The Mintek process is a large-scale batch silicothermic process for extracting magnesium that operates at atmospheric pressure. It aims to overcome issues with an earlier Magnetherm process. In the Mintek process, the furnace must operate above 1600°C, potentially as high as 1800°C, to achieve an economically acceptable rate of magnesium extraction while maintaining low energy consumption. Several factors like temperature, feed recipe, slag depth relative to furnace diameter, and reactions in the arc attachment zone influence the process.
10. 4
Paper1
A BRIEF STUDY OF ENERGY EFFICIENCY IN ALTERNATIVE
ROUTES OF EXTRACTION OF ALUMINIUM AND MAGNESIUM
Mainak Saha1
1
Department of Metallurgical and Materials Engineering,National Institute
of Technology,Durgapur,India
(mainaksaha1995@gmail.com)
ABSTRACT
The production of Aluminium from its ores at present relies on the Bayer
(alumina production) and the Hall- Heroult (Al production) process. The
cost associated with alumina production and apparent disadvantages like
high energy consumption of the Hall-Heroult process have led to intensive
research to find alternative routes for Al production. The direct has been
thoroughly investigated as an alternative technique. Another alternative
includes the indirect carbothermal reduction route where alumina (or
aluminous ores) is first reduced to intermediate Al compounds before
reduced further to Al. Magnesium is a light metal that used in structural
applications, and also used as additive in chemical and metallurgical
industries. The current dominant route for producing magnesium is via the
Pidgeon and electrolytic processes,requiring intensive energy and high
capital costs.So,energy efficiency of alternative routes for Mg extraction
such as Mintek and solid oxide membrane processes(SOM) along with
Direct carbothermal process(only) for Al extraction is discussed and
highlighted in this paper.
KEYWORDS:
Direct carbothermal process for Al;Mintek process for Mg;Solid Oxide
process for Mg;Hall-Heroult’s process;Magnetherm process for Mg;Arc
Attachment Zone(AAZ)
11. 5
1.DIRECT CARBOTHERMAL REDUCTION PROCESSES FOR
EXTRACTION OF Al
The direct carbothermal reduction of alumina to aluminium has potential
for greater
greenhouse gases emission, compared to the Hall Heroult process. It is well
established that
direct carbothermal reduction of iron oxide, through blast furnace
technology, offers far
greater productivity and energy efficiency than any comparable electrolytic
process. One major difficulty associated with prior art carbothermic
reduction processes is that they have relied
upon electrically heated furnaces as a major source of energy. It is well
known that when electricity is produced by combustion of fossil fuels, only
a minor portion of the heat value of the fuel is recovered as electricity.
Accordingly, prior art electrically powered processes for carbothermic
reduction of aluminum fromAl are relatively wasteful of energy. In the first
step of this method aluminum carbide is produced, and a second step in
which aluminum carbide and alumina are reacted at a temperature higher
12. 6
than that of the first step to yield aluminum. Heated gaseous carbon
monoxide evolved in both steps is used to preheat the reactants. However,
rather than providing a stack reactor, first step is performed in a low
temperature zone and the second step in a high temperature zone, with the
two zones being at different locations on generally the same level. The
reaction is performed by circulating a stream of molten slag through
successive low and high temperature zones. The reactants are not heated by
partial combustion of carbon so that a major proportion of energy
requirements must be met by electricity. The production of aluminum by
the electrolytic Hall–Heroult process suffers from high energy
requirements, the release of perfluorocarbons, and vast greenhouse gas
emissions. The alternative carbothermic reduction of alumina, while
significantly less energy-intensive, is complicated by the formation of
aluminum carbide and oxycarbides. In the present work, the formation of
Al, as well asAl2 OC, Al4 O4C, and Al4C3was proven by experiments on
mixtures of Al2O3and activated carbon in an Ar atmosphere submitted to
heat pulses by an induction furnace. Thermochemical equilibrium
calculations indicate that the Al2O3-reduction using carbon asreducing
agent is favored in the presence of limited amounts of oxygen. The
temperature threshold for the onset of aluminum productionis lowered, the
formation of Al4C3is decreased, and the yield of aluminum is improved.
Significant further enhancement in the carbothermic reduction of Al2O3 is
predicted by using CH4as the reducing agent, again in the presence of
limited amounts of oxygen. Inthis case, an important by-product is syngas,
with a H2/CO molar ratio of about 2, suitable for methanol or Fischer–
Tropsch syntheses.Under appropriate temperature and stoichiometry of
reactants, the process can be designed to be thermo-neutral. Using
alumina,methane, and oxygen as reagents, the co-production of aluminum
with syngas, to be converted to methanol, predicts fuel savings ofabout
68% and CO2 emission avoidance of about 91%, vis-a-vis the conventional
production of Al by electrolysis and of methanol bysteam reforming of
CH4. When using carbon (such as coke or petcoke) as reducing agent, fuel
savings of 66% and CO2. Emission avoidance of 15% are predicted.
Preliminary evaluation for the proposed process indicates favorable
economics, and the required high temperatures process heat is readily
attainable using concentrated solar energy.
13. 7
Experimental tests:
A differential thermal analysis method had been applied
to study the aluminum–oxygen–carbon system at reduced
pressures at 1700–2200ͼC.A differential thermal analysis method had been
applied
to study the aluminum–oxygen–carbon system at reduced
pressures at 1700–2200ͼC
. The results indicated that the
direct reduction according to Eq. (1) did not occur.
Instead, Al was proposed to be formed by the following
steps occurring at progressively higher temperatures in the
order listed, resulting in the overall reaction
Al was proposed to be formed by the following
steps occurring at progressively higher temperatures in the
order listed, resulting in the overall reaction
In the present work, the carbothermic reduction ofAl2O3 mixed with
activated carbon was examined initially by thermogravimetry coupled with
gas chromatography of
gaseous products, and by heating the above mixtures in an
induction furnace.
14. 8
By thermogravimetry:
A mixture of Al2O3 and active carbon (Fluka 5105, ca. 85%) (molar ratio
1:3) in a graphite crucible was placed into the sample holder of a high-
temperature thermogravimeter (Netzsch STA 409) under a constant Ar
flow of 200 ml/min. Evolved gases were sampled every 3 min for gas
chromatography (MTI Micro GC P200, equipped with a MS 5A column
and a TC
detector). The temperature was raised at a rate of 40ͼC/min until 1550ͼC,
and then kept at this temperature for 10 h. The observed weight loss was
54.0% of that required and the amount of CO determined by gas
chromatography was 15% of the theoretical. The residue in the crucible
was identified by XRD to consist mainly of Al4C3. No Al was detected on
the walls of the reactor. This analysis was performed using a Philips X’Pert
MPD/DY636 instrument, and identification of peaks was carried out with
the Philips Analytical Software for XRD. In a similar experiment, the
reactants were kept for 17 h at
1766ͼC, resulting in a weight loss of 71.4%, again without
formation of Al.
Thermochemical equilibrium calculations were performed
using the CET85 and FactSage program codes,
15. 9
assuming closed systems. Results were expressed as mole
fractions against temperature, all at 1 bar pressure. Products
with mole fractions of less than 10
5
were not considered.
Reaction enthalphies were calculated using the data of the
NIST chemistry web-book
. Substantial reduction of
Al2O3
to Al was found to occur only above the melting point
of Al, 933.5 K, and to be almost complete only close to the
boiling point, 2767 K. Four different cases were examined
with respect to CO2
emission avoidance, fuel saving, and
economics—all for a proposed annual plant supplied with 0.102*106ton
Al2O3
—with initial reactants described by:
(a) Al2O3+4CH4+0.4O2
;(b)Al2O3+4C+0.6O2;
(c)Al2O3+28.9C+13O2
;(d)Al2O3+4C.
The carbothermic reduction of Al2O3
is simulated with
an initial reaction mixture of Al2O3+4C.
The excess of unused carbon appears as the product C(gr).
The yield of Al according to Eq. (8) is 85.4%.
