Fine Intergranular Surface Cracks in Bloom Casting*

By Manf red M. WOLF**

Synopsis cracking features had been clearly brought about:

The occurrenceof fine intergranular surface cracks is investigatedfor the i) crack areas show a considerable enrichment in
case of bloom section 260 X 260 mm cast on a low head curved mold type Cu, Ni, Sn and Sb (e.g., Cu-rich phase of 85 %
four strand machine with 8 m radius, based on EAF steel supply. Cu, 5 % Sn, 4 % Ni, 5 % Fe and 1 % Sb);
Crack incidencedependson residual elementsprimarily and can be reduced
ii) enhanced crack severity with increasing Cu-
belowcritical levels by limiting the Cu content to max. 0.22 % provided
and Al-contents;
the application of optimized casting conditions.
iii) cracking also increases at higher surface tem-
The latter consist of a casting speedlimit at about 1.0 mlmin, connected
with a minimum mold powder consumptionof 0.3 kg/m2. Furthermore, peratures (e.g., 1 200 °C vs. 1 000 °C), and with
high stability of mold level control and pouring stream directedtoward the increasing impingement density of spray cooling.
meniscusas well as soft indirect secondarycooling below the mold are also The above alloy effects on hot shortness are, of
important, and, at low casting speeds, an increased mold taper appears to course, particularly relevant in case of EAF-steelmak-
be beneficial,too. ing with high residual level from poor scrap quality.
On the other hand, soft mold cooling by reduced water velocityand in- However, the cracking can also be observed in case
creasedwall thicknessas well as wall plating and also total spray cooling of BOF-steel where rather an effect of mold powder
intensity do not show significant effects. Replacementof mold powder by behavior was found decisive,7'8~and crack frequency
oil lubrication indicates a lesser sensitivity to residual content but is not
also reduced by mold EMS9~-beside a reduction in
practical due to occurrenceof large depressions. No effectof C-content be- spray cooling intensity.? 9> The change-over to the
comes apparent but increasing Al-content seems to enhance crack suscep-
tibility. air/mist-type spray cooling gave significantly reduced
cracking, too.lo>
The fact that such intergranular cracking can occur
I. Introduction also in BOF-steel, i.e., without high residual levels,
In bloom casting, the requirements on surface indicates close similarity with other fine surface
quality are steeply increasing-particularly in view cracks but more distinct orientation, i.e., either
of the application of hot charging in order to save transversely or longitudinal.9'12~ The general feature
energy. Thus, it is necessary to avoid any surface of such cracking is the coincidence with " depressed"
defects on the as-cast blooms which can not be scaled- surface areas-although sometimes hardly visible by
off in the reheating furnace. eye and only detectable by a coarser subsurface micro-
Longitudinal face cracks and transverse (near-) structure. Thus, the beneficial effect of mold plating8)
corner cracks which are the prominent surface defects might equally well be explained by a reduced de-
in slab casting, are of less frequent concern in bloom pression frequency owing to the higher wall tempera-
casting. However, there is one type of surface crack- ture of plated molds. In this respect, the mold wall
ing still persistent which appears as network of fine thickness may be a further factor, although the
intergranular cracks (Fig. 1), thus also termed " net- cracking has been reported even with plated block
work " cracks. Such cracks have been described molds of 60 mm wall thickness when mold powder
early in literature and originally attributed to copper selection and strand cooling was not optimized.13~
pick-up from the mold wall"2>; consequently, mold
plating was introduced as a countermeasure. The
similarity in the formation mechanism of star cracks
in slab casting is thus obvious.
Nevertheless, the cracks were also observed using
newly plated molds-although to a lesser extent and,
subsequently, attributed to the grain boundary
infiltration of Cu and Sn at the surface after pre-
ferential oxidation of iron during scaling.3j In ac-
cordance with a similar crack mechanism during hot
rolling,4,5~such cracks were also termed " hot short-
ness " cracks; other terms often used are " checking ",
" crazing "
, " hair ", or " shatter " cracks.
In a fundamental study involving laboratory
simulation as well as plant trials6~ the following Fig. 1. Network of fine intergranular cracks.

Presented to the 105th ISIJ Meeting, April 1983, S156, at The University of Tokyo in Tokyo. Extended version received April 1,
1983. QC 1984 ISIJ
K CONCAST Service Union AG , Todistrasse 7, CH-8027 Zurich, Switzerland.

