WEAR OF HOT ROLLING MILL ROLLS: AN OVERVIEW
S. Spuzic*, K. N. Strafford*, C. Subramanian* and G. Savage**
* Department of Metallurgy, University of South Australia, The Levels,
South Australia 5095
** BHP Research - Melbourne Laboratories, Mulgrave, Victoria 3170
Rolling is today one of the most important industrial processes because a greater volume of material is
worked by rolling than by any other technique. Roll wear is a multiplex process where mechanical and
thermal fatigue combines with impact, abrasion, adhesion and corrosion, which all depend on system
interactions rather than material characteristics only. The situation is more complicated in section rolling
because of the intricacy of roll geometry. Wear variables and modes are reviewed along with published
methods and models used in the study and testing of roll wear. This paper reviews key aspects of roll wear
control - roll material properties, roll pass design, and system factors such as temperature, loads and
sliding velocity. An overview of roll materials is given including adamites, high Cr materials, compound
rolls and high speed tool steels. Non-uniform wear, recognised as the most detrimental phenomenon in
section rolling, can be controlled by roll pass design. This can be achieved by computer-aided graphical
and statistical analyses of various pass series. Preliminary results obtained from pilot tests conducted using
a two-discs hot wear rig and a scratch tester are discussed.
Rolling steel at elevated temperatures ("rolling", hereafter) is one of the most important industrial
processes, for a greater volume of material is worked by rolling than by any other technique [1,2]. Key
tools in the rolling process are the rolls themselves. These tools have to withstand severe extremes of
temperature and load. In addition to the obvious need for resistance to breakage, there is the continuing
component of roll wear that is critical to the economics of fabrication, and the geometrical tolerances of
the rolled products.
From the earliest days of metal working by rolling (some 500 years ago), rolls were used as plain (flat) or
as grooved (calibre) rolls . Rolling between flat rolls enables the shaping of rectangular profiles only.
Rolling with grooved rolls enables the sophisticated and straightforward manufacture of enormous variety
of profiles. A disadvantage of calibre rolls, supplementary to complications in their production and
maintenance, is the significantly higher wear when compared with plain rolls. In an attempt to maintain
overall simplicity in roll geometry, the complexity of rolling mills has been increasing over the years,
involving the application of various types of universal stands.
Conditions in groove rolling are clearly more severe when compared with the rolling of flat sections. The
diversity in temperature, pressure and stress fields, along with slip gradient in calibre rolling result in
accelerated wear of grooved rolls. On account of this Corbett  has queried whether costly secondary
rolling mills (with calibre rolls) could be replaced by continuous casting, as has been the case for primary
(blooming-slabbing) mills. Present manufacturing trends indicate however that the use of grooved rolls
cannot be avoided and, indeed, various combinations of trains with calibre respective plain rolls will be
applied. Even after the introduction of universal intermediate and finishing mills, there is a need for calibre
rolls upstream in the rolling process. Currently there is a tendency to increase drafts at secondary mills to
compensate for the absence of primary mills [4,5]. In response to these various demands, it is important to
improve the resistance to wear of roll materials used in structural mills.
In general, rolls contribute some 5 - 15 % of overall production costs ie A$ 1-5 per ton of rolled product
and per rolling mill stand. However, interruptions of the production process - shut down on account of
wear related problems - cause additional significant expenses [2,6]. Considering the relatively high
proportion of roll costs in overall production costs, the ability to predict roll performance, especially in the
domain of wear, becomes more important.
A principal aim of this present review is to critically discuss accumulated knowledge relating broadly to
the question: "how does the hot wear of calibre rolls occur?" An area of specific interest concerns abrasive
Published in Journal
Volume No 176 (1994)
pp 261 - 271
wear at elevated temperatures. The review is confined to hot wear of grooved rolls for producing the
medium and heavy structural sections.
A further aim of this survey is to provide direction and focus to an experimental program. Roll wear
factors and modes (based on their perceived significance) are reviewed and an outline provided concerning
the methodology and models applied in examination of these phenomena. The research proposals and pilot
experiments conducted using a two-discs rig and a scratch tester are rationalised and described.
2. VARIABLES AND MODES OF ROLL
2.1. Roll Wear Criteria and Rates
The properties required of rolls differ in detail at the various stages of the rolling line. While at the initial
stage of rolling process the heat shock resistance and the general strength of the material are the
dominating requirements, at later stages, resistance to abrasive wear becomes the most important
In addition, there are different perceptions of "roll wear" amongst rolling mill operators, designers and roll
producers. Roll "performance" is a vague term that may concern: mass or length of hot steel produced per
millimetre of machined roll diameter, or per mass of roll material used; or total tons of the product
processed per set of rolls or per roll change. Calibre rolls have wear resistances of about several hundreds
tons of hot rolled steel per millimetre of machined roll diameter (t/mm), whereas typical flat rolls have
"lives" about 5.103-20.103 t/mm [1,7,8]. Such a variety of roll wear criteria makes objective comparison
of roll performances impossible. This problem might be overcome by defining roll wear in terms of
"radial wear (mm) per sliding path" .
With calibre rolls, non-uniform wear (Fig. 1) should be observed, because the points of extreme wear
determine the total depth of subsequent machining . Generally, differences in wear (eg along the groove
meridian, among the roll calibres, between the top and bottom roll, among the different stands) decrease
roll life and obscure the rolling process.
