Thermit welding

of chromium-vanadium

rail steel

Thermit welding of chromiumvanadium rail steel

H. D. Fricke (U.S. Thermit Inc.)

Thermit welding techniques are' discussed with particular reference to short preheat cycles and its suitability to the welding of high-strength chromiumvanadium rails. The composition of the Thermit steel used is adjusted to match the phase-transformation characteristics of chromium-vanadium steels. To avoid a martensitic or bainitic structure in the finished weld, the cooling cycle must be controlled.

1. INTRODUCTION
Thermit welding has been applied as a regular welding process since 1898, when the aluminothermic reaction between a metal-oxide and aluminum was discovered by Hans Goldschmidt. This type of reaction is characterized by its exothermic nature. Due to the difference of free energy between a metal-oxide and aluminum-oxide, sufficient heat is generated to produce liquid steel or any other metal and alloy without applying energy from outside. All components for the aluminothermic process are used as granules or powders. Once the reaction is initiated by an external heat source, the aluminum reduces the oxygen from the metaloxide, which will result in a liquid superheated metal and aluminum-oxide (AI203)' Since these two components are different in density, they separate automatically and the liquid metal can be utilized for different welding applications, or just for the production of special metals or alloys. Thermit welding had its most important large-scale technical application during the early parlofthis century as a

process for joining heavy cross-sections for machinery, millhousings, rudderstocks, stern frames and pinions, but the welding of rail sections has become more and more important with the widespread installation of continuous welded rail. While the majority of rail steel is still rolled according to the AREA or VIC specifications, the continuing development towards faster speeds, higher axle loads and increased train frequencies has led to increased requirements for rail steels with improved mechanical properties. The common carbon-manganese rail steels do not satisfactorily withstand these parameters and show gross plastic deformation and abrasive wear after relatively short periods of time. This has led to the development of special alloyed rail steels with higher yield and tensile strength. During the course of development of these alloyed rail steels, one major demand has always been taken into consideration, that is weldability by both the in-plant welding as well as field welding processes such as Thermit welding.

2. RAILSTEELTYPES
Since the time the first alloyed rail sections were rolled by various steel mills, extensive research has been done in our metallurgical laboratories to develop Thermit portions for this new type of rail. The welding parameters had to be changed in order to avoid critical cooling rates in the weld, as well as the adjacent heat-affected rail sections, since, generally speaking, alloyed rail steel has different transformation temperature characteristics which will result in embrittled bainitic or martensitic microstructures where the

Table I. Chemistry of various alloyed rail steels

Chemical Composition Manufacturer A C,% 0.55 to 0.75 0.65 0.65 to 0.80 0.68 to 0.78 0.76 Si,% Max. 0.70 Mn,% 0.80 to 1.30 1.00 0.80 to 1.30 1.10 to 1.40 0.37 Cr,% 0.80 to 1.30 1.00 0.70 to 1.20 1.00 to 1.30 1.49 Mo,% Max. 0.15 V,% Max. 0.30 Max. 0.20 P,% Max. 0.03 Max. 0.03 Max. 0.03 Max. 0.04 S,% Max. 0.03 Max. 0.03 Max. 0.03 Max. 0.03

B C

0.50 0.30 to 0.90 Max. 0.35 0.84

D E

0.02

0.02

63

Vanadium in Rail

Steels

350
NE E

Process:

SI<V

DEFLECTION, 50

in

0.90
'iii

0.3

0.4

0.5

0.6

2
"0

45
300

!:: 2
::i en en w en a: w Q. Q. ::>

40

250

en en w a: 35 In a: w Q. Q.