Research Article ( 351)

( 352 ) Transactions ISIJ, Vol. 24, 1984

In the present case study, cracking problems " multi stage mold " or
, briefly, " MS-mold "). As
became apparent only when a heavier section, i.e., shown in Fig. 2, these plates are cooled by spray
260 x 260 mm was introduced for the manufacture nozzles form behind; no direct cooling of the strand
of seamless tubes. Thus, an investigation was started with corner sprays is applied in case of 260 x 260 mm.
considering the factors of known influence as de- In order to check whether cracking might be related
scribed above. to eventual copper pick-up from the plates, several
casts were also made with direct spray cooling
II. Experimental Procedure instead.
The main caster data are listed in Table 1. For The tests comprised 21 heats. The tested para-
all quality steels like seamless tubes, the submerged meters and their range are listed in Table 3. All
pouring technique is applied. Deviating from the the tests were carried out by casting 260 x 260 mm
smaller sections with chamfered corner geometry, the on one strand only, with smaller sections simul-
mold for the large size of 260 x 260 mm has a corner taneously cast on the other three strands.
radius of 20 mm. Steel superheat in the tundish was maintained
The range of casting speed for each size is given mostly between 25 °C and 40 °C. Regarding steel
in Table 2, together with the spray cooling intensity analysis, Fig. 3 shows that C-content ranged from
in Zone I which is assumed to be most decisive for 0.09 to 0.42 %, and Al-content was between 0.003 %
cracking. This figure does not include the cooling and 0.016 %. The Cu-content went from 0.08 up
in the top zone (Zone IA) which is indirect only, by to 0.48 %, with the corresponding levels of Sn and
means of 300 mm long cooling plates of Cu (i.e., Ni as illustrated in Fig. 4 (the Sb-analysis was not
available). Since intergranular cracking can be
Table 1. Main caster data.
enhanced also by the precipitation of (Fe, Mn)-
oxysulfides,l4~ the Mn/S-ratio is evaluated in both
figures, too : they range from 17 to 116, with higher

Fig, 2. Sketch of mold and top zone coolin g plates (" MS-
mold ").

Table 3. Range of test con ditions.

Table 2. Section sizes, casting speed and spray cool-

ing intensity in Zone IB.

Research Article
 Transactions ISIJ, Vol. 24, 1984 (353)

values for the medium carbon steels around 0.2 % C was performed for all four faces of every slice inde-
(Fig. 3). There is also a tendency to lower values pendently. Furthermore, also the defect location,
with increasing Cu-content (Fig. 4) but the scatter i.e., corner, off-corner and mid-face was monitored.
too wide for a meaningful correlation. Despite the substantial amount of defect readings
For the evaluation of crack severity, a sample (JV.-400) a proper multiple regression analysis could
slice was cut from each bloom (bloom length about not be performed in view of the many variables in-
5 m) and etched in 20 % hydrochloric acid in order volved. Thus, the evaluation had to be based rather
to remove scale and reveal eventual intergranular on mean value comparison which yields not more
cracking. The severity was rated based on standard than a " trend analysis " and semi-quantitative cor-
samples with the grades from 0 to 4 (O=no cracks, relations at the most.
4=strong network of cracks up to 2 mm deep). This
III. Test Results
In the first test casts, efforts concentrated on the
optimization of secondary cooling. Then, further
attention was focused on the mold area. In both
respects, alloy effects were given due consideration.

1. Effects of SecondaryCooling
In the first tests, direct spray cooling was applied
in Zone IA instead of the cooling plates in order to
check whether copper pick-up from this support
system would be a factor. Thus, only 0.1 l/kg was
applied in the top zone to match the soft intensity of
the indirect plate cooling, and 0.3 I/kg for the total
secondary cooling. As indicated by the results in
Fig. 5, the average crack indices of 1.30 and 1.33 for
plates and sprays respectively are virtually identical.
During these tests, differences in the preferred
defect position could be observed. Table 4 shows
the relative frequency of defect location, i.e., how
frequently defects were observed at corner, off corner
and mid-face location for the slices checked. As
can be seen, the mid-face location is distinctly less
plagued by defects in case of the cooling plates, thus,
copper pick-up might be ruled out. Rather the
corner and off-corner regions are the preferred crack
Fig. 3. Range of C- and total Al-contents as well as Mn/S locations for both cases, in case of direct cooling
ratio for the test heats.