2.2. Roll Wear Modes
Before arriving at a selection methodology for materials and/or other system parameters to mitigate roll
wear, it is useful to identify what wear processes are possible. Various interactions can affect the wear
modes, eg the loss or transfer of surface material may lead to loss of fit or alignment (in addition to
generating debris, which is a potential abrasive), with resultant changes in loads and friction, generating
subsequent wear damage . Experience has shown that roll wear rate increases rapidly after production
of a specific amount of rolled steel; hence roll changes should be conducted after rolling a characteristic
tonnage, to avoid catastrophic wear.
To identify wear processes, it is necessary to refer to some of many published wear classification systems.
An example of the useful approximation to wear categorisation is shown in Fig. 2 .
Abrasion is among the dominant ever-present components of the total roll wear process [12,13].
Generally, abrasion can be conceptualised as two-body or three-body wear. Considering the common
presence of oxide scales of high hardness and low plasticity, on hot steel surfaces, three-body wear would
be expected to be an important mode of roll wear [6,11]. Further sub-classification distinguishes micro-
ploughing, micro-cutting, micro-fatigue and micro-cracking . Depending on the shape and hardness of
the abrasive particles, there occurs either micro-cutting or plastic deformation of the surface and sub-
surface layers by sliding; sub-surface deformation can result in crack formation . Such a sub-
classification generally has not been recognised in research of roll wear, and hence a further specific
analysis of abrasive mechanisms is needed pertaining to this field.
Thermal fatigue is widely known as a mechanism of roll wear [6,12,15,16]. Any point on the roll surface is
alternately heated by contacting hot steel and cooled by water. Consequently, compressive and tensile
stresses are generated at a frequency of the roll rotation. If the compressive stress exceeds the compressive
yield limit of the roll material during the heating stage, the outer layer will deform plastically. On the
contrary, in cooling, a high tensile stress will be imposed on roll surface without plastic deformation,
because the ductility of the roll material is not sufficient at these lower temperatures. When the fatigue
limit of the material is reached, crack nucleation begins and the characteristic overall "firecrack" pattern
will result. Heavy bending stresses will accelerate the process of crack propagation.
Various forms of corrosion attack eg fatigue or stress corrosion are also responsible in some measure for
roll wear . On the other hand, Ohnuki  has highlighted the benefits of oxides formed by high
temperature oxidation at roll surfaces: at specific temperatures, dependent on the roll material grades, a
hard and smooth black magnetite scale film is formed on the roll surface and excellent resistance to
abrasion wear obtained.
Adherence ('seizure') of bands of oxide layers derived from rolled steel surfaces has been reported by some
authors [17, 18]. This phenomenon defined as "the transfer of part of the work material surface to the roll
surface, causing loss of product quality" occurs more readily at high reduction passes. Intermittent seizure
leads to hot rolled material "pick up", roll surface roughening and increase in friction coefficient.
Occasionally, associated uniform "prints" can be identified as periodic damage along the finish product
Rolls can wear through slip during initial passes and hence the bite must be improved by increasing
friction or by decreasing the reduction per pass. Knurling, produced on grooved roll surfaces to increase
the angle of bite, can be rubbed out by skidding and the biting ability can thus be adversely affected [4,19].
2.3. The Classification and Properties of Roll
Materials: Selection of the Roll Materials
Roll materials can be classified in several ways:
i) roll manufacturing route - via casting, forging, double pouring, etc. Vertical spin casting is used today
by the majority of manufacturers of rolls to be used for flat products. In contrast, rolls for heavy structural
products require grooves so great that a useable layer thickness could only be obtained by a roll involving
a single pour. With the development of sleeves for universal stands, spin casting has also been applied for
structural mills and, today, most rolls for universal stands have been centrifugally cast . Universal mill
finisher rolls (channels and beams) are press-fit, glued or shrunk on to the shaft. Stresses to be induced by
the fit will be influenced by the roll material choice. Spheroidal irons and high carbon steels are generally
adequate for both horizontal and vertical rolls. Some forged and composite vertical sleeves are in use .
ii) roll material chemistry;
the range of chemical analysis is:
0.3%≤C≤3.8%; 0.2%≤Si≤2.5%; Mn≤2.5%; Cr≤30%; Ni≤5%; Mo≤4%
Further alloying elements are P, S, V, W, Nb, Ti.; The contents of P and S are carefully controlled
iii) microstructure of the roll working surface;
Depending on chemical composition and heat treatment, roll materials have a microstructure consisting of
a matrix of pearlite / bainite / martensite / austenite, with a greater or lesser proportion of
carbide/cementite, and graphite of a particular distribution and size . The properties of roll materials
are significantly influenced by the quantity and distribution of carbides (MxC) which can comprise up to
40% by volume of the roll microstructure. Cementite has exceptional abrasion resistance at room
temperature and excellent wear resistance at the elevated temperatures. In general, there is a tendency to
strive for the maximum quantity of carbide in the roll structure compatible with adequate ductility and heat
shock resistance .
Classical categorisation has divided roll materials into two main groups i) cast iron rolls and ii) steel rolls.