17. 11
This paper calculates the theoretical minimum energy by assuming the
reactants enter and the by products leave the system at room temperature
and that molten aluminum leaves the system at 960°C. Changes in the
operating temperature of a cell have a minor effect on the theoretical
energy requirements.For example, operating changes of 100°C in aHall-
Héroult cell, operating in the range of700°C to 1,100°C, result in less than
a 1% change. Some studies assume that the gasesevolved during reduction
leave the system at the molten metal temperature. In these studies, the
theoretical minimum is 2.5% to 3% higher.Theoretically, it is possible to
capture all theenergy associated with these gaseous emissions.Three energy
factors must be examined in the production of aluminum; energyrequired to
drive the reduction reaction forward, energy required to maintain the
system at constant pressure and temperature, and energy required to change
the temperature of the reactants and/or products. The thermodynamics and
chemical equilibrium of reactions aredescribed by the Gibbs equation: ΔG
= ΔH - TΔS. The energy required to drive the reaction forward is the Gibbs
free energy (ΔG). For alumina reduction, ΔG is less than the heat of
reaction (ΔH) and additional energy (ΔH - ΔG) must be added to the
system to maintain the system temperature. Otherwise, the system would
cool as the reaction progresses. (Reduction cells operate at atmospheric
18. 12
conditions and no pressure change results from the reduction.) The energy
required to change the temperature of reactants and products is calculated
from their heat capacities (Cp). The theoretical Hall-Héroult reaction is
assumed to occur under perfect conditions, where there are no reverse
reactions, no parasitic reactions consuming additional anode carbon, no
limitations to the ionic species reacting at the electrodes, and no heat or
energy losses external to the system. The energy required to drive the
reaction forward (ΔG) is 5.11 kWh/kg, the thermal energy required to
maintain the system temperature is 0.49 kWh/kg and the thermal energy
associated with the molten aluminum is 0.39 kWh/kg Al. The theoretical
minimum energy requirement is 5.99 kWh/kg Al. Faraday.s law provides
the minimum amperage requirement for electrolytic reduction, 2,980 Ah/kg
Al. The Gibbs free energy divided by the Faraday amperage provides the
minimum voltage required to drive the reaction forward. Cell voltage and
current efficiency are variables that are controllable by design and they
determine the electrical power required for reducing alumina. In practice,
electrolytic cells have significant inefficiencies and operate above the
minimum voltage requirement. This excess voltage provides the thermal
energy required to maintain system equilibrium (ΔH- ΔG) and to produce
molten material (Cp).In actual carbon anode cell operations, current
efficiencies of less than 100% result from reverse oxidation reactions
between part of the aluminum metal that is dissolved in the cryolite and
carbon dioxide gas. . The high current efficiency of existing technologies
leaves little opportunity for lowering amperage. Since current efficiency is
high, lowering cell voltage presents the best opportunities for improving
energy efficiency. Carbon is consumed during the reaction process and
gives the process a lower theoretical energy requirement (7.32 kWh/kg Al)
than the direct reduction of alumina. If an 85% thermal and 95% reaction
efficiency were assumed a carbothermic reactor would require 9.07
kWh/kg Al, a 40% reduction in energy. Additional energy reduction could
come from capturing the fuel value of the carbon monoxide byproduct.The
carbothermic reaction results in the generation of carbon-based greenhouse
gases (GHG), mainly carbon monoxide (CO), at twice the rate of the Hall-
Héroult reaction.However, the carbothermic process requires electricity
only for heating and not for electrolysis. The total greenhouse gas (GHG)
19. 13
emissions from .utility-to- metal. for the carbothermic process are roughly
8% less than a modern Hall-Héroult cell.
Primary Aluminum Electric Energy Consumption 1900 to 2000, shows the
significant electrical energy
20. 14
improvements made between 1900 and 2000. Onsite electricity use varies
from less than 13 kWh/kg Al for state-of-the-art plants up to more than 20
kWh/kg Al (U.S. plants in 1995 averaged 15.4 kWh/kg Al (The Aluminum
Association,1998)). Compared to theoretical values, U.S.facilities are
averaging roughly 35% energy efficiency. There is a minimum cell
amperage required to produce aluminum (2980 Ah/kg Al).
21. 15
2.MINTEK PROCESS FOR EXTRACTION OF Mg
The Mintek Process is a large scale batch silicothermic process operating at
atmospheric pressure. It attempts to overcome the productivity and
operational difficulties associated with an earlier large batch process called
the Magnetherm process. Pechiney developed the Magnetherm process in
the 1950s and operated at 1550 degC under vacuum.In this process ,to
achieve an economically acceptable level of magnesium extraction and rate
of Mg(V) generation, the furnace operating temperature needs to be
increased above that employed in Pechiney’s Magnetherm processes .
Electrical energy consumption could, therefore, be higher .In the process
during DC-arc smelting of magnesium-containing raw materials at
atmospheric pressure is influenced by several factors, including
temperature, choice of feed recipe, slag depth relative to furnace diameter,
and the reactions in the arc attachment zone (AAZ). As indicated
previously, the furnace operating temperature needs to be above 1600°C; it
could be as high as 1800°C in order to achieve an economically acceptable
degree of magnesium extraction and rate of extraction and to allow
continuous slag tapping, yet keep the consumption of furnace electrical
energy as low as possible. As the slag bath is not agitated externally,
uniformity of the slag temperature could become an issue: segregation of
high-melting slag components and their possible freezing on the furnace
side-walls and/or on the hearth could take place. Were this to happen,
22. 16
continuous slag tapping might be interrupted.A simple way to overcome
such a potential difficulty is by optimizing the ratio of slag depth to furnace
internal diameter. Electrical energy consumption could, therefore, be higher
.In the process during DC-arc smelting of magnesium-containing raw
materials at atmospheric pressure is influenced by several factors, including
temperature, choice of feed recipe, slag depth relative to furnace diameter,
and the reactions in the arc attachment zone (AAZ). As indicated
previously, the furnace operating temperature needs to be above 1600°C; it
could be as high as 1800°C in order to achieve an economically acceptable
degree of magnesium extraction and rate of extraction and to allow
continuous slag tapping, yet keep the consumption of furnace electrical
energy as low as possible. As the slag bath is not agitated externally,
uniformity of the slag temperature could become an issue: segregation of
high-melting slag components and their possible freezing on the furnace
side-walls and/or on the hearth could take place. Were this to happen,
continuous slag tapping might be interrupted.A simple way to overcome
such a potential difficulty is by optimizing the ratio of slag depth to furnace
internal diameter. A schematic of the 100kg/h magnesium pilot plant is
shown in the next page . Theequipment consists of a 1.5 MW (10kA)
power supply, a DC arc furnace, a raw materialfeed system, a magnesium
condenser, a combustion chamber, a gas-cleaning system, and certain other
ancillary equipment. The airtight feed system consists of three surge
binsn(upper bins) and three weigh bins (lower bins), varying in size from
60 l to 550 l. Each pair of upper/lower bins is dedicated to deliver a specific
raw material to the furnace. The feed system is designed to feed 250 to 500
kg/h hot dolime (at up to 900°C), 40 to 80 kg/h ferrosilicon, and 15 to 60
kg/h aluminium to produce 50 to 100 kg/h magnesium vapour. The upper
bins are provided with argon and vacuum lines for purging, after charging
batches of raw materials to them. Airtight valves are positioned between
each pair of bins. Vibratory feeders are used to charge the ferrosilicon and
aluminium, while a rotary feeder is employed to feed the dolime into the
furnace.The furnace consists of a refractory-lined cylindrical shell and a
conical roof. The furnace has an internal diameter, at the slag level, of
about 1200 mm, and a shell diameter of 1900 mm. The hearth area is lined
with carbon blocks, while magnesia- based refractories are used for the hot
face, above the slag level (due to the erosion of the side wall refractory,
23. 17
neutral slag recipe was employed to warm-up the furnace and to form a
freeze lining, as will be discussed later) . The anode connection consists of
two graphite rods, cemented into the carbon hearth-blocks. The graphite
rods are further linked to the anode cables via water-cooled copper pins.
The furnace shell is equipped with water spray cooling. Water-cooled
panels are used at the conical roof, which is lined with alumina castable.
The roof contains a central entry port for the graphite electrode, an offgas
duct, and a feed port. The electrode seal consists of a flexible stainless-steel
bellow. The off-gas port contains vermiculite based bricks, behind the
alumina hot-face castable.
