Accepted Manuscript

Premature failure analysis of forged cold back-up roll in a continuous tandem

mill
Hamid Reza Bakhsheshi Rad, Ahmad Monshi, Hassan Jafari, Mohd Hasbullah
Idris, Mohammed Rafiq Abdul Kadir
PII:
DOI:
Reference:

S0261-3069(11)00253-6
10.1016/j.matdes.2011.03.078
JMAD 3719

To appear in:

Materials and Design

Received Date:
Accepted Date:

21 December 2010
31 March 2011

Please cite this article as: Bakhsheshi Rad, H.R., Monshi, A., Jafari, H., Idris, M.H., Abdul Kadir, M.R., Premature
failure analysis of forged cold back-up roll in a continuous tandem mill, Materials and Design (2011), doi: 10.1016/
j.matdes.2011.03.078

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Premature failure analysis of forged cold back-up roll in a continuous

tandem mill
Hamid Reza Bakhsheshi Rad1,2, Ahmad Monshi2, Hassan Jafari1, 3, Mohd Hasbullah Idris1,
Mohammed Rafiq Abdul Kadir1
1

Materials Engineering Dept., Faculty of Mechanical Engineering, Universiti Teknologi

Malaysia, Skudai 81310, Johor, Malaysia

2
3

Materials Engineering Dept., Islamic Azad University, Najafabad branch, Isfahan, Iran

Materials Engineering Dept., Faculty of Mechanical Engineering, Shahid Rajaee Teacher

Training University, Tehran 16785-136, Iran

Cooresponding author:
Hamid Reza Bakhsheshi Rad
Rezabakhsheshi@gmail.com
Bhamidreza2@live.utm.my
Add: No. 801, U8A, Kolej Perdana, UTM, Skudai 81310, Johor, Malaysia
Tel.: 0060-147382258
Fax.: 0060-75534610

Abstract
In this paper, premature failure of a forged back-up roll from a continuous tandem mill was
investigated. Microstructural evolutions of the spalled specimen and surface of the roll were
characterized by optical microscopy, X-ray diffraction, scanning electron microscopy and
ferritscopy, while hardness value of the specimen was measured by Vickers hardness testing.
The results revealed that the presence of pore and MnS inclusion with spherical and oval
morphologies were the main contributing factors responsible for the poor life of the back-up
roll. In addition, metal pick up and subsequently strip welding on the surface of the work roll
were found as the major causes of failure in work roll which led to spalling occurrence in the
back-up roll. Furthermore, relatively high percentage of retained austenite, say 9%, in outer
surface of the back-up roll contributed spalling due to conversion of this meta-stable phase to
martensite and creation of volume expansion on the outer surface through work hardening
during mill campaign.
Keywords: A. Ferrous metals and alloys; F. Microstructure; H. Failure analysis
1. Introduction
In rolling mill operation a four roll high stand tandem mills including two work rolls and two
back-up rolls are used to reduce force and power of work roll as well as increase the accuracy
and thickness uniformity of thin sheets. Back-up rolls are the main trait of hot and cold roll
mills which decrease unintended bending and support the work rolls, enabling them to endure
higher loads without failing [1]. Work roll reduces the thickness of strip by plastic
deformation which creates by high compressive stress via the rolls. Generally, steel material
used for back-up roll is refined in electric arc furnace followed by vacuum degassing. The
produced ingot is then forged and subsequently differential heat treatment is performed in
order for the material to withstand the campaign in milling. Repeated loading under bending
and compressive stresses, severe friction and wear under corrosive environments at high
temperatures are some conditions that back-up rolls should endure during mill campaign
[2,3].
The main reason for premature failure of the forged back-up roll can be the combined effects
of mechanical and metallurgical factors. Mechanical factors include rolling parameter
misalignment, uneven roll surface, lubrication, bearing, rolling speed seizure, insufficient
stock removal during grinding and the experience of operators [4,5]. Metallurgical factors
2

comprise the presence of nonmetallic inclusions, localized overloading, casting defects,