Their essential differences can be summarised as follows: 1) Cast iron rolls (often used at finishing stands)
are brittle materials that possess high wear resistance. 2) Steel rolls are often used in primary and
secondary mills because they exhibit the following (superior) properties in comparison with cast iron rolls:
"higher" coefficient of friction (leading to a better "bite"); "higher" strength, so they can withstand
"higher" bending and torsional stresses; and uniform hardness to a sufficient depth from the surface .
Recently, highly alloyed ferrous materials, notably the high chromium irons and steels, as well as high
speed tool steels, have been promoted as roll materials with extremely good wear resistance [5,22,23,24].
Table 1 shows a classification of roll materials based on route of manufacture and structure of the working
Categorisation of roll materials can be based on an understanding of the interrelations of mechanical
properties, microstructures and chemical compositions, especially C content, see Fig. 3 . The most
important metallurgical parameters that influence abrasive wear of steel are microstructure (ie type of
matrix and dispersed phases), hardness at working temperature and interstitial element content (%C and/or
When considering the selection of roll materials, in relation to general wear problems, more emphasis is
often put on improving the strength and hardness of metallic alloys, rather than their ductility .
However, ductility can decrease significantly with increasing strength or hardness. Roll cracking
susceptibility due to thermal fatigue and abnormal thermal shocks is mainly affected by fracture
toughness, thermal conductivity, thermal expansion and ductility. Hence, a compromise between strength
and ductility has to be looked for.
The practical importance of the tool material can be outlined if specific problems in roll choice are
considered. For example, to select materials for rolls in a universal medium section mill, the following
criteria were applied: 1. The roll surface defect history (thermal cracks or other wear modes); 2. Product
categories (light, heavy); and 3. Mechanical properties required (stresses expected). Generally, materials
offering the best combination of resistance to mechanical stresses, thermal fatigue and wear, are various
centrifugally cast steel base grades (adamites). Thus adamites were chosen commencing with a material
with a hardness of 300HB for the entry group of stands, and ending with 420HB for the finishing rolls.
Further improvements were sought by use of high Cr alloy cast irons and alloyed bimetal centrifugally
cast spheroidal irons .
A quite different situation appears in mills producing heavy structural products, where the loads are high
and the roll geometry is such that stresses become severe as the roll diameter is reduced. For the break-
down rolls, toughness, as well as resistance to thermal shock and wear, is required . Here it is apparent
that there is a conflict in material requirement and, generally, mono bloc steel-based rolls are selected. In
these situations material commonly ranges from cast carbon steel (eg 0.5% C, with hardnesses of 30 Shore
C, tensile strength 650 MPa and ferrite-pearlite structure) to adamite rolls (eg 2.3%C, Cr-Mo alloyed, with
hardenesses of 50 Shore C, tensile strength 500 MPa, and a microstructure of spheroidised carbides
dispersed in a pearlitic matrix) .
In some modern mills, spheroidal iron is favoured for the intermediate stands, for it exhibits the strength of
the higher carbon steel roll grades and the wear resistance of the alloy iron grades. For the same
intermediate stands, but with deep groove rolls such as angles and channel rolls, a high carbon steel roll or
even a graphitic steel would be a better choice. The reason for this is that a steel roll can be heat treated to
maintain a uniform hardness to a sufficient depth - much greater than it is possible with either the
spheroidal iron or alloy grain irons. Grain iron roll material ("indefinite chill") has an outer chilled face on
the roll body causing finely divided graphite at the surface, which increases in amount and in flake size as
distance from surface increases . The structure of alloy grain irons depends critically on cooling rate
during casting and they generally lose hardness very rapidly from the surface to a depth of ca 80-100mm.
Hence, a roll made from iron will wear non-uniformly .
Rail and large structural intermediate stand rolls are generally always fabricated from steel, although some
success has been achieved using pearlitic acicular spheroidal iron rolls .
2.4. Rolling Loads and Stresses
Hot wear rate is directly proportional to the normal pressure on roll surface. Average rolling pressures can
be considered to be in range 100 - 300 MPa. The corresponding cyclic stresses, amplified by thermal
cycles, in roll surfaces are estimated to amount to ±500 MPa [7,17].
Cyclic loads result in material fatigue and other forms of surface deterioration. Besides the well-known
mechanisms of cyclic softening and corrosion fatigue, there is growing evidence of the damaging
influence of tensile stress during the contact fatigue, leading to cracking and pitting . In addition, it has
been found that cyclic pre-stressing has a significant influence on the material removal process in sliding
wear . A related question is the actual distribution patterns of the rolling loads and stresses. The
observed non-uni-form wear, particularly the unsymmetrical wear of the symmetrical calibres, illustrates
the significance of the irregular pressure distribution. The appreciation of these contrasts in stresses has led
to the introduction of different materials for top rolls and bottom rolls in the practice of section rolling.
If metal flow is not appropriately allowed for in roll pass design, metal is unnecessarily forced to exert
additional localised pressure and wear on the groove walls. There are in fact basic principles for roll pass
design in existence, where roll wear is among the main criteria. Higher loads and draft non-uniformity
should be applied in the initial passes and they ideally should diminish in the finishing passes [2,31].
Rolling loads and stress distributions should be designed to ensure stable and uniform wear of finishing
calibres. Corrosion fatigue can be suppressed by proper design considerations aimed at reducing stress.