~ ~ I~ ..J

~ 0.80 1-'

z w

I-

30 ::> 200 '-105 2 4 6 LOAD CYCLES 8 106 2 x 106

Z 0 U 0.70 z 0 III a: <t U 0.60

Fig.1 Fatigue tests of welds made with chromiumvanadium steel profile S54 rails. Test frequency 5.55 sec-1; stress limit (00) calculated from section modulus WxF

0.50

10

12 DEFLECTION,

14 mm

16

cooling rates, after completing a weld, are not sufficiently reduced. Table I demonstrates the chemical composition of five alloyed rail steels from five different manufacturers which, to my knowledge, represent the various types of alloyed rail on the market today. The chemical analyses shown in the Table were taken either from rail samples or reflect a complete range for each element as indicated by the manufacturer. Whereas the approach to develop a rail steel with higher strength and better wear characteristics superior to the carbon-manganese steel varies with each manufacturer, the Thermit portions are formulated for the application to any one of these rail steels. The Thermit test welds with these alloyed rail steels were performed with different procedures, ranging from no preheat at all (Self-preheat process), 1to 2 min preheating (SKV),

Fig.2 Correlation between carbon content and deflection of Thermit welds with an average carbon content of 1.15% (1 m (3.3 ft) span between supports) and 5 to 10 min preheating (SoW). The welds were subjected to various degrees of testing, such as slow bend test, fatigue test, hardness tests as well as to intensive investigations ofthe microstructure.

3. PHYSICAL TEST RESULTS

Fatigue tests were carried out according to the specifications of Deutsche Bundesbahn (DB) and Schweizerische Bundesbahn (SBB). The distance between the two supports was 1200 mm (47.2 in) with a two-point load applied to the weld. The load points were 150 mm (5.9 in) apart. The minimum stress limit was 100 N/mm2 (14.5 ksi), with a test

1000

~.~m
I~
Iv
.

1832
Thermit Steel

800
0

'\.
1\
..\

'm
'Ill
1112 w'

u
w' a: 600
::>

I<t a: w Q. 400 2 w

a: ::> I<t a: w Q. 2 w

I-

200

I- ~Ms
M
I

~i
L

~
~8N3~
1

752

I-

392

'~~~I
10
L-L.J...J

0.1 sec
L...L..I

102
~

103
~

104 102

0.01 min

0.1

1 TIME

10

Fig.3 CCT diagram for Thermit steel, welded on chromium-vanadium steel rail. (Therm it steel composition: C, 0.7%; Si, 0.5%; Mn, 1.2%; AI, 0.35%. Tensile strength: 1100 N/mm2 (160 ksi)) (Taken from H. D. Fricke, H. Gunterman, N. Jacoby, ETR, 4 (1976) pp. 199-208) 64

Thermit welding of chromium-vanadium

rail steel

frequency of 5.5 cycles/sec. Welds were tested up to 2.0 million cycles and in case of no failure, discontinued. Figure 1 shows the Woehler curve which resulted from tests made with SKY welds on chromium-vanadium steel rail sections, profile S54. Slow bend tests showed increased fracture loads of about 10% higher in comparison to carbon-manganese steel welds, but as anticipated somewhat lower deflection. As in the case of welds made with heat-treated carbon-manganese rail steel, the AREA specifications calling for 25.4 mm (1 in) deflection in the commonly known AREA test arrangement, cannot always be met. This can be attributed to the fact that elements such as chromium-vanadium and molybdenum will

increase the strength by refinement of the interlamellar spacing, as well as the precipitation of carbides, but will also result at the same time in a decrease of ductility. Figure 2 shows the correlation between carbon content and deflection of welds made with chromium-alloyed rail steel.

4. TRANSFORMATION CHARACTERISTICS
Figure 3 shows the typical chemistry of a Thermit steel and its corresponding CCT diagram. The steel was austenitized for 30 min at 950C (1742F) resulting in a grain size of ASTM No. 3, which is comparable to the grain size in the

1000
..... Rail Steel U.S.A. 132 Ib/yd (66 kg/m) -

1832

800

1472
Act = 720C

u
w'

~ 400 I200

a: :> I- 600 a: f-w Cl.

IC:M' 100

1.00 V

w' a:
1112.

u..

