July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 541

THE DESIGN OF ELECTRIC REDUCTION FURNACES

FOR FERRO-ALLOY PRODUCTION
By J . J . Coetzee*, B.Sc., B.Sc.(Eng.) (Visitor)

SYNOPSIS
T he history o f the application o f the electric arc furnace to the extraction o f m etals from
refracto ry ores is briefly discussed.
Difficulties associated with the use o f prebaked electrodes and the high cost o f carbon and
graphite, led to the invention o f the continuous self-baking electrode. T he application o f this type
o f electrode to ferro-alloy furnaces is discussed.
F ollow ing on a description o f a typical furnace crucible design, the factors involved in de te r­
m ining the furnace characteristics such as electrode size, spacing etc., are analyzed, ft is show n th at
the physical and chem ical properties o f the raw m aterials to be used m ust be taken into account
w hen designing a furnace. Because o f the peculiar properties o f S outh A frican reductants, optim um
fu rn ace p roduction is achieved a t a som ew hat higher burden resistance than in E urope and Am erica.
T h e successful application o f the continuous self-baking electrode to ferro-alloy furnaces is
d ependent on an understanding o f the baking process and the problem s involved in conducting
large electric currents to the electrode.
T he use o f closed-top type furnaces, in order to extract and clean the fum es involved in the
process, introduces new problem s o f furnace design and o f the p reparation o f the raw m aterials to
be used.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

TABLE OF CONTENTS
IN TR O D U C TIO N :
FERRO-ALLOY FURNACE DESIGN
GENERAL
FURNACE CRUCIBLE OR HEARTH
FURNACE CONFIGURATION
ELECTRODE DESIGN
ELECTRICAL LEADS, BUS-BARS, BUS-TUBES, ETC.
CLOSED TOP FURNACES
CONCLUSION
ACKNOWLEDGEMENT
REFERENCES

INTRODUCTION
T he developm ent o f the m odern submerged arc electric smelting furnace followed
o n the discoveries o f the electric dynam o during the last century.
Credit m ust go to M oissan1, who, recognizing the value o f the concentrated
h eat in the electric arc for reducing the oxides o f the m ore refractory ores, first
d em onstrated the use o f the new tool in the extraction o f a variety o f metals from
their ores.
U p to th at tim e the blast furnace had been successfully applied to the production
o f ferro-m anganese but was found capable o f m aking only a com paratively low grade
o f such alloys as ferro-chrom ium and ferro-silicon. The alloys o f tungsten, m olyb­
denum , cobalt, etc. depended for their production on such processes as the alum ino-
therm ic and silico-thermic.

‘ R esearch and D evelopm ent E ngineer, A frican M etals C orp oration, Ltd.
 542 Journal of the Sou th African Institute of M in in g and M e ta llu rg y July 1 9 ^ 1
The d e sig n o f e le c tric red u ctio n f u r n a c e s f o r fe r r o - a llo y r e d u ctio n — J . J . C o e tz e e

The industrial application o f M oissan’s w ork was soon taken up in France,

G erm any and Italy. However, H erould, w orking in France and Germ any, can be
considered to be the pioneer in this respect.
A lthough the first electric arc furnaces were intended for the recovery o f scrap
iron and steel, they were soon applied to the production o f calcium carbide, for
which product a fast growing dem and was then developing. Indeed, the calcium
carbide furnace was the prototype o f the electric ferro-alloy furnace. Calcium carbide
units are possibly the largest furnaces operating in the w orld today, single furnaces
w ith ratings o f 40 mVA and m ore being no exception.
Because carbide furnaces, like phosphorus furnaces, operate at a relatively high
secondary voltage com pared w ith ferro-alloy units, their secondary currents are
com paratively small for equal kVA ratings and therefore larger^ capacity units can
be readily designed.
The electric furnace o f the early days was entirely dependent on carbon or graphite
electrodes. As these electrodes were available only in com paratively small sizes, the
furnace capacities were limited. To solve this difficulty the so-called ‘packet’ furnace
was developed. In these units a num ber o f rectangular electrode blocks were assembled
into a large rectangular electrode. In this way very large electrode currents could be
handled. U nits o f this type with two or m ore phases, usually had the electrodes
arranged in a straight line to ease construction and operation difficulties. M any of
these ‘packet’ furnaces are still in use today.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

The dem and for bigger electrodes to handle increasingly larger currents
stim ulated the carbon and graphite electrode industry to im prove their raw m aterials
and m anufacturing m ethods, in order to m ake these massive products o f equal high
quality to the electrodes o f m ore m odest size.
It is now possible to obtain excellent quality prebaked electrodes, carbon or
graphite, o f 40 in. diam eter th at will take currents o f 56,000 am peres in the case of
carbon, and 70,000 am peres in the case o f graphite electrodes.
The availability of large prebaked electrodes had caused the ferro-alloy industry
in countries where such electrodes are obtainable a t an econom ical price, to depend
entirely on them for their furnaces. This is particularly true for the U nited States of
A merica where the raw m aterials for these electrodes, viz. petroleum -coke and pitch,
are abund an t and cheap. M odern oil-refinery processes do not, however, make
petroleum -coke available and the quality o f m any cokes is poor. Thus we find that
even in the U nited States m any new furnaces are now designed to use the self-baking
electrodes.
Furnaces equipped w ith prebaked electrodes have the draw back th at they cannot
be operated continuously because o f the necessity to add new electrode lengths as
these are consum ed. This is very rarely possible with the furnace on load.
This difficulty was largely overcome by the developm ent o f the continuous self­
baking electrode in 1909. This electrode, which is form ed o f a steel outer casing of
circular cross-section and w ith radial ribs extending inwards, is filled with a carbon-
ta r paste. It was invented by the Swedish engineer C. W . Soderberg, w orking in
N orw ay and bears his name.
The ‘green’ carbon paste consists o f a m ixture o f different sizes o f calcined
anthracite and a tar binder. It is so designed as to become fluid and to com pact in
the electrode m antle at the tem peratures prevailing ju st above the electrode contact
 July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 543
T he d e sig n o f e le c tric re d u ctio n fu r n a c e s f o r fe r r o - a llo y re d u ctio n — J . J . C o e tz e e

clamps which conduct electric current to it, norm ally a few feet above the furnace
bed. As the electrode is slowly consum ed the paste is gradually subjected to increasing
tem peratures and the lighter fractions o f the tar start to distil off. The vapours are
forced dow nw ards through the pores o f the hot, sem i-baked electrode by the im per­
vious plastic mass o f paste above. C racking and carbon deposition take place with
the result th at the electrode density increases. As the now m ore dense electrode
moves dow n into the furnace burden it is progressively subjected to higher tem pera­
tures. The result is th at the last few inches a t the lower tip consists o f an impure
form o f graphite.
Over the years continuous im provem ents in electrode design and to the paste
have taken place, so th at we are today able to design an electrode o f up to 60 in.
diam eter th a t will handle as m uch as 85,000 amperes.
Sdderberg electrodes are being used m ostly in carbide and ferro-alloy furnaces.
The cost o f these is usually not m ore than 3d. to 4d. (2}c to 3c) per pound o f paste
which is ab o u t one-half to one-third th at o f prebaked carbon electrodes, and one-fifth
the cost o f graphite electrodes. The truly continuous nature o f operation on Soderberg
electrodes and the low cost, m ake this electrode particularly attractive for use in
furnaces o f large capacity.
A n excellent description o f a furnace with Soderberg electrodes has been given
by D r W. Bleloch) in a paper published in the Journal o f the Chemical, Metallurgical
and M ining Society o f South Africa. The interested reader is referred to this paper
for a general description o f such a plant.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

