get a cell count would by the help of colour etching.
Figure 19: Sample after Nital- etching for 20 seconds.
Colour etching preparation
You colour etch your sample to form a thin film on the surface which creates an optical interference
effect. When colour etching the samples to reveal the grain structure and size of the material an
etchant based on picrin- acid was used. This etchant has been proven to be the most effective to
26
reveal the cell structure of the material according to experiments conducted at Scania. In the images
one can tell where the concentration of Si is the highest, it is seen by a range of white to light blue to
dark blue in colour in the images. Si has negative segregation and will segregate to the first solidified
area, revealing which area solidified first. The areas that are browner in colour have less Si and have
probably solidified later. The focus is to reveal the grainsize and grain density of the samples.
500 ml of picrin-acid was measured in a glass vessel and 125 grams of sodium hydroxide weighed and
dissolved in the acid. The reaction is very exothermic so not all the sodium hydroxide was added at
once, but it was added as it was dissolved. The solution is very basic with a pH of around 13. The
solution was heated up to 90 °C with an induction stove using a thermometer to establish the
temperature. The temperature fluctuated between 85-95 °C, because of difficulties to fix the
temperature. A magnetic stirrer in the bottom of the vessel was used so the solution was well mixed
and homogeneous. The sample surface was submerged in the solution and left submerged first for
approximately nine minutes. The sample was rinsed in water, then with ethanol before being dried
with a blow drier. The sample was then investigated with a LOM to see if the sample needed more
etching or if it was sufficient. The etching time needed is very temperature and concentration
dependent. Being hard to get the exact temperature the time needed to etch can vary greatly. The
guidelines given state that for 80 °C 10-25 minutes is needed to get a clear cell structure and less for
higher temperatures. 90 °C was chosen per instruction from a supervisor at STC with long experience
of etching.
After investigating the sample using the LOM it was concluded that the sample was under etched and
is was etched again for three more minutes and then investigated again. This procedure went on
until the sample was properly etched. This time to etch a sample varied between 20-30 minutes
depending on time, temperature and the concentration of the etchant that became less effective
over time making the task a lot more difficult than projected. Figure 20 shows the corner of a sample.
Here both the segregation of Si as well as the grains can be seen after colour etching.
Figure 20: Corner of a sample after colour etching. The blue colour comes from SI segregation to the first to solidify region.
27
Image J An imaging software programme ImageJ is used to enhance the images. The programme has many
usable features such as adjustment of the colour and contrast to easier see the grain-boundaries and
the eutectic cells that we are investigating. By adding a scalebar in the axio-software that is
connected to the microscope. Length measurement can be taken in the image that is needed to
calculate grains per area and the average grain size. Figure 21 shows an enhanced image and an
image where the grain boundaries are filled in with red colour to easier identify the lines.
Grain density A way of determining if there was enough inoculant in the melt and compare the thermal analytic
results with the microstructure of the sample is by investigating the grainsize and the average grains
per area also known as grain density. These two measurements should correlate with each other.
Increased difference in eutectic minimum temperature should give us fewer grains per area and
therefore larger grains. The number of grains per area (Nf) is calculates using equation 4. F is the area
of the rectangle that you are investigating, Ni are the grains inside the rectangle and Nw are the
grains that intersect the sides of the area. The grains that are in the corners are not counted by
themselves but two adds up as one. To easier see the grain boundaries they are filled in with a black
line using windows paint.
𝑁𝐹 =𝑁𝑖 + 0.5𝑁𝑤 + 1
𝐹 Equation 4
Figure 21: Shows how lines are drawn using software paint to easier see the grain boundaries when calculating
size and density.
Grain size The average grain size can be measured by using the mean intercept method. A line is drawn across
the area of investigation and the length of the line is divided by the grain boundaries crossed. Four
lines are drawn, and an average length/ diameter of the grains are established. This method is most
reliable if the size of the grains does not vary too much.
28
Figure 22: How the lines are drawn to calculate the average grain size.
For both above calculations the sample has be colour etched to reveal the grains- boundaries and the
imaging software image J is used to measure and to enhance the images. Two images are taken for
each sample investigated and an average value for each sample is taken from the average of the two
images. A total of 12 samples are etched and examined. An example of this can be seen in figure 22.
Additional testing Other tests were conducted to see how different foundry practices affects the inoculation level of
the melt.
• How carbon and wrought iron additions will affect the levels of inoculant
• How increased oven temperature will affect inoculant fading.
• To ensure good measurement procedures, test how de-slagging will affect the results.
• How the inoculant levels change as the level of melt decreases in the oven. This will be done
both taking samples from the ladle as well as from the oven.
• Testing the difference in DTEmin between the inoculant used in the cups and used in
production.
Chemical analysis For each oven, a chemical analysis is done with a spectroscope. A sample for analysis is first taken in
the oven and then from the first ladle. This is done for every oven, but we selected the following
samples to be able to find some correlation. The composition of the charge for some selected
samples were investigated to see if a correlation between the composition of the charge and
nucleation level could be established.
29
Thermocalc The thermodynamic simulating software Thermocalc is used to create both a metastable phase-
diagram that will simulate what happens in the cup containing tellurium that promotes the formation
of cementite. A stable phase-diagram will also be simulated that should be more accurate for the
melt solidifying in the other cups. The diagrams are binary Fe-C diagrams which do not consider all
the alloying elements, they will only be relative accurate when considering the carbon equivalent
(CE) which takes into consideration P and S.
To consider the alloying elements the Scheil- Gulliver model for solidification will be done using the
Thermocalc software to calculate TL for the melt. This model will also be done but by using the
carbon equivalent for both stable and metastable solidification. Carbon will be picked as a fast
diffuser when running the calculations. An average CE was established from 12 samples of CGI. The
amount P and S present in the melt is very low and the difference between them insignificant so the
CE can be considered the same for both. An average for the alloying elements for the three different
alloys is also taken and used for the calculations.
Comparison of TL for cups To be able to minimise the cup usage, time, and equipment the TL measurement using an ordinary
tellurium cup is compared with the TL registered using the testing cup without inoculant. These two
cups are connected to two different computers that considers the volume of the cup which is
different for the two cups and which is needed to calculate the TL. If a correlation can be derived
how the temperature differs between these two cups, only two instead of three cups needs to be
used for this process.
