ISIJ International
Online ISSN : 1347-5460
Print ISSN : 0915-1559
ISSN-L : 0915-1559
Regular Article
Ripple Marks on Cast Steel Surfaces
Hatto JacobiKlaus Schwerdtfeger
Author information

2013 Volume 53 Issue 7 Pages 1180-1186


Surface features developing on steel in static ingot casting were investigated using iron melts of 25 kg with 0.09% carbon. The main type of feature was the normal ripple marks which form directly at the meniscus line by the overflow of the melt over the tip of the solid shell. Their spacing and depth (groove depth) was measured, and both were found to decrease with increasing superheat and increasing casting speed (rate of meniscus rise). From the comparison with other spacing data published in the literature, it seems that ripple formation is influenced at least by two main factors. One is heat flow and the other is wave motion. The latter can be understood quantitatively. A different type of surface marks forms at strong cooling of the meniscus when freezing occurs already at the flat surface of the meniscus. This happened in casting in a helium gas phase. Their spacing and depth also decreases with increasing superheat. But they are wider and deeper, respectively, than those of the normal ripple marks developing in an argon gas phase. A third feature are transverse depressions which develop at a later stage, after solidification. Ripple mark formation may play a role also in continuous casting. According to a recent investigation by Tacke1) horizontal surface marks on thick slabs have non-normally wide spacings, compared to regular oscillation marks, at low casting speed. Tacke’s data were discussed with respect to their possible relationship to ripple marks.

1. Introduction

Ripple marks, also called surface marks, wrinkles, striations, grooves or other, are defects developing on the surface of static ingots during casting. They run, with certain spacing and depth, parallel to the meniscus surface. An example is shown in Fig. 1. Ripple marks have been known for a long time from ingot casting and were investigated extensively in the early eighties of the last century and before. Then, engagement with them had become smaller. But recently interest has been revitalized1,2) because it is believed that they play a role also in continuous casting.

Fig. 1.

Example for ripple formation on ingot surface. This work. Experiment 74. Iron with 0.09% carbon. Casting speed 0.70 cm s–1, superheat 55 K, furnace atmosphere 50 Torr argon. Another defect, denoted as transverse depressions, is present on this ingot in the form of comparatively broad indents with large spacing.

The experiments of this study were carried out more than thirty years ago when the authors were associated with each other at the Max Planck Institut fuer Eisenforschung in Duesseldorf. They were not published at that time because both authors changed their affiliation, and for other reasons. One reason was that with the advent of continuous casting the interest in static ingot casting had become smaller. Another was that a good theoretical interpretation of the observations could not be produced at that time. The latter situation still persists. But, on the other hand, the demand for satisfactory cast surface quality keeps increasing, and, consequently, also the need for a better general understanding of the various mechanisms making surface defects. Since our data seem to be a good supplement to the existing knowledge on surface mark formation, and in view of the recent interest in such marks, we decided to publish them.

According to the reported experimental investigations3,4,5,6) there are several types of non-smoothness features at the surface of static ingots which form for different reasons. The main types are the following:

Ripple marks: These marks look as shown in Fig. 1. They have also been called type I marks5) or primary marks.6) Ripple marks form directly at the meniscus line. The tip of the solidifying shell grows into the curvature at the edge of the meniscus and stiffens the contour which, therefore, does not bend back to the mold during further rise of the meniscus. The next liquid metal, arriving at the meniscus from below, flows over the original meniscus and causes the so-called overflow. The phenomenon repeats itself. Each overflow makes a ripple.

Crust marks: This type is a perverted version of the normal ripple marks. It develops, when at low casting speed, low superheat and particularly strong surface cooling the solidification occurs already at the horizontal surface area of the meniscus. The next melt coming from below breaks through the skin (or crust) floating on the meniscus, makes an overflow and pushes solid skin material to the mold wall. These marks are also horizontal, but have wider spacing than the normal ripple marks. It is suggested to call them horizontal crust marks. In the literature, they have been denoted as type III features or laps.5) German foundry engineers called them cold run marks (German: Kaltlaufmarken).