FERRO-ALLOY FURNACE DESIGN

General
The field o f ferro-alloy furnace design is so large th at one cannot cover it all
in a paper o f this nature. It is therefore proposed to limit our deliberations to some
problem s concerning the design o f three-phase ferro-alloy furnaces equipped with
Sdderberg electrodes. Such furnaces are rarely o f less than 3,000 kVA or m ore than
15,000 kV A capacity.
They m ay be o f the open top type, allowing the gases to burn on the surface o f
the furnace bed, o r the closed top type w ith equipm ent to w ithdraw the unburned
gases and clean them from dust before ejecting through a flare to atm osphere. The
closed type furnace is generally successfully em ployed for the production o f pig-iron
and ferro-m anganese. Furnaces for ferro-silicon an d ferro-chrom ium are alm ost
always o f the open top type. M any attem pts have been m ade to convert furnaces
operating on these alloys to the closed type, in order to facilitate the extraction and
cleaning o f the furnace gases, but w ithout m arked success. Ferro-chrom ium , and
particularly ferro-silicon furnaces m aking a high-analysis alloy, need constant w orking
o f the burden top to prevent crusting an d severe ‘blowing’. I f this is neglected,
operation becomes erratic and results in p o o r m etal recovery and high power con­
sum ption. Nevertheless, it is a m atter o f tim e until suitable fume-collecting m ethods
and operating techniques are developed to produce these alloys in submerged arc
furnaces w ithout the discharge o f large volumes o f dust to atm osphere. Proper
preparation o f the raw m aterials to ensure good porosity o f the bed will be one o f
the pre-requisites to successful closed-furnace operation on these alloys. In some
cases, particularly for ferro-silicon m anufacture, recourse is taken to a rotating or
oscillating furnace hearth in an attem pt to prevent the form ation o f hard impervious
crusts on the bed.
 544 Journal of the Sou th African Institute of M in in g and M e ta llu rg y July 1961
The d e sig n o f e le c tric re d u ctio n fu r n a c e s f o r fe r r o - a llo y re d u ctio n — J . J . C o e tze e

Furnace crucible or hearth

The form of a furnace hearth is determ ined by the particular arrangem ent of
the electrodes. W here these are arranged in a straight line, or nearly in a line, the
crucible is usually designed as a rectangular tank. W here the three electrodes are
placed at the apexes o f an equilateral triangle, the crucible is alm ost always designed
as a cylinder.
The furnace shell, enveloping the refractory hearth lining, invariably resembles
a shallow tank 10 to 15 ft deep. Therefore submerged arc furnaces are also called
‘ta n k ’ furnaces. Because o f the high tem peratures prevailing in these furnaces, ade­
quate cooling o f the crucible shell m ust be provided. W ater cooling o f the sidewalls
and resting the shell on a grid o f steel beam s to allow free access o f cooling air to
the bottom , generally suffice.
Furnace hearth linings for units m aking the m ore com m on alloys last as m any as
10 to 20 years, w ithout m ajor repairs. F o r this reason proper provision m ust be
m ade during the design to allow for, not only the initial expansion o f the hearth
lining when brought to operating tem perature, but also the gradual ‘grow tth’ o f the
refractories over the years w ithout dam age to the shell. One m ethod is to design the
tan k as a portion o f an inverted cone, the sidewalls having a taper o f 8° to 10° to the
vertical. This wall-slope converts radial pressure to a vertical force, m oving the outer
layer o f refractories up tow ards a greater diam eter o f the shell. A diagram o f such a
furnace shell w ith lining is shown in Fig. 1, which also illustrates how the furnace
lining is made. Such furnace shells have been in operation for nearly 10 years at the
K ookfontein W orks o f African M etals C orporation, Ltd. and show no visible dis­
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

tortion. Shells are usually m ade with riveted joints because welded joints o f norm al
quality do not last long under the therm al conditions experienced in electric furnaces.
However, completely welded shells o f cylindrical design with semi-spherical bottom s
are in successful use. These dem and a very high standard o f welding followed by
proper stress-relieving.
The internal carbon lining o f tank furnaces can be m ade o f pre-baked and shaped
carbon bricks or p u t in as a ram m ed, m onolithic carbon paste. The latter is much
less costly and in our experience lasts ju st as long. A pre-baked carbon block lining
for a 24 ft furnace shell would cost in the neighbourhood o f £6,000 (R 12,000), whereas
a similar paste lining could be installed for less than £2,000 (R4,000).

Furnace configuration
Successful furnace design is a m atter o f determ ining the best dim ensional relations
between such factors as furnace electrical characteristics, hearth diam eter and depth,
electrode diam eter and spacing.
An adequate balance m ust be obtained between the m ost desirable characteristics
o f the furnace from m etallurgical considerations and providing these with electrical
equipm ent at a reasonable cost. In the physical arrangem ent o f items such as trans­
form ers, bus-bars, charging chutes, electrodes, etc. a keen appreciation o f operational
and m aintenance procedures is necessary. W here a furnace o f a capacity not pre­
viously encountered is to be designed, the engineer makes use o f experience gained
on other furnaces m aking the product it is desired to produce in the new unit.
One o f the m ost interesting articles on furnace design is th at published by W. M.
Kelly in 19583. In this an attem pt is m ade to apply the concept o f electrode periphery
resistance, developed by F. V. A ndreae over 30 years ago, to the design o f submerged
arc furnaces for different products and at different electrode power densities.
 July 1961 Journal of the Sou th African Institute of M in in g and M e ta llu rg y 545
T he d e sig n o f e le c tric re d u ctio n fu r n a c e s f o r fe r r o - a llo y re d u ctio n — J . J . C o e tze e

W hilst this concept merits intensive study by everyone concerned with submerged
arc furnace design, the relations established by the author between the ‘k ’ factor
and electrode pow er density for different ferro-alloys, cannot be applied w ithout
regard to the raw m aterials used.
Let us consider the im plications o f a too rigid adherence to Kelly’s proposals.
A ndreae’s form ula for the electrode periphery resistance is :
. E 77 D
= ~1
where k = E lectro d e periphery resistance
E = E lectro d e to ground voltage
D = E le c tro d e diam eter in inches
I = E le c tro d e current.
By writing the form ula
E
A:=//77_D
we note that
I Electrode to ground voltage
C urrent per inch o f electrode periphery
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

CEMENT.