Results For each test taken only one DTEmin could be established, due to time limitation when taken test in
the production. No standard deviation can therefore be established for the test which limits the
reliability of the tests and sources of error are hard to establish. The point of the results is to show if
this measuring technique is sufficient and the value of the results are less important.
Level on inoculation in oven Table 4 shows the starting average level of inoculation in the oven and the max in min values
registered for the grades of CGI at T=0. The CGI400 average is based on 19 melts, the CGI450 is based
on four samples and CGI425 on five, but due to the very similar chemistry a combined average can be
used. Table shows that CGI425 and 450 have similar values and that CGI400 has a higher average
value.
Table 4: Max, min, median and average DTEmin starting value for CGI400, CGI425, CGI450 and combined.
DTEmin [°C] CGI400 CGI425 CGI450 Combined
Max 12.8 7.4 7.3 12.8
Min 2.1 3.6 1.2 1.2
Median 6.3 5.0 4.7 6.3
Average 6.7 5.0 5.4 6.2
30
Inoculation fading over time Figures 23-27 shows the DTEmin over time also known as the inoculation fading. This was done for
six different ovens for different length of time seen in the graphs. The samples were taken as
mentioned previously in the section about experimental method. The figures show an increase of
fading at first, but after two to three hours most show a decrease in DTEmin.
Figure 23: DTEmin vs time. Illustrates inoculant fading over an eight-hour period.
Figure 24: DTEmin vs time. Illustrates a five-hour test on inoculant fading.
7,37,7 8,6
8,3 8,2
5,5
6,5
5
5,5
6
6,5
7
7,5
8
8,5
9
0 1 2 3 4 5 6 7 8 9
DTE
min
[°C
]
Time [h]
4,3
6,6
5,2
6,3
4,8
4
5
6
7
8
9
10
11
0 1 2 3 4 5 6
DTE
min
[°C
]
Time [h]
31
Figure 25: DTEmin vs time. Illustrates inoculant fading over a four-hour test.
Figure 26: DTEmin vs time. Inoculant fading over a two-hour test.
Figure 27: DTEmin vs time. Inoculant fading over a two-hour test.
5,7
6,6
10,4
4
5
6
7
8
9
10
11
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5
DTE
min
[°C
]
Time [h]
7,1
8
4
5
6
7
8
9
10
11
0 0,5 1 1,5 2 2,5
DTE
min
[°C
]
Time [h]
1,2
6,5
0
1
2
3
4
5
6
7
8
9
10
0 0,5 1 1,5 2
DTE
min
[°C
]
Time [h]
32
Saturation point Figures 28-32 shows the temperature where the eutectic undercooling is reached for the different
levels of inoculation. The samples were taken from the same oven over an eight-hour period. The
lower value of inoculant of the two cups being compared shows the level of inoculation additions
needed. All test conducted shows that when comparing the cup containing 0.05wt% inoculant with
the cup containing 0.1wt% the saturation point is meet which is reached when the difference is less
than 3 °C. This was for the CGI450 composition.
Figure 28: TEmin vs inoculation to find the saturation point, test 1.
Figure 29: TEmin vs inoculation to find the saturation point, test 2.
1140,9
1146,8
1146
1148,1
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
0 0,05 0,1 0,15 0,2
TEm
in [
°C]
Inoculation [wt%]
1138,2
1144,6
1146
1147,7
1136
1138
1140
1142
1144
1146
1148
1150
0 0,05 0,1 0,15 0,2
TEm
in [
°C]
Inoculation [wt%]
33
Figure 30: TEmin vs inoculation to find the saturation point, test 3.
Figure 31: TEmin vs inoculation to find the saturation point, test 4.
Figure 32: TEmin vs inoculation to find the saturation point, test 5.
1139,2
1142,5
1145,2
1147,1
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
0 0,05 0,1 0,15 0,2
TEm
in [
°C]
Inoculation [wt%]
1140,8
1144
1146,4 1145,8
1140
1141
1142
1143
1144
1145
1146
1147
0 0,05 0,1 0,15 0,2
TEm
in [
°C]
Inoculation [wt%]
1139,5
1145,21145,8
1147,5
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
0 0,02 0,04 0,06 0,08 0,1 0,12 0,14 0,16
TEm
in [
°C]
Inoculation [wt%]
34
Testing inoculation additions to the ladle Previous test showed that 0.05wt% inoculant should be added to reach the saturation point. 12
meters of Odermath inoculant wire was cut and placed in the bottom of 8 ladles. Every meter
contains 120 grams of inoculant, 12 meters give you 1440 grams of inoculant. Every ladle contains
1800 kg of melt which means the inoculant additions are 0.078wt%. The inoculant exchange will not
be 100% so a little bit more than 0.05wt% was added as told. These eight tests are conducted on 2
different ovens, three on the first seen in figure 33 on ladle 3-5, where the starting inoculation level
on the oven was recorded. The starting value is denoted as ladle zero in figure 33. Five tests on the
other seen in figure 34 on ladle 1-5. One of the cups had a hole in it so ladle number four is not
shown on the figure 34.
The figures show that changing the order of the cups for the samples had no evident effect on the
result. Figure 33 shows that even with additions of inoculant to the ladle, the DTEmin still increased
for all the ladles tested. The test was conducted on CGI400 and with an average starting value of 6.7
one can assume that the inoculation additions to the ladle seen in figure 34 also had no effect to the
DTEmin.
Figure 33: DTEmin for the five tests taken for the different inoculant additions.
5,3
8,1
8,2
8,1
0
1
2
3
4
5
6
7
8
9
0 1 2 3 4 5
DTE
min
[°C
]
Skänk
35
Figure 34:DTEmin for the five tests taken for the different inoculant additions.
Grain size and density Figure 35 shows how the grainsize changes as the DTEmin for the cup with and without inoculant
additions increases. This value is taken by using the intercept method on images taken from samples
that has been colour etched. Several images are taken from each sample and an average is derived.
Samples are taken with different DTEmin values go generate the plot. The figure shows that an
increase of DTEmin will increase the grainsize.
Figure 35: Grainsize vs DTEmin, how DTEmin effects the size of the grains.
The 𝑅2= value for the exponential fitted curve is 0.9338 and the model is based on equation 5.