Fig. 5. Crack index res.mode of top zone cooling.

Table 4. Average frequency (%) of defect location as

function of secondary cooling mode in Zone
IA (NN400).

Fig. 4. Range of Cu-, Sn- and Ni- contents as well as Mn/S

ratio for the test heats.

Research Article
( 354 ) Transactions ISIJ, Vol. 24, 1984

equally the mid-face position. The reasons are not ening, although the effect does not appear to be very
immediately evident but the coincidence of cracking significant.
with surface depressions was clear in many instances. From these comparative trials, the beneficial effects
In further tests with direct cooling by spray nozzles, of cooling plates on strand containment became
total spray cooling intensity was varied within a clearly apparent (Table 6). Allowing up to 5 mm
relative wide range (Fig. 6) . Nevertheless, the defect side bulging-which is at least permissible for lower
index does not indicate any clear trend under the carbon steels to prevent interdendritic cracks-, the
test conditions applied : while the curve for powder critical casting speed is nearly 1.3 mlmin for the
C reflects the expected trend, the one for powder plates whereas for sprays the limit is 1.0 rn/min for
B would just give the inverse dependence. Thus, lower and 1.1 m/ min for higher cooling intensity.
other factors must be more important to defect forma- Of course, the best strand support is provided by a
tion. Nevertheless, spray cooling intensity in Zone rigid roller apron but, necessarily, also more expensive.
IB and II was maintained at a low level down to Consequently, the cooling plates have been main-
0.21/kg in total, as to realize a high strand surface tained on all further test conditions.
temperature up to 1 030 °C at straightening. This
also provided adequate safety against transverse 2. Mold Effects
crack formation. In this context, the crack index In earlier tests, deep oscillation marks have led
was evaluated independently for each face in order to a preferentially transverse course of cracking
to find out whether the straightening action would which was aggravated particularly in case of low
enhance the cracking incidence. As can be seen carbon contents, i.e., around 0.1 % C. Such cracking
from Table 5, the outer radial side (fixed side) shows largely disappeared when the oscillation frequency
the lowest average crack index which may be due was increased to the maximum of 120 cpm.
to the compressive surface stresses during straight- The main attention was devoted to the selection
of mold powders, and also to the uniformity in
(manual) distribution. Mold powder consumption
was monitored, too. Mold powder performance was
mainly judged by the appearance in general. In
this respect, powder A gave consistently the most
favorable results-at least up to a casting speed of
about 1.0 m/min which coincides with a minimum
permissible powder consumption in the order of
0.3 kg/m2 (Fig. 7). The data for powder A bring
about also quite a clear relationship between crack
index and Cu-content as shown in Fig. 8. Powder
A has a viscosity of 7.5 poise at 1 300 °C and the

Fig. 6. Crack index as function of total spray cooling for

two mold powders.

Table 5. Evaluation of crack index for each face* on

the first seven test heats (N=136).

Fig. 7. Mold powder consumption res. casting speed for

powder A.

Table 6. Effect of secondary cool ing mode in Zone IA

and casting speed on side bulg ing. (Mold
length 800 mm)

Fig. 8. Crack index as function of Cu-content for powder A.

 Transactions Is", Vol. 24, 1984 (355)

typical composition of 29 % CaO, 33 % Si02, 16 % This measure resulted in substantial improvement as