I'

, \ It
P",

100 l(9100\\I t -1\

transformation Completeferrite/pearlite

~
2 w !-

I
752

a: w Cl.

i-- Ms
1 sec

at 2i\ "nd 20 sec \

M
10

'\

r-,.\
102

\ \ \"
103 10 102

392
104

1 min

TIME Fig.4 CCT diagram for carbon-manganese (AREA) rail steel. (Composition: C, 0.79%; I\1n, 0.88%; Si, 0.11%. Tensile strength: 900 N/mmz (130 ksj)) (Taken from W. Ahlert, ETR, 18 (1969), pp. 468-520)

1000

I -= .--1.
""

u
w a:

800

',
"\

Cr-Mn Alloyed Rail

Steel of Special Grade

(
1

1832

Act'" I 600
I

725C

\.

1472
u.. 0

~ ~ I-

"
A

fi 100.. " Complete

P

a: w Cl.
400

Ferrite/Peal rlite Transformation at 580 C and 220 sec 1112

~
!;t

'\

\"
Commendable 752

a: w

Cl. 2

BMs '\"" 40 M i
I ro..

Cooling Curve \. 392 104 102

200

1 sec

10
I

102
I

103 10

1 min TIME

Fig.5 CCT diagram for 1% chromium steel. (Composition: C, 0.77%; Cr, 1.07%; Mn, 1.06%; Si, 0.39%. Tensile strength: 1080 N/mm2 (157 ksi)) (Taken from W. Ahlert, ETR, 18 (1969), pp. 468-520)

65

Vanadium in Rail Steels

Thermit steel adjacent to the rail. Even though the maximum temperature in a weld is significantly higher, the total time interval during which the weld will remain in this high temperature range is considerably shorter, consequently resulting in the same grain size. A time/temperature curve, taken at the fusion line between the Thermit and the rail steel is plotted in the CCT diagram. It demonstrates that in the area with the highest cooling rate, the microstructure transforms fully into pearlite. This time/temperature curve, shown in Fig. 3, was taken in a weld made with the SKY Thermit welding process, where the rail ends were preheated for 2 min. Welds made with the SoW Thermit welding process, which requires a longer preheating time of up to 10 min, naturally resulted in slower, even less critical cooling rates. The time/temperature curve of a weld made with the self-preheat (SPH) Thermit process with no external preheat, is almost identical to the one measured with the SKY process. This process is accomplished through a different weld design as well as a different

flow and larger amount of the liquid steel, where one-third of it is used only for the purpose of preheating and washing the rail ends. Furthermore, it should be mentioned that the chemical analysis of welds made with the SPH process is different, with higher silicon and lower manganese content. This higher Si content of up to 1.5%is due to the different raw materials used for the manufacture ofthe molds, where the Si is absorbed by the liquid Thermit steel from the mold. Figures 4 and 5 show the CCT diagram for a regular AREA carbon-manganese steel and for a 1% chromium rail steel. Other diagrams for the different variations as shown in Table I, have been published in several papers. The CCT diagram for a 1% chromium steel demonstrates that for a temperature range from 800C (1472F) to 500C (932F), at least 200 sec are required for fully pearlitic transformations, whereas the same temperature interval for a regular carbonmanganese"steel, according to AREA specifications, might be crossed within 50 sec.

Temperature, 800-400 C

1 cm (0.39 inl
1400

(1472-752 F)
1... 2... 3...
-

?
w 0::
::>

1000

1\CJ \
: I
.

ci?

Standard practice, maid top removed Maid top removed, wrapped with insulation Full maid removed, air jet applied

23min 37 min 6.3min

u..