Y = 126.83e0.1968x Equation 5
7,6
8 7,7
8,4
0
1
2
3
4
5
6
7
8
9
10
1 2 3 4 5
DTE
min
[°C
]
Skänk
0
200
400
600
800
1000
1200
1400
0 2 4 6 8 10 12
Gra
insi
ze [
µm
]
DTEmin [°C]
36
Figure 36 shows how the grain density changes as the difference in undercooling between the two
samples increases. Equation 5 is used to calculate the grain density and the samples for investigation
have been colour etched to reveal the grains structure as when calculating the grain size. An average
for each sample is taken by investigating several images. The figure shows that with increased
DTEmin the grain density will increase.
Figure 36: Grain density vs DTEmin, how increased DTEmin effects the density of the grains.
The 𝑅2= value for the fitted exponential curve is 0.8545 and the model is based on the equation 6.
y = 56.988e-0.36x Equation 6
Addition of scrap iron The thermal analysis before and after 100 kg of scrap (wrought) iron is added to melt weighing
approximately 10 000 kg that has been sitting for eight hours can be seen in figure 37. The figure
shows that addition of iron will decrease DTEmin excessively.
Figure 37: DTEmin before and after 100 kg of wrought iron additions.
0
5
10
15
20
25
30
35
40
0 2 4 6 8 10 12
Gra
in d
enis
ty [
n/m
m2
]
DTEmin [°C]
6,5
3
2,5
3
3,5
4
4,5
5
5,5
6
6,5
7
0 20 40 60 80 100
DTE
min
[°C
]
Scrap Iron [kg]
37
Effects of inoculants used Difference in the eutectic minimum temperature between the inoculant provided by Electro-Nite and
the inoculant used by Scania for the CGI production line can be seen in table 5. This was done by
comparing the sample cup without inoculant additions with one cup containing 0.05% SMW605 and
one cup containing 0.05% 9N23. It shows that there is a difference in what value of the eutectic
minimum temperature that they portray.
Table 5: Shows the difference in TEmin temperature for the two inoculants used.
SMW605 9N23 DT
1148.2 1146.8 1.4 1146.8 1144.6 2.2 1147.5 1142.5 5.0 1146.3 1144.0 2.3 1146.0 1145.2 0.8
Effects of slagging Figure 38 shows the effect of slagging has on the DTEmin which is important to document too ensure
measurement accuracy. The melt was heated up to 1400 °C and the test was taken the same time as
the operators take the spectrometer sample and the TL for the melt. The second test was taken after
the melt had been heated up additionally to 1460 °C and the slag had been removed. The figure
shows that there is no difference in the DTEmin if the sample is taken before and after de-slagging if
they are taken within in a short time interval.
Figure 38: DTEmin before and after slag is removed from the melts surface in the oven.
Effect of carbon additions to DTEmin Figure 39 shows DTEmin before and after the addition of four cups of carbon addition to lower the TL
that was at 1144 °C. The second sample was taken 15 minutes after the carbon addition. The figure
shows that 15 minutes after carbon has been added it will not have any effect on the DTEmin.
5,2 5,4
0
1
2
3
4
5
6
7
8
9
10
0 1
DTE
min
[°C
]
De-slagging
38
Figure 39: Shows DTEmin before and 15 minutes after carbon additions.
Increased temperatures effect on DTEmin Figure 40 shows the DTEmin before and after the oven has been idling at a temperature of 1570 °C.
The figure shows that an increased temperature of the oven will lead to a rapid decrease of the
inoculant level.
Figure 40: DTEmin vs time. The fading of inoculant over one hour at 1570 °C.
Comparison of nucleation level between ladle and oven Figure 42 shows a comparison of the level of inoculation when taking samples from the oven after
each ladle removed and samples taken from each ladle taken from the oven. The starting value
“zero” is equal to the starting value in the oven when no ladle has been removed which should be
the same value as the first ladle will have. The measurements from the two different lines are not
from the same oven so the value of the numbers should not be compared but the “behaviour/
4,74,5
0
1
2
3
4
5
6
7
8
9
10
0 1
DTE
min
[°C
]
Carbon addition
7,5
20
0
5
10
15
20
25
0 0,2 0,4 0,6 0,8 1
DTE
min
[°C
]
Time [h]
39
tendency” of the level of inoculation should. Both lines in the figure show the same tendency to
increase and decrease of the inoculation level.
Figure 41: DTEmin vs ladles taken from oven. The trend between the test taken from the ladle and the oven.
Chemical analysis Table 6 shows the chemical composition of ten CGI400 batches taken both from the oven and from
the first ladle using a spectroscope. Table 7 shows the chemical composition of five CGI425 ovens
and table 8 the composition for three CGI450 ovens. The average composition for the three different
types of CGI can be seen in table 9. The samples for the spectroscope are taken when the melt has
been heated up to 1400 °C. When comparing the composition for each grade of CGI no correlation
could be seen between amount of each alloying element and DTEmin.
Table 6: CGI400 composition for 10 samples.
C [wt%]
Si [wt%]
Mn [wt%]
P [wt%]
Cr [wt%]
Mo [wt%]
Cu [wt%]
Sn [wt%]
Ti [wt%]
DTEmin [°C]
3.63 2.22 0.343 0.025 0.0245 0.0042 0.864 0.0615 0.00401 12.8
3.71 2.23 0.338 0.028 0.023 0.0037 0.852 0.0614 0.00464 7.1
3.76 2.34 0.348 0.027 0.024 0.0045 0.853 0.0585 0.00547 6.2
3.56 2.18 0.378 0.027 0.039 0.0098 0.851 0.0545 0.00635 2.3
2.68 2.22 0.405 0.028 0.025 0.0033 0.886 0.587 0.00683 5.2
3.61 2.25 0.402 0.029 0.021 0.0031 0.916 0.0655 0.00737 2.1
3.58 2.25 0.434 0.028 0.026 0.0040 0.899 0.0576 0.00868 7.8
3.68 2.26 0.299 0.028 0.024 0.0049 0.885 0.0545 0.00403 4.4
3.62 2.26 0.391 0.247 0.039 0.0097 0.901 0.0545 0.00616 6.7
3.60 2.29 0.399 0.028 0.026 0.0051 0.896 0.0558 0.00696 5.7
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5
DTE
min
[°C
]
Ladles taken from oven
Oven Ladle
40
Table 7: CGI425 composition for 5 samples.