The tensile component of stress causes stress corrosion. Residual compressive stresses, deliberately
introduced through an adequate heat treatment, will suppress crack initiation and growth, as well as stress
corrosion and fatigue . Geometrical aspects (eg roll diameter) have considerable effect on tool life via
associated stress and heat concentration [7,32].
Considering the feasibility of influencing stresses and loads as process variables that can be controlled by
roll pass design, this established correlation between stresses and surface deterioration must be an
important aspect of research into roll wear.
2.5. Aspects of Rolling Temperature
Rolling temperatures vary mainly between 800oC and 1200oC [7,17,18]. Roll surface is heated initially to
approximately 650oC while it is in contact with the hot slab, and subsequently cooled by water to around
50oC during the same cycle . Flash temperature could in fact rise above 800oC due to the friction-
generated heat .
An important aspect is the effect of surface temperature on the formation of oxide scales on roll surfaces.
Cast steel and adamite develop hard and smooth magnetite black scale films above 400oC; this process is
followed by a rapid decrease in wear and friction coefficient, f. As the surface temperature rises, the scale
transforms from magnetite to wustite. Above 650oC and higher, the surface layer matrix is softened and
deformed plastically. Hence, cast steel and adamite materials should be used in the temperature range from
500 to 650oC. For analogous reasons, grain cast iron rolls should be used between 600 and 700oC. This is
because the eutectic carbide and primary crystallised graphite in the surface act to halve the effective area
available for magnetite scale formation. Above 600oC, the wustite scale from the counter-body adheres to
the cast iron to form a black scale, thereby reducing wear rate. At 750oC and above plastic deformation
occurs in the surface layers, increasing the surface roughness .
An additional phenomenon associated with temperature is thermal fatigue [16, 17, 27]. At higher
temperatures in the use of roll materials (eg 600oC), acceptable stress levels in fatigue are generally lower
when compared with behaviour at room temperature. By the use of a counter rotating discs test, Ohnuki
 has examined the surface-layer thermal fatigue of hyper-eutectoid cast steel and adamite. At contact
stresses of ca 250 MPa and temperature cycles 100 - 500oC, the thermal fatigue limit was found to be in
about 1 - 4 104 revolutions. The surface layer fracture originated at a depth 0.1 mm below the surface; that
is where the contact shear stress reaches a maximum, and this effect terminates the protective action of the
Roll surface temperature depends also on the deformation zone geometry. This again indicates the
importance of roll pass design.
2.6. Cooling and Lubrication
Although it is clearly understood that temperature variation around and across the roll surface is
detrimental, extremely wide differences in cooling practice remain unsatisfactorily explained . Water
is widely used as a coolant in section rolling; it is applied either on its own or within an aqueous dispersion
or emulsion . Experience in rolling practice has demonstrated that idle rolls should not be cooled,
because of the damaging effect of the increased thermal gradient.
Lubrication is widely considered as a method for the improvement of roll life. Mineral oils as well as
organic fats and oils were tried as lubricants in the early stages of rolling technology development. Light
mineral oils (compounded with additives like lanolin) exhibited low staining propensities. In the hot
rolling of steel plates, heavy and possibly contaminated mineral oils were initially used: this practice was
understandably abandoned . Many authors confirm that significant decreases in roll wear have been
achieved by use of various special lubricants, although some limitations are reported relating to their use at
high temperatures (above 900oC) and the influence of hydrocarbon fluids in particular, on roll
performance. Several authors however indicated that, in hot rolling of steel, lubrication sometimes
paradoxically increases roll wear [1,34,35]. Schey has discussed some aspects of the health and
environmental risks, associated with lubrication practice in hot rolling 
Williams has discussed the possibility of roll cooling by air and indicated that glass might have attractive
lubricating properties .
2.7. Hot Rolled Steel
The chemistry, mechanical properties and dimensions of the rolled product undoubtedly all have a strong
influence on wear. Particularly rapid wear of calibres occurs during the rolling of alloy steels. This is
partly explained by increase of deformation resistance of the work piece, especially during rolling at lower
Magnee  and Sheasby  have discussed details of the environment of the deformation zone
especially concerning oxide layers present and their effect at rolling temperatures, thus: i) haematite
(Fe2O3) of hardness 1050 HV subjects the roll surface to a severe abrasion; ii) magnetite (Fe3O4)
generates abrasive particles of hardness 450 HV; iii) wustite (FeO), of hardness 300 HV, acts as a
The observed effects of temperature on wear of roll materials via oxide influence may be explained as
follows: i) the oxides formed in region 400 - 600 oC are Fe2O3 and Fe3O4 and increase abrasive wear; ii)
in contrast, within the range of 600-900 oC, the progressive formation of wustite increases the lubricating
effect of the scale and effectively decreases roll wear; iii) at temperatures >900oC, the formation of
magnetite and haematite leads again to increase in abrasive wear .
2.8. Relative Slip and Sliding Distances
Increase of wear loss with relative slip and sliding distance is a proven effect. Detailed studies of metal
flow during rolling of rectangular profiles have revealed the elaborated laws of metal flow in this,
relatively simple, case of rolling [1,2,31]. In calibre rolling, the situation is more complex due to variations
of all wear factors along the groove meridian and the resulting non-uniform wear. For example, the
differences of roll diameter in the axial direction, amplified by variances in area reduction, cause a
significant gradient of relative slip along the groove width. Wear in calibres is increased further by metal
flow, perpendicular to the rolling direction . Various pass sequences have been developed in attempts
to minimise these variations and optimise the roll wear resistance. Figure 4 depicts the variety of roll
designs used in rolling of channel sections .