A1203, 6 % F and 5 % Na20. illustrated by Fig. 14.
Several control tests performed with oil instead of
slag lubrication gave quite low defect incidences, 3. Effects of Steel Composition
particularly the ones performed with a thick wall The effect of C-content is evaluated in Fig. 15,
tubular mold (Fig. 9). However, at the same time with no apparent influence. On the other hand, the
several deep transverse face depressions were ob- Al-content in Fig. 16 shows some increase in cracking
served which present a potential hazard to breakout incidence at higher levels.
safety. Thus, oil lubrication is not a practical solu- As indicated in Fig. 3 already, the higher Al-
tion for large blooms-even when applying a gas content applies more to steels with higher C-content
shroud for maintaining high steel cleanness. The which, therefore, should render the latter more crack
tests with reduced mold water velocity (Fig. 10) sensitive. However, this steel group is, on the other
gave no clear trend. Also the tests with increased hand, characterized by a lower Cu-content and also
thickness of the Cr-plating (Fig. 11) did not show a higher Mn/S-ratio which both may compensate
any significant improvement. the unfavorable effect of the increased Al-content,
Some rather adverse results in the latter test, even although the effect of Mn/S-ratio does not appear to
with low Cu-content of only 0.15 %, were found at- be very significant when comparing the crack indices
tributable to the rare occurrence of rhombic bloom for the second and third range of C-content in Fig.
deformation with 3.4 and 4.5 % (Fig. 12) whereas 15 under the largely identical conditions of Cu- and
the usual conditions being only 0.6 % in the average, Al-content as listed in Table 7. But, at least the
and 1.0 % for sprays respectively 0.9 % for plates inverse correlation of Cu- and Al-content can provide
as maximum values. On the rhombic blooms, the a valid explanation why no effect of C-content is
main defect occurrence was found indeed on the apparent in the present case.
obtuse corners which points to the important effect
of local cooling rate as function of strand/mold-contact Iv. Discussion
intensity. The latter context may also explain the Owing to the preferred crack location in depressed
beneficial effect of mold taper (Fig. 13) although any surface regions, this aspect will be discussed first.
change in bloom geometry was not obvious at the
standard conditions, already characterized by low 1. Shell Contractionand DepressionFormation
values of rhomboidity as well as side bulging (Table 6). Depression formation of the strand shell in the
The final tests of this investigation were devoted mold is a very complex phenomenon. Basically the
to optimized pouring stream configuration in re-
placing the straight tube by one with lateral orifices.

Fig. 9. Crack index as function of Cu-content Fig. 11. Crack index for different
for oil lubrication and two different Fig . 10. Crack index as function of mold thicknesses of mold wall

wall thicknesses of the tubular mold. water velocity for two mold powders. Cr-plating.

Fig. 12. Crack index and rhombic Fig. 14. Crack index res. pouring tube
deformation. Fig. 13. Crack index as, mold taper. geometry.
(356) Transactions Is", Vol. 24, 1984

Table 7. Average Cu- and total Al-contents as well as

Mnf S-ratio for three groups of C-content in
case of heats cast with powder A and cool-
ing plates at max. 1.0 m/min.

Fig. 15. Crack index as. C-content.

Table 8. Optimized casting conditions for 260x 260
mm to reduce intergranular crack occur-
rence.

Fig. 16. Crack index us. total Al-content.

strand shell contracts in more or less regular intervals

whenever the compounded effect of shell thickness
and shell strength has reached an extent where the
shell is able to detach from the mold wall against the 2. Structural Effects and Crack Mechanism
action of the ferrostatic pressure.l5-20) Consequently,
" soft " mold cooling with controlled heat flux is As a consequence of depression formation, the
necessary in particular for shrinkage-sensitive steel solidification structure undergoes local coarsening,
which leads to enhanced crack sensitivity as already
grades to avoid this very undesirable " buckling"
effect,21-23) Therefore, it is important to reduce outlined in previous work.24) Furthermore, a coarse
initial shell growth. For instance, in static casting solidification structure also increases grain size of
experiments a K-factor of about 12 mm/min°5 was the subsequent transformation structure (Fig. 17).
derived as a critical threshold.22) Assuming the order Upon enrichment of residuals during scaling, local
of 25 mm as typical buckling distance,19>the section concentration reaches critical levels earlier if grain
260 X 260 mm would indicate a shell thickness exceed- boundary extension is reduced by coarse grain diame-
ing 1.6 mm as being critical since depressions and ter. Thus, the necessary precondition for crack
cracking were observed even on the blooms cast at formation is-under given conditions of residual
1.4 m/ min. content and other casting parameters-a certain
In order to reduce the initial shell thickness in critical grain diameter.
favour of greater growth uniformity and avoidance Of course, the real defect formation is more com-
of depressions, the optimized casting conditions are plex since also the extent of scaling being of influence.
listed in Table 8. Regarding the comparatively en- For instance, scale formation is found to be about
hanced crack incidence on 260 X260 mm as com- three-times as high for a 0.1 %C-steel when com-
pared to the smaller sizes, the higher casting speed pared with a 0.7 % C-steel3) ; on the other hand, a
appears to be the most prominent factor in favor maximum of network cracking at 0.25 % C was
of the latter ones. reported in the same work. This was explained with
As cracking is most frequently observed in the the critical surface temperature of around 1 100 °C
corner region, this must be attributed to enhanced at mold exit for the latter steel group whereas lower
contraction at the large corner radius of 20 mm in C-steels (with close to 1200 °C) would give less re-
case of 260 x 260 mm whereas the smaller sections sidual concentration owing to more rapid diffusion
being cast with a chamfered corner. Thus, a change along grain boundaries. Diffusion rate is thus im-
to the latter configuration should be contemplated, portant for critical residual enrichment, and the
too, since its positive effect on shell growth uniformity critical temperature range generally found to be be-
has been empirically established from the experience tween 950 °C and 1 150 °C.2s)
of several other casters. From the above, one might conclude that a soft