24. 18
A radically new condenser designed based on certain principles in order to
allow continuous (semi-
continuous) operation of the facility for extended period of time. It allows
the removal of the condensed magnesium along with any oxides that might
be present. Flow restrictions and blockages in the condenser system are
also dealt with online without interrupting magnesium production. The new
condenser set-up includes an elbow section, a crucible, a secondary
condenser, a stirrer, and plungers for the clearing of blockages. The top
section of the condenser (elbow) incorporates a thermally insulating
refractory material, between concentric 3Cr12 stainless steel sleeves
(ferritic steel developed by Southern Africa Stainless Steel Development
Association, SASSDA), with an inner diameter of 400mm. The elbow
forms the connection between the furnace and the condenser crucible. The
condenser crucible consists of a cylindrical steel shell, a dished bottom, and
a flat top plate. The crucible is 1.2 m in diameter, 1.3 m high, and is made
of 28 mm thick 3CR12 steel plate. The top section is widened to fit the inlet
and outlet into a relatively small diameter crucible. The crucible contains
an underflow/overflow arrangement, consisting of an inclined spout and an
overflow box, in order to remove magnesium under sealed
conditions.During operation the magnesium level is kept between the
bottom and the middle of the overflow box, for effective operation of the
condenser. The impeller of the stirrer is engineered to create a vortex in
order to pull in oxides and magnesium vapour. The condenser crucible
contains three baffles to break up the centrifugal motion that is induced by
the stirrer, and to enhance mixing. A secondary condenser is connected
behind the crucible and consists of a cylindrical 3CR12 steel pipe, 400mm
in diameter, and 1.7 m long. The surface area of the secondary condenser is
designed such that it is sufficient to condense all the magnesium vapour
produced in the furnace. The condenser system also includes two propane
burners for preheating of the condenser crucible and the secondary
condenser, and for maintaining the condenser temperature at 650°C to
750°C during magnesium production. Three hydraulic plungers are
included for the on-line cleaning of blockages in the crossover duct
between the furnace and the condenser, in the elbow, and in the secondary
condenser. The off-gas system consists of an argon reservoir (mild steel,
350mm diameter), equipped with a flap valve arrangement devised to
25. 19
prevent a large amount of air from entering the condenser via the
combustion chamber, and a pressure release disc. The combustion chamber
is refractory lined and equipped with a pilot burner to oxidize any
uncondensed magnesium, carbon monoxide and hydrogen. The reverse-
pulse bag-filter is intended to separate the oxidized magnesium from the
off-gas, and to vent only clean air and argon to the atmosphere. Auxiliary
equipment includes a condenser cleaning station, drill machine-mud gun
assembly for slag tapping, etc. The dolime is heated up in an electrically
operated kiln. The kiln temperature varies between 1000-1100degC, and
was rotated at 3.0-3.5rpm. The feed rate is varied in order to match the
furnace requirements and ranged from 250-400kg/h. The heated dolime is
collected in refractory-lined transfer bins and then discharged into the
furnace feed system. Temperatures taken in the dolime upper hopper
indicated a dolime temperature of 550-630degC, as compared to 640-
750degC measured at the kiln discharge. The furnace total feed rate
averages about 525kg/h (10.7%FeSi, 5.5% Al, the balance being dolime)
and the feed duration is about 2.5 hours, giving a batch size of 1300kg. The
feed recipe is selected based on the relative ease of tapping of the resulting
slag at the temperatures employed. Commercially, the recipe will be largely
dictated by the overall economics of the process. About 30 tons of
magnesium producing recipe have been processed over 22 feeding- tapping
operations, where 24 tons of slag is produced. The furnace energy losses
averages about 360kW. The operating power is about 780kW, on average,
and varies between 680 and 870kW, depending on the feed rate, furnace
energy losses, and the dolime temperature, as measured in its weigh bin.
An operating voltage of 180- 200V is employed during the smelting period.
The powervoltage operating ranges results in an arc length of 100-250mm
(shorter arc at the end of the batch where the slag depth is 430-460mm,
longer arc at the beginning of the feeding period where the slag depth is 80-
150mm). It is interesting to note that silicon utilisation (proportion of the
silicon metal in the feed that reacts to produce magnesium vapour) varies
between 84 and 88%. The lower value has been derived from the slag mass
and its average SiO2 content, while the higher figure is based on the Si-
content in the tapped residual FeSi and assuming that all the iron in the feed
reports to the residual FeSi. In addition, the electrode consumption is
determined at 0.89kg/MWh. Note that a 200mm diameter graphite
26. 20
electrode is used, and the electrical current density averages 13.7 A/cm2
during the campaign. This electrode consumption value is relatively low
and could contribute towards improving the overall plant availability in a
commercial installation, as electrode addition would be less frequent. In
addition, the usage of special magnesium coalescing agents (AlF3, for
example) in the refining stage may be significantly reduced, as less carbon
would be present in the crude metal.
3.SOLID-OXIDE MEMBRANE PROCESS FOR EXTRACTION OF
Mg
The Solid Oxide Membrane (SOM) process can be alternative to
electrolytic process. In this process, reduction of magnesium oxide
dissolved in fluoride-based electrolytes (MgF2-CaF2-MgO) is carried out
by passing electric current at 1150 to 1300 degC. When electrical current is
applied, magnesium oxide dissociates into magnesium and oxygen. Oxygen
ions are pumped out through Yttrium Stabilised Zirconia (YSZ) membrane
to the anode. There are several scientific challenges in the SOM process.
The energy efficiency of the present process is attributed, in part, to the
combination of two steps, one in which a metal oxide is reduced to form
highly reactive metal and another in which a metal compound of interest is
reduced while the highly reactive metal is oxidized. In one or more
embodiments, it takes significantly less electrical energy to reduce the
highly reactive metal and use it to produce the desired metal (or metal
compound) from the metal oxide of interest than it does to reduce the metal
compound of interest to produce the desired metal in an alternate, single-
step process. This is because in the first step of the proposed multi-step
process, the energy used to reduce the metal oxide to form the highly
reactive metal compound is less than energy that would otherwise be used
27. 21
to directly reduce the metal of interest in a comparable single-step
electrolysis process (such as a direct SOM electrolysis which is energy
intensive). And, in the second step of the proposed multi-step process, no
substantial amount of energy is needed to reduce the metal compound of
interest because the highly reactive metal facilitates a chemical reaction. In
sum, less energy is used to reduce a selected amount of the metal of interest
than would be used in a comparable single-step electrolysis process. In one
or more embodiments, it takes significantly less electrical energy to reduce
the highly reactive metal and use it to produce the desired metal (or metal
compound) from the metal compound of interest than it does to reduce the
metal compound of interest to produce the desired metal (or metal
compound) in an alternate, single-step process. This is because in the first
step of the proposed multi-step process, the energy used to reduce the metal
oxide to form the highly reactive metal is less than energy that would
otherwise be used to directly reduce the metal of interest in a comparable
single-step electrolysis process (such as a direct SOM electrolysis which is
energy intensive). And, in the second step of the proposed multi-step
process, no substantial amount of energy is needed to reduce the metal of
interest because the highly reactive metal facilitates a chemical reaction. In
sum, less energy is used to reduce a selected amount of the metal of interest
than would be used in a comparable single-step electrolysis process. The
reported energy demand is 10 kWh/kg Mg.
The chemical stability of an oxide compound can be reflected by the Gibbs
free energy while the
enthalpy formation determines the minimum energy requirement of a
process. The Gibbs energy of magnesium oxide is lower than iron oxide,
which indicates that magnesium oxide is more stable and requires more
energy to extract the metal. The difference between the theoretical energy
requirements and the actual energy usage reflects the different aspects of
28. 22
the process route. The energy efficiency of ferrosilicon electric arc furnace
is only 51.5% while the efficiency of coal fired furnace is 12%. The
Pidgeon process urgently needs improvement in order to reduce energy
consumption and greenhouse gas emissions. Improvement that has been
proposed includes utilisation of cleaner energy source (e.g. natural gas or
producer gas) and integrating small smelters to improve raw material and
energy efficiency. The Solid Oxide Process(SOM) process has a lower
Global Warming Potential compared to the Pidgeon process (47.3 kg
CO2/kg Mg to 62.7 kg CO2/kg Mg), but suffers from low productivity and
high capital costs. The development of new high productivity
pyrometallurgical processes by CSIRO and Mintek may provide the route
by which Western countries, such as Australia, could develop an
environmentally and economically acceptable route to Magnesium
production.
30. 24
CONCLUSIONS AND DISCUSSIONS:
As compared to Bayer’s and Hall-Heroult’s process ,energy requirements
carbothermic production of Al can reduce energy requirements 35% below
those of the best Bayer-Hall technology.It has been reported that the
implementation of carbothermic reduction processes in aluminium
production may lead to energy savings of up to 21 %, GHG emissions
reductions of up to 52 % and exergy efficiency increase of up to 10
percentile points. Additionally, the prospect of utilizing concentrated solar
energy to provide process heat can render the primary aluminium
production truly sustainable. In case of Mg extraction considering Mintek
process ,one way of minimizing energy requirements is by charging the
dolime and/or magnesia as hot as possible. The chemical stability of an
oxide compound can be reflected by the Gibbs free energy while the
enthalpy formation determines the minimum energy requirement of a
process. As seen in the Gibbs energy of magnesium oxide is lower than
iron oxide, which indicates that
31. 25
magnesium oxide is more stable and requires more energy to extract the
metal. The difference
between the theoretical energy requirements and the actual energy usage
reflects the different
aspects of the process route. The energy efficiency of ferrosilicon electric
arc furnace is only
51.5% while the efficiency of coal fired furnace is 12%. The Pidgeon
process urgently needs
improvement in order to reduce energy consumption and greenhouse gas
emissions.