Fig. 1

In Fig. 2 we show an idealized ^yiew o f a furnace producing an alloy such as

high carbon ferro-m anganese.
The ‘k ’ factor now expresses the resistance in ohm s between the elctrode surface
and the m etal b ath to the current per square inch or the total electrode current
considered as leaving the electrode in a circum ferential ring one inch wide.
The optim um furnace voltage, E can be considered as constant and the electrode
current / is kept constant as nearly as possible by the furnace regulators which move
the electrode up or down, thus keeping the furnace resistance ‘k ’ constant. The
furnace resistance ‘k ’ is m ade up o f the slag resistance Rs, plus the burden resistance
R b• W hereas the burden resistance per unit length is dependent mainly on the physical
and electrical properties o f the reductant, it is obvious th at with small coke o f high
electrical resistivity for instance, the electrode m ust operate with its tip very near to
the slag surface. The opposite is true for large coke o f high conductivity.
 546 Journal of the Sou th African Institute of M in in g and M e ta llu rg y July 1 9 6 1
T h e d e sig n o f e le c tric red u ctio n fu r n a c e s f o r fe r r o - a llo y re d u ctio n — J . J . C o e tz e e

By establishing the optim um values o f furnace voltage and electrode current in a

num ber o f furnaces in this country, m aking the same alloy from raw m aterials of
very similar physical and chemical properties, it has been proved th at the ‘k ’ factors
o f these units are a constant despite fairly wide variations in electrode power densities.
This is contrary to K elly’s findings th a t the ‘k \ factor decreases with electrode power
density for optim um production o f any particular alloy.
We use m ostly small coke o f high ash content with an electrical resistivity
som ew hat higher than th a t found on A m erican and E uropean cokes. The result is
th a t our electrodes tend to ride too deep in the furnace. By increasing the electrode
voltage (E ) or the percentage coke in the burden this tendency can be com bated. We
find, therefore, th at the best production figures on furnaces in this country are obtained
with ‘k ’ factors som ew hat above those shown by K elly’s curves. The properties of
the constituents m aking up the furnace burden m ust therefore be considered when
deciding on w hat ‘k ’ factor should be used. By the use o f these relationships and
rem em bering th a t S8derberg electrodes operate well on current densities o f from
25 to 45 am peres per square inch o f cross section, and th a t voltages around 90 to 160
are custom ary between phases for ferro-alloy furnace transform ers, the designer
can determ ine the required electrical relations and electrode diam eter. The higher
electrode current densities are norm ally used in furnaces m aking the less volatile
alloys such as silicon and chrom ium . F o r m anganese furnaces the lower values
contribute to operating at a lower electrode-tip tem perature and thus a lower stack-
loss o f the m etal. By designing the furnace transform ers to give any secondary voltage
from 90V to 160V in steps o f not m ore th a n 8 volt ap art, a very flexible arrangem ent
is achieved.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

The next step is to determ ine the electrode spacing and furnace diam eter. The
relations proposed by K elly are indicated in Fig. 3.
According to this diagram it is proposed to space the electrodes at such a dis­
tance th at if D R is the assum ed reaction zone diam eter around one electrode, working
on the raw m aterials and m aking the alloy envisaged, then D F = 2 D R and

E S = ^ ^ D R = 0 - m D R.

The problem therefor resolves itself into determ ining the reaction zone diameter,
D R . To say th at D R is prop o rtio n al to the square-root o f the kilow att input o f the
furnace appears reasonably close to the tru th but leaves the designer with the problem
o f finding a value for D R on an existing furnace working on the raw m aterials to be
used in the new furnace, on which to base his calculations. This inform ation is essential
to ensure a reasonable m easure o f success with the new furnace design. This over­
simplification o f the relations between furnace param eters can lead one to serious
errors unless proper care is taken. The reaction zone diam eter m ust o f necessity depend
on the electrode diam eter and the burden porosity. It is also related to the electrode
current density. Furtherm ore, for alloys such as ferro-silicon it is an advantage to
have the reaction zones overlapping m uch m ore than shown in the diagram , i.e.,
the electrode-spacing is sm aller th an th a t given by the form ula. It being impossible
to determ ine the exact dim ension o f the reaction zone in any particular furnace, the
assum ption th at the relations between DF, D R and E S are as indicated in the diagram
m ay easily lead to either a too wide electrode spacing or to a too small furnace dia­
m eter.
A nother m ethod is to determ ine the furnace diam eter by calculating the therm al
load per square foot area and relating this to experience gained on other furnaces
 July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 547
T h e d esign o f e le c tric re d u ctio n f u r n a c e s f o r fe r r o - a llo y re d u ctio n — J . J . C o e tze e

m aking the same product. Once again it is necessary to emphasize th a t raw m aterial
characteristics and preparation play a very great role and one may find th at therm al
loadings outside o f the ranges previously used sometimes yield excellent results.
All these considerations point to the great value o f pilot plant runs on any new
raw m aterials before starting the design o f furnaces to utilize them.

Electrode gear
The design o f the electrode gear o f a Soderberg electrode varies considerably
between the different suppliers o f this type o f equipm ent. There are, however, some
com m on problem s and we propose to discuss these, referring to the typical electrode
arrangem ent shown in Fig. 4.

D F
D R
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

Fig. 3 Fig. 4

The electrode casing is freely suspended inside the cooling m antle and cooling
air is blown dow n the annular space between these two cylinders. To prevent this air
from leaking out o f the top, a seal is introduced ju st above the suspension yoke.
Where the annular space is usually only a m atter o f 2 to 3 in. wide, any accum u­
lation o f dirt, tar and pieces o f the packing m aterial in the top seal, will soon prevent
the free m ovem ent o f the electrode casing in the cooling m antle. It is necessary to
slip the electrode at frequent intervals during operation through the cooling m antle
and contact clamps in order to m aintain a constant length below the clamps. N or­
m ally some 24 in. are consum ed in 24 hours. The best results are obtained if the
electrode is slipped ab o u t 1 in. every hour. This m ovem ent m ust be kept as sm ooth
as possible otherwise dam age can be done to the electrode, causing breakages and
operating interruptions.
It is im m ediately obvious th a t the electrode casing should be a perfectly circular,
sm ooth and gas-tight tube. Small holes will cause ta r vapour to escape ju st above the
contact pads where the green paste enters the h o t zone and these vapours then
 548 Journal of the Sou th African Institute of M in in g and M e ta llu rg y July 1961
T h e d e sig n o f e le c tric r e d u ctio n f u r n a c e s f o r fe r r o - a llo y re d u ctio n — J . J . C o e tze e

condense and carbonize on the casing surface. This will not only cause trouble when
slipping the electrode through the cooling m antle and pads, but may cause such an
increase in the tem perature o f the green paste inside the casing, by obstructing the
flow o f cooling air, th a t the gases escaping from the to p o f the paste catch fire and
can cause serious damage.
A nother point a t which the Soderberg electrode frequently causes operating
difficulties is the contact clamps. W hen one rem embers th a t currents o f the order o f
50,000 am peres have to be carried from the copper bus-bars and tubes into the
electrode paste, it can be realized th a t this is often a form idable task under the hot
and dusty conditions prevailing above the furnace bed. Experience has shown that
the current density in the contact area between the contact pad and electrode casing
should n o t exceed 12 am peres p er square inch, and is preferably around 8 amperes
per square inch. It will be realized th at this requires large contact pads for furnaces
operating at a low voltage and therefore high secondary current.
A m easure o f relief is possible by m aking the delta o f the transform er secondary
leads in the bus-tube ring above the contact pads and n o t in the electrode itself. This
reduces the current to be carried by the contact clam ps from double, to I • 732 times
the transform er phase current.

Electrical leads, bus-bars, bus-tubes, etc.

