Engineering Failure Analysis 7 (2000) 55±67

www.elsevier.com/locate/engfailanal

Life of rolls in a cold rolling mill in a steel plantÐ

operation versus manufacture
A.K. Ray a,*, K.K. Mishra c, G. Das b, P.N. Chaudhary c
a
MTE Division, National Metallurgical Laboratory, Jamshedpur 831007, India
b
MTC Division, National Metallurgical Laboratory, Jamshedpur 831007, India
c
IRM Division, National Metallurgical Laboratory, Jamshedpur 831007, India
Received 11 December 1998; accepted 19 January 1999

Abstract

Failures of rolls occur due to improper manufacturing and operational parameters. Prematurely failed (spalled)
roll samples collected from a reputed steel plant were examined for their chemistry, inclusion content,
microstructures, carbide characteristics, hardness and retained austenite content. The residual stresses were also
measured on the inner and outer surfaces of the spalled roll pieces. The higher content of retained austenite was
primarily responsible for the spalling of indigenous rolls for which subzero treatment has been recommended.
Several suggestions have also been made for smooth operation of the mill and consequently for the life extension of
work rolls. # 1999 Elsevier Science Ltd. All rights reserved.

Keywords: Fatigue; Life extension; Retained austenite; Rolling mill; Spalling

1. Introduction

The work rolls used in cold rolling mills are expected to perform under very arduous operating
conditions. With the advent of faster mills and the increase in load, the quality of rolls needs continuous
improvement. Cold mill rolls should withstand mechanical and thermal shock, must have excellent wear
resistance and very little plastic deformation.
The premature failure of a cold mill roll increases not only the cost of the rolls but also the down
time of the mill, a€ecting productivity. The reasons for premature failure of forged and induction
hardened alloy steel work rolls can be attributed to the combined e€ects of mechanical and

* Corresponding author. Tel.: +91-657-426091; fax: +91-657-426527.

E-mail address: [email protected] (A.K. Ray).

1350-6307/00/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 5 0 - 6 3 0 7 ( 9 9 ) 0 0 0 0 4 - 7
56 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67

metallurgical factors. Mechanical factors include misalignment, uneven roll surface, bearing seizure and
insucient stock removal during grinding. Metallurgists have considered failures arising out of localised
overloading, temperature gradients due to insucient cooling, the presence of non-metallic inclusions,
casting defects and phase transformations [1].
The present study [2] was taken up to examine the manufacturing and operational parameters
responsible for the poor life of a cold mill work roll used in a ®ve stand 4 Hi tandem mill in an Indian
steel plant. It was observed that the imported rolls were giving an average life of 35,000 tons whereas
indigenous rolls were failing after rolling less than 15,000 tons on an average. The aim of this work was
to analyse the reasons of failure from a manufacturing as well as an operational point of view and to
provide suitable recommendations for extending the life of the rolls.

2. Background

It was observed that the three contributing factors responsible for the poor life of a work roll were as
follows:

1. Spalling.
2. Cracking.
3. Metal pick up and wrapping.
Failures of types `a' and `b' are partly manufacturing and partly operational, whereas type `c' is mostly
operational. Amongst these failures, spalling is considered to be the major cause of roll failure in a cold
rolling mill. A work roll in service is subjected to a set of stresses. First of all, there are internal stresses
resulting from the heat treatment procedure. On top of these internal stresses are those caused by
operation. Spalling is mainly classi®ed into two types [3]:

1. Spalling due to fatigue fracture of which the origin is at the roll barrel surface or just below the
surface.
2. Spalling due to fatigue fracture of which the origin is inside the roll body.
Failure of type (a) is related to problems of operational origin whereas type (b) is mostly of
manufacturing origin. Factors [4] responsible for spalling are given below.

2.1. Operational factors

1. Overloading
2. Insucient stock removal
3. Thermal stresses due to insucient roll cooling
4. Severe bruising of rolls
5. Mill chatter, improper crowning and mechanical alignment problems
6. Sharp temperature gradient on the surface of the rolls (e.g. near strip edge)
7. Pickling quality of hot rolled coil
8. Wrong grinding operation
9. Hydrogen pick up from soluble oil used as roll coolant.
 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67
Table 1
Detailed speci®cation of investigated roll

Roll size 585  1420 mm B.L.