Research Article
 Transactions ISIJ, Vol. 24, 1984 ( 357)

Fig. 17. Austenite grain size vs, primary dendrite spacing Fig. 18. Regions of high temperature embrittlement relat-
for 0.15 %C-steel.25> ed to structural effects, schematic and adapted
after. 14)

spray cooling intensity leading to high strand surface solidification structure and enhanced microsegregation
temperatures could be deleterious-as also shown by formed in depressed surface areas.
laboratory simulation.6> However, strong cooling
raises the thermal stress level and also may, at a given V. Conclusion
surface temperature, enhance total scaling by con- (1) The formation of fine intergranular surface
tinually removing newly formed scale. cracks in bloom casting is strongly affected by the
A further negative effect of strong secondary cool- residual content but, like in case of star cracks, not
ing, particularly observed in case of BOF-steels with a necessary precondition.
low residual content can be the additional embrittle- (2) Rather, the main precondition for crack
ment due to precipitation formed by microalloys formation appears to be a coarse microstructure lead-
(Al, Nb, V, B) in connection with reduced surface ing to critical concentrations of segregated solutes
temperature as well as strong thermal cycling. This and/or precipitates along austenite grain boundaries.
probably may lead to a combined effect of region II- (3) Consequently, in order to assure best surface
and III-embrittlement for network cracking-as illus- quality, the aim of a fine solidification structure is of
trated in Fig. 18, and with the following consequences highest priority, achievable by maximum uniformity
to the formation of surface cracks in general: in the initial shell growth and avoidance of depression
(1) Region I mostly concerns steels with strong formation. All other measures, in particular a soft
segregation tendency (high C-steels and fully aus- spray cooling intensity, are of secondary importance
tenitic stainless grades in the first place), and crack only.
formation of intergranular nature (rather than inter-
dendritic) since microsegregation being highest at Nomenclature
locations of mismatch in dendrite orientation, al- Ir: relative spray cooling intensity (10-3)
though this effect can be blurred by subsequent Is : specific spray water consumption (i/kg)
grain boundary migration away from its original L : length of spray cooling (m)
position during cooling.27,28) .P: width of spray cooling (m)
(2) Region II is either caused by Cu/Sn/Sb- Q: mold powder consumption (kg/m2)
enrichment or (Fe, Mn) S-0 precipitated along grain Ve: casting speed (m/ min)
boundaries whereby a critical minimum grain diame- VM: mold water velocity (m/s)
ter is required, usually resulting as a consequence of W: spray water flow rate (m3/min)
coarse primary structure.
Acknowledgements
(3) RegionIII results from precipitations of micro-
alloys (A1N, Nb(C, N), BN) enhanced by proeutec- The cooperation with the operating team at
toid ferrite formation where solubility of precipitates STEFANA/Ospitaletto Works, headed by its tech-
is distinctively lower in ferrite, and also by coarse nical director Mr. 0. Bonetti, as well as the support
austenite grain size. of the experimental work by the personnel of CON-
Consequently, longitudinal cracks may form either CAST ITALIANA/Milan are gratefully acknowl-
in region I or II, depending on steel type which edged.
deviates from the generally assumed preference for Particular thanks are due to Dr. H. G. Suzuki of
region I,12,29) Star and network cracks are pre- Nippon Steel Corporation/Oita Works for helpful
dominantly forming in region II, and transverse discussions on the mechanisms of surface cracking.
cracks in region III but mutual overlapping into
regions I I and III appears to be also possible. How- REFERENCES
ever, it is not excluded that the very initiation of 1) R. Rosegger: Radex Rdsch., (1964), No. 5, 277.
microcracks would coincide with region I for all 2) K. G. Speith and A. Bungeroth: Stahl u. Eisen, 84 (1964),
types of surface cracks--in connection with coarse 1297.

