Journal of Materials Processing Technology 153–154 (2004) 596–602

Weldability of austenitic manganese steel

J. Mendez, M. Ghoreshy, W.B.F. Mackay, T.J.N. Smith, R.W. Smith∗
Department of Materials and Metallurgical Engineering, Queen’s University at Kingston, Kingston, Ont., Canada K7L 3N6

Abstract

Hadfield’s manganese steel, nominally Fe–1.2%C–13%Mn, is an alloy of inherent toughness, work-hardening characteristics and
excellent resistance to some types of adhesive and abrasive wear. However, due to its low yield strength, it may be deformed markedly
before its work-hardening become effective. In certain applications, such as railroad crossings and rock-crushers, this can be a disadvantage.
In practice, when this deformation becomes excessive, welding is employed to restore the casting to its original dimensions. During welding,
precautions have to be taken to avoid overheating and the attendant carbide precipitation which may lead to subsequent early failure.
Three different electrode compositions were used to overlay-weld austenitic manganese steel cast in the form of rail heads. Two of the
electrodes were obtained commercially and the third was of novel chemical composition and was produced in our laboratory. Mechanical
tests were then carried out to simulate the battering deformation likely to result from in-service exposure. The procedure highlighted the
work-hardening characteristics and resistance to plastic flow of the weld deposit and base material, one of which consisted of the standard
Hadfield’s alloy whilst two others had minor transition element additions. The electrode containing molybdenum produced a weld overlay
which showed better work-hardening characteristics and deformation resistance than those of the other two commercial electrodes studied.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Austenitic manganese steel; Weldability

1. Introduction appropriate welding rod compositions in order to achieve

longer service life under the increasingly severe operating
Hadfield’s manganese steel, with a composition of conditions.
Fe–1.2%C–13%Mn, normally has a structure of metastable Welding electrode compositions have been the subject of
austenite which is obtained by water-quenching the steel research since the early 1920s when one of the first electrode
from an annealing temperature of 1050 ◦ C. This austenitic patents for manganese steel was issued to Churchward [1]
alloy work-hardens rapidly under repeated impact and dis- with the compositions of 1.0–l.25% C and 3.0–13.0% Mn.
plays remarkable toughness. This property makes the steel Other patents have been issued since then but the general
very useful in applications where heavy impact and abrasion trend has been to reduce the carbon content and add some
are involved, such as within a jaw crusher, impact hammer, nickel to help avoid martensite formation. The smaller car-
rail-road crossing (frog), etc. bon content was intended to dilute the relatively high carbon
Due to the low yield strength of unalloyed manganese content of the partially fused parent metal and so reduce or
steel, when used for rail-road components such as frogs, prevent carbide precipitation which could lead to embrittle-
points and crossings, significant deformation may accrue un- ment if the frog was not to be heat-treated after welding.
der the enormous impact loading present during rail-road However, the effects of a number of other alloying additions
service. This causes undesirable dimensional changes to oc- to the welding electrodes have been studied, e.g. Cr, Ni and
cur. In practice, rail grinding is used to maintain the geom- even increased Mn [2]. In particular, molybdenum additions
etry of the component but, eventually, overlay welding is are claimed to produce weld metal that is modestly superior
employed to restore the original dimensions. However, the to that with nickel additions at the same carbon level. But
weld repair of a worn frog is expensive and incurs consid- this superiority would go unnoticed except in applications
erable traffic dislocation. Hence the search for a modified where the higher yield strength associated with the presence
Hatfield alloy and an improved rebuilding procedure using of molybdenum is utilised.
To avoid carbide precipitation during welding, it is imper-
ative to keep the temperature of the steel below 300 ◦ C [3].
∗ Corresponding author. Due to the very small thermal conductivity of Hadfield’s
E-mail address: smithrw@post.queensu.ca (R.W. Smith). steel, the temperature in the base metal near the welding