Last, but not least, it should be noted that the rolled bar is much longer at finishing stage when compared
to the break-down stage of rolling. In roll pass design practice, this is compensated by corresponding
distribution of loads and stresses.
3. METHODS AND MODELS USED IN ROLL
3.1. Aspects of Wear Testing
The 'holy grail' of engineers is a rapid, inexpensive, accelerated laboratory wear test, which will accurately
predict the performance of materials for very specific applications without expensive industrial trials. On
the other hand, scientists are looking for a test method that entirely isolates a given wear mechanism, or
variable, from all others .
The wear literature contains literally hundreds of different machine designs, and it would appear that the
number of "one-of-a-kind" purpose-built machines far exceeds the number of "standard" machines.
Various systematisations are attempted; eg, DIN 50320 classifies wear tests by the type of "tribological
action", which leads to a list of 13 tribo-systems involving sliding, rolling, impact, fretting and abrasion in
various media and motions .
Blau  suggests the following questions when a wear testing device is to be designed: 1. To what use
will the test data be put (basic research, quality control, simulation, materials selection, standard
development)? 2. Does the needed data already exist (Is it cheaper to purchase this data or to carry out
one's own testing)?
Several parameters should be used to measure wear during the same experiment. Generally, wear tests
exhibit significant variations in reproducibility .
Other methods relevant to background data collation are: chemical analyses, mechanical testing at room
and/or elevated temperatures, thermophysical property tests, metallographic examination, corrosion tests
and non destructive tests.
A roll wear test is frequently conducted by measuring the loss of sample weight per unit of surface area
after the sample has been in moving contact with a standard hardened surface under defined conditions
(temperature, pressure, time, speed). The contact might be sliding or rolling with or without abrasive, and
one material can be tested in contact with identical, similar or dissimilar materials. Although such
laboratory wear tests attempt to, but rarely mirror service conditions, they have been used for ranking
materials. Thus, while different types of wear test may permit classification of roll materials, there is
always the need to compare wear test results with performance from the real service life of actual rolls .
A significant number of authors have used a two-discs wear test to simulate roll wear at high temperatures
(Fig. 5). A two-discs apparatus enables a separation of three essential aspects in roll wear investigation:
load, slip and temperature. This device enables simulation of the following ranges of the key variables:
specific working pressure: 50-250MPa; sliding speed: 0-3m/s; temperature of "hot steel" disc: 25-900oC;
thermal cycling options exhibit a satisfactory flexibility .
One of advantages of this two-discs method is the ability to simulate cyclic wear involving long sliding
distances. The degree of similarity with hot rolling is in fact sufficient for many purposes (eg evaluating
roll materials, assessing the role of lubricants, etc). Obviously the costs and running times of such
experiments are much lower than they would be for even semi-industrial trials. However the method is not
without problems: the main disadvantages - 1. the stress state in the roll material is only a rough
approximation to real conditions; 2. the ratio of elastic to plastic deformation of the counter body is much
higher than in the real situation; 3. with increase of plastic deformation there is a corresponding decrease
in the ease of cyclic testing over long sliding distances; 4. Lundberg  has stated that the test roller is
exposed to heat radiation from the heated disc during the heating/idling period, unlike in the real rolling
process; this is, however, because of Lundberg's particular design of hot wear rig: the heating of the 'steel'
disc ∅500mm x 45mm by the propane burner takes ca 30 minutes.
Matuda  and Goto  have conducted a block-on-ring test in addition to the two-discs method. The
temperature of the counter-body materials (stainless steel and carbon steel) was 800oC. The wear of
graphitic cast iron was examined and the following conclusions were reached: i) the coefficient of friction
generally decreased when the amount of graphite in a structure increased, ii) the wear initially decreased
but subsequently increased with the amount of graphite in the microstructure, except when the counterface
material was carbon steel (S45C) under low stress; iii) for carbon steel and low stress conditions, wear
increased regularly with the amount of graphite.
Some authors have examined the wear resistance of roll materials by using a small, laboratory scale, hot
rolling mill (eg one-tenth size of a production mill roll diameter). Such an experimental mill gives the
closest approximation to actual industrial conditions. These tests are of course significantly more
expensive to conduct when compared with other tests [22,24,34]. Matuda  has effectively adopted
these three wear tests to rank roll materials: two-disks, block-on-disc and laboratory hot rolling mill.
Among the scientific methods focused on the assessment of abrasive wear, the scratch test is an example
of an attempt to isolate one wear mechanism from others . Such scratch tests allow studies on abrasion
without the super positioning of other major wear mechanisms such as surface fatigue or adhesion. This
method has been used for the investigation of wear behaviour of hard multiphase metallic alloys subjected
to abrasive wear at elevated temperatures up to 1000 oC .
From the viewpoint of hot roll wear, the two-discs testing device and scratch tester can be considered as
two extremes in experimental approach. The first is the comprehensive method focused on simulating the
real situation as closely as possible. The second technique, single-point scratch tests, enables isolation and
identification of specific wear mechanisms and particular wear factors.