temperature gradients due to insufficient cooling and phase transformations [5,6]. It was
observed [6] that spalling, cracking, metal pick up and subsequently strip welding are three
critical factors responsible for the poor service life of back-up and work rolls during milling
operation.
Spalling can be classified into two types. The first type is surface initiated spalling, which is
identified by fatigue path accompanied arrest marks, originate from thermally crack or
mechanically indentation, and subsequently fatigue path propagates circumferentially
opposite to the direction of roll rotation. The second one, subsurface initiated spalling, which
is recognized by the presence of a concentric fatigue pattern (fish eye) on the fracture surface
with arrest marks in the form of oval pattern, originates from a material defect and propagates
in different directions away from the initiation site usually within a single plane of
propagation [3]. In the present study the premature failure of a forged back-up cold roll used
for continuous tandem cold strip rolls was investigated. The main factors affecting the
premature failure of the forged back-up roll were determined and analyzed [3,7].
2. Experimental procedure
A spalled sample from a forged back-up roll used in a 5 stand 4 Hi tandem mill was
investigated. The chemical composition and detailed specifications of the forged back-up and
work cold rolls are given in Table 1 and 2, respectively. The sample was washed thoroughly
with running distilled water, rinsed and ultrasonically degreased with acetone and dried.
Afterwards, it underwent microstructural and fractographic examinations. To prevent
converting back up roll being scraped and any phase transformation during cutting as well as
to obtain reliable results several methods were investigated for sampling; consequently a ring
(Fig.1a) measuring 1365 mm in diameter, 15 mm in thickness and 100 mm in depth was
carefully removed during 24 hours from a tandem mill back up roll by heavy duty lathe
machine (INNSE 62"). Several samples (Fig.1b) were then cut from the ring for further
microstructural and microhardness experiments.
Table 1
Table 2
Figure 1

Microstructure of the samples prepared from the working surface and spalling specimen was
characterized via optical (OM) and scanning electron microscopes (SEM) equipped with
energy dispersive spectrometry (EDS), while microhardness value of the samples was
measured by Vickers hardness testing method using 10 Kg force. For microstructural
examination of phases, the specimens were etched in Vilella's reagent (1 g C6H3N3O7, 5 mL
HF, 95 mL CH3CH2OH) and Berahas reagent (3 g K2S2O5, 10 g Na2S2O3, 100 mL H2O).
Fischer MP3 ferritscope was employed to measure the amount of ferromagnetic martensite phase of the samples. X-Ray diffraction (XRD) analysis was carried out on the
samples to measure retained austenite content by measuring the integrated intensities of
(111) and (110) diffractions using equations (1) [8]:
V = 1.4l / (I + 1.4l )

(1)

where V is the volume fractions of austenite and I and I are the integrated intensities of
(111) and (110) peaks, respectively.
3. Results and discussion
3.1. Hardness profile measurement
Hardness profile, Fig. 2, was measured at every 5 mm distance from the working surface of
back-up roll. As can be observed, hardness value of the samples decreased from 607 to 489
HV with increasing the distance from the working surface. However, there was a fluctuation
in hardness value between 607 to 553 HV until the 17 mm distance from the surface of the
back-up roll. The result revealed that the hardness value of the roll was almost close to the
normal, whilst the main reason of fluctuated behavior in hardness profile may attributed to
the softening some parts of working surface [3,4]. Fig 3 shows the microstructure of the
softening regions indicating a large amount of precipitated carbide caused by superficial
tempering due to localized heating.