0924-0136/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmatprotec.2004.04.033
 J. Mendez et al. / Journal of Materials Processing Technology 153–154 (2004) 596–602 597

zone could exceed 300 ◦ C during welding. Therefore all

welding procedures should be such as to maintain local
temperatures below 300 ◦ C. This makes arc welding the
only recommended process for welding and hard facing
manganese steel, because of the relatively short period of
heating involved. According to Avery et al. [4], to prevent
embrittlement of the base metal, the temperature of the
zone 12 mm from the weld should not exceed 250 ◦ C.
The present work was undertaken to determine the
changes in the microstructure of the weld and heat-affected
zone of various Hadfield’s steel for which the com-
position had been modified slightly. Molybdenum- and
nickel–chromium-bearing electrodes were used to verify
the superiority of the former in the mechanical properties Fig. 1. Cutaway view of railhead sand mould.
of the weld deposit as claimed by Avery et al. [4]. A third
electrode bearing chromium and similar to the one used by
only the head of the rail was cast as rail component for this
Canadian Pacific Electrode [5] for building up worn frogs
welding study.
and crossings, was used also.
Olivine sand was selected as a old material due to its
In order to quantitatively test the various electrodes and
low reactivity properties for austenitic manganese castings
base metal combinations, two sets of apparatus were con-
[8]. To obtain sound castings, it was necessary to use an
structed. A rail/wheel impact simulator [6] was designed and
insulating riser sleeve.
constructed to apply a repeated impact to the specimens. A
Three quartz tubes 4 mm I.D. were arranged in the sand
second piece of equipment was constructed in order to pro-
in such a way that their closed end were located at 6, 12
duce massive deformation in the test specimen of base metal
and 18 mm, respectively, from the surface of the rail to be
and weld deposit by dropping a known weight on the sample
welded, as shown in Fig. 1. These tubes provided the holes
from a fixed height.
for thermocouples to be inserted to record the temperature at
The effects of these two methods of testing on the
the given location during welding at the surface of the rail.
work-hardening characteristics and the resistance to plastic
Before welding, the as-cast rail heads were first austenitised
flow of the alloys were investigated.
at 1150 ◦ C for 2 h [9] in a controlled argon atmosphere to
avoid decarburisation [10,11] and then quenched in water.
The use of a submerged arc as the welding process made
2. Experimental procedure it necessary to use electrodes of the wire type. Table 2 shows
the electrodes that were available commercially in 3 mm di-
The raw material used in alloy preparation was low car- ameter size. The other was the molybdenum electrode which
bon rail-stock to which were made alloy additions to obtain had been produced in quartz tube of 4 mm I.D. This has
the desired nominal compositions, as shown in Table 1. The been done by forcing liquid metal into the quartz tube by
charge was melted in an induction furnace and cast into a means of vacuum. The cast rods were welded autogenously
dried sand old from a pouring temperature of 1450 ◦ C. The (no filler metal) to each other by gas-tungsten welding to
old pattern consisted of a rail head of 4.2 in height, 12.5 cm provide sufficient electrode length for a 17.5 cm weld pass
in length and 6 cm in width. Due to the low thermal con- along the full length of the rail head. An automatic sub-
ductivity of austenitic manganese steel [7], it was unneces- merged arc welding machine was used in the experiment.
sary to cast any of the web section of a standard rail. Thus, To obtain suitable welding conditions, the following criteria
were established:
Table 1 (a) the temperature of the zone beyond 12 mm from the
Nominal and actual alloy compositions for railhead casting (wt.%) weld surface should not exceed 300 ◦ C, and
Classification C Mn Mo V (b) the weld pool should be approximately 2 cm wide.
Standard Hadfield (R3) 1.2 12 – –
Standard Hadfield (R3)a 1.19 12.15 – – Table 2
Low carbon–1%V (R9) 0.8 12 – 1 Nominal composition of the welding electrodes
Low carbon–2%V (R7) 0.8 12 – 2 Classification C Mn Cr Ni Mo Si Fe
Low carbon–2%V (R7)b 0.82 12.8 – 1.93
Low carbon–1%Mo (R10) 0.8 12 1 – Chromiuma 0.23 16.54 16.61 0.91 0.05 0.56 65.1
Nickel–chromium 1.0 13.7 4.7 3.7 – 0.4 76.5
aModified Hadfield’s steels. Molybdenum 0.8 13 – – 1 – 85.2
bActual analysis results. Total other elements is less 0.5% and the
a Canadian Pacific Electrode [4].
remainder is Iron.
598 J. Mendez et al. / Journal of Materials Processing Technology 153–154 (2004) 596–602

Table 3 The other testing equipment used was the Weight Drop
Welding parameters for the two electrodes Machine. This device was used to study the effect of massive
Diameter Volt Ampere Burn-off Traverse Deposition and rapid deformation on the weldment. For this test, a 72 kg
(mm) rate speed rate load was dropped from a height of 2.5 m onto a specimen
(cm/min) (cm/min) (g/min) which had been glued on an anvil.
2.78 43 105 109.2 11.4 40
4 43 105 44.5 9.1 38.9