W here such high currents are used at relatively low pressures in ferro-alloy
furnaces, great care m ust be devoted to avoiding unnecessary induction losses. This
is the m ore essential because it is n o t possible to site the transform er or transform ers
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

feeding the furnace near the point o f pow er utilization, i.e. the electrodes, due to the
high tem peratures near the furnaces. By suitably interleaving the two leads from one
leg o f the transform er, the inductive losses can be held at a low value up to a point
adjacent to the tw o electrodes taking pow er from these leads.
It is after this point, where it is unavoidable to have leads carrying pow er parallel
to each other th a t the heavy losses occur. D u rre r4 gives calculations for a 10 mVA
furnace w orking at 130 volt transform er secondary phase voltage, which show that
about 28 per cent o f the total furnace reactance occurs in the movable copper leads
between the busbars and electrodes, whilst alm ost 50 per cent occurs in the parallel
electrodes and associated iron masses in the furnace. There is thus a voltage drop o f
35 -5V from the transform er to the electrode. The furnace should therefor operate
with the shortest possible electrodes consistent w ith good operating conditions in
order to achieve a high pow er factor. The designer can contribute to this by keeping
the parallel copper leads to a m inimum length.

Closed top furnaces

The design o f a closed top furnace introduces a num ber o f new problem s. The
gases evolved in an electric reduction furnace m aking ferro-m anganese contain about
70 per cent carbon-m onoxide and are thus highly explosive and poisonous.
The gas-offtake system m ust operate at all times under a slight positive pressure
to prevent th e inhalation of air into the system and reduce the risk o f explosions
occurring. However, by so doing there is the constant danger o f unburned CO gas
accum ulating in w orking areas. Proper ventilation o f all such possible points is
therefore necessary.
The furnace ro o f m ust further be provided with seals at the electrodes where
these pass through it and w ith feed-pipes to introduce the raw m aterials. One way to
 July 1961 Journal of the South African In stitute of M ining and M e ta llu rg y 549
T h e d e sig n o f e le c tric red u ctio n f u r n a c e s f o r fe r r o - a llo y reduction — J . J . C o e tz e e

simplify the ro o f arrangem ent is to charge the raw materials around the electrodes,
thus elim inating the feed-chutes and electrode seals.
It is essential to provide closed-type furnaces with raw m aterials o f better physical
properties than can be tolerated in open furnaces. Furnace control should also be
more strict. One can imagine w hat the result will be if a severe ‘boil’ occurred in a
closed furnace.
The introduction o f a ro o f necessarily m eans longer electrodes and thus a
poorer power factor. W hilst this does not place an unbearable penalty on the con­
sumer paying for pow er on a kilow att m axim um demand charge, those who have to
pay on a kilovolt-am pere basis m ust introduce larger power factor correction
apparatus to reduce costs. W hilst it takes only a few hours to put an open furnace
back in production when an electrode is broken by accident, the same occurrence
in a closed-top furnace will usually put the unit out of commission for a day or two.
Unless use is m ade o f the energy in the gases collected from a closed furnace,
there is very little to recomm end it to the operator other than the elim ination o f a
fume nuisance.
CONCLUSION

A lthough the m odern ferro-alloy furnace can be designed with a m easure of

confidence th at satisfactory m etallurgical results and economical operation will be
achieved, the art has not yet developed to such a degree that the designer can guaran­
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

tee results with new and unknow n raw m aterials. For this reason pilot plant tests
are absolutely essential.
The continuous, self-baking Soderberg electrode is not only m uch m ore.econo­
mical than pre-baked electrodes for use in large ferro-alloy furnaces but it also offers
truly continuous operation. This advantage will be appreciated by the ferro-alloy
furnace operator.
ACKNOWLEDGEMENT

Acknowledgem ent is m ade to A frican M etals Corporation Lim ited, for their
permission to publish this paper and to my colleagues who assisted me with its
preparation.
REFERENCES
1. M o is s a n , H . ‘Der E lektrische O fen,’ Berlin, 1900.
2. B l e l o c h , D r W il l ia m . ‘T he electric sm elting o f iron ores and steels and recovery o f chrom ium
and vanadium .’ Journal o f the Chemical, M etallurgical and Mining Society o f South A frica, 49
(1949), 363.
3. K e l l y , W . M . ‘D esign and construction o f the subm erged arc furnace.’ N atio n al C arbon C om ­
pany’s C arbon and G raphite News, 5 (1948), N o. 1.
4. D u r r e r , P r o f D r R . and V o l k e r t , D r P h i l . G. ‘D ie Metallurgie der F errolegierungen.’
Springer-Verlag, Berlin, 1953, 87.
‘ British P aten t 733963.

Contributions to Discussion
Dr Wm. Bleloch (Past President): ‘M r Chairman, gentlemen, Southern Africa
has been for m any years an exporter o f the ores o f chromium and manganese. Large
capital sum s are invested in electro-metallurgical plants in the U .S.A . and Europe
to treat these ores. We cannot I think fairly attem pt to process all our ores or even
the greater p art o f them , even if we were able to do so, or even if we so processed
them to semi-finished products such as ferro-chromium, ferro-m anganese, etc. But
 550 Journal of th e Sou th African In stitu te of M in in g and M e ta llu rg y July 1961
T he d esign o f e le c tric r e d u ctio n fu r n a c e s f o r fe r r o - a llo y r e d u c tio n — J . J . C o e tz e e

in view o f our considerable im ports o f capital goods, especially those fo r generation

and transm ission o f electric power, we should in equity find purchasers in the U.S.A.
and Europe for at least some processed products from our raw ones, if our selling
prices are competitive.
M r Coetzee’s valuable review o f the subm erged-arc furnace for ferro-alloy
production is m ost opportune for two reasons. It is presented at a stage when the
South A frican gold m ining industry has attained the greatest rate o f production in
its history, and when very large extensions to the steel industry and the production
o f bulk pow er are being planned or have already been p u t in hand.
A m ong other m atters, M r Coetzee has ably discussed the question o f the
peculiarities o f our cokes and coal chars when used as the reducing agents in sub­
merged arc plants.
In particular the au th o r’s rem arks on K elly’s ‘k ’ factor are illuminating. W hen
one considers the incredible shapes assum ed a t times by the battle-scarred working
ends o f Soderberg electrodes, their wide departures from the cylindrical renders the
‘k ’ factor concept a little tenuous; but no dou b t some com fort for designers is forth­
coming from the graphs published by Kelly and earlier by Andreae.
1 have taken the liberty o f reproducing the chart referred to in M r Coetzee’s
text (Fig. I). It seems to me th at looking at the curves for high carbon ferro-chrom ium
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

Fig. 1
 July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 551
T he d e sig n o f e le c tric red u ctio n fu r n a c e s f o r fe r r o - a llo y r e d u ctio n — J . J . C o e tz e e

way up on high at the right, and th a t for standard ferro-m anganese way dow n low
on the left, both uncom fortably rem ote from the dotted theoretical curve in all its
straightness and firmness, th at this chart could be better entitled ‘A nything can
h appen’. M r Coetzee is quite right in saying th at it all depends on the raw m aterial.
A lthough K elly’s chart is a serious and praisew orthy effort to correlate the
perform ances o f m any designs o f subm erged-arc furnaces under a great varity o f
conditions, it is becom ing apparent th at to tackle the problem o f furnace design by
basing calculations on assum ed processes a t the w orking end o f the electrode, o f
which relatively little is at present know n, leads to relatively few results o f practical
im portance. V ariations in raw m aterials are ap t to cause such calculations to fail.
On the other h an d it is becoming apparent th at by designing the electrical side
o f any given subm erged-arc operation so th a t there is a reasonable chance under
steady-state condition o f obtaining the m axim um dissipation o f w att-ful current in
the arc p ath, the effect o f variations in the raw m aterials on furnace output and
conditions and on grade o f product, can be greatly reduced, if not entirely elim inated.
F o r any given design o f arc furnace the m axim um power output th at can possibly
be p u t into the secondary circuit, Wmax= —— where E is the L.T. voltage between
phases, and X is the 50 c.p.s. phase reactance in apparent ohms. This occurs when
and only when r = X , r being the ohm ic resistance o f the arc including the rela­
tively small ohmic resistance o f the secondary leads, contact shoes, etc. W hen
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