Scrap size 555  1420 mm B.L.
Surface hardness 95±97 Sh D (Amr)
Depth of hardened layer 15±20 mm
No. of stands 5
Types of stands All 4 Hi
Average no. of dressings on a roll before 45±46
scrapped
Bearings Morogoil bearing on backup rolls
Drive All rolls are driven by 7500 HP DC motors
Cooling water Quantity and pressure 18001/mm, 6±12 kg/cm2 with a pH of 7±8
Lubricant with cooling water Palm oil system A±8 to 12% system B±15 to 20%
Maximum speed 1830 m/min
Pickling practice Sulphuric acid with inhibitors at a line of 80±120 m/min
Roll material Forged steel (Induction hardened). A typical composition is as follows: C=0.85, Mn=0.35, Si=0.50, Cr=2±3.5,
M=0.30.

57
58 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67

Table 2
Performance data for the year 1993

Indigenous roll Imported roll

Date of scrapping Life obtained (tons) Reason for scrapping Date of scrapping Life obtained (tons) Reason for scrapping

15.2.93 20,116 Small spall 2.2.93 51,088 Depression

21.2.93 4536 Spalled at middle 9.2.93 56,167 Depression
4.3.93 3822 Spalled at middle 16.2.93 59,415 Metal pickup
28.6.93 9419 Spalled at middle 9.3.93 39,457 Depression
6.7.93 25,484 Small spall 29.3.93 52,226 Pickup and spall
12.7.93 5593 Spalled at middle 5.4.93 40,359 Low hardness
23.7.93 2684 Spalled at middle 22.4.93 13,031 Pickup and depression
19.9.93 3162 Spalled at middle 31.7.93 32,371 Low hardness
21.9.93 9990 Spalled at middle 23.8.93 9118 Small spall
22.9.93 2367 Spalled at drive end 31.8.93 8438 Spalled at one end
9.10.93 6374 Spalled at one end 23.9.93 24,218 Spalled at one end
24.10.93 10,557 Small spall 10.11.93 42,465 Pickup
4.11.93 22,816 Small spall 10.12.93 23,770 Depression
5.11.93 14,601 Spalled at one end 21.12.93 49,737 Depression

2.2. Manufacturer's factors

1. Non metallic inclusions
2. Casting and pour defects such as coarse grained structure or clustered carbides
3. Heat treatment faults such as poor carbide morphology, coarse martensite, retained austenite and
residual stress.
Cracks develop in regions between the top hardened surface and subsurface, a few microns below the
®nal grinding stage [5±7]. Due to thermal gradients, cracks propagate and failure occurs. Cracks are also
initiated due to compressive stresses experienced during contacts between roll and strip while rolling.
They grow under fatigue conditions and failure is not usually premature [8]. It is also believed that the
inclusions are centres from where the cracks start building up and propagate resulting in spalling [3].
Failure also occurs due to metal pick up and wrap-up of coil with roll. Due to breakage of the strip,
the material gets wrapped up and the temperature increases much above the speci®ed temperature,
leading to failure.

3. Methodology

A visit was made to the manufacturer's premises to examine critically the process sequence involved in
the manufacture of a forged roll. Several visits were also made to the steel plant for collection of roll
performance data, cause of premature failure and operational parameters. Samples of prematurely failed
rolls, from indigenous as well as imported makers, were examined for their chemistry, microstructure,
hardness and residual stress.
The detailed speci®cation of the rolls investigated in the present work is given in Table 1. These rolls
are induction hardened forged steel quality and are used in stand 2 of a 5 stand 4 Hi tandem mill. The
 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67 59

Fig. 1. (a) and (b) Thermally spalled indigenous roll; (c) thermally spalled imported roll.
60 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67

Fig. 2. (a) Presence of martensitic needles and higher percentage of retained austenite in indigenous roll. The white region is
retained austenite; (b) presence of tempered martensite and lower percentage of retained austenite in imported roll. Evenly distribu-
ted globular shaped undissolved carbide (also appearing white) is clearly revealed.

data on the lives obtained from various suppliers, indigenous as well as imported, along with cause of
premature failures were collected. These data are revealed in Table 2.
The methodology adopted focusses on metallographic features. The etching reagent (4%
picral+CH3OH) was used to examine the retained austenite. A RIGAKU X-ray di€ractometer was
used to determine the % retained austenite on the inner and outer surfaces of the spalled pieces.
Inclusion analysis was conducted to compare the quality of indigenous rolls with imported rolls.
Residual stress was measured on the inner and outer surfaces of the failed pieces using a X-ray stress
analyser AST-2001. The failed surfaces were also examined in the scanning electron microscope (SEM)
Jeol JSM-840 A to study the fractographic features of the spalled pieces.
Besides this, mill operation was carefully observed in order to suggest the ways and means to extend
the life of a work roll.
 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67 61

Fig. 3. XRD pattern of an indigenous roll (outer surface).