3.2. Wear Models
Lim and Ashby  state that for each of the major wear mechanisms wear rate can be expressed in terms
wi = wear rate, m3/m; F = normal force, N;
v= sliding velocity, m/s; To= initial temperature, oC; M = material properties (eg yield strength, MPa,
Each factor from the equation (1) can be expressed approximately as a vector with n components:
vi = (vi1, ..., vin), .............................. (2)
and each component can be considered as a variable.
There is an obvious necessity to reduce the number of variables in (1) and (2). Theoretical disciplines
contain relations involving some of the above variables. Some of the components can be considered as
discrete variables, while others can be assumed as approximately constant values. Figure 6 summarise the
factors influencing the roll wear.
According to published knowledge, this problem has been approached by whole range of wear models
from simplified, general wear equations to "comprehensive" formulae for special cases of hot wear of roll
materials. Here the criteria for optimising the number of arguments in any model can be found in
Archard-Holm's equation for worn volume during sliding-adhesive wear is an example of a fundamental
W = KFL(Rp)-1........................................ ....(3)
W = worn volume, mm3; K= wear coefficient;
Rp= proof stress, MPa; F= normal force, N; L= sliding distance, mm.
In the open literature can be found more comprehensive attempts to model special cases of hot roll wear.
Thus Felder  has suggested that the wear mechanism (abrasion or thermal fatigue) in rolling depends
on a relationship involving the shear factor m (defined in ) and the deformation zone geometry.
Magnee  has proposed the following model:
AD2(v)[(W/Wc) - 1].........(4)
= wear in kg/m2
A = an intrinsic constant of the roll material
D2(v)= quadratic function of speed, associated with the kinetic energy of the process;
W = strain energy in a roll material induced by displacement of the friction force [J]
Wc= critical energy [J]; when W>Wc, then wear phenomenon occurs.
W/Wc= f1(M,F,To); D2(v)=f2(Vo,Ve)
Vo; Ve, = hot steel velocity; respective, roll peripheral velocity, m/s.
Variations of the wear rates in groove rolling can be calculated by introducing in Magnee's model the
relevant values of Vo, Ve, To and F corresponding to their variations along the groove meridian. Magnee
has demonstrated how parameter A from equation (4) can be used to evaluate roll materials, especially in
case of Cr-Mo addition level. A propitious change of roll material, characterised by significant decrease of
parameter A, will result in the decrease of wear.
In model (4) the bulk stress state and stress history of the roll material have not been taken into a count.
4. A RESEARCH PLAN AND PILOT TESTS: RATIONALE AND DETAILING
4.1. Research Proposals
Critical consideration of the foregoing review suggests that a useful main focus of research would concern
hot abrasive wear of grooved rolls, although interactions with other wear modes should also be evaluated
within a structured programme. The following aspects are worthy and appropriate to be addressed:
i) building a logical structured concept (eg, wear rate criteria, definitions, dependent and independent
variables, limitations); ii) categorisation of roll wear factors, permitting the focusing of the research onto
most significant and feasible parameters; iii) analysis and selection of the methods to be used for simulated
roll wear studies; iv) a laboratory experimental program assessing the effect of materials, load, roll stress,
slip ratio and temperature; v) the upgrade of Magnee's model  for roll wear above 1000 oC and the
relative slip below 0.3 m/s via statistical analyses of the data gained during the step iv); vi) the correlation
between roll pass design and roll wear.
A first hypothesis in this research program is that Magnee's model, defined by the equation (4), describing
hot roll wear up to 900 oC, can be extended to temperatures above 1000oC and to the relative slip below
0.3 m/s without the appearance of cuspidal points. Testing of this hypothesis will be realised by means of
hot wear rig via statistical analysis of the effects and interactions of temperature, normal load and relative
slip. The stochastic approach will improve approximation to real hot wear situation when compared with
original deterministic form of the Magnee's model.
Following the analogy with stress corrosion and cyclic softening of steel materials, a second hypothesis to
be tested in the program is that the bulk stress state and stress history have significant influences on
abrasive wear. The effect of bulk stress on abrasive wear can be evaluated by means of a single point
scratch tester. Here exists the potential to examine additional aspects of the interaction with oxide films.
A third hypothesis is that the various adopted pass sequences of rolling, published in the literature, in fact
are associated with optimised roll wear resistance, because these pass sequences were developed during
centuries of trial-and-error industrial rolling practice and corrected many times towards the same aim of
minimised roll wear. In this context it is believed that there is a correlation between the parameters of roll
pass design and the non-uniform wear of the grooved rolls. Testing of this hypothesis would exceed the
resources foreseen at this stage and requires a separate research programme.
Within this overall rationale it has been decided to use two test methods: i) two-disks hot wear rig and
ii) scratch tester. Two types of roll materials are to be assessed: 1) a material already used in rolling
practice with known wear resistance in industrial application: adamite, and 2) high speed tool steel, which
may be expected to have better wear resistance than adamite.
4.2. Preliminary Experiments
Some preliminary experiments were carried out using a scratch tester. Hypo and hyper-eutectoid steel
specimens were bent to obtain tensile or compressive stress and then scratched in situ under a normal load
of 25-40 N. In harmony with the phenomenon of increased plasticity of a material exposed to a tri-axial
compressive state, the scratches produced on surfaces under compression exhibited a higher level of
plastic deformation, when compared with scratches made on surfaces exposed to tension. The scratches on
the surfaces under tension were deeper when compared with scratches made on surfaces under
compression or unstressed surfaces. A tensile state favours the micro-cutting abrasive sub-mechanism and
that can be accounted for the observed increase in scratch depth of more than 15%. The domination of
micro-ploughing at surfaces under compression favours strain hardening of the worn layers. Further
research is proposed on the effects of bulk temperature of the worn material; .