Improvement that has been proposed includes utilisation of cleaner energy
source (e.g. natural
gas or producer gas) and integrating small smelters to improve raw material
and energy
efficiency.The electrolytic process has a lower Global Warming Potential
compared to the Pidgeonprocess (47.3 kg CO2/kg Mg to 62.7 kg CO2/kg
Mg), but suffers from low productivity and high capital costs.In case of
Mintek process (for Mg) for prompt tapping from DC arc smelter and for
low cleaning and refining costs the condensed metal should contain as little
solids as possible. Magnesium extraction and extraction rate are dependent,
in part, on the operating temperature and the feed recipe used; they need to
be optimized in order to improve the economics of the process. For this
purpose, the furnace needs to be operated in such manner as to produce and
deliver almost pure magnesium vapour to the condenser.
CO + Mg(V) = MgO + C [1]
In this method, the coalescence of the condensed magnesium droplets is
adversely affected by the presence of carbon. Conde nsation to a liquid
phase reduces the energy requirements in the cleaning and refining stages,
as re-melting of the crude magnesium is not required.
ACKNOWLEDGEMENT:
The author would like to express appreciation for the support of the
Department of Metallurgical and Materials Engineering,NIT Durgapur.
References
32. 26
[1].S. Namboothiri, M. P. Taylor, J. J. J. Chen, M. M. Hyland, M.
Cooksey, “Aluminium Production Options with a Focus on the Use of a
Hydrogen Anode: A Review”, Asia-Pacific Journal of Chemical
Engineering, Vol.2, (2007), pp. 442-7.
[2]I.M. Morsi, K.A. El Barawy, M.B. Morsi, S.R. Abdel-Gawad,
Silicothermic reduction of Dolomite Ore under Inert Atmosphere, Canadian
Metallurgical Quarterly, Vol. 14, No. 1,(2002),p.15.
[3]. Brooks, G., Trang, S., Witt, P., Khan, M.N.H , Nagle, M,
'Carbothermic Route to Magnesium', Journal of Minerals, Metals and
Materials Society, Vol.58, no. 5, p.51-55.
[4]. R.T. Jones, Computer Simulation of Pyrometallurgical
Processes, Proceedings of the Twentieth International Symposium on the
Application of Mathematics and Computers in the Minerals,Industries, Vol.
2,( 1987),p.265.
[5]. Cherubini, F., Raugei, M. , Ulgiati, S. (2008), 'LCA of
Magnesium Production.Technological Overview and Worldwide
Estimation of Environmental Burdens',Resources Conservation &
Recycling, vol.52, no. 8-9, p. 1093-1100.
[6].
Jenkins, D.H., Sheehan, G.J. ,Frost, M.T. (2009), 'Piloting the
Australian MagnesiumProcess', Mineral Processing and Extractive
Metallurgy (Trans. Inst. Min Metall. C),vol.118, no. 4, p. 205-213.
[7]. Kipouros, G.J. & Sadoway, D.R. (1987) Advances in Molten
Salt Chemistry. [8]. Koenig, R., Baker, G. , Goebel, B. (2002) 'The
Stanwell Magnesium Project – Providing New Environmental Standards
for Magnesium Production', Green Processing Conference 2002, Cairn, The
Australian Institute of Mining and Metallurgy p. 85-89
[9]. R.A. Sharma: “Method for Producing Aluminum Metal from
Aluminum Trichloride”, US Patent 6,066,247, 2000. [10]. S. Wilkening:
“Process for Producing Aluminium by Molten Salt Electrolysis”, US Patent
4,919,771, 1990. [11]. K.M. Tomaswick: “Low Temperature Aluminum
Production”, US Patent 6,428,675, 2002. [12].
J.C. Terry, A. Lippman, R. F. Sebenik, , H. G. Harris: “Reduction of
Aluminum Chloride by Manganese”, US Patent 3,900,312, 1975.
[13]. H.N. Sinha: “Production of Anhydrous Aluminium Chloride”,
US Patent 4,264,569, 1981. [14]. H.P. Mueller, H. Alder, G. Zhuber-
33. 27
Okrog: “Process for the Production of Aluminium Chloride”, US Patent
4,289,735, 1981.
Paper2
LOW COST SYNTHESIS OF SINGLE WALLED CARBON
NANOTUBES FROM COAL TAR USING ARC DISCHARGE
METHOD
Mainak Saha
Department of Metallurgical and Materials Engineering , Naitional Institute
of Technology(NIT) , Durgapur,India
Email:mainaksaha1995@gmail.com
ABSTRACT
There are various methods such as arc discharge, laser ablation, chemical
vapour deposition (CVD), template-directed synthesis for the growth of
CNTs in the presence of catalyst particles. The production of carbon
nanotubes in large quantities is possible with inexpensive coal as the
starting carbon source by the arc discharge technique. It is found that a
large amount of carbon nanotubes of good quality can be obtained in the
cathode deposits in which carbon nanotubes are present in nest-like
bundles. For more than two decades, now, there has been extensive
research on the production of carbon nanotubes (CNT) and optimization of
its manufacture for the industrial applications. It is believed that they are
the strong enough but most flexible materials known to mankind. They
have potential to take part in new nanofabricated materials. It is known
that, carbon nanotubes could behave as the ultimate one-dimensional
34. 28
material with remarkable mechanical properties. Moreover, carbon
nanotubes exhibit strong electrical and thermal conducting properties. This
paper primarily concentrates on the optimising such parameters related to
the mass production of the product. It has been shown through Simplex
process that based on the cost of the SWNT obtained by the arc discharge
technique, the voltage and the current should lie in the range of 30 - 42 V
and 49 - 66 A respectively. Any combination above the given values will
lead to a power consumption cost beyond the final product cost, in turn
leading to infeasibility of the process. Strong expectations exist for future
use of carbon nanotubes as composite materials in a large number of
industries. Production cost and control of the purity and properties of such
materials will influence the impacts nanotubes on the chemical, computer
and construction industries. Coal properties in this case are also important.
Weak bonds and mineral matter in the coal play an important role in the
formation of the nanotubes.
Keywords: Coal tar; Carbon Nanotubes; SWCNT; Simplex; Optimization
1.Introduction
In 1985, Drexler.et.al proposed a molecular bearing consisting of two
graphitic nanotubes of different diameter, which are concentrically
arranged. It was a virtual operation inside a computer. This dream,
however, has become more realistic by the discovery of carbon nanotubes.
There had been revolutionizing researches on the pro- duction of Carbon
Nanotubes from last twenty years and optimising its manufacture for the
35. 29
industrial applications. It has been thought that they are the strongest but
most agile materials known to humankind, and thus have potential to take
part in new nanofabricated materials as additives. It has been shown that
carbon nanotubes could behave as the ultimate one-dimensional material
with remarkable mechanical properties. Carbon nanotubes exhibit strong
electrical and thermal conducting properties. Study of the past researches
on the production of carbon nanotubes from coal revealed that most of the
researches concentrate on producing CNT by arcing electrodes, produced
separately by mixing the crushed coal with coal tar followed by molding
process. However, the only process till date that has shown a positive
adaptation to direct coal application is thermal plasma jet technique.With
the extensive research in the production of the Carbon Nanotubes, the
requirement of optimising the process parameters are realised. This paper
concentrates on the optimising such parameters related to the mass
production of the product. It has been found through calculation that the
only determining parameter in the arc discharge technique is the power of
the equipment, in terms of voltage and current, and it is seen that on
optimisation of these parameters, the cost of the process reduces
drastically.