C [wt%]
Si [wt%]
Mn [wt%]
P [wt%]
Cr [wt%]
Mo [wt%]
Cu [wt%]
Sn [wt%]
Ti [wt%]
DTEmin [°C]
3.68 2.21 0.398 0.0266 0.0238 0.00382 0.891 0.0664 0.00588 4.0
3.62 2.21 0.342 0.0255 0.0255 0.0072 0.897 0.0586 0.00427 7.4
3.62 2.20 0.314 0.0273 0.03 0.0106 0.928 0.0693 0.00406 3.6
3.60 2.30 0.327 0.0259 0.0236 0.00226 0.958 0.0744 0.00611 5.2
3.51 2.22 0.269 0.0259 0.0217 0.00291 0.939 0.0704 0.00658 4.7
Table 8: CGI450 composition for 3 samples.
C [wt%]
Si [wt%]
Mn [wt%]
P [wt%]
Cr [wt%]
Mo [wt%]
Cu [wt%]
Sn [wt%]
Ti [wt%]
DTEmin [°C]
3.64 2.30 0.304 0.027 0.0245 0.00653 0.99 0.0854 0.00481 6.3
3.57 2.29 0.336 0.027 0.032 0.00324 0.988 0.0824 0.00652 1.2
3.66 2.37 0.321 0.0258 0.0231 0.00631 0.978 0.0896 0.00623 7.3
Table 9: Average composition for CGI.
Charged material Table 10 shows the average amount of each material that is added to each charge of the different
compositions of CGI. This is based on seven charges of CGI400, three of CGI 425 and four CGI 450. It
shows that CGI450 and 425 have more pig and scrap iron in the charge and a lower DTEmin.
Table 10: Average amount of the ingoing material in a charge for the different CGI compositions and the average DTEmin for the samples used..
Grade of cast iron Pig iron [kg] CGI- returns [kg] Scrap iron [kg] DTEmin [°C]
CGI450 2903 2672 3602 5.4
CGI425 2947 2763 3680 5.2
CGI400 2624 2621 3394 6.8
Thermocalc Figure 42-43 show the stable and the metastable phase diagram for a Fe-C alloy. Table 11 shows the
both liquidus temperature and the temperature when the melt has completely solidified when using
the Scheil- Gulliver model with a CE of 4.23%. It shows the different phases, the difference in
eutectic and eutectoid temperatures and when the two systems are completely solidified. The
metastable solidification which needs increased undercooling has a lower temperature for all the
above.
Average
C [wt%]
Si [wt%]
Mn [wt%]
P [wt%]
Cr [wt%]
Mo [wt%]
Cu [wt%]
Sn [wt%]
Ti [wt%]
CGI400 3.64 2.24 0.37 0.027 0.0027 0.0052 0.88 0.058 0.0060
CGI425 3.61 2.23 0.32 0.027 0.025 0.0054 0.92 0.068 0.0054
CGI450 3.62 2.32 0.32 0.027 0.027 0.0054 0.99 0.86 0.0059
41
Figure 42: The stable Fe-C phase-diagram constructed using Thermocalc.
Figure 43: The metastable phase-diagram constructed using Thermocalc.
Table 11: The liquidus temperature and the temperature when the metal is completely solidified for the stable and metastable system using Thermocalc.
Solidification TL [°C] 100% solidified [°C]
Stable 1165.8 1153.4
Metastable 1165.8 1148.2
42
Comparison of cups Table 12 shows the difference between the TL measurements between the normal testing cup used
to establish the inoculation level and the tellurium cup used in production. Ten samples were used
for the experiment with a combination of the different grades of CGI, for this experiment the grade
does not matter because it is the difference between the values that are of importance.
Table 12: Difference in liquidus temperature for the testing cup with no inoculant additions and the tellurium cup.
TL, testing cup [°C] TL, tellurium [°C] DT [°C] 1139.2 1137 2.2 1137.3 1136 1.3 1140.6 1134 6.6 1139.1 1142 2.9 1140.5 1145 4.5 1140.9 1139 1.9 1141.9 1140 1.9 1141.4 1145 3.6 1140 1140 0 1140.5 1143 2.5
The average temperature difference between the two cups were 2.75 °C.
Discussion
Average inoculation level The average inoculation level was established for all CGI compositions. For CGI400 the average was
6.7 and the median 6.3 when analysing 19 tests at time t=0. For CGI425 and CGI 450 the value was a
bit lower with a median and average of 5.0 for CGI425 and a median value of 4.7 and average of 5.4
for CGI450. The results for CGI450 and CGI425 are only based on nine measurements combined. The
only reason that fewer tests has been done on these compositions is that more CGI400 products are
produced at Scania and there for more ovens are run that test can be conducted on. CGI425 has 25%
lower average and CGI450 20% lower average then CGI400.
A few tests gave very high numbers, the highest being 12.4 and some low, the lowest being 1.2. No
clear explanation can be given to these deviating results. The very low ones could be due to
operators adding carbon just before the test without notifying the tester. This would give a what is
called a “fake” inoculation level. This would only last a few minutes before dropping. One thought
was that perhaps slag had been trapped in the testing cup, but no slag particles could be seen in the
microscope.
For a melt to be considered to have a sufficient level of inoculant the value of DTEmin should at least
be below three and preferable below two. Both the values of 6.7 and 5.2 are far from ideal but the
values were to be expected. The Sintercast ads inoculants to every ladle so it would be concerning if
this was not the case. This is the average starting value in the oven and it is not until the melt has
been poured into the ladle that the level will be adjusted just before being cast.
Saturation point The starting inoculation level for the melt is too low as mentioned previously so the level of
inoculation addition to reach the saturation point is of most interest. The tests conducted shows that
for all tests with DTEmin with starting values ranging from 3.2-6.4 the saturation point could be
found between the cups with 0.05 and 0.1wt% additions as seen in table 13. This shows that an
43
addition of 0.05wt% inoculant is needed to reach the saturation point where DTEmin is less than
three. This also proves that a big difference in DTEmin like three and six is not so big when it is
converted to inoculation additions.
Table 13: DTEmin for the five tests taken for the different inoculant additions. The t denotes hours after the first measurement was taken.
Testing of inoculant additions to the ladle The results from this test was not at all as expected. When conducting test on the first oven the
starting DTEmin for the oven was 5.3, because of technical difficulties the first sample was taken first
on ladle three. Ladle three, four and five all showed similar values but the DTEmin was higher than
the starting value in the oven after inoculation had been added. When samples are taken, the first
cup without inoculation is filled just as the ladle is moving from the oven to the casting floor. The
second sample is taken when the TEmin can be identified by the software and the melt for the cup
with inoculations is taken roughly at the same time as the ladle reaches the cast floor, approximately
1 minute later. One reason could be that perhaps the inoculant had not had time to mix properly
when the first sample was taken, leading to low TEmin values. For the second test the order of how
the samples were taken were switched, now the cup with inoculant was filled first. The result was
the same. When comparing figure 33 and 34 the difference between the lowest and the highest
value is 0.6 which can be neglected as measurement error.