Testing of adamite (1.8%C) by means of two-discs tester (steel disc was K1054) has been commenced
involving variations of the normal force (60 - 160N) and relative slip (0.2-0.6m/s) at temperatures between
760 and 960oC. The sequential experiments are conducted to evaluate the resolution capability for two
levels of force: 60N and 100N at 860oC and slip 0.5 m/s. With 10 runs the testing method has enabled the
measurement of the wear gradient with 80% confidence. This confidence limit can be increased by
increasing the number of runs or test duration. A further series of experiments was conducted at two level
of temperatures: 860oC and 1000oC with constant force of 100N and slip of 1m/s. After 20 runs it was
concluded with 90% confidence, that wear does not decrease with temperature increase from 860oC to
1000oC. Other authors [7,41] claimed that wear decreases with temperature increase from 700 to 900oC.
A separate group of runs, made with clock-wise rotations of both discs, was compared with a situation
where the roll disc and steel disc were rotated in opposite directions. The treatment with clock-wise
rotations of both discs exhibited significantly higher wear rate. This phenomenon explains the increased
wear in some calibres during the section rolling, where the particular components of metal flow have the
direction opposite to the corresponding components of the roll velocity.
Further tests are programmed by means of factorial design, to evaluate the significance and interactions of
main wear parameters.
Steel rolling is recognised as one of the most important industrial processes. Rolling using grooved rolls
(as a category different from groove-less rolling) is the most common practice in production of steel
sections. Key tools in this process are the rolls that contribute up to 15 % of production costs. A main
cause of roll consumption is due to continuous wear, complex process where mechanical and thermal
fatigue combines with impact, abrasion, and corrosion. The necessity to compensate for the non-uniform
wear during machining is an additional aspect of roll consumption. An area of specific interest is
concerned with abrasive wear within the environment of rolling in grooves, where the nature of the
deformation zone can accelerate roll surface deterioration.
The published research into roll wear is mainly concerned with the effects of roll material, oxide scales,
temperature, normal force and sliding velocity. Though a selective application of numerous roll materials
is decisive for roll consumption, there is no general index for determining their resistance to wear. Several
authors have examined how temperature has affected the wear of various roll materials via corrosion,
formation of oxide films and fatigue. Detrimental influences of the rolling load and relative slip on hot
wear are clearly proven. On the other hand, the published works give very little, if any, information on the
possible ranking and interactions of wear factors. In addition, there is a lack of knowledge about the
influence of bulk stress in roll material on abrasion.
The published wear models indicate that deterministic approach dominates in applied mathematical
definitions. Wear equations are based on deterministic relations of heat transfer, diffusion, plastic
deformation and tribology. These published models of roll wear are mainly focused on the relatively
simple case of flat rolling and there is no information relating to possible statistical interactions and
variations in wear. A literature search has revealed no stochastic approach in modelling wear.
Roll pass design exerts vital influence on the non-uniform wear of rolls via control of loads, stresses and
relative slip. In addition, the groove wall inclination has a direct influence on the radial machining required
to regenerate the groove geometry. The metal flow in grooves is determined by the interactions of
subsequent stages of the whole process. An important question is: why does a particular location on the
roll experience more wear than others, when the whole surface of the roll is made from the same material?
The pilot experiments carried out by means of scratch tester have proven that the abrasive wear differs
depending on wether the specimens are under tension or compression. Steel material exposed to bulk
tensile stress shows a decreased resistance to abrasive wear.
In experimental testing of roll wear, the two-discs devices are widely used in various configurations. There
are certain advantages such as extended test duration and stability within the Hertzian stresses. However, it
has been reported that there are some limitations: the method simulates hot wear in flat rolling only and
there are certain difficulties regarding the use at temperatures above 1000oC. Pilot trials have confirmed
the device reliability, especially for ranking of materials by exposing them to identical conditions of hot
wear simulation. With the recent method improvement, the experiments above 1000oC appear to be
The use of other techniques is still widely reported. To improve the reliability of ranking roll materials, at
least two separate testing methods are to be applied, along with the appropriate standard metallurgical
The work was supported by the BHP Steel Long Products Division, Whyalla South Australia, and BHP
Research – Melbourne Laboratories, Clayton, Victoria. We also than Study Adviser Ms Trish McLaine,
who corrected grammatical errors and provided critical evaluation of the text.
