1. Introduction

The supermartensitic steels have been developed based on the standard 11-14% Cr martensitic steels by reducing the carbon content to below 0.02% to improve weldability, increasing the Ni content to ensure a ferrite-free microstructure, and alloying with Mo to improve corrosion resistance ^{[1, 2]}. The microstructure of these steels consists mainly of “soft” low carbon martensite, but some finely dispersed residual austenite and ferrite may also be present. The supermartensitic stainless steels have comparable mechanical properties to the standard martensitic grades and provide more economical alternative to the austenitic and duplex stainless steels in some oil and gas pipeline applications ^{[1, 2, 3]}.

PWHT is usually recommended for welded joints of these steels in order to temper the martensite, thus improving the toughness and ductility ^{[1, 2]}. Because of the increased Ni content, the AC1 temperature is usually lower than 600˚C (1110˚F), and correspondingly lower than the typical tempering range for martensitic steels (650-750˚C). Tempering above the AC1 will result in reformation of some austenite and subsequent formation of “fresh” martensite upon cooling. The presence of untempered martensite and potential for a high fraction of retained austenite following postweld heat treatment can compromise the mechanical properties of the weldment.

Thus, it is of interest to precisely determine the AC1 temperature in supermartensitic steels under actual heat treatment conditions. The simulation experiments on welding and tempering supermartensitic steels, reported by Akselsen et al. ^[3], did not determine the AC1 temperature by dilatometric analysis during tempering between 600 and 800˚C, although some microstructural evidence of newly formed martensite was found. The difficulties in defining the effective AC1 temperature have been attributed to the slow diffusion and transformation kinetics at that temperature, and to the competitive influence of the alloying elements ^[2]. Another important issue in welding supermartensitic steels is determining the MS and MF temperatures under actual welding conditions. These are necessary in order to correctly select preheat and interpass temperatures. Based on the above discussion, the aim of this study was to investigate the on-heating martensite-to-austenite (M-A) and on-cooling austenite-to-martensite (A-M) transformations that occur during welding and postweld tempering of supermartensitic steels, and to determine the respective transformation temperatures. Actual (in situ) and simulation (in vitro) welding and heat treatment experiments were performed on the high alloy grade 12Cr-6.5Ni-2.5Mo steel. The phase transformation temperatures were determined correspondingly by single thermocouple differential thermal analysis (STC DTA) for the in situ experiments and by dilatometric analysis (DA) for the in vitro experiments.

The STC DAT is a recently developed technique for in situ investigation of the solid-liquid and solid-state phase transformations that has tremendous application potential in welding and other thermal and thermo-mechanical applications ^[4]. It is based on single thermocouple measurement and computerized acquisition of the thermal history in particular area of the metal, subjected to thermal processing. The reference thermal cycle, which in the classic DTA is acquired by the second thermocouple measuring the thermal history of a reference specimen with no phase transformations, is replaced in the proposed methodology by a calculated reference thermal cycle. The thermal effects of the phase transformations, which are associated with consumption or release of heat, result in significant changes of the temperature difference (∆T) between the measured and reference cycle. These changes appear as local deviations over the T(∆T) and ∆T(t) curves, whose beginnings and ends coincide with the start and finish of the phase transformation (Figure 1). Compared to the available techniques for in situ thermal and differential thermal analyses, the new technique possesses greater sensitivity to the transformation heat of reaction and allows more precise determination of the transformation start and finish temperatures. It has been shown that the STC DTA has greater sensitivity and equal accuracy to conventional dilatometric analysis, and allows reliable determination of the solid-liquid and solid-state phase transformation temperatures under actual welding conditions ^[4].

2. Experimental Procedures

The chemical analysis and mechanical properties of the 12Cr-6.5Ni-2.5Mo supermartensitic steel used in this investigation are given in Table 1.