One would assume that the inoculant had enough time to dissolve because when the inoculant is
added in production it is feed quickly into the melt and then casted shortly thereafter. In this case
the melt is poured on top of the inoculant which should lead to good mixing and a homogeneous
melt. One thought is that the inoculant is oxidized too quickly. A little bit of melt is almost always left
in the ladle and when the inoculant is added to the left-over melt fading will start to occur. The time
from when the inoculant is added to when the melt is poured is just a few minutes. One last reason
that perhaps the easiest is when the melt is poured in excess temperature of over 1500 °C the
inoculation fading will be very rapid which is seen in other test conducted. This together with the
other factors mentioned is most likely the reason for the poor inoculant exchange to the melt.
When comparing the inoculation index on the Sintercast software, no increase could be detected on
the ladles that were treated with additional inoculation compared to the ladles that was not. This
tells us as well that this way of adding inoculant is not working.
The test shows that this is not a sufficient way of adding inoculants to a melt.
Inoculation fading over time When analysing the results of how the DTEmin changes over time some trends can be noticed. For
the first two hours for all the test conducted there is an increase in DTEmin, the level of inoculant has
decreased, fading has occurred. Up to three hours fading has occurred in most of the test, but after
four hours after first seeing a fading of inoculation a decrease of DTEmin can be seen in all tests. This
was not expected. If we look at the eight-hour test conducted seen in figure 23 a steep increase in
DTEmin can be noticed up to 2.5 hours from T=0. Between 2.5- 6 hours a small decrease can be seen.
Inoculant additions t=0, DTEmin [°C] t=2.5, DTEmin [°C] t=4, DTEmin [°C] t=7, DTEmin [°C] t=8, DTEmin [°C]
0-0.05 5.9 6.4 3.3 3.2 5.7
0.05-0.1 0.8 1.4 2.7 2.4 0.6
0.1-0.15 2.1 1.7 1.9 0.6 1.7
44
This small difference could be due to measurement error, but after six hours a large decrease can be
seen that cannot regarded as a measurement error.
One possibility why there is a decrease in DTEmin after a period could be because of the erosion of
the lining in the oven. The lining used in the ovens are provided by Calderys and is of the type SILICA
MIX Q 16. This ceramic as the name entails is based on Si which can also be found in the inoculation
agent. It is a known fact that the lining of the oven will erode over time and high temperature,
stirring and long exposure to high temperature will increase erosion. This lining is changed every six
weeks as an average and if the weight of lining that is eroded is divided by the ovens run
approximately 1.1-1.2 kg of the ceramic is goes into the melt. This would amount to 0.012wt% of the
melt. So, one thought is that when the oven has been sitting for a long period of time, the melt has
cooled down, reheated for a longer period Si from the lining will be realised to the melt.
Other reasons could be that over time the melt at the surface start to oxides increasing the
nucleation points, especially after the slag is removed that protects the melt. Another is that slag
particles dissolve in the melt, but this has not been seen in the samples viewed in the LOM.
The increase in DTEmin if comparing the starting value with the highest is not that big. A bigger
difference was expected, the largest difference seen was 5.3 and for that test the starting value was
very low, below the saturation point.
Grain size and density When investigating the microstructure of twelve different samples with a DTEmin ranging from high
to low, one can clearly see the correlation. With increased difference in eutectic minimum
temperature between the two cups means less inoculation particles in the melt, which leads to fewer
nucleation points. Fewer nucleation points will lead to fewer grains that will have space to grow
larger. That is what the figure 35 and 36 shows us. Increased grain size should lead to a decrease in
grain density, which the figures show. The line fitted for the data points in figure 35 has an R2 value
of 0.9338 which is very high. The R2 tells us how well a model fits the data points. This means that
there is a 93% chance that by using this model that other points would be on that line. Using
equation 5 given an x value it will give a y value, so the point ends up on the line. This equation for
the model is exponentially growing, which means that the grainsize will grow exponentially with
increasing DTEmin. The line shown figure 36 has an R2 of 0.8545 which is also high and is also
exponentially fitted curve. No previous compression between DTEmin and grain size and density has
not been found to compare with.
Addition of scrap iron The addition of scrap iron is for ovens that have been sitting, idling for a longer period. This is done
to replenish the Si that has burnt off without altering the composition. Figure 37 shows that by
adding wrought iron you will decrease the DTEmin for the melt. By this addition you create
movement, this from the temperature difference between the melt and the iron. This will create
convection stirring which will homogenise the melt and the Si will work as nucleation points.
Effects of inoculants used When comparing the DTEmin for the two different inoculants a difference can be seen in the results.
The difference is rather small and could have to do with the accuracy of the measurement
equipment. The test was the difference is 5 °C cannot on the other hand be blamed on that. For best
accuracy, the inoculant used in production should be used. Different inoculants will have different
saturation points so the one used in production should be used to be able to establish inoculation
additions.
45
Effects of carbon additions As mentioned previously additions of carbon can give a fake nucleation level but it is presumed to be
short lived. During other test when measuring the DTEmin operators added carbon without notifying
the tester and it was noted that after the addition DTEmin would decrease. When adding the carbon
to the oven it is done by throwing granulated carbon on the surface of the melt in the oven and
increasing the temperature of the oven to stir the melt with help of convection. The carbon yield is
very poor and a lot of it gets stuck to the slag at the surface, this will lead to the fake nucleation level
when taking the sample from the top of the oven. But as the test showed, 15 minutes after stirring is
enough time for the level to even out. The dangers of this is that when the thermal analysis is done
on the casting floor from the ladle and there is fake nucleation level an insufficient amount of
inoculation will be added to melt.
Increased temperatures effect on DTEmin It is a known fact that increased temperature and time have a negative effect on the inoculation
level. As can be seen in figure 40 the DTEmin increased a lot faster and higher than had been seen
when the oven was left at 1400-1460 °C. The DTEmin went from 7.5- 20, stating the fact that the
fading of the nucleation level is very temperature dependent.