Figure 2
Figure 3

3.2. Microstructure
Fig. 4 shows optical microstructures of the back-up roll material as a function of depth from
the roll surface. Figs. 4a and b represent the microstructures of the samples cut from 5 mm
and 15 mm distances from the surface of the back-up roll showing high percentage of
tempered martensite, approximately 87% and 67% volume fraction, retained austenite,
around 9.5% and 9.2% volume fraction, respectively, and chromium carbide. Due to the
presence of medium amount of carbon and alloying elements in chemical composition of the
steel, retained austenite usually coexists with tempered martensite in microstructure after
continuous cooling process [3,7]. However, the microstructure of the sample selected from 15
mm distance from the surface exhibited small amount of non-tempered martensite and
bainite, owing to a decrease in cooling rate with an increase in distance from the surface of
the roll. The reason for transforming austenite to bainite is accessibility of sufficient time for
carbon to have long range diffusion in the microstructure. The microstructure of the samples
in 55 mm and 70 mm depths from the roll surface comprised low percentage of volume
fractions of martensite, spheroidised pearlite, retained austenite and a considerable amount of
undissolved carbide, as seen in Figs. 4c and d. At the inner layer, precipitation of undissolved
carbide occurred in the microstructure due to sufficient heat and time during continuous
cooling of the material.
Figure 4
3.2.1. Retained austenite
The content of retained austenite from surface to depth of the back-up roll was determined by
the XRD analysis. Fig. 5 exhibits the variation of retained austenite content versus depth from
surface of the back-up roll. The content of retained austenite was measured as relatively low
as 5.3% at inner layer, while it was almost high in the outer layer, about 9.5%. The optical
microstructure evolutions of the specimens in 5 and 75 mm distance from the surface of the
roll are shown in Figs. 6a and b, respectively. It can be seen that the content of retained
austenite (light area) in outer layer is higher than inner layer of the roll. As a matter of fact, in
differential hardening method, the surface barrel of back-up roll was subjected to
austenitization process by back-up roll differential heat treatment furnace. Subsequently the
surface barrel was heated to about 900C for a few hours which lead to a decrease in
temperature gradient from the outer layer towards the inner layer. This will result in the outer
5

layer having a higher temperature compared to the inner layer causing an increase in the
solubility of chromium carbide in the outer layer. This will increase the concentration of
alloying elements in the solution which leads to martensite start temperature (MS) decreased
significantly [9]. The martensite finish temperature (Mf) will subsequently fell to under room
temperature which brought about a large amount of non-transformed austenite microstructure
at the outer layer [10,11]. During mill campaign, this meta-stable phase can transform into
martensite through work hardening which creates external pressure due to volume expansion
and causes spalling.

Figure 5
Figure 6

3.2.2. Carbide characterization

Fig. 7 shows SEM microstructures of the back-up roll material as a function of depth from
the roll surface in the region away from spalled sample. It was revealed that all samples
comprised carbides with different morphologies, sizes, volume fractions and distribution in
the matrix of tempered martensite. Figs. 7 a and b indicate the microstructure of the sample in
5 mm depth from the surface of the back-up roll. As can be seen high volume fraction of fine
spherical carbide homogeneously has distributed in tempered martensite matrix. This
microstructure increases wear resistance of the steel and subsequently improves service life
of the back-up roll. Figs. 7c and d represent the microstructure of the sample selected from a
depth of 20 mm from the surface of the roll. It can be seen that significant volume fraction of
irregular shaped carbide compared to the previous sample has distributed uniformly in the
microstructure. Figs. 7e and f exhibit the microstructure of the inner layer (40 mm distance
from the surface) of the roll encompassing coarse needle shaped carbide with lower volume
fraction and heterogeneously distributed in the matrix. It is believed that the needle shape of
the carbide acts as the center for crack initiation and propagation. Moreover, intersection
point of the coarse needle shaped carbide can be an impact in accelerating crack initiation and
propagation. Therefore, it causes poor service life of the back-up roll.
Figure 7