Transverse depressions: These horizontal indentations are also seen on the ingot surface shown in Fig. 1. They are much coarser than the ripples and crust marks. Such depressions do not form at the meniscus but at some distance below the meniscus line, by distortion of the shell which is brought about by mechanical stresses developing in the growing shell. In the literature they have been called type II5) marks.

All these three features were observed in the present investigation, but emphasis of the investigation was on ripple marks.

There are other kinds of horizontal surface striations. Fine horizontal lines, called secondary marks, have been observed, on low melting alloys, between the primary marks.6) The mechanism of their formation is unclear.

2. Experimental Method

The experiments were carried out in a vacuum furnace using 25 kg iron melts (high purity grade) with addition of 0.09% carbon. After preparation of the melt the casting was performed, in the furnace vessel, under a controlled gas phase which in most cases was argon at 50 Torr (50 mm Hg, 6666 Pa) and in some cases helium at 50 Torr. The mold for receiving the melt consisted of a water cooled chill plate at one wide side which was made of copper and had a smooth metallic surface, and a ceramic structure at the other sides as well as at the bottom. The latter contained the channel for the feeding of the melt. A schematic sketch of the arrangement is given in Fig. 2. The internal horizontal cross section of the mold was 103 × 115 mm2. The width of the copper plate was 190 mm, the height 350 mm. The melt was added into the feeding channel via a small tundish placed above the upper opening of the feeding channel, then entered into the interior mold space through a hole at the center of the bottom plate, and it then rose upwards with the casting velocity vc which was controlled by the diameter of the orifice in the tundish. There was a thermocouple inserted into the mold space directly above the opening hole in the bottom for measuring the temperature of the melt when it enters the mold. The superheat of the melt was deduced as ΔT (=Tmelt at mold entrance –Tliquidus).

Fig. 2.

Sketch of setup used in present study of mark formation at the surface of a solidifying static steel ingot.

Spacings of marks at the ingot surface were determined by measuring the distance, along the height direction, containing a certain number of marks (5 to 10) and division by the number. This was performed all along the height of the ingot except at its upper end where the spacing usually was smaller. Then an average spacing value was computed. The depth of the grooves between the marks was determined with an electronic profile measuring device, called “Conturograph” (Feinpruef GmbH Goettingen). This had a needle which is moved over the ingot surface thereby sensing the surface profile. The signal is amplified, stored and then plotted on paper. The groove depth is defined as the average height difference between maxima and minima on the surface profile. It was determined by drawing tangents over neighboring maxima and measuring the depth between tangent and minimum. Measurements were taken at several locations of the ingot. Then the average value was computed.

3. Results on Ripple Marks

Both ripple spacing and depth depend on superheat and casting speed. The obtained data are listed in Table 1. In Fig. 3 ripple spacing is plotted against superheat, for three groups of casting speed range, and in Fig. 4 against casting speed, for groups of superheat. Both, increasing superheat and increasing casting speed, cause a decrease of ripple spacing. For the application of the data it is useful to fit them with a formula. There is indication from the data points that the dependence on superheat is approximately exponential with approach to some constant value A at high superheat. That is, a function of the type λripple = A + B exp(–CΔT) was taken with B depending on casting speed. The coefficients A and C were determined by trial, finding A=0.42 mm and C=0.0054 K–1. Figure 3(b) shows the data of Fig. 3(a), now plotted as (λripple – 0.42) against superheat ΔT. Note, that the ordinate is in log scale. It is evident that the points can be represented satisfactorily by straight lines proving that the assumed form of function serves adequately. The dependence of B on casting speed was also taken to be exponential, that is B = D exp(–Evc), and E was determined as E=0.54 s cm–1, see Fig. 4(b). The resulting final expression for representing the present data comes out to be   

λ ripple =0.42+2.85exp(-0.0054ΔT-0.54 v c ) (1)
where λripple is in mm, ΔT in K and casting speed (rate of rise of meniscus) vc in cm s–1. Figure 5 gives a comparison of the spacings, as computed with this formula, with the measured values.
Table 1. Data on ripple spacing and depth (type I ripples, primary ripples) as obtained in present work. 25 kg ingots of iron with 0.09% carbon. Melting was carried out in a vacuum furnace and subsequent casting in the vacuum vessel under 50 Torr argon.
No. of experimentcasting speed vc , cm s–1superheat ΔT, Kripple spacing, mmgroove depth, μmdepression spacing, cm
700.5922.3630not measured
951.65371.53not measured3.38
Fig. 3.