2.Method of synthesizing SWCNTs from non-metallurgical grade coal
fines and coal tar through arc discharge technique
There are various methods such as arc discharge, laser ablation, chemical
vapour deposition (CVD), template-directed synthesis and the use of the
growth of CNTs in the pre- sence of catalyst particles.Special ambient gas
is re quired for the fabrication of SWCNTs, in order to prevent the
oxidation of carbon at high temperature. The production of carbon
nanotubes in large quantities and other nanomaterials as by-products is
possible with inexpensive coal as the starting carbon source by the arc
discharge technique. It has been found that a large amount of carbon
nanotubes of good quality can be obtained in the cathode deposits in which
carbon nanotubes are present in nest-like bundles which is determined
using Scanning Electron Microscope(SEM) and Selected Area Diffraction
Pattern(SADP) in high resolution Transmission Electron
36. 30
Microscopy(HRTEM). Compared to other methods, the arc discharge is the
simplest, cheapest and easy to implement and it has been used because of
its potential merits to make a massive production. The mineral matter in
raw coals may also play an important part in the formation process of
carbon nanotubes.The proposed procedure for carrying out arc deposition
of SWCNTs from non-metallurgical grade coal fines and coal tar which is
obtained as one of the by-products from a steel plant are as follows. the
coal is crushed and sieved with 150 μm mesh.Then it is mixed with coal tar
and molded to form coal rods used for anode with a graphite cathode for
the arc discharge process.Iron mesh wire is used between electrodes and it
is interestingly observed through TEM formation of carbonaceous matter is
found to occur on wire mesh. Coal or coal-derived carbons show
macromolecular structures rather than the lattice structure of graphite. In
their chemical structures there exist many weak binding linkages between
carbon polymeric units such as polymerized aryl structures. In the fast
paralyses process under arc plasma conditions, these weak linkages will be
rather easily to release a variety of reactive fragments of hydrocarbon
molecules such as alkynes and aromatic species. In place of heavy
hydrocarbons, pure compound, toluene may be used as the pure substrate to
establish the reaction system for the production of carbon nanotubes.
Toluene is fed by a mist-spray feeding system with a carrier gas as 9:1
mixture of nitrogen and hydrogen at 100 ml/min, and following the reaction
at 750°C catalysed by 9.8% (by weight) ferrocene, Carbon nanotubes are
found in the carbonaceous product deposited on inner wall of a quartz tube
and at the exit of the tube. The product is observed by scanning electron
microscopy and analyzed by temperature programmed oxidation
experiments to identify the presence of carbon nanotubes. Based on the
reaction system and reaction conditions with toluene, the production of
nanotubes is examined by using heavy hydrocarbons such as asphaltene
and maltene fractions from natural asphalt. Under selected reaction
conditions including the reaction temperature and the amount of the
catalyst, carbon nanotubes with a diameter of 30 - 60 nm are found.
3.Optimisation of voltage and current used in the SWCNT synthesis
using arc plasma deposition process
37. 31
It is known that Voltage and Current constitute the Power requirements for
an electrically operated machine. Same in the case of the production of
CNT from the arc discharge method, the electrical power signifies the
characteristics of the arc that is generated between the two electrodes, in
this case, one coal-based and the other, graphite. For the optimisation
process, the total cost of the input materials must be lesser than that of the
output product.
Field/Co. Cost of grades of coal in
rupees/tonne
A B C D E F G
ECL 36
90
35
90
16
80
13
50
10
10
79
0
56
0
ECL/
Mugma
36
90
35
90
19
50
16
10
12
90
96
0
62
0
ECL/Rajma
hal
- - - - 13
30
11
30
91
0
BCCL 36
90
35
90
16
30
13
50
10
80
86
0
61
0
CCL 36
90
35
90
15
90
13
00
10
30
82
0
59
0
NCL 36
90
35
90
14
30
12
00
96
0
75
0
56
0
SECL 36
90
35
90
13
70
11
40
95
0
74
0
56
0
MCL 36
90
35
90
13
70
11
40
95
0
74
0
56
0
(A—Graphite/high quality anthracite; B—Anthracite (C:H > 30); C—
Anthracite (C:H ~ 26 - 30); D—Semi-anthracite; E—Semi-bituminous; F—
Bituminous; G—Low quality bituminous).
Basic price of run of mine non-long-flame non- coking coal
38. 32
(Courtesy : Coal India Ltd)
Package (gram) SWNT DWNT
P
u
r
it
y
(
$
)
Arc
CNT*
($)
High purity*
($)
1 210 83 210
10 1600 700 1600
50 6850 3050 6850
100 12400 5500 12400
500 Call Call Call
1000 Call Call Call
Carbon nanotube price list
*The High Purity CNT (more than 90% pure) is not achievable by Arc
Discharge Method, as per different researches, and thus the cost of the Arc
CNT, made from Arc Discharge Method, is always less than the former.
However, DWNTs are not the common by-product from the Arc Discharge
Method.
(Courtesy: Coal India Ltd)
The rate of anode feed in the experiment is taken to be 10mm for 2 hrs. The
length of the electrode in the ex- periment is 75 mm. So, the total
experiment time equals 15 hrs. For that time the 45 units of electricity is
con- sumed (1 KWhr = 1unit of consumed electricity). As per the
regulations of Calcutta Electricity Supply Corporation (CESC), the rate is
Rs.2.7/unit for the first 25 units consumed, then at the rate of Rs.3.3/unit
for the next 35 units. Therefore, Total cost = 25*2.7+20*3.3=Rs.133.5 So,
39. 33
the total input cost for the production of CNT is Rs.133.5 (neglecting the
cost of coal as too small as compared to the power consumption cost) .The
amount of CNT produced in experiment is about 20mg. The cost above is
given in $, so we consider 1$ = Rs 50 and the package unit is in grams.
Thus, the total cost comes to be 20*83*50/1000= Rs 83. Therefore, it can
be understood that as the input cost is greater than the output cost,
optimisation is essential in this case. The coal-based electrode
specifications for the arc dis- charge process are found from the
experimental works in the past. With the aim of the maximum cost
involved in the production of the coal-based electrode, the electrode with
maximum volume is selected, as it is the electrode that involves the
maximum amount of coal and thus the cost. Out of the several independent
researches the specification of diameter 10 mm and length 200 mm is
chosen as it gives the maximum volume as compared to the other
electrodes in the other researches.
Therefore, electrode volume = (3.14/4)*100*200=15707 cubic mm
Bank density of coal is 1346 kg/cubic metre
So, Bank Weight of the powdered coal used = 1346*1.57*10^-5=0.021 kg
Maximum Cost of the coal is taken to be Rs.3.5/kg
Therefore, Cost of the coal used = 3.5 × 0.021 = Rs0.073
The cost of power is to be calculated next. In a typical experiment, the
current was taken to be 100 A and the voltage was taken to be 36 V.
Therefore, Power = V × I = 30 × 100 = 3000 W = 3 KW
Power(P) = IV
LogP=logV +logI
Taking, logP = x1; logV = x2; logI = x3
In all the experiments minimum voltage and current are taken as 30 V
and 50 A respectively.
V>=30,I>=50
logV>=1.47,logI>=1.69
Therefore, x1=x2+x3,x2>=1.47;x3>=1.69
Considering the power consumption rate as Rs 2.7/unit,
The total power can be calculated as 15 × 2.7 × P = 40.5 P
For optimum result, 40.5P<=83
40. 34
implies,P<=2050W
or,logP<=3.31
Applying Simplex Algorithm,
min x1=x2+x3
x2+x3=3.31
x2=1.47
x3=1.69
The area enclosed by the three constraint lines gives the range of
feasibility of the variables.
Paper3
Study of Austenitic Stainless Steel Castings
Mainak Saha, Department of Metallurgical and Materials Engineering
National Institute of Technology (NIT) Durgapur, West Bengal, INDIA
Abstract: Steel casting is a specialized form of casting involving various
types of steel. Steel castings are used when cast irons cannot deliver
enough strength or shock resistance.Examples of items that are steel
castings include: hydroelectricturbine wheels, forging presses, gears,
railroad truck frames, valve bodies, pump casings, miningmachinery,
marineequipment, turbochargerturbines and engine cylinder blocks.Steel
41. 35
castings are categorized into two general groups: carbon steels and alloy
steels.Steel is more difficult to cast than iron. It has a higher melting point
and greater shrinkage rate, which requires consideration during mold
design. Risers should be given more capacity to draw from as the metal
cools and shrinks. Attention should be paid to the thickness of mold
cavities, as thinner areas will cool quicker than thicker areas, which can
create internal stress points that can lead to fracture.Molten steel is also
less fluid than molten iron, making it more difficult to pour and fill intricate
gaps in a mold cavity. Molten steel is also more likely to react with internal
mold surfaces, making for more unpredictable results.In this paper, the
quality of austenitic stainless steel castings will be discussed.