EDS analyses and quantitative microprobe of the carbides are shown in Fig. 8. Based on the
quantitative microprobe, it can be concluded that the stoichiometry of the carbides in all
samples were M7C3 due to the atomic percentage ratio of carbon to other elements is almost
equal to 3:7 [12,13]. The EDS analysis of carbide in outer layer of the back-up roll (5 mm
distance from the surface) include 8.57% C, 87.22% Fe, 2.08% Cr, 0.53%Mo and 0.61%Mn.
The atomic percentage of carbon was about 30.30%, whilst the entire atom percentage of the
other associate elements was 70.20% which indicated that the stoichiometry of the carbide
was (Fe,Mn,Mo,Cr)7C3 (Fig. 8a). In the middle and inner layers, 20 and 40 mm distance from
the surface respectively, the (Fe,Mo,Cr)7C3 carbide contain around 8.48% C, 87.07% Fe,
4.01% Cr and 0.44%Mo (Fig. 8b and c).
Figure 8
3.2.3. Non-metallic inclusion
Fig. 9 shows the presence of inclusion in the samples of 10 and 15 mm distance from the
surface of the back-up roll. The EDS analysis of working surface exhibited that the
stoichiometry of inclusion was manganese sulfide (MnS) with globular and oval
morphologies, as given in Figs. 9a and b. The inclusion contain 40.83% S and 59.17% Mn
while, the atomic percentage of it was 54.12% and 45.88% for S and Mn, respectively
(Fig. 9d). Generally, the inclusion can be originated from steel making process and it is
believed that during plastic deformation inclusion can act as a critical location for
concentration of dislocation and stress, so it causes crack to initiate in inclusion/matrix
interface [14, 15] subsequently it propagates and spalling occur in the outer layer. Stress
concentration is created in the interface as a result of two main factors, namely the difference
in thermal expansion coefficients of inclusion and matrix and the concentrations of
dislocation due to work hardening.
Fig. 9c also indicates the presence of pore in the 12 mm distance from the surface of the
back-up roll. The pore distributed randomly near the spalling surface during the rolling
operation which might be the location for crack initiation. From this position sub-surface
crack initiated and propagated within a single plane in all directions away from the initiating
site [4,6]. Since the roll is continuously dressed up during milling process, the shear and
residual stresses produced a high accumulated stress. After initiation, the crack propagated
circumferentially until the strength of the surrounding material was reduced to such a degree
that large spalling occurred (Fig. 11).
7

Figure 9
3.3. Metal wrapping and strip welding
Figs. 10a and b show the metal wrapping and strip welding occurred on the surface of the
work roll during mill campaign, respectively. The strip welding located at the surface of the
work roll measured as 250 mm length and 130 mm width. Metal wrapping is the most
important factor for failure of back-up and work rolls which is attributed to improper
operating techniques. Generally the initial operating temperature of the back-up roll in
continuous tandem mill is between 90 to 110 C. However, it decreases significantly to about
55 C by continuous cooling liquid (spray) of water and oil during the milling campaign. But
when metal wrapping occurs in work roll/ back-up roll contact zone due to the fluctuation in
thickness of strip within the roll bite, compressive stress increases locally and subsequently
causes a significant increase in roll surface temperature [16]. This condition leads to rupture,
welding and even melting of the local region of the strip on the surface of work roll [6]. Fig.
10c demonstrates a typically round localized indentation (roll mark) formed on the surface of
the back-up roll having a maximum diameter of approximately 75 mm and a depth of 3 mm.
The reason for happening this defect can be attributed to the strip welding occurred on the
surface of the work roll which caused severe plastic deformation on the surface of the backup roll. Afterwards, with continuing milling service, surface cracks initiated from the
localized indentation area and propagated radially and circumferentially until spalling
occurred.
Figure 10
3.4. Spalling
Fig. 11a indicates overload and bruise areas as well as roll marks on the surface of spalling
sample of the back-up roll. The hardness of the bruise area was measured as 564 HV showing
around 27 HV lower than that of the base material, owing to an increase in local temperature
beyond the tempering temperature of the roll material during mill campaign. The main
sources of bruising in this spalled sample were metal wrapping (strip welding) and
fluctuation in thickness of strip which can be prevented by controlling operating techniques.
Figure 11