2192

,\,\

I-

111

"\:- \.
\\

--".:::::

w
1472

<I: et w 1- 600

ii

:2 w

I,

--- --- ----------"""""'

........... 1

~ I:2 w

<I: 0:: w
12

752

I-

200

' 3

0 Fig.6 Thermal

10

20

30 40 TIME, min welds cooled by different

50 methods 1832

cycles of SKY Thermit

1000 0 800

Key

X SKVweld
Self-preheat (SPH) weld 1472

u 0
w 0::

.. f
I I 1112

u..
0

w ~ I<I: 0:: w 1:2 w I-

~ <I: 600
0:: w 1:2 w

3cm
752

I-

400

200

I I I I I I I I
4 8 12 THinE,min

t:J9
16 20

392

Fig.7 Temperature/time curves for SKY and SPH welds 3 cm (1.18 in) from end of rail

66

---

-------

Thermit welding of chromium-vanadium

rail steel

A comparison between Figs. 3 and 5 shows that control ofthe cooling rate is most important in the heat-affected zone of the rail steel where precautionary steps have to be taken in order to avoid embrittled microstructures. This is most important with the preheated types of Thermit welds, where the excess weld metal, the so-called riser on top of the rail

head will be removed either by hot chiselling or shearing after solidification of the weld. Immediately thereafter, the sheared off upper part of the mold containing hot steel and slag should either be placed back on to the top of the weld or the weld, as well as the adjacent rail sections, should be covered with an insulating blanket. Furthermore, head and base of the rail on both sides of the mold-half should be evenly post heated for approximately 2 min. Figure 6 shows three thermal cycles of SKV welds, cooled by different methods. Optimum results are accomplished by wrapping the weld with an insulating blanket after the excess metal has been removed. If the SPH process is applied for Thermit welding, the head riser does not have to be removed until the weld cools to approximately lOOC (2lOF) or less. Until that time the mold remains in place and will protect the weld and heat-affected zone from cooling too fast. Insulating blankets are not required, as the wide mold halves overlap the weld and rail section on both sides by 9.5 cm (3.74 in). However, in order to build up a heat wall and prevent the rail sections from cooling too fast, a postheating treatment should be applied in the same manner as that recommended for the SKV and SoW processes. Figure 7 shows the cooling curve of a Thermit weld made with the SPH process, where the molds remained in position for the entire cooling period.

~. THERMALTRANSFORMATION OF THE RAIL STEEL

Depending upon the gap size between the two rail sections to be welded as well as the total preheating time, the width of the heat-affected zone is developed differently. Figures 8, 9 and lO show some macro-etched cross-sections which are characteristic for each process. The total width of the heat-affected zone is bordered by the two light vertical lines, where pearlite has been spheroidized but not as yet austenitized- Figure 11 shows the microstructure in the transition zone from the weld to the heat-affected rail steel.

Fig- 8 Macroetched cross-section of a SPH weld

Fig.9 Macroetched cross-section of a SKV weld 67

Vanadium in Rail Steels

Fig.10 Macroetched cross-section of a SoW weld

Fig.11 Microstructure of the transition zone between rail and Thermit steel X100

I Welding Method
I

Preheat!ng Time 5min 2min "

f
"

0 M
> 400 J: IX: W aI ~ 350 :> 2
Cl) Cl) w is 300 IX: <t J: 250

.A --"-

. .-- - Se.." SPH

Cr-Mn Alloyed Steel Rail Head

KV~F

~..
\,,(

~/

.
\
\ "' ..-"-~." 6" .,.:

"

.\
'. /.'. \
'

0 M
400 >
"',

. . ~'.,'~ .,." '. \ .. "

"

I /J "-"'.\ i '. '

I
I

E
IX: w aI

'..' \.

..i

v, I i . / .'''''-.1 \ - r;' :. \,i;

,.

"

..

A,

"

\" ! \ '--1 1\ I "-.. \

'

. ,\,...~

\ 'A.A" ~ ..
.

... \ io.--. \ .',1. /\ I / \ ,. \

I
\

. I',

\.. .,'-.

'.- /

\: I

-~,!

"I i

,.,

I "

".

..
,

\: "

" ~ \ I '.-'-'\ I \'

'\'.

I, . .~, \ .--.
I

; \,/ \

\ .

SPH SoW SKV-F

16 12

\" ,/ - V ;/

HAZ Thermit Weld HAZ ~

..