The in situ welding investigations were carried out by gas tungsten arc (GTA) welding on specimens cut from tube with 219 mm diameter and 11 mm wall thickness. The GTA process was chosen in order to provide a stable welding arc with lower levels of electromagnetic noise and consistent heat transfer conditions into the welding zone. The in situ heat treatment experiments were performed using specimens cut from the autogenous GTA welds. A specialized furnace with controllable treatment parameters and inert atmosphere (100% Ar) was used for that purpose. The welding and heat treatment parameters are given in Tables 2 and 3, respectively. The temperature measurements during the in situ experiments were performed using Ni-Cr/Ni (type K) and W-Re/W (type C) thermocouples with a diameter of 0.25 mm. The thermocouple wires were placed into two-hole ceramic insulators and their tips welded together by low current GTAW in argon. The thermocouples were then capacitor discharge welded into 1.8 mm holes located in the weld metal, HAZ and base metal. Manual plunging of thermocouples into the molten pool was also conducted. The thermal histories were recorded using an eight-channel computer-based data acquisition system (DAS) with sampling rates of up to 1000 s-1. Typical sets of in situ digitally recorded thermal histories during welding and heat treatment are shown in Figures 2 and 3. The in-vitro experiments were performed using a Gleeble™ 3800 thermal simulator. The first series of these experiments were designed to evaluate the sensitivity and accuracy of the proposed STC DTA methodology by directly comparing it to the DA. For that purpose, thermal cycles with heating rates comparable to the in situ tests and free cooling from the peak temperature were applied to laboratory scale specimens with dimensions 3mm x 12mm x 90mm. The free cooling allowed simultaneous application of the DA and the STC DTA. A typical example of that procedure is shown on Fig. 4. The parameters of the in situ recorded welding thermal histories and those of the in vitro tests are given in Tables 2 and 4, respectively.

The second part of the in vitro experiments was designed to verify the temperature range of the M-A transformation in the investigated steel, and determine the conditions for avoiding the backward A-M transformation during cooling after the tempering heat treatment. For that purpose, a wide range of heating rates was simulated over laboratory scale specimens, covering the typical heat treatment and welding conditions. The parameters of the simulated heat treatment and welding histories are given in Table 4.

3. Results and Discussion

The phase transformations occurring under actual welding conditions in the weld metal are presented in Figure 5 as revealed in situ by the STC DTA technique. The weld solidifies as δ-ferrite in the range 1480 to 1460˚C, Figure 5a. The δ-ferrite transforms to austenite between 1380 and 1180˚C, Figure 5b. The main A-M transformation occurs in the range from 230 to 190˚C (Figure 5c) and is followed by a series of small thermal effects of the austenite decomposition almost down to room temperature, Figure 5d. These phase transformations are accompanied by significant thermal effects that are easily recognizable over the T - ∆T curves, plotted by the STC DTA methodology. The phase transformation temperatures determined here lie completely within the transformation ranges for this type of steel, as reported by Akselsen et al., ^[3].

The in vitro verification of the STC DTA technique by dilatometric analysis is presented in Figures 6 and 7. The two methods yield a comparable temperature range (1320 to 1250˚C) for the δ-ferrite to austenite transformation, Figures 6a and 6b. The A-M transformation is accompanied by a large thermal effect, which required utilization of separate reference curves for more accurate determination of the transformation start and finish temperatures, Figures 7a and 7b. The extensive heat release and the large dilatation during the martensitic transformation coincide very well (240 to 130˚C), Figures 7b and 7c. It is seen in Figures 7a and 7c that a very slow heat release effect precedes the beginning of expansion by about 30˚C. A similar effect during the formation of acicular ferrite in high strength weld metal has been reported previously ^[4]. This effect needs further clarification, but most probably it is related to the different principles (heat of reaction versus dilation) that the two methods are based on.

The influence of the heating rate on the temperature range of M-A transformation is presented in Table

5. It is clearly seen that increasing the heating rate from typical heat treatment to typical welding values (three orders of magnitude) results in shifting the AC1 – AC3 range up by about 300 ˚C.