3. Results and discussion

Using the 2.78 and 4 mm diameter electrodes, a number of
weld beads were laid on the rail heads at different burn-off 3.1. Thermal analysis
rates and traverse speeds. Also for every welding pass made
the temperature in the rail head at 6, 12 and 18 mm from Due to the very low thermal conductivity of austenitic
the welded surface were recorded using a three channel manganese steel (7.6% of the thermal conductivity of a low
chart recorder. The welding parameters used are shown in carbon steel) [7], it was important to evaluate the maxi-
Table 3. mum depth of the rail head which was unaffected by the
The sequence of sample preparations for post-welding “informal” heat-treatment of overlay welding height from
studies is shown in Fig. 2. Deformation studies were carried the surface. The results from the temperature measurement
out in two ways. The rail/wheel impact simulator, Fig. 3, are shown in Fig. 4.
was used in an attempt to simulate in the specimen the de- Fig. 4 shows the maximum temperature reached in the
formation of a frog under service conditions. The specimen thermocouples located at various defects from the upper sur-
was placed in the sample holder and then mounted in the face of the rail. Although the temperature recorded by the
fly-wheel of the machine, Fig. 3b. last thermocouple was lower than the values of the steady
The fly-wheel was rotated at 30 rpm. Spacers were used state region, this did not influence the outcome of the tem-
under the specimen to change the height of the sample after perature survey. The plateau in the temperature suggests that
a given number of impacts. The maximum specimen height a steady state condition was reached and maintained for ap-
above the surface of the sample holder was 2 mm for base proximately 7 cm of rail.
metal and 3 mm for weldment, and the static applied load
was 22,370 N. 3.2. Compression test
During the test, the specimen was periodically removed
from the equipment and the overall length measured. The Fig. 5 shows the results obtained from the compression
difference in length was taken as a measure of the amount tests on the samples from the base metal using the rail-
of plastic flow occurring under the conditions and duration way simulator. The values shown in the figure are the aver-
of testing. age of four specimens with a typical standard deviation of
0.14–0.28. Both the alloys R7 and R9 proved to be the most
resistant materials to plastic flow of those tested. The stan-
dard Hadfield’s steel (R3) appeared to have a deformation of
approximately 25% over the strongest material tested, R7.
Specimen R10 showed the least resistance to deformation of
all the steels tested. At the end of the test (240 impacts with
railway simulator), the surface hardness of the deformed
sample was recorded and the average value for each speci-
men calculated, Table 4. Specimen R3 appeared to have the
best work-hardening characteristic, followed by R10. This
might be due to little resistance of these alloys to plastic flow
in comparison with R7 and R9. Microstructural examination
of deformed samples of R3 and R10 showed no significant
differences between the two alloys. This indicates that the
addition of 1% molybdenum to the low carbon Hadfield’s

Table 4
Average hardness of Hadfield’s manganese alloys
R3 R7 R9 R10

Hardness (Rc)a 49 45 45 47
Standard deviation 1.41 1.11 0.75 1.11
a Average of seven hardness determinations.
Fig. 2. Specimen preparation for work-hardening and plastic deformation.
 J. Mendez et al. / Journal of Materials Processing Technology 153–154 (2004) 596–602 599

Fig. 3. Rail/wheel impact simulator: (a) general view; (b) specimen location on flywheel marked ‘S’.

steel (R10) does not alter the formation of twins apprecia- by the molybdenum electrode rose rapidly and continued
bly, Fig. 6. The microstructural information is in accordance to work-harden slowly with increasing number of impacts.
with the similarity in the plastic deformation characteristics In contrast, there was no significant differences in the
exhibited by these alloys, Fig. 6. The microstructure of R7 work-hardening characteristic of the nickel chromium and
and R9 in which V was added to the low carbon Hadfield’s chromium deposits, although the former showed slightly
steel shows a sharp decrease in the number of twins; al- lower hardness values consistently. The maximum hard-
though both alloys display many, the dendritic structure and ness of both weld deposits was unaffected by continued
its associated carbide segregation, together with the forma- testing.
tions of twins seems to account for high resistance of these The evaluation of the plastic deformation characteristics
alloys to plastic deformation, Fig. 7. of the weld deposit as well as that of the weldment was car-
The result of compression testing on the weldment speci- ried out on 5 mm ×10 mm ×20 mm specimens, Fig. 2, using
mens (weld deposit and base metal combined) as they were a maximum height of 2.5 mm above the sample holder and,
subjected to an increasing number of impacts is shown in as noted earlier, a maximum static applied load of 22,370 N.
Fig. 8. The hardness value shown in this figure are the
average of seven hardness determinations with a typical
standard deviation of 0.39–2.67. In the “as-welded” condi-
tion, the molybdenum deposit was slightly harder than the
nickel–chromium and chromium weld-rod deposits. The
same trend was observed throughout the duration of the
test. However, the hardness of the weld metal deposited