Fig 11a indicates roll marks on the spalled sample as well. This kind of indentation can create
surface cracks by acting as stress concentration factors. The entire roll surface experiences a
high tensile and compressive stresses cycle during each roll revolution. Therefore, the
concentration of stress at any roll marks could develop a surface crack. Fig. 11b shows the
fatigue path with beach mark on the back side of the spalling sample which consists of two
parts, namely shiny (rubbed) and dark (oxidized) areas. As the drawn lines for the
fractography show, fatigue path initiated and propagated circumferentially as result of
overload on the surface of the spalled sample, as indicated in Fig. 11a.
Figs. 12a and b show fatigue cracks which probably initiated from inclusion or pore defects
on the sub-surface of the back-up roll where stress concentrated. As can be clearly seen in
Fig. 12c, the presence of dimples at the fracture surface of this spalling sample is an evidence
for the ductile fracture mechanism which can be attributed to the existence of inclusions and
pore in the structure. It is reported that [17] inclusions and pore can produce dimple on
fracture surface due to creation of the sites for crack nucleation.
Figure 12
4. Conclusions
(1) The presence of spherical and oval MnS inclusions and pores can be acted as the centers
for crack initiation at the inclusion or pore/matrix interfaces resulted in crack propagation,
subsequently spalling occurred. Modifying steel making process by new methods such as
electroslag remelting (ESR) or ladle refine furnace with VOD function (LFV) is a
recommendation for minimizing the defects.
(2) The other reason for premature failure of the back-up roll was spalling which created by
metal wrapping and strip welding in work roll/ back-up roll contact zone. Wrapping and strip
welding can be reduced by controlling operating techniques and thickness of product being
rolled.
(3) Relatively high percentage of retained austenite was found one of the main reasons caused
poor service life of the roll during mill campaign. This meta-stable phase converted to nontempered martensite through work hardening and created volume expansion led to external
pressure and spalling occurrence. Triple temper or sub-zero treatment are recommended as
appropriate remedies to reduce the content of retained austenite.

(4) Heterogeneously distributed coarse needle shaped carbide in the inner layer of the backup roll with (Fe,Mo,Cr)7C3 stoichiometry can be considered as another crack initiation source
and poor service life of the roll.
References
[1] Kang X, Xia D.L, Campbell J, Li Y. Development of cast steel back-up roll. J Cast Met
Res 2006; 19: 66-71.
[2] Fractography, vol. 12, ASM Handbook, ASM Int, OH; 1992.
[3] Pantazopoulos G, Vazdirvanidis A. Fractographic, Metallographic Study of Spalling
Failure of Steel Straightener Rolls. J Fail Anal and Preven 2008; 8: 509514.
[4] Wu Q, Sun D.L, Liu C.S, Li C.G. Analysis of surface and sub-surface initiated spalling of
forged cold work rolls. Eng Fail Anal 2008; 14: 401-410.
[5] Li H, Jiang Z, Tieu K.A, Sun W. Analysis of premature failure of work rolls in a cold
strip plant. Wear 2007; 263: 14421446.
[6] Ray A.K, Mishra K.K, Das G, Chaudhary P.N. Life of rolls in a cold rolling mill in a steel
plant operation versus manufacture. Eng Fail Anal 2000; 7: 55-67.
[7] Azevedo C.R.F, Neto J.B. Failure analysis of forged and induction hardened steel cold
work rolls. Eng Fail Anal 2004; 11: 951-966.
[8] Lee K, Shin H.C, Jang Y.C, Kim S.H, Choi C.S. Effect of isothermal transformation
temperature on amount of retained austenite and its thermal stability in a bainitic Fe3%Si
0.45%CX steel. Scripta Mater 2002; 47: 805809.
[9] Akhbarizadeh A, Shafyei A, Golozar M.A. Effects of cryogenic treatment on wear
behavior of D6 tool steel. Mater Design 2009; 30: 32593264.
[10] Flvio J, Sinsio D, lisson R , Emmanuel O, Antnio M. Performance of cryogenically
treated HSS tools. Wear 2006; 261: 674685.
[11] Akhbarizadeh A, Golozar M.A, Shafeie A, Kholghy M. Effects of austenizing time on
wear behavior of D6 tool steel after deep cryogenic treatment. J Iron Steel Res Int, 2009; 16:
29-32.
[12] Michaud P, Delagnes D, Lamesle P, Mathon M.H, Levaillant C. The effect of the
addition of alloying elements on carbide precipitation and mechanical properties in 5%
chromium martensitic steels. Acta Materialia 2007; 55: 48774889.