Dependence of ripple spacing on superheat and casting speed. Iron with 0.09% carbon. a) Plot of data versus superheat. b) Plot of same data in half-logarithmic scale indicating that a formula of the type λ = A + B exp(–CΔT) can be used to represent the data.

Fig. 4.

Dependence of ripple spacing on superheat and casting speed. Iron with 0.09% carbon. a) Plot of data versus casting speed. b) Plot of same data in half-logarithmic scale proving that also the dependence on casting speed can be taken to be exponential.

Fig. 5.

Diagram showing the quality of Eq. (1) for representing the measured values of ripple spacing.

The data on groove depth (ripple depth) are illustrated in Fig. 6. Also groove depth decreases with increasing superheat and increasing casting speed. The change with casting speed is non-uniform, first being strong (between 0.3 to 0.4 cm s–1) and then small. A simple formula cannot represent this dependence.

Fig. 6.

Dependence of groove depth on superheat and casting speed. Iron with 0.09% carbon.

4. Discussion of Results on Ripple Marks

It is evident from the given data that, according to our measurements, ripple (type I mark, primary mark) spacing clearly decreases with increasing superheat and increasing casting speed (rate of rise of meniscus). These effects are due to the influence of casting speed, superheat and cooling on the formation of the solid skin at the edge of the meniscus. At high casting speed and high superheat the tip of the skin may be located in the lower part or below the curved edge of the meniscus. With decreasing casting speed and decreasing superheat the tip moves upwards and into the curvature, thus enhancing overflow.

Previous research on ripple formation on standing ingots: The previous data on ripple spacing, obtained for different metals and alloys, Table 2, only partially agree with each other and with the present data. The effect of superheat measured in the present work is in the same direction but much stronger as reported by Wray.5) Flemings et al.6) state that there is no significant effect of superheat. But their data points indicate a tendency for decrease of spacing with increasing superheat. According to the works of Thornton3) and of Wray5) ripple spacing decreases with increasing casting speed, similarly as observed in the present investigation. But according to Flemings et al.6) there is an increase with increasing casting speed. In Fig. 7, the data on ripple spacing (type I marks, primary marks) obtained in the four investigations are shown together for the comparison.

Table 2. Effect of superheat and casting speed (speed of meniscus rise) on ripple mark spacing as observed in present and preceding studies.
studymaterials of cast metal and (of mold)ingot cross section (chill side × other side, cm2)effect of superheat on ripple mark formationeffect of casting speed on ripple mark formation
Saucedo, Beech, Davis4)carbon steels, stainless steel, aluminum (mild steel or water cooled copper)7.5 × 7.5“in general, the severity of rippling was greatly reduced by increasing superheat.”“it was found that the lower the casting speed, the higher the degree of rippling.”
Thornton3)lead (cast iron)5 × 5decrease of casting speed and decrease of teeming temperature enhance ripple formation
Thornton3)steel, 0.4 to 0.6%C (cast iron)10 × 7not reporteddecrease of spacing with increasing casting speed
Wray5)lead (copper)7.6 × 2.2decrease of spacing with increasing superheatdecrease of spacing with increasing casting speed
Stemple, Zulueta, Flemings6)tin-lead alloys (copper)2.0 × 3.6“no significant effect of superheat on spacing was found.”increase of spacing with increasing casting speed
present worksteel, 0.09% carbon (copper)10.3 × 11.5decrease of spacing with increasing superheatdecrease of spacing with increasing casting speed
Fig. 7.

Mark spacing plotted against casting speed. Comparison between the data measured in present work, and measured by Thornton,3) by Wray,5) and by Stemple, Zulueta, Flemings.6) Materials and size of ingot used in the different studies are listed in Table 2.