Comparison of the nucleation level between oven and ladle Only two tests have been done and they show the same trend seen in figure 41, but the increase and
decrease does not occur after the same ladle removed. The highest value can be seen in what is left
in the oven, that is poured in the last ladle to be discarded. The decrease of DTEmin can probably be
explained by fading over time, the increase is harder to explain. It could perhaps have to do with the
increased stirring after the oven has been tilted back and forth and the extra space in the oven leads
to more movement of the melt. Also, the reasons presented previously about erosion of the lining is
valid here. When comparing the samples taken from the oven and from the ladle itself, the same
tendencies can be seen, and these are taken from two separate ovens. The time factor for moving
the equipment from the oven to the casting floor limited the possibility of taken these
measurements from the same oven. The samples from the ladle are taken before any inoculation
additions.
If the inoculation level changes between each ladle, the one-step Sintercast method is not very
reliable. If each level has changing levels of inoculant you cannot base the additions on
measurements done on the previous ladle for the next. You can end up with a melt that does not
have enough inoculant which will decrease the mechanical properties of the material.
Chemical analysis When comparing the chemical analysis for the selected samples no correlation can be seen between
the samples in terms of variation of chemical composition and the DTEmin. The variation between
the samples are relatively small due to strict min and max levels set for each element that this small
variation should not influence the DTEmin sufficiently. The elements that was analysed was the
elements that have a min and max value set by Scania. Other trace elements were detected by the
spectrometer, but due to their small percentage they were neglected.
46
Material charged When no correlation could be seen by DTEmin and the chemical composition an idea came that
perhaps it could have something to do with the charge. Of the ingoing materials is known that scrap
(wrought) iron and pig iron have a positive effect on the inoculation level. An interesting observation
was made when comparing the different CGI compositions. On an average over 10% more pig iron
and 5% more wrought iron goes into CGI425 and CGI450 compared to CGI400 when comparing what
the weight of the ingoing materials. But when dividing the CGI returns on the total amount of weight
of a charge for the different compositions there was only roughly a 1% difference between CGI425
and CGI450 compared to CGI400. CGI400 having 1% lower amount of scrap and pig iron in the
charge, which should even if it is a very small difference contribute to the improved inoculation level.
Wrought Iron shows in our test conducted that it has a positive effect on the inoculation level and
though not tested in this project it a well-known fact that adding pig iron will increase the inoculation
level. It is an old foundry trick, that one ads pig iron to a melt that is “dead”. When looking at the
average inoculation level for the two, CGI425 has a 28% lower average value and CGI450 has 20%
lower. These higher starting values of inoculation probably comes from the higher amount of pig and
wrought iron being used.
The recipe for each melt is altered depending on what is left in the oven and what composition it
had. The software used to calculate amount of the different materials in the charge takes this into
account and alters the amount of each material going into the charge to ensure each meets the
standard requirement set by Scania. This makes it hard to compare different ovens. When analysing
these four main components making up the charge does not take into consideration the alloying
elements effect on the DTEmin.
Thermocalc When the liquidus temperature is taken in production to establish carbon content, tellurium cups are
used as mentioned previously which promotes metastable solidification. The eutectic temperature as
can be seen in figure 43 is 1148 °C. This is at equilibrium and, but it is not a process in equilibrium. To
get a solidification process started there must be undercooling, the temperature must be below the
solidification point. The temperature therefore where the melt starts to solidify is below the liquidus
temperature or the eutectic temperature for a eutectic melt. A system is in equilibrium when the
Gibbs free energy is equal to zero, then it is the most stable state. This is not on the phase changing
boundary, here there is great potential for a decrease in Gibbs energy which means that the system
is not in equilibrium.
The stable phase diagram is what should be seen in the other cups not containing tellurium. As seen
in the samples taken graphite particles can be seen and not cementite. Samples containing tellurium
was attempted to be cut in half, but due the hardness of the cementite it was not successful.
When producing CGI products, the TL should be in the range of 1140- 1145°C, for the samples that
was used to get an average CE for the Thermocalc measurements the average TL was 1143°C. This is
lower than the temperature both seen in the phase-diagram at a CE of 4.23 and in table 11 showing
the TL using the Scheil-Gulliver model. The same TL was derived both for the single point equilibrium
calculations and the non-equilibrium Scheil-Gulliver model.
The phase diagram can be explained as mentioned that even though this is the starting temperature
of solidification, the melt will not have completely solidified until a below this temperature where
the undercooling is greater. When the CE is so close to the eutectic point, the liquidus temperature
47
and the eutectic temperature will almost have the same value. This would in that case correlate well
with a TL of 1143°C.
Thermocalc is a very good for thermodynamic calculations and they have very good databases that
are based on empirical data and research. It is a good tool, but these are just predictions and not
reality and other thermodynamic simulation software will get other results depending on their
databases.
Comparison of cups When looking at the comparison of the TL for the two cups, the difference is not very large. The
average difference 2.74, the problem is that there are some samples were the difference is quite
large, 6.6. This would not be tolerated when there is only a 5 degrees interval. The software that
calculates the TL using the cups uses an equation that considers the volume of the cup, which is
different for the two cups. That together with the tellurium additions is what sets the cups apart.
Perhaps with more testing a correlation can be established between the cups but for the most
accurate reading of the TL the cups with tellurium should be used hence that is what they are
designed for.
Overall discussion When analysing the results gathered from all the different test conducted, is this a good method to
establish the inoculation level in the melt? I would say yes. The microstructure analysis shows that an
increase in DTEmin will have a negative effect on the grain size and density which will have a
negative effect on the mechanical properties of the finished product. If the result would not show
this there would be something wrong. The R2 is very high which increases the credibility of the
model used for the grain size and density.
A correlation can be suspected between the saturation model and the grain size model. The
saturation model has a logarithmic appearance, which is the invers of the exponentially fitted curve
for grain size. After the saturation point has been meet in a melt the TEmin difference between the
two testing cups are low, it reaches a minimum. The same can be seen for the grain density, it first
has a sharp decrease in grain density that later slows down not unlike the saturation of a melt but its
invers. A saturated melt will lead to an increase in grain density, which means a decrease in grain
size.
The starting level of inoculation is to low, which was no surprise. I gather that there are three
different alternatives to increase the level. Either a base inoculant treatment should be given straight
into the oven before being poured into the ladle. The benefits of this is that it is simple and easy to
do. The downside of it is that it must be mixed into the melt and due to fading time will be a factor
and should be poured into a ladle and cast right away.