r—y 5 w
'
_ F2
rr m a x —

As any installation has a significant inductive reactance and a very low, usually
negligible, capacitative reactance, the furnace circuit has a lagging power factor
equal to cos 0, 0 being the angle o f lag, it being assued th at current and voltage
approxim ate to the sine wave.
The actual pow er inp u t or useful pow er attainable on any existing installation
Wu = Wmax, 2 sin 0. While the m axim um possible pow er Wmax, is therefore dependent
only on E and X which are built into the design, Wu the actual or useful power depends
on the value o f 0 obtained on load, and on the wave distorition. This correlation
m ight lead one to suppose th at design for suppression o f the 50 c.p.s. reactance to
the m inim um is desirable, thereby m aking cos 0 approach unity.
In practice this is not so because in large arc furnaces the question o f arc stability
and wave distortion becomes o f im portance. A t low values o f 50 c.p.s. reactance and
arc currents o f the o rder o f 40,000 apm eres or higher the arc is unstable and po o r
furnace conditions result. There is, in my view, for any given furnace design on opti-
y
m um value o f = cot 0 at which the furnace is m ost likely to be virtually independent
o f the nature o f raw m aterials, provided th at the designed limits o f operating secondary
voltage are met. These last-m entioned param eters are set by the desired size o f furnace,
the expected range o f secondary voltages and the desired economic output, which
in turn have determ ined the m ain features o f the particular transform er design.
A rc furnace plants neglecting these considerations have been designed in the
past and have given p o o r operating results, overriding any considerations o f the ‘k '
factor. The design calculations to meet the desired value o f - ^ = c o t 0 for any given
Jx.
operating voltage range and m axim um continuous secondary current can be extremely
 552 Journal of the Sou th African Institute of M in in g and M e ta llu rg y July 1 9 6 1 '
The d e sig n o f e le c tric red u ctio n f u r n a c e s f o r f e r r o - a llo y r e d u ctio n — J . J . C o e tze e

complex and tedious, despite the Herculean labours o f A ndreae in compiling com­
prehensive tables* which cut short m uch o f the com putations based on the geo­
m etrical m ean distances of the various circuit com ponents such as inter-laced bus­
bars, flexible bus-tubes, etc.
The net result o f these considerations has been th a t some m odern large arc
furnace plants having transform er capacities of 40 mVA or more have been designed
w ith high reactance, even, I understand, as high as 50 per cent in- order to achieve
optim um perform ance w ith stable conditions. The subject is indeed a deep one, which
cannot be adequately dealt with shortly. It is w orthy o f a separate paper, which I
hope M r Coetzee will one day present to us.
The problem is further com plicated by the wide variation o f the current wave
form in the secondary circuit from the pure sine wave on which calculation of
reactance and pow er factor are ideally based. It is well know n in arc furnace practice
th at the arc itself m ay have little or no angle o f lag or lead between current and
voltage, but m ay yet show a low pow er factor because o f the heavy distortion o f the
current wave from the approxim ate sine wave of the voltage. For this reason it is
usual to connect at least the secondary in delta in order to suppress as far as possible
the effects o f the third harm onic which cannot be readily done if the secondary is
connected in star.
W hen very large pow er inputs are attem pted in submerged arc furnaces for
m etallurgical products, the design m ust provide for correspondingly large currents,
because in general the operating voltage o f such furnaces is below 175. To cite an
example if it is desired to operate a furnace for the production o f pig iron by reducing
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

a 55 per cent haem atite ore w ith coal char with a slag volume of say 0-65 ton per
ton o f iron produced, and with an electrical input o f 20 megawatts, the operating
voltage is not likely to exceed 160, and the operating line am perage will then be
of the order of 100,000. This current requires 60 in. diam eter Soderberg electrodes
weighing up to 50 tons each. The m echanical design of furnaces of this size therefore
becomes com plicated because o f the very large currents necessary to attain the
required pow er input at the necessarily low voltage.
In an attem pt to avoid some o f the complexities due to such very large arc currents
smelting furnaces have been built to operate on the open arc principle at m uch
higher voltages than those attained in the subm erged arc design. In this type of
operation a long arc is draw n from each of six electrodes in line using a three-phase
supply. O perating voltages as high as 350 to 400 volts are obtainable and a 20 megawatt
input under these conditions is obtainable at a phase current of only 20 kilo-amperes
instead o f som ething like five times th at value necessary to dissipate 20 megawatts
on the submerged arc principle. The general operating characteristics o f high voltage
6-electrode arc furnaces are shown in Fig. 2.
It is for this reason that the size o f subm erged arc furnaces which operate at
inherently low voltages is limited in practice. As is well know n carbide submerged
arc furnaces can be built in m uch larger unit size than those for production o f ferro­
manganese. Carbide reduction is inherently a high voltage operation, whereas the
m aking o f ferro-m anganese in the subm erged-arc proceeds at a m uch lower voltage.
The open-arc type o f smelter furnace has been developed to a unit size o f 20
m egawatts a t the Sorel plant of the Quebec Iro n and Titanium plant in C anada. A

*See for instance, Journal du Four electrique, N o .’s 2 and 3, 1951; and N o. 2, 1952.
 July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 553
The d e sig n o f e le c tric red u ctio n f u r n a c e s f o r f e r r o - a llo y r e d u ctio n — J. J . C o e tz e e

:
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

r e s is t a n c e m ill i- ohm s
Fig. 2

com parable subm erged-arc unit size is the type built a t M o-i-R ana. The Sorel units
take 21 to 22 kilo-am peres, while those a t M o-i-R ana take up to 100,000 amperes,
for approxim ately the same pow er input.
Originally the subm erged arc furnace was built to provide for three star or
delta connected electrodes in a line arrangem ent. W ith increasing size and electrical
loading it was found that when arranged three-in-line the electrodes could not be
m ade to share the load equally. This phenom enon is due to the m utual inductance
between the leads and electrodes resulting in one or other o f the outer phases having
a higher reactance th an its counterpart. The high reactance electrode will be either
one or the other o f the outer o f the three phases depending on the phase rotation. In
the operation o f large three-in-line units fairly frequent, e.g. daily, phase—rotation
reversals were carried out, in order to even out as far as possible the perform ance
of the outer phases in any given furnace.
This out-of-balance led to the arrangem ent o f the three electrodes at the apices
o f an equilateral triangle w ith the delta connection as close to the electrode as possible.
A m ore even load a t each electrode, and, as m ight be expected, a higher pow er factor,
was obtained under any given set o f circum stances with this triangular arrangem ent,
which in the field of steel-melting has been developed into very large unit sizes, e.g.
40,000 kVA. But it has recently been proposed* to instal a saturable reactor o r trans­
d u c e r o f special design in each phase o f the secondary of the arc furnace circuit
so that the following can be secured:

* B ritish P aten t 833963.