Flemings et al.6) observed the meniscus motions during filling of the mold which they termed “waves”. They believe that surface wave formation was the major cause for formation of their primary ripples. This matter can be elaborated by looking into wave theory.7) Usually, waves forming at the surface of liquids are gravity induced. It can be shown that the frequency f of the simple (one-nodal) standing wave in a rectangular vessel is given by the equation   

f= 1 2π πg W =8.835 1 W (2)
where g is the gravity constant (981 cm s–2), W the length of the broad side of the vessel in cm, and f in s–1 (Hz). The distance between two crests (or two troughs) at the wall of an oscillating meniscus, rising in the vessel with velocity vc, is vc/f, and this distance is the mark spacing on the narrow side of the ingot if marks are made by this wave motion. Hence,   
λ wave =1.132 v c W (3)
with λ in mm, vc in cm s–1 and W in cm. In the experiments of Flemings et al.6) the cross section of the mold cavity was 3.6 × 2.0 cm2. Hence, inserting W=3.6 cm in Eq. (3) yields   
λ wave =2.15 v c . (4)

The curve according to this expression is given in Fig. 7. It is seen that the curve is sufficiently close to the experimental data points to conclude that in Flemings’ experiments the ripples were indeed made by gravity waves.

The function given by Eq. (1) which represents the present measurements, is also drawn in Fig. 7 for two selected values of superheat, ΔT =20 and 84 K. The data of Thornton3) and of Wray5) are rather near to these curves.

In summarizing, it is confirmed that ripple formation on the surface of standing ingots is rather complex. There can be several influencing factors, and different mechanisms can operate. In one of them, control of spacing is by heat flow. Superheat has an effect and also the substrate is of influence.3,4) In another evolution, control of mark spacing is by fluid flow (wave motion), and this phenomenon seems to be understood quantitatively, see Eq. (3). In neither of these mechanisms interfacial tensions (surface and/or melt/solid) appear to play a role. But this may be the case in still another mechanism, proposed by Tomono, Kurz and Heinemann8) in which the spacing is taken to be controlled by the profile of the meniscus. In their experiments and also in experiments conducted by ourselves and reported by one of us9) ripple spacing was much smaller if casting occurred in an oxidizing gas phase, e.g. air, and this was attributed to the decrease of surface tension of the cast steel by dissolved oxygen.

It is not known at present under which circumstances one or the other of these influences mainly acts and determines the ripple spacing.

5. Formation of Transverse Depressions on the Surface of the Ingots Cast under Argon

The data on the depression spacing are included in Table 1, and Figs. 8 and 9 show plots versus superheat and versus casting speed. Depression spacing is much larger than ripple spacing which is seen qualitatively already on the photo given in Fig. 1. There is only small influence of superheat and casting speed. The ripple marks run rather undisturbed through the depression indentations, see Figs. 1 and 10, which proves that the ripples were present on the surface when the depression formed. It has already been stated in the introduction that these depressions develop due to thermal stresses in the solidifying shell.

Fig. 8.

Plot of transverse depression spacing versus superheat. Iron with 0.09%C.

Fig. 9.

Plot of transverse depression spacing versus casting speed. Iron with 0.09%C.

6. Some Observations on Castings Carried out under Helium. Mark Formation at High Meniscus Cooling

A few experiments were carried out using a helium gas phase. The intention was to explore the effect of an increase of heat transfer to the mold utilizing the higher heat conductivity of helium compared to that of argon. Of course, also the cooling at the free meniscus is faster in helium. Results on ripple spacing and depth are given in Table 3. Figure 10 shows ingot surfaces obtained at two values of superheat. The horizontal striations are similar in principle to the normal ripple marks considered in the preceding part of the paper. Their spacing decreases also with increasing superheat. But they are deeper and wider at low superheat than in argon gas. Figure 11 explains what is assumed to be the difference between the normal ripple formation (formed in argon and treated in Sections 3 and 4 of the paper) and that in the helium gas phase. The coarser surface structure of the ingots made under helium is believed to be caused by freezing at the flat surface of the meniscus, due to the strong cooling capacity of helium, see Fig. 11(b), whereas the normal ripple formation occurs by freezing at the curved edge, Fig. 11(a). It appears that bending of the crust occurs with lower frequency and at a larger distance from the mold wall. Therefore, spacing and depth of the marks become larger than under argon. But with increasing superheat the crust becomes thinner, and conditions approach to those under argon. Consequently, also spacing and depth of the marks approach to the values in argon.