10

[13]Hoseiny H, Klement U, Sotskovszki P, Andersson J. Comparison of the microstructures

in continuous-cooled and quench-tempered pre-hardened mould steels. Mater Design 2011;
32: 2128.
[14] Nusair Khan A, Muhammad W, Salam I. Failure analysis of bainitic steel pipeFailed
during cold working process. Mater Design 2010; 31: 26252630.
[15]Hashimoto K, Fujimatsu T, Tsunekag N, Hiraoka K, Kida K. Study of rolling contact
fatigue of bearing steels in relation to various oxide inclusions. Mater Design 2011; 32:
16051611.
[16] Gaspard C, Ballani J, Batazzi D, Adams T. Use of HSS rolls to skip the chrome plating
in cold rolling applications. Materials, Science & Technology 2004; 26-29.
[17] Salemi A, Abdollah-zadeh A. The effect of tempering temperature on the mechanical
properties and fracture morphology of a NiCrMoV steel. Mater Charact 2008; 59: 484-468.

Captions
Tables
Table 1. Chemical composition of the 3Cr-MO steel (wt.%).
Table 2. Detailed specifications of the investigated back-up and work rolls.
Figures
Fig. 1. (a) The ring cut from the tandem mill back up cold roll and (b) samples cut from the
sector of the ring.
Fig. 2. Hardness profile of back-up roll.
Fig. 3.SEM micrographs of softening region at (a) 5mm (b) 7mm from the roll surface.
Fig. 4. Microstructures of material of back-up roll as a function of depth from the roll surface;
(a) 5, (b) 15, (c) 55, and (d) 70 mm.
Fig. 5. Content of retained austenite as a function of depth from the surface of back-up roll.
Fig. 6. Optical microscopic image of retained austenite (light area) as a function of depth
from the surface of back-up roll; (a) 5 and (b) 75 mm.
Fig. 7. SEM microstructures of material of back-up roll as a function of depth from the roll
surface; (a) 5, (b) 20, and (c) 40 mm.
Fig. 8. EDS analyses of the carbides in the microstructures of back-up roll material as a
function of depth from the roll surface; (a) 5, (b) 20, and (c) 40 mm.
11

Fig. 9. SEM micrographs of samples from different distance departing from surface of backup roll; (a) 10, (b) 15, (c) 12 mm, and d) EDS microanalysis of inclusions in back-up roll.
Fig. 10. (a) metal warpping, (b) strip welding, and (c) localized indentation on the surface of
the rolls.
Fig. 11. (a) overload area, roll marks and bruise area on the surface of spalling sample and (b)
fatigue propagation accompanied beach marks on the back side of spalling sample.
Fig. 12. SEM fractograph of spalling sample revealing; (a) fatigue cracks initiate of inclusion
of sub-surface of roll, (b) crack propagate radially, and (c) fracture origin with existence of
dimples.

12

Table 1. Chemical composition of the 3Cr-MO steel (wt.%).

C

Si

Mn

Cr

Mo

Ni

Sn

As

Cu

Al

0.68 0.31 0.77 0.014 0.008 3.15 0.69 0.07 0.005 0.009 0.045 0.003

13

Table 2. Detailed specifications of the investigated back-up and work rolls.

Back-up roll size (mm)

48221525

Back-up scrap size (mm)

48221365

Back-up roll weight (Kg)

36310

Back-up scrap weight (Kg)

31680

Back-up surface hardness (HV)

551-587

Depth of hardened layer (mm)

160

Work roll size (mm)

3765585

Work roll scrap size (mm)

3765510

Work roll weight (Kg)

4865

Work roll scrap weight (Kg)

4008

Work roll surface hardness (HV)

874-952

Depth of hardened layer (mm)

75

No. of stands

Annual production (ton)

1,500,000

Maximum speed (m/min)

1150

Entry strip thickness (mm)

2-4

Exit strip thickness (mm)

0.18-3
650-1500

Coil width (mm)

14

Research highlight

Metal wrapping and strip welding in work/back-up rolls contact zone caused spalling
MnS

inclusion and pore initiated crack which propagated in milling led to spalling

Retained austenite conversion to -martensite accelerated spalling failure

Needle shaped carbide, (Fe,Mo,Cr)7C3, my cause poor service life of back-up roll

15

16

17

18

19

20

21

22

23

24

25

26

27