Fig. 4. Welding temperatures in a railhead. Fig. 5. Impact compression of Hadfield’s manganese alloys.
600 J. Mendez et al. / Journal of Materials Processing Technology 153–154 (2004) 596–602

Fig. 6. Deformed low carbon–1%Mo modified Hadfield’s manganese steel

(R10) (50×).

Fig. 8. Work-hardening characteristics of weld deposits.

of the molybdenum deposit on the four different steels ap-

peared consistently better than the other steel companies.
These showed a minimum deformation of 16.8% on R10
and a maximum of 25.3% on R7 as compared to 20–28 and
19–30% deformation for the Ni–Cu and Cr deposits, respec-
tively.
Fig. 9a–c consists of photomicrographs of deformed
weldment specimens showing the weld interface of the
different electrode composition on alloy R7. All the weld
Fig. 7. Deformed low carbon–2%V modified Hadfield’s manganese steel
deposits exhibited a cellular-dendritic structure and approx-
(R7) (50×). imately the same amount of deformation twinning. The
nickel–chromium and chromium deposit show relatively
larger inclusions located mainly along the grain bound-
The results of testing all the weld deposit steel combina-
aries. Also the latter shows larger grains. These two effects
tions are summarised in Table 5. The compression values
together may be the reason for the low plastic flow resis-
reported in this table are the average of four specimens with
tance of the chromium weld deposit. On the other hand, the
a typical standard deviation of 0.16–0.32. After the test pe-
molybdenum weld deposit, Fig. 10, shows smaller inclu-
riod (360 impacts), the plastic deformation characteristics
sions which appear to be more uniformly dispersed in the
matrix than along the grain boundary. This and the defor-
Table 5 mation twins may account for its excellent work-hardening
Average compression values of the weldment and weld deposit (%) characteristic and resistance to plastic deformation.
Hadfield’s manganese alloys Comparison tests were also carried out on the weight drop
R3 R7 R9 R10
device to study the effects of massive and rapid deformation
on the work-hardening characteristics of the weld deposits.
WM WD WM WD WM WD WM WD At the end of the test, the surface hardness values of the de-
Molybdenum 9.3 18.6 8.5 25.3 8.7 23.0 10.2 16.8 formed weld deposit were recorded for each electrode com-
Nickel–chromium 9.6 20.0 8.9 28.0 9.3 26.2 11.0 20.1 position. These showed a drop in average hardness of each
Chromium 10.6 25.1 8.8 30.3 9.4 27.1 10.5 19.9 weld deposit as compared with the average hardness values
WM: weldment and WD: weld deposit. obtained by the rail/wheel impact simulator. A decrease of up
 J. Mendez et al. / Journal of Materials Processing Technology 153–154 (2004) 596–602 601