Table 3. Data on mark spacing and depth of marks (type III features) as obtained in present work after melting and subsequent casting under 50 Torr helium. 25 kg ingots of iron with 0.09% carbon. These marks have been called crust marks in the present work.
No. of experimentcasting speed vc , cm s–1superheat ΔT, Kcrust mark spacing, mmgroove depth, μm
(average 0.56)
Fig. 10.

Surface marks formed during casting under helium gas. Iron with 0.09% carbon. Casting speed about 0.60 cm s–1. a) Experiment No. 96 with low superheat. b) Experiment No. 98 with high superheat.

Fig. 11.

Formation of surface marks in casting of standing ingots by the overflow mechanism. Edge of meniscus at two consecutive moments t1 and t2 during filling of the mold. a) Ripple mark formation in argon gas phase. Shell tip ends in curved part of meniscus. Horizontal part of meniscus is liquid. b) Crust mark formation if the horizontal part of the meniscus freezes and becomes covered by a solid crust. Schematic.

So, we made the crust marks (type III marks5)) in these castings. Transverse depressions were formed also in these experiments, see Fig. 10(b). The pitch and depth values of the marks produced in the helium gas are plotted against superheat in Fig. 12 and are compared with the data found for the normal ripple marks.

Fig. 12.

Spacing and depth of marks that developed in a helium gas phase, and comparison with the values of the normal ripple marks. Iron with 0.09% carbon.

7. Remarks Concerning Spacing of Surface Marks on Continuously Cast Slabs

The spacing λom of normal oscillation marks formed on continuously cast strands has been found in many studies to be according to   

λ om = v c f (5)
where vc is casting speed (withdrawal speed of strand) and f the oscillation frequency of the mold. Equation (5) means that one mark is made in each oscillation cycle. But spacings of horizontal surface marks, considered to be oscillation marks, do not always have this theoretical value. Recently, Tacke1) carried out measurements on the narrow sides of thick slabs cast in Dillingen. He found that spacings, under his conditions, were considerably larger than predicted by Eq. (5). His plot of “ratio measured to theoretical spacing” versus casting speed is given in Fig. 13 indicating that real values can be higher than the theoretical ones, by a factor of up to almost four. Tacke mentioned ripple mark formation similar as on static ingots, and marks made by surface waves as possible explanations for the occurrence of his widely spaced marks. His results may be discussed in the light of the findings of the present paper.
Fig. 13.

Spacing of marks on narrow face of continuously cast slabs expressed in terms of “ratio measured to theoretical spacing” plotted against casting speed, according to measurements of Tacke.1) The theoretical spacing λom is that made by the mold oscillation, Eq. (5). The ratios λripple / λom and λwave / λom were computed with Eqs. (6a) and (7a). The oscillation frequency for Tacke’s data is from 62 to 87 cpm.

The ratios λripple / λom and λwave / λom are deduced from Eqs. (1) and (5), or (3) and (5), respectively, to be according to   

λ ripple λ om = [ 0.42+2.85exp(-0.0054ΔT-0.90 v c ) ]f 1   000 v c (6)
λ wave λ om =0.00189 W f. (7)

Casting speed vc is in m min–1 in Eq. (6) as this is customarily so in continuous casting, therefore the factor 0.90 before vc instead of 0.54 in Eq. (1). Mold width W is in cm and oscillation frequency f in cpm. Both ratios are proportional to f. The ratio λripple / λom depends on casting speed, decreasing with increasing casting speed. A superheat value of ΔT=10 K (typical for continuous casting of steel) is used in Eq. (6). The width of Tacke’s mold was W=2200 mm=220 cm, and oscillation frequencies were between 40 and 125 cpm (between 62 and 87 cpm in Fig. 13). Hence, for Tacke’s conditions the ratios become   