4. Results

Fig. 1 shows the damaged portion of an imported and an indigenous work roll. Typical fatigue lines
as well as origins of cracks close to the surfaces are revealed.

4.1. Microstructure

The microstructure (Fig. 2a) of the indigenous roll shows the presence of martensite needles, which
was perhaps not converted fully to tempered martensite during heat treatment whereas the
microstructure (Fig. 2b) of imported rolls reveals tempered martensite. The white portion in the optical
micrograph (Fig. 2a) is retained austenite. This was found to be more in indigenous rolls compared to
the imported rolls (see also Fig. 2b). Globular shaped undissolved carbides which appear also white in
Fig. 2b are clearly revealed on the inner surface of the spalled imported roll.

4.2. Retained austenite

Percentage retained austenite on the outer surface of di€erent samples of indigenous as well as
imported rolls was evaluated from the XRD peaks (Fig. 3). It varied from 6 to 15% by volume for
indigenous rolls, whereas for the imported rolls, % retained austenite was 10.94% by volume. However,
62 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67

Fig. 4. (a) Microstructure of indigenous roll revealing non-uniform distribution of undissolved carbides (not strictly globular).
These carbides appear white in the micrograph; (b) carbides are evenly distributed in imported roll. These undissolved carbides are
globular in shape and also appear white in the optical micrograph.

on the inner surfaces of the spalled pieces, % retained austenite varied between 13±15% for indigenous
rolls, whereas for the imported rolls it was 10.5%.
Retained austenite being a metastable phase, appears to be the main reason for spalling of indigenous
rolls. During rolling, retained austenite possibly converts to martensite resulting in expansion of volume
which creates external pressure and causes spalling.

4.3. Carbides

Undissolved carbides were not evenly distributed and globular in shape in indigenous rolls compared
to imported rolls (Fig. 4). This may be due to inadequate soaking during the heat treatment operation,
where not all the carbides may have a proper distribution of shape and size.

4.4. Non-metallic inclusions

The steel from imported rolls was found to be cleaner than that of indigenous rolls. The inclusion
content in the imported roll varied in the range of 0.02±0.15% by volume, whereas in the case of
indigenous rolls it was 0.16±0.24%. It is believed that the inclusions are the centres from where the
 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67 63

Fig. 5. (a) SEM fractograph revealing slag inclusion (indicated by arrow) on the surface of indigenous roll; (b) SEM fractograph
showing multiple cracks (indicated by arrow) from an inclusion in imported roll.

cracks initiate. Fig. 5 shows the presence of inclusions on the spalled surface of the rolls and multiple
surface cracks initiating and emanating from an inclusion on the surface of an imported roll.

4.5. Residual stress

Residual stresses measured on a spalled piece along the radial and transverse directions are shown in
Fig. 6. It can be seen from Fig. 6 that residual stresses were tensile on the spalled (inner) surface of the
indigenous roll. For the imported roll, compressive residual stresses were observed on both the surfaces
(inner and outer) and are not considered detrimental. However, the residual stresses during rolling are
more relevant because these might have become relieved by spalling.
64 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67

Fig. 6. (a) Residual stresses measured on failed surface (due to spalling) for indigenous roll (All units in ksi); (b) residual stresses
measured on failed sample (due to spalling) for an imported roll.
 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67 65

Fig. 7. Origin and propagation of crack in work roll (schematic). (a) Radial marks emanating from fracture origin; (b) chevron
marks pointing to fracture origin; (c) river marks from fracture origin.