more or less the same as that observed by using the rail/wheel

impact simulator. Thus, it appears that rapid, massive de-
formation does not allow the structure to fully develop its
work-hardening potential.
This is particularly significant as Kotechi and Rajan [2], in
a study of the influence increased Mn and Mn–Cr contents on
the work-hardening capacity of the weld deposit, report that
a drop-weight test was found to offer better discrimination
than standard tensile or hardness testing. It should be noted
that these are effectively single load applications, whereas
the industrial situations in which austenitic manganese steels
are used typically involve repeated applications of load, e.g.
a rail wheel on a frog, a hammer crushing rock.
The subject of this paper has been the rebuilding of
austenitic manganese rail track components, although other
applications can be considered. However, it is difficult to
find economic data for such rebuilding, and the extent to
which the installation of a frog with improved properties
might be cost-effective.
Dahl et al. [12] report that, in Sweden, the cost of the
weld repair of a frog is approximately 20% compared to the
installation of a new frog. They also report on their optimum
method of laying down the weld overlay.
Often the frog is bolted to the adjoining rails for conve-
nience and to avoid the carbide embrittlement of the frog if
it is welded in place. However, Bartoli and Digioia [13] sug-
gest how such welding attachments may be achieved with-
out embrittling the frog.
A question still to be answered by the rail carrier indus-
Fig. 9. (a) Deformed weld interface of nickel–chromium weld deposit on
try is what work-hardening capacity and impact strength is
R7 (50×); (b) deformed weld interface of chromium weld deposit on R7 adequate for a heavy haul railway frog; since axle loads ap-
(50×); (c) deformed weld interface of molybdenum weld deposit on R7 pear to gradually creep-up, ‘as high as possible’ seems to
(50×). be the answer!
Of particular interest to the foundryman is the extent to
to 20% in hardness was observed in the molybdenum weld which the purchaser of (say) frogs is likely to pay an ap-
deposit. Similarly, a drop in hardness of 10% was observed propriate premium for superior frogs. In Canada, most cer-
in both the chromium and the nickel–chromium deposits. tainly, CP rail would like to be able to install a frog with
A microstructural analysis of the samples deformed with a 50% increase in the service life from the present figure
the weight drop device shows significant twinning but not of 140 million gross t. The most expensive alloying addi-
as much as expected since the amount of deformation twin- tions used in our Hadfield’s Steel Development Programme
ning developed in the microstructure using this test would be to date would increase the charge cost/frog by 200–300%.
However, many of the elements are now relatively abundant,
e.g. tungsten, and so the price would fall dramatically with
increased consumption. A detailed cost benefit analysis of
using a 200 million gross t frog, compared with the present
version would be interesting. It is clear that the initial casting
price is a small part of the total cost of a frog, i.e. purchased
casting, straightening, grinding off the decarburised layer,
explosive hardening, fitting, installation, rebuilding (usually
twice), and finally replacement.

4. Conclusions

1. It was shown that the plastic deformation and work-

Fig. 10. Deformed molybdenum weld deposit (500×). hardening characteristics of the molybdenum weld
602 J. Mendez et al. / Journal of Materials Processing Technology 153–154 (2004) 596–602

deposit were significantly better than those of the weld References

metal deposited using commercial nickel–chromium and
chromium electrodes. [1] J. Churchward, US Patent No. 1,377,543 (1920).
2. There was no significant difference between the [2] D.J. Kotechi, V.B. Rajan, Weld. Res. Suppl. (July 1998) 293–
nickel–chromium and chromium weld deposits with 298.
[3] Roll Crusher Maintenance, Rebuilding, and Repair, Pit and Quarry,
respect to work-hardening and plastic flow upon defor- October 1970, pp. 104–106.
mation. [4] H.S. Avery, et al., Weld. J. 33 (5) (1954) 459–479.
3. The low carbon–1%V and –2%V Hadfield steel ex- [5] E. Taylor, Private Communication, Canadian Pacific Railway, 1981.
hibited excellent resistance to deformation, but low [6] T. Smith, B.Sc. Thesis, Department of Mechanical Engineering,
work-hardening characteristics. Queen’s University, Kingston, Ont., Canada, February 1982.
[7] N. Tsujimoto, AFS I.C.M.J. 4 (2) (1979) 62–77.
4. The work-hardening characteristics of the weld deposits [8] C.W. Farrar, J. Iron, Steel Inst. 202 (1964) 543.
were appreciably reduced when subjected to massive de- [9] W.B.F. Mackay, R.W. Smith, CIGGT Report No. 82-6, Queen’s
formation. University, Kingston, Ont., Canada, October 1982.
5. It was found when using the molybdenum bearing [10] S.B. Sant, R.W. Smith, J. Mater. Sci. 22 (1987) 1808–1814.
electrode that the extent of deformation of the base [11] C.H. White, R.W.K. Honeycombe, J. Iron Steel Inst. 200 (1979)
457.
metal in order of increasing deformation was (1) low [12] B. Dahl, B. Mogard, B. Gretoft, B. Ulander, In situ railtrack repair
carbon–1%Mo, (2) standard Hadfield’s steel, (3) low and reclamation by welding, Weld. Rev. Int. (February 1996) 19–
carbon–1%V and (4) low carbon–2%V. However, the 22.
overall measured deformation of the weldments was less [13] M. Bartoli, M. Digioia, Manganese steel castings: new technology
in this case than when using the nickel–chromium and for welding frogs to rail, Transport. Res. Rec. 1071 (1986) 39–
43.
the chromium electrodes.