The second approach would be to ad inoculant into the first ladle poured. This way one will get good
mixing of the inoculant when the melt is poured into the ladle with the inoculant additions, and
fading will not be an issue because it will be cast right away. The only problem with this method is
that the inoculant additions for the next ladle is based on the previous one. If the first ladle has high
inoculant values the next ladle will not get enough. The Sintercast system would have to be adjusted
for this.
The test conducted when adding inoculant to the ladle was not successful and why is unclear. Adding
inoculant to the ladle or straight in the mould is a proven technique that is used in industry. The
inoculant used in the test was probably not designed for this method of additions and other factors
mentioned earlier also contributed.
48
The third approach would be to be able to connect the results given by the testing cups with the
Sintercast system if that is possible. The results so far show that 0.05wt% inoculant is sufficient to
reach the saturation point. If this information could be relayed to the Sintercast system for the first
ladle, this would solve the problem. The other ladles can then be based on the test results provided
by the Sintercast system. as mentioned previously though is the issue with alternating inoculation
levels in the ladles taken from the oven. This is problematic if the treatment for the next ladle is
based on the previous, though when conducting tests 0.05wt% was enough to inoculate DTEmin
from a range of over 6 to 3. Still, it should be mentioned.
The test has proven when the samples should be taken and what effect material additions will have
on the melt both short and long term. Pig iron is known to increase the inoculant level though not
tested for this report. Scrap iron was proven to increase the inoculant level in the test conducted. Pig
iron is expensive compared to other resources and wrought iron comes from in-house returns and
not in great volume. Another cheap way of introducing inoculant should be investigated.
The main issue will be how to incorporate this into production. This will mean extra work and
knowledge required by the operators and will add some more time to the process. Minimize the
number of cups needed for the different measurements have been tested and should be further
investigated if it is possible. Even though this extra time and money for the extra equipment, it
cannot compare to the savings it can lead to with a decrease in discarded products.
Conclusion Thru out this project the main goal was to test how well thermal analysis works on establishing the
inoculant level and if it can be used in production. Now the inoculant level is measured in the ladle
but with the help of thermal analysis the hope was to retrieve this information already in the oven.
The conclusion that can be drawn from this project is the following:
• The DTEmin is above 3 which means that the starting level of inoculant is too low in the
oven.
• 0.05wt% inoculant additions should be added to reach the saturation point.
• The tests showed that an increase in DTEmin will decrease the grain density and increase the
grain size.
• Fading of inoculant occurred when the oven was sitting idling for the first 2-4 hours.
• Increased temperature increased the rate of inoculation fading.
• Carbon additions temporarily gave a false inoculation level that was dispersed 15 minutes
after.
• Additions of scrap iron in the charge will increased the inoculant level.
Before this tool is used in further research should be done how it can best be implemented into
production.
49
Further research As mentioned in the discussion further research should be conducted to see what the best way of is
increasing the nucleation level. We have the three options mentioned, either in the oven, ladle or
that the Sintercast system accounts for the low value for the first ladle. Other options are probably
available that have not been mentioned in this report.
Further investigations should be done in how the charge material can be best altered to get a better
inoculant level in the melt. Pig iron is proven to have this desired effect, but it is not very cost
effective. Is there any other in-house discarded material that can be used?
One result that was surprising when measuring the fading effect over time is that the inoculant level
increased after a certain period. On possibility of this was the deuteriation of the lining of the oven,
but more research should be put into this matter.
As mentioned earlier is that the goal of this report was to conclude if thermal analysis could be used
sufficiently to measure the inoculation level and an answer to that has been given. Further research
must be done on the accuracy of the tests. Several samples should be taken at the same time or right
after each other to see what the standard deviation is between the results. How accurate are they
and what is the source of error? This is hard to do when working and testing in production, but by
testing this in lab one limits the variables you cannot control and that why area able to just alter one
thing and see what affect it gives.
Sources of error When filling the sand- cups there might be some unwanted factors that might affect the results.
There might some variations in the fill volume which is used for the calculations and the tap
temperature that can affect the results. The sand cups geometry will alter, first the sand will absorb
the heat which will lead to the binder to soften which will change the geometry. Also the pour-in
filling technique will introduce impurities as the metal oxidizes when in contact with air and air
entrainment when filling is also of concern [38].
When collecting samples from the oven communication is key between the one taking the samples
and the operators. It happens that there are some miscommunications that can lead to unwanted
results. Such as carbon is added before the two samples are needed to compare the results. This can
almost always be detected in the results or it is noticed just as it happens. There is still a small chance
that something like this has gone undetected.
50
Acknowledgment I would like to give a special thanks with a deep sense of gratitude to the individuals that made it
possible for me to conduct and finish my master thesis.
Isak Hollinger and Jessica Elfsberg that gave me the opportunity to do my master thesis at Scania.
My supervisors at Scania: Mitra Basirat Engström PhD, Mostafa Payendeh PhD and David Lindström
PhD who guided me and made sure I could conduct all the test necessary to finish my thesis even
though times very tough.
Fareed Khan at STC, Scania who showed me how to colour etch and analyse my samples.
To all the operators and engineers at the foundry that showed me around an helped me with my
experiments. Without you it would not have been possible.
To my supervisor at KTH Björn Glaser who helped me with my report and made sure everything was
going well.
51
References
[1] European Comission, “Energy efficiency and Co2 Reduction in Iron and Steel Industry,” [Online].
Available:
https://setis.ec.europa.eu/system/files/Technology_Information_Sheet_Energy_Efficiency_and_CO2
_Reduction_in_the_Iron_and_Steel_Industry.pdf. [Accessed 22 06 2020].
[2] Sintercast, “Compacted Graphite Iron - Mechanical and physical properties for engine design,”
Sintercast, 2014.
[3] F. Fonseca, “Porosity Analysis on Compacted Graphite Iron,” Faculdade De Dngenharia, Universidade
Do Porto, Porto, 2017.
[4] M. König, “Microstructure Formation During,” Chalmers University of Technology, Gothenburg, 2011.
[5] Sintercast, “Compacted graphite iron- Material data sheet,” Sintercast.
[6] Total Materia, “Compacted graphite iron: part one,” 2017. [Online]. Available:
https://www.totalmateria.com/page.aspx?ID=CheckArticle&site=kts&NM=492 [12. [Accessed 25
february 2020].