The experimental results presented in Table 5 and Figure 8 show that at lower heating rates (0.33 and 1 ˚C/s) the M-A transformation appears to be occurring in two parts. The lower temperature part occurs between 560 and 680˚C is accompanied by a very small dilatation effect (Figure 8), and can only be revealed by careful examination of the respective part of the dilatation curve. A proof for the existence of that transformation is the dilatation effect of the backward A-M transformation, occurring during cooling from 700 to 650˚C (Figure 8). The small magnitude of dilatation associated with these transformations show that only a small fraction of the martensite is transformed to austenite below 700 ˚C. During heating to 900˚C more martensite transforms to austenite, which is confirmed by the high degree of dilatation associated with the backward A-M transformation. No dilatation effects due to phase transformations are found when heating up to 600˚C (Figure 8 and Table 6).

The effect of “splitting” the M-A transformation might also exist at some higher heating rates. It also could be the same transformation with transformation rate going to zero at some intermediate temperature range. Obviously, more detailed investigations are needed in order to clarify the kinetics of this transformation and its dependence on the heating rate and temperature.

The results of the in situ heat treatment experiments are shown in Table 3 and Figures 9 and 10. There is a good correlation between the phase transformation temperatures, determined in vitro by DA and in situ by STC DTA. As can be seen from Figures 9a and 10, the STC DTA technique has much higher sensitivity to the small magnitude MA and A-M transformations than the DA. Even after tempering at 600�C, where the DA does not detect any volumetric changes during cooling, the STC DTA reveals decomposition of the small amount of austenite that has been formed between 580 and 600�C during heating (Figure 10b). The STC DTA technique was not capable of revealing the thermal effects of the M-A transformation occurring just below the tempering temperatures of 650 and 850˚C. The reason for that is the sharp decrease in heating rate which normally occurs when reaching the tempering temperature during furnace heating. The STC DTA application software is not able to generate suitable reference curves for such cases. It is therefore recommended, for the purpose of determining phase transformation temperatures during furnace heating, to heat a test specimen up to a temperature exceeding the particular phase transformation range.

It should be mentioned that the MF temperatures presented in Tables 3 and 4 correspond to the final temperatures of heat release and dilatation, as determined correspondingly by STC DTA and DA. Further metallographic examinations and possibly experiments involving cooling to subzero temperatures are needed in order to obtain reliable values for MF.

The in vitro and in situ experiments presented here have clearly shown that the new technique for single thermocouple differential thermal analysis has greater sensitivity and equal accuracy, as compared to conventional dilatometric analysis, for determining the phase transformation temperatures during welding and heat treatment of supermartensitic stainless steels.

4. Conclusions

1. The Ac1 temperature of the 12Cr-6.5Ni-2.5Mo steel during postweld heat treatment, as determined by the STC DTA and DA, is between 560 and 580˚C.

The MS temperature during gas tungsten arc welding and postweld heat treatment, as determined by the STC DTA and DA, is between 250 and 230˚C.
The Ac1-Ac3 transformation range during welding and heat treatment is strongly dependable on the heating rate. Varying the heating rate from typical furnace heat treatment and to welding heating rates shifts this range up by about 300˚C.
Further investigations are needed to clarify the kinetics of the austenite-to-martensite transformation and its dependence on the heating rate and temperature, and also to determine correctly the MF values for the 12Cr-6.5Ni-2.5Mo steel.
The newly developed STC DTA technique demonstrated equivalent accuracy and improved sensitivity compared to DA in determining the phase transformation temperatures during welding and heat treatment.
The in situ experiments demonstrate the potential of the new STC DTA technique for industrial application for developing and testing welding and heat treatment procedures.

Acknowledgments:

The present research has been sponsored by and conducted in the framework of the NSF – NATO Postdoctoral Fellowships for Scientists from NATO Partner Countries Program. The supermartensitic stainless steel pipe was provided by Dr. Vincent van der Mee of Lincoln Smitweld B.V.

References:

[1] Lippold J.C. and D. Kotecki, Welding Metallurgy and Weldability of Stainless Steels, to be published by John Wiley and Sons, Inc, New York, March 2005.

[2] Marshall A.W. and J.C.M. Farrar, 2000. Welding of Ferritic and Martensitic 1114% Cr Steels, IIW Doc. IX-1975-00.