 554 Journal of the Sou th African Institute of M in in g and M e ta llu rg y July 1961
T h e d e sig n o f e le c tric red u ctio n f u r n a c e s f o r fe r r o - a llo y r e d u ctio n — J . J . C o e tze e
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

Fig. 3
 July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 555
The d e sig n o f e le c tric re d u ctio n fu r n a c e s f o r fe r r o - a llo y r e d u ctio n — J . J . C o e tz e e

(i) a definite and accurate adjustm ent o f the reactance o f each phase, either
m anual or autom atic, so that there is no longer any need to arrange the
electrode at the apices o f a triangle. This may lead to "a reversion to the
older three-in-line arrangem ent which is m uch simpler and cheaper to
construct under a given set o f condition, and is for certain metals and alloys
m uch easier to operate.
(ii) a general stabilization o f the arc, and also apparently the suppression o f
arc flare, because o f the latent inductance o f the transductor in each phase.
There is a further aspect o f the result o f installing transductors in the secondary
o f the arc furnace circuits. It is well know n th at even harm onics do not exist in the
arc furnace circuits for the reason th at one o f the effects o f the even harm onic is
to m ake the two halves o f the current wave dissimilar and this is not possible with
the ordinary type o f constant speed a.c. generator. But the odd harm onics exist in
arc furnace circuits and m ay in fact be developed to a degree th at will cause m arked
wave distortion. The large mass o f unsaturated iron which exists and the arc furnace
type o f transductor tend to depress heavily the third harm onic o f frequency three
times that o f the fundam ental, and still m ore so the higher odd harm onics o f still
higher frequency. It is considered th at in this way such transductors will assist in
obviating one o f the prim e difficulties now facing large arc furnace plants taking
pow er from central netw orks, namely the form ation o f ‘light-flicker’ due to fluctua­
tions o f low frequency, e.g. 10 c.p.s. in the high-tension supply network.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

It is possible th a t our own E.S.C. will not leave us in this country for very m uch
longer w ithout enforcing suppression as far as possible o f such low-frequency waves
generated by large arc furnace plants.
M r Coetzee’s rem arks on the power factor o f an arc furnace installation are
very m uch to the point. The general effect o f pow er factor coupled w ith any given
load factor on the cost o f pow er under specific condition o f tariff can be far-reaching
as can be seen from Fig. 3.
In this connection it is evident that the lower the power factor the greater becomes
the electrode current for any given rate o f pow er input. I think M r Coetzee’s statem ent
that by a certain positioning o f the delta connection the electrode carries twice the
phase current is an error. This does not happen because the current in the two con­
nections from different parts o f delta, although seemingly feeding twice the phase
current into an electrode, is in fact feeding two currents which are 120° out o f phase
and the electrode never under any circum stances carries m ore than ^ /3 o f the phase
current.
In conclusion I would like to th an k M r Coetzee for his valuable contribution
to the literature o f the subm erged-arc furnace, which I feel sure is about to play an
im portant p art in the foreseeable industrial future o f this country.

J . P . Coetzee ^(Member):
Furnace Rotation
The speaker has referred to the form ation o f hard impervious crusts on ferro
silicon furnaces. Any ferro-silicon smelter can confirm M r Coetzee’s experience. The
suggested rem edy to smelt this product in a rotating or oscillating furnace is not
altogether a blessing as crusts are still form ed at times and this, in turn, has the
effect o f breaking the electrodes. The breaking o f an electrode is, at the best o f times,
 556 Journal of the Sou th African Institute of M in in g and M e ta llu rg y July 1961
T he d e sig n o f e le c tric re d u ctio n f u r n a c e s f o r f e r r o - a llo y re d u ctio n — /. /. C o e tz e e
\

a m ajor catastrophe in a ferro-alloy furnace. The baking o f the ‘green’ electrode is

so time consuming. Even in ferro-m anganese furnaces breaks are experienced on
account o f the rotation o f the furnaces. Q uite a num ber o f smelters in Europe simply
do not rotate their furnaces in spite o f the fact th a t they are built for this.
To design and build a rotating or oscillating furnace shell and ro o f is very m uch
m ore expensive, but has the advantage o f allowing the lining o f the furnace to wear
evenly. Personally, I think rotation o f a furnace is an extravagant luxury.

Shell design
The expansion o f the lining and the grow th o f the refractories are better allowed
for by having an air gap between the shell and the lining. This gap can be filled with
coke o r any m aterial which can be easily crushed.
Personally I do not like a shell with an 8° to 10° taper as the lining continually
has a tendency to rise, which necessitates the rem oval o f layers o f bricks from the
top from time to time. This design is the direct cause o f very poor tapholes as the wet \
clay for the plugging o f the taphole develops superheated steam at a terrific pressure
in the lining. W ith a taper in the shell, the bricks above the taphole simply move up.
A fter a few years the lining above the tapholes is ju st broken up with bricks lying at
various angles.
Electrode and shell design
The speaker’s efforts to correlate the m ain dim ensions o f a furnace on a more
scientific basis, have been m ost interesting. I w holeheartedly agree with him th at this
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

m ethod gives a good guide if considered in conjunction with adequate practical

experience o f furnace design and available raw m aterials. However, very good results
were achieved w ith conventional m ethods o f design as was confirmed by the speaker.
His rem ark th at one m ay find th at therm al loadings outside the ranges previously
used sometimes yield excellent results has been proved in a num ber o f cases. I feel
th a t particularly for ferro-m anganese and pig iron electrical furnaces, the furnace
shells are norm ally under-designed, i.e. the furnace hearth is too small. Shells th at are
too small necessitate frequent tapping, leaving the furnace relatively cold after it
has been tapped, and w ith a high rate o f production such as o f ferro-m anganese and
pig iron becom ing overheated and over-full before the tap. Slightly bigger hearths
should result in m ore even operation, lower pow er consum ption and a longer life
o f the lining. I happen to know th a t the speaker him self had the experience o f design­
ing a' furnace for 13,000 kVA but, through other lim itations, was forced to operate at
10,000 kVA. He also had the pleasant experience o f tapping the furnace only twice a
shift, yielding excellent results.

Electrical leads, bus-bars, etc.

W ith the heavy currents and the very h o t conditions prevailing in the vicinity of
the contact clamps, bus-tubes, flexible cables and bus-bar joints, oxidation occurs very
rapidly in all copper joints, with the result th at sparking is often experienced. It has
become m ore o r less standard practice in South A frica to silver-plate all copper
joints.
The distribution o f the raw m aterials through holes in the ro o f o f the furnace
round the electrodes, is o f the utm ost im portance. The conventional way o f feeding
the raw m aterials through the ro o f at three places round each electrode is disadvan­
tageous in th a t the furnace blows very strongly through the valleys in the burden.
As the furnace cannot be rabbled, this blowing cannot be stopped, and results in
 July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 557
T h e d e sig n o f e le c tric re d u ctio n fu r n a c e s f o r fe r r o - a llo y re d u ctio n — J . J . C o e tz e e

considerable loss o f energy and overheating o f the furnace gases which have to be
cooled before scrubbing.
The au th o r’s suggestion to charge the raw m aterials all the way around the
electrodes is very interesting, but it is going to be difficult to achieve in practice,
especially as electrical insulation will have to be fitted on the ro o f to avoid short
circuits and arcing through the raw m aterial and ro o f between electrodes.
F urther, I am rather w orried th at the reaction zone o f the furnace might be
raised with the piling up o f the raw m aterials round the electrodes. This can be
dangerous when the electrodes climb out because o f excessive coke on the burden
and will certainly cause the furnace hearth to cool off. The opposite may be equally
dangerous where the furnace runs with a deficiency o f coke and starts boiling. The
boiling slag will then tend to form a solid crust round the electrodes through which
the raw m aterials m ust be fed. However, w ith good furnace control, 1 think these
latter objections can be overcome, and I feel th at if M r Coetzee can accomplish
sound ;electrical insulation between phases on the roof, he deserves a ‘feather in his
cap.’
A lthough I agree th at use should be m ade o f the energy in the gas collected, and
also o f the sludge reclaimed in order to obtain m axim um benefit from the covered
furnace, I feel inclined to think th a t a covered furnace is still w orthwhile even when
the gas or the sludge are not used. The saving on m aintenance alone is fantastic,
not only on the furnace proper, but on all auxiliary plant. The m aintenance on the
electrode gear o f an open furnace can be very severe if the parts are not over-designed
beyond all proportion.
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