[7] Gjuteriteknisk handbok, “Gråjärn,” 2020. [Online]. Available:
http://www.gjuterihandboken.se/handboken/3-gjutna-material/32-graajaern. [Accessed 25 February
2020].
[8] A. T. Kuzu, W. Wu, D. A. Stephenson, M. Bakkal, J. Hong and A. J. Shih, “High- Throughput Dry and
Minimum Quantity Lubrication Drilling of Compcated Graphite Iron,” Elsevier B.V., 2016.
[9] E. Nährström, “Relation between microstructure features,,” KTH, Stockholm, Sweden, 2015.
[10] D. S. Dawson, “Process Control for the Production,” Sintercast, Kansas City, 2002.
[11] Gjuteriteknisk handboken, “Kompaktgrafitjärn (CGI),” 2020. [Online]. Available:
http://www.gjuterihandboken.se/handboken/3-gjutna-material/33-kompaktgrafitjaern-(cgi).
[Accessed 3 march 2020].
[12] H. H. Hooshyar, “Thermal Analysis of Compacted Graphite Iron,” Chalmers University of Technology,
Gothenburg, 2011.
[13] Gjuteriteknisk handbok, “Framställning av kompaktgrafitjärn,” [Online]. Available:
http://www.gjuterihandboken.se/handboken/3-gjutna-material/33-kompaktgrafitjaern-(cgi)/333-
framstaellning-av-kompaktgrafitjaern. [Accessed 15 04 2020].
[14] Sintercast, Sintercast process training- Sintercast process control, Sintercast, 2015.
[15] ASM Specialty Handbook Cast Irons, Classification and Basic Metallurgy of Cast Iron, ASM
international, 1996, pp. 4-12.
[16] Atlas Foundry Company, “Understanding Cast Irons - Compacted Graphite Iron,” 2018. [Online].
Available: http://www.atlasfdry.com/graphite-iron.htm. [Accessed 23 february 2020].
[17] S. Shao Dr. Steve Dawson M.Lampic, “The mechanical and physical properties of Compacted Graphite
Iron,” Materials science and engineering technology, p. Abstract, 15 september 2004.
52
[18] D. M. Stefanescu, “Thermal analysis- theory and application in metalcasting,” International journal of
metalscasting, vol. 9, no. 1, pp. 7-22, 2015.
[19] Electro-nite, “Thermal Analysis of Cast Iron,” Heraeus, Electro-nite, Houthalen, Belgium.
[20] M. Darwish, Driving force for solidification, Beirut: American Univeristy of Beirut.
[21] A. Tadesse, “On the Volume Changesduring the Solidification of Cast irons and Peritectic Steels,”
Royal institute of technology, Stockholm, 2017.
[22] Sintercast, “SinterCast- Compacted graphite iron,” 2020. [Online]. Available:
https://www.sintercast.com/technology/compacted-graphite-iron/. [Accessed 8 February 2020].
[23] Katholieke Univerity of Leuven, “THERMODYNAMICS,” 2020. [Online]. Available:
http://2011.igem.org/Team:KULeuven/Thermodynamics . [Accessed 13 april 2020].
[24] M. Darwish, Nucleation and Growth, Beirut: American University of Beirut.
[25] Sintercast, “The Effect of Metallurgical Variables,” Sintercast.
[26] Gjuteriteknisk handbok, “Optimering av ympningsprocessen,” 2020. [Online]. Available:
https://www.gjuterihandboken.se/handboken/3-gjutna-material/31-oevergripande-om-
gjutjaern/311-ympning-av-gjutjaernslegeringar/3112-optimering-av-ympningsprocessen. [Accessed
06 febrauary 2020].
[27] D. Stefanescu, “Theory of Solidification and Graphite Growth in Ductile Iron,” in Ductile Iron
Handbook, American Foundrey Society, 2010, pp. 4-6.
[28] Gjuteriteknisk handbok, “Ympning av gjutjärnslegeringar,” 2020. [Online]. Available:
https://www.gjuterihandboken.se/handboken/3-gjutna-material/31-oevergripande-om-
gjutjaern/311-ympning-av-gjutjaernslegeringar. [Accessed 05 february 2020].
[29] V. Alexis, J. Philippe, R. Fredric and L. Patrick, “Optimizing the inoculation of a ductile cast iron using
thermal analysis,” Brno, 2009.
[30] B. L. H. F. J. L. C. H. A. C. A. B.-P. C. M. A. S. R. R. A. L. S. S. A. D. L. Nastac, Advances in the Science and
Engineering of Casting Solidification, New Jersey: John wiley and sons, 2015.
[31] B. Domeij, “Compacted Graphite Iron: On Solidification Phenomena Related to Shrinkage Defects.,”
Jönköping University, Jönköping, 2019.
[32] M. E. Glicksman, Principles of Solidification: An Introduction to Modern Casting and Crystal Growth
Concepts, New York: Springer, 2011.
[33] Gjuteriteknisk handbok, “Ympningsmetoder,” 2020. [Online]. Available:
http://gjuterihandboken.se/handboken/3-gjutna-material/31-oevergripande-om-gjutjaern/311-
ympning-av-gjutjaernslegeringar/3111-ympningsmetoder. [Accessed 03 March 2020].
[34] Gjuteriteknisk handbok, “Optimering av ympningsprocessen,” 2020. [Online]. Available:
https://gjuterihandboken.se/handboken/3-gjutna-material/31-oevergripande-om-gjutjaern/311-
ympning-av-gjutjaernslegeringar/3112-optimering-av-ympningsprocessen. [Accessed 27 february
2020].
[35] Thermo-calc, “Thermo-calc,” [Online]. Available: https://www.thermocalc.com/products-
services/software/thermo-calc/. [Accessed 03 04 2020].
53
[36] Thermocalc, “Scheil for the Solidification Process,” 2020. [Online]. Available:
https://www.thermocalc.com/products-services/software/scheil-solidification-simulation/. [Accessed
06 April 2020].
[37] ASKCHEMICALS, “SMW605,” 2010. [Online]. Available: file:///C:/Users/carwca/Downloads/smw-
605_eng%20(2).pdf. [Accessed 07 05 2020].
[38] P. P. S Dawson, “Thermal Analysis and Process Control for Compacted Graphite Iron and Ductile Iron,”
Sintercast, 2014.
TRITA ITM-EX 2020:462
www.kth.se