[3] Akselsen O.M. et al., Microstructure-Property Relationships in HAZ of New 13% Cr Martensitic Stainless Steels, Welding Journal, Vol. 83, No 5, May 2004, pp. 160s

– 167s.

[4] Alexandrov B.T. and J.C. Lippold, 2004. Methodology for In-situ Investigation of Phase Transformations in Welded Joints, IIW Doc. IX-2114-04.

Tables

C	Si	Mn	P	S	Cr	Ni	Mo	Nb	Cu	N	Rm (MPa)	HV
0.011	0.43	0.41	0.018	<0.005	12.2	6.6	2.69	<0.005	0.046	74 ppm	883	292

Table 1: Chemical analysis (weight percent) and mechanical properties of steel 12Cr-6.5Ni-2.5Mo

Current, A	Voltage, V	Travel Speed, mm/s	Heat Input, kJ/mm	Thermocou ple type	Thermocoupl e position	TMAX, ˚C	t8/5, s
180 DC	11.5	2.06	1.005		Embedded in	1530	9.52
250 DC	12.6	2.06	1.529	W/W-Re	weld bead	1525	18.11

Table 2: In situ conditions and parameters of the digitally acquired welding thermal cycles

Specimen	Heating rate, ˚C/s	Tempering temperature, ˚C	Dwell time, min	Cooling	Cooling rate,˚C/s	Ac1-Ac3, ˚C	MS - MF, ˚C
HT1	0.3	850	15	still air	1.50	580 - 680	250 - 115
HT2	0.3	650	15	still air	1.33	not processed	240 - 115
HT3	0.3	600	15	still air	1.17	not processed	360 - 150

Table 3: In situ heat treatment parameters and phase transformation temperatures

	Heat treatment											Welding*
Heating rate, ˚C/s	0.33	0.33	0.33	1	1	1	1	5	5	15	50	367**	367* *
TMAX, ˚C	900	650	600	900	700	650	600	900	600	900	900	1350	1413
Dwell time, min	-	15	15	-	3	15	15	-	15	-	-	-	-
t8/5, s	-	-	-	-	-	-	-	-	-	-	-	9.27	11.9 7
Ac1-Ac3	not detected	565 -650	-	570 -680 700 -780	560 -680	580 -650	-	720 – 820	-	780 – 850	830 – 880	854–978***
MS - MF	230-100	210 -100	-	270 -130	250 -140	190 -120	-	240 -130	-	240 -140	240 -140	234–113***

* The heating part of welding thermal histories is equal to that of the in-situ digitally acquired one (see Table 2)

** Calculated as average heating rate in the range between 250 and 1350 0C

*** Average of 5 experiments

Table 4: Parameters of the in vitro simulated heat treatment and welding thermal histories and phase transformation temperatures

Figures

Figure 2: In situ acquired weld thermal histories in HAZ and WM Figure 3: In situ acquired heat treatment histories in HAZ and WM

Figure 4: In vitro simulation of weld thermal history and dilatometric analysis, TMAX = 1413 0C, .t8/5 = 11.97 s, free cooling a) Weld metal solidification as δ-Ferrite, TMAX=1525 ^˚C, ∆t8/5=18.11 s

b) δ-Ferrite - Austenite transformation, TMAX=1525 ^˚C, ∆t8/5 =18.11 s

c) Austenite - Martensite transformation, TMAX=1530 ^˚C, ∆t8/5=9.52 s

t8/5=9.52 s

Figure 5: In situ STC DTA. Phase transformations in weld of steel 12Cr-6.5Ni-2.5Mo

a) In vitro STC DTA

T, ⁰C

1400

1350

1300 T-Dillatation (heating) T-Dillatation (cooling)

1250

1200

170 220 Dillatation, m 270

b) In vitro DA

t8/5 = 11.97 s

a) In vitro STC DTA

b) In vitro STC DTA

Figure 7: Transformation of austenite to martensite in HAZ, TMAX = 1413 ˚C, ∆t8/5 = 11.97 s Figure 8: Phase transformations during in vitro heat treatment. Heating rate 1 ˚C/s, free cooling.