J . M eintjes and T. J . van Esch (V isitors): A lthough, as the title of his paper
infers, M r Coetzee has confined him self to the design o f electric reduction furnaces
for ferro-alloy production, the theoretical basis o f design as well as the practical
aspects o f lay-out and operation apply equally well to furnaces for calcium carbide
production. O ur experience has been m ainly in the latter field and we therefore
propose to com m ent on M r C oetzee’s observations and conclusions .as they apply
to furnaces for carbide production.

Baking o f electrodes
In describing the process o f electrode baking it is stated th at as the vapours are
driven off, cracking and carbon deposition take place with the result th at the electrode
density increases. F ro m this it should not be inferred th a t the density o f the baked
paste is greater than th a t o f the green paste, but merely th a t density increase occurs
in the initially baked paste as the electrode moves dow n into the furnace and is
subjected progressively to higher tem peratures. Typical specific gravities are 1-60 for
green paste, 1 • 33 for the past baked in a laboratory furnace and 1 • 5 for baked paste
in the vicinity o f an electrode tip.

Electrode current densities

In the largest carbide furnaces operating today currents as high as 105,000
amperes are employed w ith Soderberg electrodes o f 55-1 in. diam eter equivalent to a
c u n e n t density o f 44-0 am ps/in.2. We ourselves have obtained satisfactory operation
for m any years with 48 in. S5derberg electrodes carrying 80,000 amperes equivalent
to a current density o f 44-2 am ps/in.2. These current densities are, incidentally,
considerably higher than those given by W. M . Kelly1 for prebaked carbon electrodes
o f like diam eter.
 558 Journal of the Sou th African Institute of M in in g and M e ta llu rg y July 1961
The d e sig n o f e le c tric re d u c tio n fu r n a c e s f o r f e r r o - a llo y r e d u ctio n — J . J . C o e tz e e

Shell design
Instead o f shells with welded or riveted joints we have found th a t bolted joints,
all bolts being tightened to constant tension, give the best results. D uring the early
life o f the shell, bolts under too high a tension tend to fracture. However, by replace­
m ent o f such bolts and periodic checking on the tensions o f all bolts, a condition o f
equilibrium is soon attained under which stresses and strains are evenly distributed
throughout the shell. The slight tapering o f the shell sidewalls m entioned by M r
Coetzee for relieving radial pressure, is an im portant design feature.
In carbide furnaces the outer surface o f the shell is never w ater-cooled due, p re ­
sumably, to the explosion hazard th a t m ay arise were w ater to come into contact
with carbide spillage. Instead, the surface is often provided with massive fins which
serve the dual purpose o f reinforcing the shell and o f offering increased radiation
surface.

The peripheral ohm fa cto r and electrode immersions

We are in com plete agreem ent with M r C oetzee’s observations regarding the
7c’ factor used by W. M . Kelly, viz. with the statem ent th at the properties o f the
constituents m aking up the furnace burden m ust be taken into account when deciding
upon the value o f the 7c’ factor and th at the ‘/c’ factor does not decrease with increase
in electrode pow er density.
In a properly designed carbide furnace it is possible to obtain optim um per­
form ance as far as quality o f product and furnace operation are concerned, over a
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

wide range o f load, e.g. between full and half load in the largest furnaces operating
tod_ay. This is done by choosing E and I to m aintain a constant E ll ratio and this
also gives a constant average electrode im m ersion over the range o f load.
This practical operating procedure does not support Kelly’s contention th at
E 77 D
k —— ------- using the nom enclature in M r Coetzee’s paper— is associated with a
definite pow er density. In the exam ple quoted above, the power density (kW /sq.in.
o f electrode cross-section) will vary betw een full and half value. To us, therefore, it
would appear better to choose electrode diam eter for a carbide furnace from con­
siderations o f m axim um perm issible current density only.
It is to be noted th a t kW h/ton usage has been excluded in the definition o f ‘opti­
m um perform ance’. k W h/ton necessarily decreases with increase in load and hence
with increase in furnace size. It cannot therefore fairly be used as a criterion when
com paring the ability at varying loads to produce the desired quality o f product
w ith quiet and trouble-free furnace operation.
O ur experience with electrode im m ersion depths in this country is exactly opposite
to th at o f M r Coetzee. W hile he has found th a t for ferro-alloy production electrodes
tend to ride too deep, we experience trouble with electrodes tending to ride too high.
We attribute this to the fact th at the high ash South A frican cokes require a higher
coke concentration in the charge to provide an adequate am ount o f fixed carbon for
producing a standard quality product. The charge resistivity is thus inherently less
than in European or A m erican practice and electrodes tend to climb out in a furnace
designed for overseas raw m aterials. R iding o f electrodes can be com bated by de­
creasing electrode to hearth voltage or increasing electrode current. W here a trans­
form er is already w orking at its maxim um designed current, however, electrode voltage
can only be decreased a t the expense o f shedding load. Hence every attem pt is made
 July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 559
The d e sig n o f e le c tric r e d u ctio n f u r n a c e s f o r fe r r o - a llo y red u ctio n — J . J . C o e tze e

to increase the charge resistivity by decreasing the particle size o f the coke or by
substituting a p urer form o f carbon such, as anthracite o f which a lesser concentration
is required in the charge. T hat the latter measures have a very significant effect on
furnace operating characteristics is undisputed and confirms M r Coetzee’s observation
th at the k ' factor for the m anufacture o f any particular product is largely dependent
upon the properties o f the constituents com prising the furnace burden. Thus, in our
own particular case, our calculated k factor is only 0-7 o f the value given by Kelly.
We are at a loss to explain why, in ferro-alloy production in this country, the
tendency is for electrodes to ride too deep while for carbide production the exact
opposite obtains. O n the basis o f greater ash in South African cokes the charge
resistivity should be lower in both types o f m anufacture since, according to our
experience, the effect o f the larger p ro p o rtio n of coke in the charge overrides the
effect o f the lower inherent resistivity o f coke m entioned by M r Coetzee.
The depth o f electrode im m ersion is, o f course, governed not only by charge
resistivity—which in tu rn depends upon absolute resistivity o f coke, proportion o f
coke added and particle size—but also by furnace design. The furnace resistance
as a whole will increase with wider electrode spacing or shallower depth.