5. Discussion

The major cause of failure of indigenous rolls was spalling. The content of retained austenite is more
in indigenous rolls than in imported rolls. During rolling, retained austenite might be converted to
martensite resulting in volume expansion, thus creating internal pressure to cause spalling. The
recommendation was made to conduct sub-zero annealing. If done at the manufacturer's premises, there
is then further scope for trying higher alloying with chromium and other elements, resulting in better
performance.
Spalling also originated from subsurface fatigue cracks. Crack propagation and branching is shown
schematically in Fig. 7. Chilton and Roberts [5], Iketahe and Nohara [6] and Nakagawa et al. [7] have
pointed out that failures of many cold rolling work rolls occur due to line spalls. The excessive stress
concentration takes place on the roll surface at the portion where the sharp strip edge contacts the rolls.
This stress concentration might cause generation of fatigue cracks and result in the spalling of rolls.
Therefore, it is advisable to measure the residual stresses at di€erent stages of operation in the rolling
mill. Because of the severe hardening process during manufacturing, high residual stress is retained in
the roll which might a€ect the performance of the roll. The residual stress in a freshly manufactured roll
should be measured in order to suggest ways of extending the life of CRM (cold rolling mill) rolls.
Another cause of roll failure is due to mill abuse, number of mill wrecks or accidents occurring during
operation. These accidents are caused by strip breakage and the stock material gets wrapped up around
the roll. There might also be local metal pick up by the rolls which gets welded with the sheet where the
temperature increases much above the speci®ed temperature and rolls fail due to thermal stress. This
66 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67

could be avoided by maintaining proper pressure of the cooling water and its quantity. Therefore,
smooth mill operation is necessary for the extension of the life of CRM rolls.
The performance of the roll also depends on quality of rolling stock and its surface conditions. There
might be variation in hardness, gauge thickness and surface condition. If the strip is under-pickled,
oxide layers would be present which become a potential source of roll failure.
The surface of the roll sometimes get bruised and develops cracks and elimination of these cracks
results in considerable increase of life.
The route of the roll was studied carefully from the roll card. The expected tonnage on stand no. 2 is
500 tons per campaign and 2000 tons per campaign on stand no. 5. In practice, the hardness undergoes
variation and the sequence might change due to non-availability of a roll for a particular stand which
a€ects the overall life of a roll. However, if a proper inventory is maintained and the inspection of the
roll surface is made frequently, it would be possible to increase the life of a roll to an optimum of
40,000±50,000 tons.
Since the mill is semi-automatic, a strict discipline on the part of the operating crew is needed for
improving the roll life. This is possible only when the HRD aspect is given due attention. This study has
increased the awareness amongst the rolling crew and there is a substantial reduction in the failure of
rolls.

6. Conclusions
1. The reason for poor performance of rolls was partly operational and partly manufacturing. Since the
mill is semi-automatic, a strict discipline on the part of the operating crew is needed for improving
the roll life.
2. From the performance data, the major cause of failure of indigenous rolls was spalling. This may be
due to the presence of a higher percentage of retained austenite. Austenite, being the metastable
phase in the steel, would be converted to stable phases by subzero annealing. It was recommended
that this facility be installed at the manufacturer's plant.
3. Residual stress is another important factor which a€ects the performance of rolls. Tensile residual
stress was found on the spalled pieces of indigenously manufactured rolls. It was recommended to
measure residual stresses at various stages during manufacturing as well as during operation which
warn of premature failure of rolls.
4. The quality of input (hot rolled strip) is not uniform with regard to its metallurgical characteristics.
This needs attention for improving the performance of the rolls.
5. Recommendations have also been made for the smooth operation of the mill. These
recommendations have already increased the awareness amongst the rolling crew during the tenure of
the project and there is a substantial reduction in the failure of rolls and down time in the mill.

Acknowledgements

The authors are grateful to Prof. P. Ramachandra Rao, Director, NML Jamshedpur, for his
permission to publish this paper.
 A.K. Ray et al. / Engineering Failure Analysis 7 (2000) 55±67 67

References

[1] Sinha AN, Sivaramankrishnan CSS. Ironmaking and steelmaking 1994;2(2):154.

[2] Ray AK, Mishra KK, Chaudhary PN. Internal Report NML/CIEP/SSP/95 1995.
[3] Ray AK, Mishra KK, Chaudhary PN. Failure of rolls of cold rolling mill in steel plant. In: Proceedings of Clinic on Failure
Analysis (COFA-97), National Metallurgical Laboratory, Jamshedpur-831007, 18±19 February, 1997.
[4] Chopra K. Quality assurance in forged steel rolls. Paper presented during Forged Rolls User Meet, Ranchi, 6±7 November,
1993.
[5] Chilton JM, Robert MJ. Iron Steel Engg 1998;58:77.
[6] Iketahe S, Nohara K. Iron Steel Inst Jpn 1971;57(5):103.
[7] Nakagawa M, Hoshi A, Sano AA, Nausher Y. Iron Steel Engg 1981;58:44.
[8] Swahnelal H. Scand J Metall 1975;4:263.