from X to point Y, the number of nuclei from which solidification of
positions 1 and 2 would have mainly originated from the misalignment
of the thermocouples and the possibility of the base thermocouple
impinging on the copper stool surface and becoming detached from the
casting (e.g. thermocouple 1, cast B 2 ) . Therefore, actual thermo
couple displacements (Table 18) were determined utilising an optical ted
microscope with a gradua/ travelling stage. Inaccuracies in these
measurements (due to sample alignment problems) are likely, possibly
in the order of + 1mm. In the region nearer to the chill, where cool
ing rates are changing rapidly with position, any slight inaccuracy
in measurement of position will, be much more significant and thus
account for the greater variability in cooling rates obtained from
the output monitored from thermocouples in this region.
There is a slightly greater spread of cooling rate for thermo
couple number 8 than for thermocouple number 7 which is accentuated
by the logarithmic scale. The likely cause of this variation is the
occasional lack of total insulation above the feeder head permitting
slight radiation heat loss. This phenomenon is illustrated on Figures
51, 55/ 57 by juxtapositioning of the readings from thermocouples 7
and 8.
To carry out the statistical analysis (the results of which appear
in Table 19 and Figures 61 and 62) twenty or more readings were required
ment, readings from a range of positions had to be assigned to the
nominal thermocouple positions (Table 19). There are therefore, small
inaccuracies inherent in the mean and standard deviation figures and,
hence, the limits of significance ( | -1.96<J | ).
The aim of this work was to develop the technique used by Gadgil
and K o n d i c ^ 0 ^ to produce a variety of cooling rates to investigate
the effect of compositional variation on the structure of ingot
mould type iron. The variability in cooling rates exhibited by thermo
couples number 1 and 2 was, therefore, not particularly significant
since, at these positions (Table 8b and Figure 9), white iron struct
ures were produced.
(68)
Information indicates that complete solidification of an
ingot mould may take as much as ten hours, suggesting that the metal
within the mould wall cools almost uniformly at an extremely slow rate.
The technique adopted in the present investigation has clearly produced,
even at the greatest distances from the chill, cooling rates significantly
greater than those encountered in commercial ingot mould practice.
During the development of the experimental method, measures were
taken to reduce the minimum cooling rate produced in the castings by
increasing the section of the castings (see Appendix 1), reducing the
water flow rate and increasing the mould preheating furnace temperature.
Further reduction in cooling rates was not undertaken so as not to risk
significant heat input occurring to the casting within the furnace,
thereby incurring non-unidirectional heat flow conditions. To have
reduced the rate of heat abstraction still further would also have
In spite of the differences in cooling rate between the present
work and commercial casting and solidification of ingot moulds, the
work carried out here does, however, provide a useful general insight
into the interaction of sulphur, nitrogen and titanium in ingot mould
type irons.
5.2 THE MICROSTRUCTURES OF THE INGOT MOULD TYPE IRONS
\
During the investigation of the effects of varying sulphur,
titanium and nitrogen contents on ingot mould type iron, a variety of
structures were obtained. A number of features however, were found to
be common to the majority of the melts. These were:-
1 . pro-eutectic austenite.
2 . the formation of the carbidic eutectic in close proximity
to the chill.
i
3. the formation of 'directional' or 'streamer' graphite
(Gadgil and K o n d i c ^ 0 ^).
4 . the formation of a graphitic area between impinging streamers
and/or dendrites which has been termed the 'grey' region in
the present work, and
5. the formation of graphitic regions towards the top of the
5.2.1 Proeutectic Austenite
The proeutectic austenite dendrites are a pre-requisite of
(1 2 4 8 9)
solidification in hypoeutectic irons ' ' '
'
and were nucleated at numerous sites at the chill/metal interface (Plate 36). On subsequentcooling the austenite underwent transformation to pearlite, which is
seen in all the photomicrographs, but, in some cases, there was partial
transformation at the dendrite arm extremities to ferrite and graphite.
At the chill interface the dendrite orientation was random due to
nucleation and growth of dendrites of various orientations on contact
with the chill. This random orientation only existed over a short
distance as those dendrites with a favourable orientation for growth
rapidly crowded out those less favourably aligned (Plates 36 and 66).
The dendrites were.very directional over the majority of the casting
lengths (Plates 44, 81) and often were clearly visible in graphitic
areas towards the top of the castings. In these regions the dendrites
tended to be more randomly orientated. It is thought that this was due
to lower temperature gradients and the influence of growing eutectic
cells rather than significant departure from unidirectional heat
abstraction (Plates 69, 85, 88, 114, 127 and 128). It is clearly visible
in Plates 176 and 186 that cells have interfered with growth.
In close proximity to the chill the dendrites were very fine
(Plates 36, 37, 66, 90 and 161) but, as expected with the lower cooling
rates prevalent at greater distances from the chill, the dendrites
coarsened (Plates 44, 60, 67, 81, 105, 113 and 130). In all of the
castings, except for Gl, the primary dendrites were very prominent.
However, in Gl, the dendritic appearance is not as marked (Plates 90