Research Article
( 358 ) Transactions ISIJ, Vol. 24, 1984

3) F. Schmid and 0. Gross: Steel International, (1967), No. 1, 16) A. Grill and J. K. Brimacombe : Ironntaking Steelmaking,
11. (1976), No. 2, 76.
4) D. A. Melford : JISI, 200 (1962), 290. 17) H. Fredriksson and M. Thegerstrom: Scand. J. Met., 8
5) A. D. Nicholson and J. D. Murray: JISI, 203 (1965), (1979), 232.
1007. 18) Y. Sugitani and M. Nakamura : Tetsu-to-Hagane, 65 (1979),
6) M. H. Burden, G. D. Funnell, A. G. Whitaker and J. M. 1702.
Young: Proc. Int'l Conf. Casting Solidif. Metals, The 19) M. Wolf and W. Kurz : Metall. Trans., 12B (1981), 85.
Metals Society, London, (1979), 279. 20) M. Matsumiya, T. Saeki, J. Tanaka and T. Ariyoshi :
7) Y. Iida and H. Ooi: SteelmakingProc. AIMS, 60 (1978), 60. Tetsu-to-Hagane, 68 (1982), 1782.
8) H. J. Chang, N. H. Lu and M. C. Tai: SEASI Quart., 21) M. Wolf: Trans. ISIJ, 20 (1980), 710.
(1983), No. 1, 18. 22) Y. Sugitani, M. Nakamura and T. Watanabe : Tetsu-to-
9) H. Yasunaka, T. Fujimoto, T. Mori, K. Narita, Y. Suzuki, Hagane, 67 (1981), 1508.
I. Wakasugi, Y. Katagiri and W. Takagi : Tetsu-to-Hagane, 23) M. Kinoshita, H. Kitaoka, T. Tomura, T. Emi, T. Koshi-
68 (1982), A165. kawa, K. Sorimachi and K. Kusano : Tetsu-to-Hagane, 68
10) H. Maeda, T. Sugawara, M. Noguchi, I. Kudo, Y. Tani- (1982), A153.
fuj i and T. Yatsuzuka : Tetsu-to-Hagane, 67 (1981), 5902. 24) M. Wolf and S. Kohl : Tetsu-to-Hagane, 66 (1980), 5756.
11) K. Yoshida, T. Nagahata, T. Tomono, K. Kimura and S. 25) Y. Nun, T. Ohashi, T. Hiromoto and 0. Kitamura :
Tsujita : Tetsu-to-Hagane, 67 (1981), 1317. Trans. ISIJ, 22 (1982), 399.
12) T. Kohno, T. Shima, T. Kuwabara, S. Mizoguchi, T. Ya- 26) G. Schonewolf, G. Blohm, H. J. Grubler and S. Richter:
mamoto, H. Misumi and A. Tsuneoka : Tetsu-to-Hagane, JVeueHi tte, 24 (1979), 462.
68 (1982), 1764. 27) F. Matsuda, H. Nakagawa, S. Ogata and S. Katayama:
13) P. Colombo, F. Cristi, B. Rossa and R. Viani: World Steel Trans. I WRI, 7 (1978), 59.
& Metalworking, 4 (1982), 146. 28) B. Rogberg : Comm. by SANDVIK STEEL, Sandviken,
14) H. G. Suzuki, S. Nishimura and S. Yamaguchi: Trans. (1982).
151J, 22 (1982), 48. 29) J. K. Brimacombe and K. Sorimachi : Metall. Trans., 8B
15) R. E. Aseev, V. N. Zhuchin and V. M. Kondrashin: Steel (1977), 489.
in the USSR, 2 (1972), 284.

Research Article