λ ripple λ om = [ 0.42+2.85exp(-0.054-0.90 v c ) ]f 1   000 v c (6a)
λ wave λ om =0.028f. (7a)

These ratios are indicated in Tacke’s diagram, Fig. 13. The curve for λripple / λom is always below the experimental points. Over most of the considered vc range it is below one. So, at first glance, it appears that ripple marks cannot have developed on continuously cast strands according to the (heat flow) mechanism operating during casting of static ingots. But this conclusion may not be correct, because the applicability of these data to continuous casting conditions is not sure. In continuous slab casting solidification of the steel at the meniscus is against a slag layer rather than against naked copper (as in the present experiments), and the substrate influences mark spacings.3,4) In the experiments conducted by Thornton3) the use of dressings (organic materials) in static mold casting increased ripple spacing in comparison to the case when no dressing was applied. According to Saucedo et al.4) low thermal conductivity mold dressings reduce and high thermal conductivity dressings enhance ripple formation. So, the presence of the casting slag substrate in continuous casting may possibly increase ripple spacing above the curve for λripple / λom in Fig. 13 and shift the curve closer to (or to the inside of) the scatter region field of the experimental points.

The wave mechanism has been discussed broadly by Tacke. Although the ratio λwave / λom is in the range of the measured values, see Fig. 13, there are other indications, according to Tacke, that waves are not involved in making these marks (absence of preferred presence of wave frequency in spectral analysis of the spacing range and in level fluctuations measured with the NKK sensor).

Thus, it remains unclear whether the wide mark spacings on the continuously cast slabs are heat flow ripples or wave marks, and there may be another way in which these marks originate. Possibly, two or more mechanisms are operating simultaneously and influence each other in a complex and unknown manner.

8. Conclusion

The present investigation which was carried out on iron ingots of 25 kg with 0.9% carbon, and other data reported in the literature indicate that ripple formation on the ingot surface in static casting may be caused mainly by a heat flow mechanism or by fluid flow (wave motion of the meniscus). But it is not known under which conditions the first or the second influence controls ripple spacing and depth. There is interest, of course, whether ripple marks may form also in continuous casting. In a recent investigation1) non-normal surface marks with wider spacing were observed in thick slab casting, and it was discussed whether these marks could be ripple marks. The question was considered, in some detail, in the present paper. But the matter is complex and could not be resolved, and it remains unclear whether wide marks on continuously cast slabs are heat flow ripples, wave marks, or some kind of mark made in still another way.


The authors wish to thank Professor Tacke for his interest in this work and for discussion, and for providing Fig. 13 for the frequency range 62 to 87 cpm (selected data from his Fig. 11 in ref.1)).

  • 1)   K.-H.  Tacke: CCC08, 4th Int. Conf. Continuous Casting of Steel in Developing Countries, CSM, Beijing, (2008), 146.
  • 2)   K. C.  Mills,  P. E.  Ramirez-Lopez and  P. D.  Lee: Sano Symp. Proc., The University of Tokyo. International Research Center for Sustainable Materials, Institute of Industrial Science, The University of Tokyo, (2008), 158.
  • 3)   D. R.  Thornton: J. Iron Steel Inst. 183 (1956), 300.
  • 4)   I.  Saucedo,  J.  Beech and  G. J.  Davies: Proc. 6th Int. Vacuum Metallurgy Conf., ‘Special Melting and Metallurgical coatings’, Am. Vacuum Soc., New York, (1979), 885.
  • 5)   P. R.  Wray: Metall. Trans. B, 12B (1981), 167.
  • 6)   D. K.  Stemple,  E. N.  Zulueta and  M. C.  Flemings: Metall. Trans. B, 13B (1982), 503.
  • 7)   L. D.  Landau and  E. M.  Lifschitz: Lehrbuch der Theoretischen Physik VI, Hydrodynamik, Akademie-Verlag, Berlin, (1974).
  • 8)   H.  Tomono,  W.  Kurz and  W.  Heinemann: Metall. Trans. B, 12B (1981), 409.
  • 9)   H.  Jacobi: Stahl Eisen, 96 (1976), No. 20, 964.
© 2013 by The Iron and Steel Institute of Japan