 J. A. Schey, Tribology in Metalworking - Friction, Lubrication and Wear, ASM, Metals Park , 1983
 M. Chausevich, Rolling of metallic materials, Veselin Maslesa, Sarajevo, 1983
 R. E. Beynon, Roll Design and Mill Layout, Association of Iron & Steel Engineers, Pittsburgh, 1956
 R. B. Corbett, Rolls for the Metalworking Industries, Iron and Steel Society, Warrendale, 1990
 M. Hashimoto et al., Nippon Steel Technical Report, No.48 (1991) p. 71.
 S. E. Lundberg, Journal of Materials Processing Technology, 36 (1993) p. 273
 A. Magnee C. Gaspard, M. Gabriel, C.R.M. Report, No.57 (1980) p. 25
 J. C. Thieme and S. Ammareller, Walzwerks-walzen, Climax Molybdenium, Zurich, 1965
 A. P. Chekmarev , Proizvodstvo Oblegchennikh Profilei Prokhata, Metallurgya, Moskva, l965
 ***, Wear resistant surfaces in engineering; a guide to their production, properties and selection,
Department of Trade and Industry, Her Majesty's Stationery Office, London, 1986
 K-H. Zum Gahr, Microstructure and Wear of Materials, Elsevier, Amsterdam, 1987
 R. V. Williams, and G.M. Boxall, Journal of The Iron Steel Institute, Volume 203 (1965) p. 369
 V.B. Ginzburg, Steel Rolling Technology; Theory and Practice, Marcel Dekker Inc., New York, 1989
 S. Jahanmir, in N.P. Suh (ed.), Fundamentals of Tribology, Proceedings of the International
Conference, The MIT Press, London, 1981, p. 455
 I. M. Dugan, Factors Which Influence Roll Specifications, Association of Iron and Steel Engineers,
 P. Harper, (1988) "The Water Cooling of Rolls." Iron & Steel M., No 34 (1988) p. 3
 A. Ohnuki, K. Hasuka, K. Nakajima, T. Kawa-nami, Advanced Technology of Plasticity, Vol. I,
 O. Kato, and Kawanami, T. Journal of Japan. Soc. Technol. Plast., Vol. 28 (1987) p. 264
 K. H. Schroeder, and W. Eilert, The 2nd International Conference on Steel Rolling, Dusseldorf, 1984
 J. C. Werquin, and J. Bocquet, Der Kalibreur, Heft 48, (1988) p. 31
 *** Catalogues: Aakers, Davy Rolls, Kubota, Goentermans Peiper, Hitachi, Midland Rollmakers
 U. Matuda, K. Sakamoto, Y. Sugimoto, K. Goto, T. Hino and M. Unno, Basic Characteristic of Roll
Material in Hot Rolling, CAMP-ISIJ, 4 (1991) p. 438
 T. Kudo, H. Okura, T. Koizumi, T. Kawashima, T. Kurahashi, Basic Characteristic of Roll Material
in Hot Rolling, Camp ISIJ, Vol 4, (1991) p. 442
 M. Ooshima, Y. Sugimura, Y. Sano Mechanical Working and Steel Processing Proceedings,
Cincinnati, 1990, p. 31
 M. B. Peterson and W.O. Winer, Wear Control Handbook, The American Society of Mechanical
Engineers (ASME), 1980
 F. Camplani, Der Kalibreur, Vol. 52,1990, p. 37
 K. Suzuki, K. Takahashi, T. Nishi, H. Kohira, M. Hori, Transactions ISIJ, Vol. 16 (1975) p. 106
 W. H. Cubberly, P. M. Unterweiser, D. Benjamin, C. W. Kirkpatrick, V. Knoll, K. Nieman, (ed.)
Metals Handbook Ninth Edition, Vol. 3;
Properties and Selection: Stainless Steels, Tool Materials and Special Purpose Metals , American Society
for Metals, Metals Park, 1980
 J. Zhang and Z. Wu, Journal of Ching Hua University, V 26, No. 6, (1986) p. 86
 S. Karmakar, A. Sethuramiah, Wear of Materials, Volume II, ASME (1991) p.193
 Z. Wusatowski, Fundamentals of Rolling Pergamon Press, Oxford, 1969
 I. V. Kragelsky, V. V. Alisin, Friction Wear Lubrication - Tribology Handbook , Mir Publishers,
 A. Ohnuki, Japan Society for Technology of Plasticity Vol 25; 285(1984) p.936
 J. Kihara, K. Doya, K. Nakamura, The 2nd International Conference on Steel Rolling, Dusseldorf,
 W. L Roberts, Hot Rolling of Steel, Marcel Dekker Inc., 1983
 J. S. Sheasby, W. E. Boggs and E. T. Turkdogan, Metal Science, Vol.18 (1984) p.127
 P. J. Blau, Friction and Wear Transition of Materials; Break - in, Run - in, Wear -in, Noyes
Publications, Park Ridge, 1989
 K. Goto and T. Mase, Journal of the Iron and Steel Institute of Japan (1991) p.107
 H. Berns, A. Fisher, J. Kleff, Wear of Materials, Volume II, ASME, (1991) p. 661
 S. C. Lim, and M. F. Ashby, Acta Metallurgica, Vol 35, No. 1, (1987) p. 1
 E. Felder, Revue de Metallurgie (1984), p. 931
 S. Spuzic, C. Subramanian, K. N. Strafford, 2nd International Conference on Interface, Ballarat,
Figure 1: Wear of the ∅ 650 mm rolls for 200 mm channel section 
Figure 2: Schematic diagram of the four main wear mechanisms 
Figure 3. Tensile strength and hardness of various roll materials
Tensile Strength MPa
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Grain Cast Iron
Carbon wt %
Fig. 4. Roll pass designs for channel sections 
Fig. 5. Schematic illustration of the disc-on-disc wear test
∅ 100 ∅ 40
All dimensions in mm
Fig. 6: Factors influencing hot roll wear