Fig.12 Hardness distribution in the 350 :i: weld and adjacent rail section for :> SPH, SKY and SoW welds. {Tensile 2 Cl) strength: 1080 N/mm2 (157 ksi)) Cl) w (U.S. Thermit Corp./Th.G.AG, 1979)
300 i5 IX: <t J:
250

12

16

DISTANCE

FROM CENTER OF THERMiT

WELD, cm

68

Thermit welding of chromium-vanadium

rail steel

metallurgical standpoint is to be considered as a casting, this difference in hardness will balance itself out as soon as the rail is exposed to regular train service, and the rail and the Thermit steel will eventually have the same end hardness. The most heat-affected zone of the rail, next to the fusion line with the weld, having a maximum grain growth of the austenite, shows the highest peak in hardness, whereas the spheroidized pearlite results in the area of lowest hardness. 7. SPECIAL PRECAUTION WHEN WELDING ALLOYED RAIL STEEL Despite the fact that all types of alloyed rail steel with chemistries such as those indicated in Table I show very good weldability characteristics for the different Thermit welding processes, there are some points that require special attention. 7.1 Hot shearing If the excess metal on top of the head is removed by shearing, the waiting time between the pouring of the Thermit into the mold and the start of the shearing operation, should be longer than normal if a single blade type shear is used. If the waiting time is not long enough, it will result in hot tearing in the center of the weld (Fig. 13). This can be eliminated by applying the double blade shear in which two blades shear the weld from opposite sides. Furthermore, tests have shown that the tendency for hot tearing will be lessened with shorter preheating periods. 7.2 Torch cutting If torch cutting is necessary, which will be the case whenever no abrasive saw is available, and the rail ends have to be cut for preparation of the gap for Thermit welding, the rail ends must be preheated according to the instructions of the different rail manufacturers. The preheating temperature should be 500-600C (930-111OaF)for at least 0.2 cm (0.08 in) on both sides of the projected torch cut, and approximately 50C (120F) at a distance 1.0 m (approx. 3 ft) on both sides. If the Thermit weld is not to be made immediately, the rail sections have to be preheated for another 5 min at a temperature of approximately 500C (930F). Arc welding cannot be performed without preheating to 5000~600C (930-11 10F), and this temperature should be maintained during the whole welding operation.

Fig.13

Hot tearing after shearing Thermit steel too early

6. HARDNESS
The heat pattern as shown in Figs. 8 to 10 is reflected in the hardness profiles taken in the head parallel to the running surface, see Fig. 12.The hardness of the weld is slightly higher than that of the unaffected rail steel. As the rolled steel has a tendency to work-harden more than the weld, which from a

DISCUSSION

ON

Thermit welding of chromiumvanadium rail steel

H. D. Fricke
F. B. FLETCHER (Climax Molybdenum) I would like to take issue with one statement you made in your presentation: You attribute the smaller deflections you observe in alloy steels to a decrease in ductility ofthe steel due to alloying. We find that, in general, one cannot conclude that the addition of chromium, vanadium, and molybdenum decreases the ductility. In fact, the tensile ductility of steel is not necessarily adversely affected by the addition of these alloying elements. The explanation for the apparent decrease in deflection you observe for alloy steels lies in the fact that there is a measurably greater degree of local deformation under the loading points in the softer unalloyed steels. The alloy steels

do not deform under the loading points as much as the carbon manganese steels, so they appear to give smaller deflections. H. D. FRICKEEven though the weld is called a weld, it is basically a casting, and that should also make a contribution to the ductility we measured. S. MARICH (BHP Melbourne Research Labs) Is there an optimum hardness of the weld metal relative to that of the rail? H. D. FRICKE Our experience has been that the rail steel work hardens faster than the weld metal and eventually, after a certain gross tonnage has passed over the joint, we end up with the same hardness in the rail as in the weld. To take this into account, the weld should, as a general rule, be about 20 points Brinell harder than the rail initially. I must emphasize, though, that this experience is based on carbon-manganese rails. The same situation may not hold in alloy rails and welds, where contributions of the various alloying elements make the situation much more complex.

69

Produced by Mid-County Press, 2a Merivale Road, London SW15