Keywords: austenitic stainless steels, liquidus, dendrite, liquidus,
multimeter
42. 36
I. Material For Investigation
Twenty-two casts of austenitic stainless steel were produced as 10 kg
air melts with compositions which approximate to AISI 304, 306, 309, 310
and 316. In the 306, 309 and 310 series the main composition variable was
Ni whereas in the 304 and 316 steels Ν was varied in the range 0.03 to
0.2%. The steels were chosen as being representative of grades of
commercial significance in the region of the phase diagram ofinterest. The
exact compositions were chosen to span the solidification boundaries as
determined from the Rivlin and Raynor phase diagram. In addition, a
further 6 alloys were produced as 5 kg air melts in order to investigate the
effect of Si on the liquidustemperature.Theliquidus temperature for each
melt was determinedusing the Land Checkpoint system.350-400 g of metal
was poured into the Checkpoint ceramic crucibles which are equipped with
Pt/Pt 13% Rh thermocouple heads in silica sheaths. The signal from the
thermocouple was displayed on a Thurlby intelligent multimeter. The
claimed accuracy of the Land Checkpoint cups is ±0.5°C and the sensitivity
of the multimeter was 1 pV. For each melt 2 liquidus determinations were
made to determine the reproducibility of the technique. The maximum
difference was 2°C. The results for the second determination are given in
Table 1 together with the chemical analysis of the sample which was
43. 37
determined by a combination of X-ray fluorescence and wet chemistry on
drillings.In each case, immediately after the second liquidus measurement
the first 22 casts were poured into 100 mm X 50 mm slab moulds. The
ingots were sectioned longitudinally, ground and etched in aqua
regia(3HC1:1HN03:4H20). This revealed that all the steels exhibited fully
columnar primary and secondary grain structures. An example of the
secondary grain structure is shown in Fig. 10. The ingots were sectioned
transversely at mid-height and specimens mounted for metallographic
examination. Ferrite contents were determined magnetically approximately
10 mm below the cast surface using a Fischer Ferritescope. The
microstructures were revealed by etching in the
following reagent:-
20 g Ammonium hydrogen difluoride
0.5 g Potassium metabisulphite
100 ml Water
Etching time: up to 10 min
This etchant stains chromium rich austenite blue and nickel rich
austenite yellow whilst ferrite remains white, or black if the ferrite
networks are particularly fine. The etch clearly distinguished between the
primary dendrites and interdendritic spaces in each steel. However, in the
high nitrogen series, Casts 15-22, the colour of the primary phase did not
vary systematically with nitrogen content. Some typical structures are
shown as black and white micrographs in which the location of the ferrite,
either dendritic or interdendritic is clearly visible. In some areas of Cast 19
the 2 typesofferrite appeared to be connected, whereas in other areas they
were divorced. The position of the ferrite with respect to the primary
dendrites was used to determine the first phase to separate and hence the
solidification mode. For example, the presence of ferrite in the primary
dendrite is indicative of mode Β solidification whereas interdendritic ferrite
alone is typical of mode C solidification. Fully austenitic casts are
designated mode C/D. These results are presented in A particularly
interesting structure was seen in Cast 8 where regions of primary ferritic
solidification coexisted with regions of primary austenite. Neither
martensite nor o-phase were detectedmetallographically in any of the steels.
44. 38
The stain etch micrographs were used to estimate both the primary and
secondary dendrite arm spacings.
These were measured by lineal analysis across 50 intersections. The
results are presented in Table 4 where it is immediately obvious that both
spacings decreased with increasing alloy content: this was particularly
noticeable in the high nitrogen steels, Casts 15-22.
Selected areas of each cast, identified by microhardnessmarks
wereanalysed using a CamecaCamebax SX50 microprobe and colour maps
of the distribution of Cr, Ni, Μη, Si and Ρ obtained for Casts 1-18 and Cr,
Ni, Μη, Si and Mo for Casts 19-22. The conditions used were as follows:-
Casts 1-14 Casts 15-22
HV 15 kV 15 kV
Current 100 mA 250 mA
Step length 2 pm 2 pm
Dwell time 50 ms250 ms
Some typical maps for ferritic and austenitically freezing casts are
presented in Figs. 17-19. Of particular interest was Cast 1 where some of
the interdendritic spaces were enriched in all elements; a typical region is
marked A . In addition, linescans were obtained across primary dendrite
arms and across interdendritic spaces for each element. Three characteristic
probe traces were observed and schematic representations. In general the
results for the primary ferritic casts, i.e. those exhibiting dendritic ferrite,
agreed with the results from stain etching. In these steels the primary
dendrites were enriched in Cr and depleted in Ni although in Cast 17 the
exact centre of the dendrite was depleted in Cr. The interdendritic regions
were enriched in nickel but slightly depleted in Cr. In several casts Cr was
enriched towards the centre of the interdendritic spaces and this resulted in
the precipitation of interdendritic ferrite in Casts 19 and 20.
In casts solidifying initially as austenite but containing interdendritic
ferrite the centres of the dendrites were depleted in all elements. The
interdendritic regions contained ferrite, evident from an increase in Cr and
a depletion in Ni and Mn in these areas. Again the results confirm the
results of stain etching.
In fully austenitic casts all elements segregated together which
implies single phase solidification leading to enrichment in all elements
45. 39
towards the interdendritic spaces. These casts were therefore designated
fully austenitic, i.e. mode D, instead of C/D deduced from stain etching.The
solidification modes deduced from the ΕΡΜΑ results are compared with
those from stain etching in. The cause of the variable response to stain
etching observed in Casts 15-22 was not evident from the ΕΡΜΑ
results.Liquidus Temperature
As outlined in Section 2 there are several possible methods of
calculating the liquidus temperature of austenitic stainless steels, ranging
from simple binary depression summations to more complex
thermodynamic approaches. Several of these methods have been used in
the present investigation and the results are compared with the measured
temperatures in Table 6.
The Andrews method of summation of binary depressions gave an
average error of + 17.8°C which is slightly outside the range for
acceptability. This is not surprising since the method was only designed for
dilute solutions although Howe3 found reasonable agreement when one
element was non-dilute. Very good agreement was found when using the
FeNiCr phase diagram in conjunction with binary depressions for residual
elements. The error when using the Chuang and Chang diagram (C&C in
Table 6) was slightly lower than obtained from the Rivlin and Raynor
diagram (R&R in Table 6), -4.9°C cf. -7.4°C.
However, fewer liquidus temperatures were determined from the
Chuang and Chang diagram and there is no real evidence from this work to
suggest that either diagram is superior. The use of both diagrams gave
acceptable results. Liquidus calculations were then made using MTDATA,
a thermodynamic approach developed by NPL and ThermoCalc, a very
similar method proposed by the Royal Institute of Technology, Sweden.
Several versions of MTDATA are now available to British Steel and these
are designated BS1, BS2 and S90 in
Initially calculations were performed using BS1, the original 12
element version (Fe, Cr, Ni, C, Ρ, S, Si, Mo, Mn, Ti, Ν, O) used by Howe
in an earlier ECSC projects. This gave relatively large errors
(Av. -36.9°C for the first 14 steels) and the error was particularly large for
the 2% Si steels, Casts 1-5. The error is plotted against Si content in Fig. 23
where a clear relationship can be seen. The temperatures were therefore
recalculated using BSl for all elements except Si with the effect of Si being
46. 40
incorporated as a simple binary depression. The liquidus temperatures
calculated in this way are presented under BSl* in Table 6. It can be seen
that the error was much reduced with an average of -7.7°C which is
comparable with that obtained using the phase diagram. A plot of measured
liquidus against the predicted value is shown in Fig. 24. Large errors (up to
-99°C) were also found using ThermoCalc17 with once again the largest
errors being associated with the high Si steels. The large error associated
with ThermoCalc is not surprising since both methods employ the same
source thermodynamic data.The thermodynamic data for Si-containing
systems was critically reassessed by NPL, and this resulted in an improved
version of MTDATA designated BS2 in Table 6. As part of this work Casts
23-28 were made in order to investigate liquidus temperatures in relevant
Si-containing ternary and quaternary systems. It can be seen in Table 6 that
the average error was reduced to -22°C but was still highest for the Si
containing steels and alloys (Casts 1-5 and 23-28). When these are omitted
the error is reduced to -19°C, similar to that found with the Andrews
method.As part of the project a subcontract was undertaken by NPL in
order to improve MTDATA particularly for high nitrogen steel . The
average error of the predicted liquidus temperature for all the steels and
alloys was -22.4°C but this was only reduced to -21.5°C when the high Si
melts are omitted.It can be seen from the above discussion that reasonable
agreement between measured and predicted liquidus temperature can be
obtained by using a relatively simple approach which involves a
combination of the FeNiCr phase diagram with the subtraction of binary
depressions due to residual or dilute elements.The use of MTDATA as it
stands at present did not result in acceptable errors due largely to the
element Si. Omitting Si from the MTDATA calculations following by
subtracting the binary depression did result in acceptable errors of similar
magnitude to the phase diagram approach. Although a limited
programmeof work carried out at NPL resulted in improved predictions, the
error due to Si was not entirely eliminated. Clearly additional fundamental
thermodynamic work, outside the scope of the present project,is required
before acceptable predictions can be obtained from MTDATA. Since the
majority of stainless steels contain appreciable Si contents in the range 0.5-
2% the use of MTDATA cannot be recommended for the prediction of
47. 41
liquidus temperatures. It is therefore recommended that the phase diagram
approach should be used.