Determination o f electrode spacing and furnace diameter

We have interpreted K elly’s procedure for determ ining the electrode spacing
and furnace diam eter rather differently. M r Coetzee assumes th a t electrode spacing
is calculated from the reaction zone diam eter and rightly states th a t this param eter
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

is a very difficult one to determine. However, Kelly makes electrode spacing the
independent variable and reaction zone diam eter the dependent variable. H e gives no
procedure for, nor does he him self attem pt to calculate the reaction zone diam eter
as a basis for fixing electrode spacing. On the contrary, he gives a m ethod for calcu­
lating electrode spacing based on design d ata o f satisfactorily operating existing
furnaces and on the data he supplies in support o f his calculations his m ethod certainly
gives a sufficiently adequate answer. H aving fixed the electrode spacing, he goes on to
calculate the furnace diam eter from a consideration o f m inimum, maximum and what
he considers to be the optim um furnace reaction zone.
Thus, if D r is the reaction zone diam eter around each electrode o f diam eter D E;
S, the electrode spacing centre to cen tre;
and D /, the furnace diam eter;
Kelly shows that,
Drmin.=S, and D f 2-16 S, actually 2-15 S;
Dr max.= 2 S — DE, and D rmax = (2-31 S — DE) which is an error and should be
(3-15 S — De)\
Dr optimum= 1 •1 6 S actually 1 • 15 S, and Df optimum= 2 -3 S.
Kelly goes on to suggest th a t the diam eter inside the furnace lining should be
approxim ately 1-15 Dr optimum or 2-65 S.
The mean o f and D fmax.= 2-65 S —^ and Kelly’s choice o f a practical
D f is thus greater th an this m ean by an electrode radius.
This procedure does seem to be som ew hat artificial and am ounts to determ ining
the furnace diam eter from the outer boundaries o f circles concentric with the elec­
 560 Journal of the Sou th A frican Institute of M in in g and M e ta llu rg y
T h e d e sig n o f e le c tric r e d u ctio n fu r n a c e s f o r fe r r o - a llo y red u ctio n — J . J . C o e tze e

trodes and passing through the centroid o f the equilateral triangle, irrespective o f
w hether these circles actually represent the true reaction zones around electrodes
or not. Kelly, however, assumes th a t electrodes have been correctly positioned by
calculating the spacing in the m anner in which he proposes and th at it therefore
follows th a t the electrode reaction zones will coincide with his optim um zones.
We have applied the form ula D f = 1 • 1 5 x 2 -3 5 = 2 -6 5 S as suggested by Kelly
as well as Df = 1 •05 X 2 • 3 S = 2 ■4 S (i.e. K elly’s original 2-3 S increased by only 5 per
cent), to calculate the furnace diam eter o f two carbide furnaces.
On a nom inal 20 M W furnace these give diam eters which are, respectively, 11-6
per cent and 1 • 1 per cent higher than the actual inside dim ension at a height o f 1 ft
above the tap-hole.
On a 25 M W furnace the calculations give diam eters which are, respectively,
9-9 per cent higher and 0 -4 per cent lower th an the design dimensions.
Since K elly determ ines electrode spacing from design data o f existing furnaces
and a proportionality factor, it appears to us th at it would be logical, simpler and
w ould elim inate all uncertainty with respect to the factor by which iS m ust be m ulti­
plied to give D f, if the furnace diam eter were calculated directly in a similar manner.
Thus according to Kelly,
if S x, S 2 are electrode spacings centre to centre on two furnaces
P 1; P2 are the respective kW loads
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

and if D f X, D f2 are the respective furnace diam eters

W hile this relationship is K elly’s own theoretical basis for calculating furnace
param eters, its applicability should be tested against existing furnace data. On the
two furnaces quoted above, it gives a result for the 25 M W furnace which is 3 per cent
too low. F o r an increase from 30 to 40 M W ’s in carbide furnaces o f a certain European
designer, it gives a result very close to actual.
This procedure will eliminate some o f the objections which M r Coetzee makes
to K elly’s m ethod but it rem ains true th a t design d ata calculated only by com parison
o f furnace sizes, cannot take into effect differences in burden com position and porosity.
C harge resistivity is a particularly im portant factor which m ust be considered when
designing a furnace to operate on particular raw m aterials.

Contact shoe current densities

W ith regard to contact shoe current density it is usual in carbide m anufacture
to regard 20 am ps/sq.in. as the safe upper limit and m ost furnaces operate at ab o u t 15
am ps/sq.in. These densities are virtually double those given by M r Coetzee for ferro­
alloy production and perhaps result from the higher electrode current densities on
carbide furnaces. This limits the width o f shoes th a t can be fitted around an electrode
perim eter and it is generally accepted th at increasing the contact shoe area by in­
creasing the height is not very effective for raising its current-carrying capacity.
 July 1961 Journal of the Sou th African In stitu te of M in in g and M e ta llu rg y 561
The d e sig n o f e le c tric re d u ctio n f u r n a c e s f o r f e r r o - a llo y re d u ctio n — J . J . C o e tze e

J iv ? ^ n S,n° d o u ,b t th f operation on low contact shoe current densities pays handsom e

dividends in reduced shoe maintenance.
We found M r C oetzee’s paper b o th interesting and stim ulating. To condense so
he set out tH c h ie v e g ^ ^ task and he succeeded adm irably in w hat

REFERENCE

L pany*’s C a rb o n fU rn aC ’ N a ti° n a ' C a rb ° n C o m ‘

BOOK REVIEWS

M INES IN THE SPINIFEX

Geoffrey Blainey. Angus and Robertson, Ltd., Sydney, 27s. 6d.

This is the story o f the lead, zinc and copper mines o f the M ount Isa and Clon-
tT r iv r f ^ A ? Uffe^ l anu- " lterest naturally centres round the developm ent of
Reproduced by Sabinet Gateway under licence granted by the Publisher (dated 2012.)

e M ount Isa itself. The history o f M ount Isa is an alm ost incredible story o f courage
™ v 'slo,n' " r ° f obstinate refusal to give up in the face o f defeat. F o r 25 years, from
° a ? ’ .t h e ^ m p a n y was always in the red, desperately switching production
°,mt) to zinc t0 copper and back again as the depression, fluctuating m etal prices
and the exigences o f w ar dicated. In the next ten years it came hom e with profits
exceeding £A21,000,000, and is now firmly established as a m ajor world producer of
lead-zinc and copper. If the book has a m oral, it is th at in the m id-tw entieth century
economy, m an cannot any longer afford to play around, as he did only a few years
ago, with the future o f a M ount Isa and its vast m ineral resources.

T im iS 0^ B'failley w roteuthe b,°°,k under the ^ o n s o rs h ip o f M ount Isa Mines,

fr 1 s ct com es through the pages a little easily, especially when he deals
w ith labour relatm ns and such m atters as state taxation, it is only to be expected
good photographs 6 an eXC6llent jo b o f tellin8 an exciting story and picking

All. mining men' dream o f pegging a K londike on their next fishing trip. Y our
reviewer hopes his K londike will not be quite so far from anywhere else or in quite
such inhospitable tropical wastes as M ount Isa.
R.A .L.B.

Afrikaans M ine Ventilation Terms— M ynventilasieterme. Supplem ent to the Journal

I(I shilling).
'n T Kelvin
/ f TH ouse,0C! e!,y °f Soilth A frica’ M ay 196L viii PP-= approx. 10 cents
Johannesburg.

... T £1S. list, containing approxim ately 800 w ords com m only used in mine ventilation
with their A frikaans equivalents, should satisfy a long-felt need. It should be con­
sidered as an initial list and will be revised in due course.
J. de V. L.