II. Solidification Structure
Austenitic stainless steels can solidify initially as austenite or ferrite.
In order to rationalise the ΕΡΜΑ work in this project the casts analysed
were grouped according to the characteristics shown by the linescans
obtained in each case. It was assumed that Cr partitions to the ferrite phase
and Ni to the austenite phase, when both phases are present. Three types of
elemental distribution were identified on this basis.They are compared with
the results obtained from stain etching.
(1) This type of linescan is illustrated schematically in Fig. 20. This
behaviour is consistent with mode Β solidification, i.e. primary ferrite with
interdendritic austenite. In these casts the dendrite centres were depleted in
Ni and enriched in Cr, indicating primary ferrite precipitation. Within the
austenite between the primary dendrites the Ni content rose moving away
from the dendrites and the Cr concentration fell. This indicates that in this
part of the structure, austenite began to be precipitated as a separate phase
while the ferrite was still precipitating.At this stage ferrite and austenite
must be simultaneously in equilibrium with, and therefore in contact with,
the liquid. (The diagram usually used to illustrate mode Β solidification is
misleading in showing primary ferrite dendrites completely covered or
enveloped by austenite.)In addition to this general pattern, seen most
clearly between the primary dendrite arms, there were also areas in which
all elements increased rapidly in concentration. These represent the final
stages of solidification, in which the composition balance in the last pools
of liquid was such as to precipitate only austenite, and the levels of solute
element rose rapidly due to segregation. In several cases, for example
Cast 19, it was evident from the probe maps, that these highly concentrated
areas had precipitated ferrite in the final stages of solidification.
(2) This type of linescan is illustrated schematically, and is consistent with
mode C solidification, i.e. primary austenite with interdendritic ferrite. The
centres of the dendrites were depleted in all elements. The interdendritic
spaces, however, contained ferrite precipitated in the final stages of
solidification, marked by elevated levels offerrite formers, especially Cr,
and reduced levels of austenite formers.It can be seen that there was an
48. 42
enrichment in the austenite formers around the ferrite, and also in some
cases a variation in concentration of austenite and ferrite formers within the
ferrite. This implies that the austenite formers were being rejected from the
ferrite as it formed. Ifall the austenite had been fully solid at this stage,
rejection of austenite formers in this way would have required a
considerable amount of solid state diffusion, making the creation of such a
large zone of enrichment difficult. The observation of such a zone therefore
indicates that the final stage of solidification consisted of austenite and
ferrite precipitating simultaneously from the interdendritic liquid, in a
similar manner to that described in case above,rather than the final liquid
solidifying as ferrite instead of austenite. As in case (1) the representation
ofthe ferrite as dendrites is possibly unrealistic but it is essential that at this
stage both ferrite and austenite are in equilibrium with the liquid.
(3) This type of linescan is illustrated in Fig. 22 and is consistent with
mode D solidification, i.e. fully austenitic. All the elements showed similar
distributions, being relatively depleted in the centres of the dendrites and
rising smoothly to a peak in the interdendritic spaces. This distribution
reflects the effects of segregation taking place during the solidification of a
single phased structure.
The level of segregation will be determined by the mode of
solidification and also by the coarseness of the structure. In general steels
solidifying initially as austenite will exhibit higher levels of segregation to
the interdendritic spaces than steels solidifying initially as ferrite. Thus the
highest levels of interdendritic Ρ segregation were found in Casts 4,10 and
14, see Table 5, all of which solidified austenitically. A similar effect was
not found in the high Ν series of steels where low levels of Ρ segregation
were observed in Casts 15-18, whatever the primary phase. Ρ was not
analysed in Casts 19-22 but again low levels of Si and Mn segregation were
found. This could be due to a strong interaction between Ν and Cr or Mo
reducing segregation generally. Alternatively it could result from the much
finer primary and secondary dendrite arm spacing resulting in faster
homogenisation during cooling.
It can be seen in Table 4 that in general dendrite arm spacing
decreased with increasing alloy content.
Thus Type 310 structures were finer than Type 309 which in turn were
finer than Type 306. However, the effect was most marked in the high
49. 43
nitrogen steels where an increase in Ν from 0.03 to approximately
0.10resulted in a reduction in primary and secondary spacing by a factor of
about 2.
By using Ni and Cr equivalent compositions (Equations (1) and (2))
it is possible to use the phase diagram to predict the solidification mode.
The modes determined in this way using the Rivlin and Raynor phase
diagram. It can be seen that good agreement was found between the
predicted and actual primary phases in 18 of the 22 casts. If the mixed
structure in
Cast 8 is identified as primary austenitic the structures of 19 of the
22 casts were predicted accurately (86.4%). The best agreement was found
for the lower alloy steels (Casts 15-22) with the largest deviation being
found in Cast 11, a Type 310 stainless steel. An improved fit would be
obtained by removing the kink in the mode B/C boundary such that the
boundary is represented by a straight line of slope
Cr/Ni = 1.5 over the full range of compositions investigated. The revised
boundary is shown as a dashed line. In this way the only major error was
for Cast 11 whose composition is brought closer to the phase boundary.
Similarly, the mode C/D boundary would be better represented by a line of
slope
Cr/Ni = 1 and the mode A/B boundary by a straight line of slope Cr/Ni = 2.
Similar boundaries have been recommended by other workers1· 10-12.
The kink in the liquidus trough is not present in the phase diagram of
Chuang and Chang,.
However, on comparing (Rivlin and Raynor diagram) and making
adjustments for the different scales employed, it is clear that the liquidus
trough is closer to the Cr axis which is the opposite of the required effect.
MTDATA was then used for the prediction of solidification mode. Initially
a ternary model was used employing the same Ni and Cr equivalent values
used above. The predictions of solidification mode using this method are
presented in Table 7. A notable success of MTDATA used in this form was
the prediction of the mixed mode of solidification observed in Cast 8.
A similar diagram is constructed using MTDATA.
It can be seen that the phase boundaries are straight lines defined by the
following equations:-
ModeA/B, Ni = 1.68 + 0.443 X %Cr
50. 44
Mode B/C, Ni = -1.0 + 0.75 X %Cr
Mode C/D, Ni = -2.82 + 1.0 X %Cr
The similarity between the slopes of these lines and those
recommended for use on the phase diagram is evident. When MTDATA is
used in this ternary form good agreement was found between predicted and
actual primary phase in 20 of the 22 steels investigated (90.9%). Cast 18
was very close to the mode B/C boundary and therefore the one major error
was for Cast 21 in which primary ferrite was predicted for a steel
undergoing primary austenitic solidification. On several occasions austenite
precipitation closely followed the initial separation of ferrite, generally
within 2°C. These steels are all classified as mixed mode B/C . If we
assume that mode B/C can indicate either primary austenite or primary
ferritic solidification then the number of correct predictions of primary
phase were as follows:-
BSl - 11 out of 14 (78.6%). The errors were all too austenitic.
BS2 - 16 out of 22 (72.7%). The errors were all too ferritic.
S90 - 21 out of 22 (95.5%). The one error in Cast 18 was too ferritic.
Examination of the result revealed that austenite precipitation was
predicted 7.5°C below that for the primary ferrite and therefore the steel
was close to the mode B/C boundary.
It can therefore be seen that MTDATA in its final revised edition
S90 gave improved predictions to those obtained using the original phase
diagram of Rivlin and Raynor. There were no major errors with the only
minor error being close to being classified as mixed solidification.
However, it must be remembered that the phase diagram would be
improved by extending the mode B/C boundary represented by Cr/Ni = 1 .
5 over the full range of compositions investigated. This would result in a
straight line boundary between primary ferrite and primary austenite as
observed with MTDATA in its ternary form.
The use of MTDATA did not result in any of the steels being
classified as fully austenitic or mode D. In fact 6 of the steels were ferrite
free and all of these were classed as mode D after ΕΡΜΑ found no
evidence of segregation indicative of the presence offerrite on
solidification. The 2 phase diagram methods resulted in successful
prediction of fully austenitic structures in 2 of the steels, Casts 5 and 10.
51. 45
The unmodified diagram also resulted in Casts 12-14 being predicted to be
fully austenitic whereas they actually contained interdendritic ferrite.
Modification to the mode C/D boundary resulted in correct predictions of
mode C solidification. However, it should be recognised that
transformation and homogenisation will occur below the solidus and
therefore the metallographic observations at room temperature are not in
serious conflict with the predictions of MTDATA. This is particularly true
in the high nitrogen, fully austenitic steels,
Casts 18 and 22, where the very fine structure may have resulted in faster
homogenisation.
Residual ferrite
The ferrite contents of each steel are presented in Table 2 where they
are compared with the predicted values using a DeLong diagram2 and the
modified Espy equations15. Using the DeLong diagram for the low Ν
steels, i.e. Casts 1-14, 15 and 19, the actual ferrite content was always
lower than the predicted value, i.e. the steels were more austenitic. The
maximum discrepancy was + 6.7% in Cast 2. For the higher Ν steels, i.e.
Casts 16-18 and 20-22, closer agreement was obtained with a maximum
discrepancy of +1.2% for Cast 20. The relationship between actual and
predicted ferrite is shown in Fig. 27. Linear regression gave the following
relationship:-
Actual ferrite = 0.61 (Predicted ferrite) - 0.41
The variance explained was 74%.
A better relationship between actual and predicted ferrite content was
obtained using the Espy formulae with a maximum discrepancy of + 5%,
again for Cast 2. The relationship is shown in Fig. 28. Linear regression
gave the following equation:-
Actual ferrite = 0.76 (Predicted ferrite)-0.04
The variance explained was 78%.
A linear regression of composition against actual ferrite content was
performed and this gave the following equation:-
% Ferrite = 1.45 Cr + 1.6 Mo -1.4 Ni -41.9 Ν -5.7
The variance explained was 82%. The relationship between actual
and predicted ferrite using the equation where the close agreement can be
seen. However, the equation does not contain a factor for Si which is a
52. 46
known ferrite stabiliser. Its use outside the range of compositions studied
cannot be recommended.
It can be seen from the results that a good agreement between actual
and predicted ferrite contents can be obtained over a wide range of
composition and solidification mode using a traditional approach,
particularly that recommended by Espy. The agreement is generally within
± 3% which is acceptable for predicted ferrite contents. It is therefore
considered unnecessary to investigate more complicated methods
employing, for example, MTDATA, over the range of compositions
examined.
Despite the comments made where there is no conflict between
solidification mode and residual ferrite content was found in the present
investigation. This is largely due to the observation that the mode
boundaries can be represented by straight lines over the range of interest
and not the curves depicted in. Thus, mode boundaries and iso-ferrite lines
on the DeLong or Espy diagram are both straight.
Acknowledgement
The author would like to express appreciation and gratitude for the
department of Metallurgical And Materials engineering , NIT Durgapur
,especially to the foundry laboratory , without whose support and guidance
, it would not have been possible to create a review.
References
[1]. Η-U. Lindenberg, Stahl und Eisen, 104, (5), 1984, ρ 51, BISI22871.
[2]. C. J. Long and W.T. DeLong, Welding Jnl. Supplement, 1973,52, ρ
213 (s).
[3]. A.A. Howe, Ironmaking and Steelmaking 1988,15, No. 3, ρ 134.
[4]. K.W. Andrews and W.E. Bardgett, BS Internal Report
3806/X94/2/55,1955.
[5]. K.W. Andrews, Note submitted to the Alloy Phase Diagram Data
Committee of the Metals Society, 1981.
[6]. V.G. Rivlin and V.G. Raynor, Int. Met. Rev. 1980,25, ρ 21.
[7]. Y-Y. Chuang and Y. Austin Chang, Metall. Trans. 1987,18A, ρ 733.
53. 47
[8]. A.A. Howe, 'Segregation and Phase Distribution During Solidification
of Carbon, Alloy and Stainless Steels', ECSC Agreement No.
7210.CF/801, Draft Final Report, December 1989.
[9]. A Guide to Solidification of Steels', Jernkontoret, Stockholm, 1977.
[10].Õ. Hammar and U. Svensson, 'Solidification and Casting of Metals',
Metals Soc. Pubi. 192, 1979, ρ 401.
[11].G. El Nayal and J. Beech, Mat. Sci. and Tech., 1986,2, ρ 603.
[12].Ν. Suutala, Metall. Trans. 1982,13A, ρ 2121.
[13].T.A. Siewert et al, Welding Research Supplement, Dec. 1988, ρ 289 s.
54. 48
Paper4
Estimating annual sale of a refractory coating enabling fuel saving in
furnace operation.
Mainak Saha
Department of Metallurgical and Materials Engineering, NIT
Durgapur
Email:mainaksaha1995@gmail.com
The most basic model for insulation on a pipe is shown below. r1 show The
outside radius of the pipe r2 shows the radius of the Pipe+ insulation.
Heat loss from a surface is expressed as
H = h X A x (Th-Ta)
Where
h = Heat transfer coefficient, W/sq.m-K , H = Heat loss, Watts
Ta = Average ambient temperature, ºC
Ts = Desired/actual insulation surface temperature, ºC
Th = Hot surface temperature (for hot fluid piping), ºC & Cold surface
temperature for cold fluids piping)
For horizontal pipes, heat transfer coefficient can be calculated by:
h = (A + 0.005 (Th – Ta)) x 10 W/m2-K
For vertical pipes,
h = (B + 0.009 ( Th – Ta)) x 10 W/m2-K
55. 49
Courtesy : Bureau of energy efficiency
Pipe insulation
Courtesy : Bureau of energy efficiency
k = Thermal conductivity of insulation at mean temperature of Tm, W/m-C
tk = Thickness of insulation, mm
r1 = Actual outer radius of pipe, mm
r2 = (r1 + tk)
56. 50
The heat flow from the pipe surface and the ambient can be expressed as
follows
From the above equation, and for a desired Ts, Rl can be calculated. From
Rl and known value of thermal conductivity k, thickness of insulation can
be calculated.
57. 51
x ECONOMIC THICKNESS OF REFRACTORY
COATING
Insulation of any system means capital expenditure. Hence the most
important factor in any insulation system is to analyse the thermal
insulation with respect to cost. The effectiveness of insulation follows the
law of decreasing returns. Hence, there is a definite economic limit to the
amount of insulation, which is justified. An increased thickness is
uneconomical and cannot be recovered through small heat savings. This
limiting value is termed as economic thickness of insulation. Each industry
has different fuel cost and boiler efficiency. These values can be used for
calculating economic thickness of insulation. This shows that thickness for
a given set of circumstances results in the lowest overall cost of insulation
and heat loss combined over a given period of time.
Courtesy : Bureau of energy efficiency
The simplest method of analysing whether you should use 1” or 2” or 3”
insulation is by comparing the cost of energy losses with the cost of
insulating the pipe. The insulation thickness for which the total cost is
minimum is termed as economic thickness.The curve representing the total
cost reduces initially and after reaching the economic thickness
corresponding to the minimum cost, it increases.
The determination of economic thickness requires the attention to the
following factors.
58. 52
i. Cost of fuel
ii. Annual hours of operation
iii.Heat content of fuel
iv. Boiler efficiency
v. Operating surface temperature
vi. Pipe diameter/thickness of surface
vii. Estimated cost of insulation.
viii.Average exposure ambient still air temperature
Determination of Economic Thickness of Insulation
Courtesy : Insulation and Refractories – British Energy Efficiency Office
59. 53
Courtesy : Bureau of Energy Efficiency
The total cost in lower when using 2” insulation, hence is the economic
insulation thickness.
60. 54
x Simplified formula for heat loss calculation
S = [10+(Ts-Ta)/20] x (Ts-Ta)
Where
S = Surface heat loss in kCal/hr m2
Ts = Hot surface temperature in oC
Ta = Ambient temperature in oC
Total heat loss/hr (Hs) =S x A
Where A is the surface area in m2
Based on the cost of heat energy, the quantification of heat loss in Rs. can
be worked out as under:
Equivalent fuel loss (Hf) (kg/Yr) =Hs x Yearly hours of operation
------------------------------------------
GCV x ήb
nnual heat loss in monetary terms (Rs.)
= Hf x Fuel cost (Rs./kg)
Where
GCV = Gross Calorific value of fuel kCal/kg
ήb= Boiler efficiency in %