Weldability charts for constructional steels

Transcription

1 IOP Conference Series: Materials Science and Engineering Weldability charts for constructional steels To cite this article: J C Ion and M F Ashby 0 IOP Conf. Ser.: Mater. Sci. Eng View the article online for updates and enhancements. Related content - A simplified model for TIG-dressing numerical simulation P Ferro, F Berto and M N James - Electron microscopy and microanalysis of steel weld joints after long time exposures at high temperatures D Jandová, J Kasl and A Rek - Hot cracking of welded joints of the 7CrMoVTiB 0-0 (T/P4) steel J Adamiec This content was downloaded from IP address on /05/08 at 04:50

2 IOP Conf. Series: Materials Science and Engineering 3 (0) 00 doi:0.088/ x/3//00 Weldability charts for constructional steels J C Ion and M F Ashby Division of Materials Science, Luleå University of Technology, SE-9787 Luleå, Sweden Department of Engineering, University of Cambridge, Cambridge CB PZ, UK john.ion@ltu.se Abstract. The weldability of materials is still a poorly understood concept; a quantitative assessment remains elusive. The variables associated with welding are reduced here into two groups - processing parameters and material properties - from which two characteristic indices are defined and used as the basis of weldability charts. For the case of constructional steels, a carbon equivalent characterises both heat affected zone hardenability and the maximum hardness developed after solid state phase transformations. The welding process is characterised by its energy input. A mathematical model is used to establish relationships between the indices, which are displayed on charts as contours of microstructure and hardness.. Introduction Despite extensive research, the weldability of steels is still a poorly understood concept. It may be defined in terms of various criteria: hardenability, sensitivity to cracking, limits on distortion etc. Even within the context of the hardenability of steels, different interpretations are applied: a maximum acceptable hardness, or a microstructure containing a given amount of martensite. Unlike the mechanical behaviour of materials, which follows well established rules that can be used to define material performance indices [], a numerical measure of weldability remains elusive. Attempts have been made to model the effect of weld energy input on the properties of the heat affected zone (HAZ) of carbon steels, which led to the development of welding diagrams [-4]. Such treatment enables weld properties to be displayed graphically in terms of the principal process variables. Here we define weldability in terms of two quantities that characterise i) the material being welded and ii) the welding process. The material index describes both the hardenability of the steel and the hardness of solid state transformation products, and is based on composition through the use of a simple carbon equivalent formula. The process index characterises the property of the welding process most relevant to weldability - its energy input. Figure shows a chart constructed using these indices, in which welding processes are displayed based on their energy, and steels are indicated according to typical ranges of carbon equivalent. Below we describe how the indices can be used with analytical models of thermal cycles induced in the HAZ adjacent to the fusion line (where hardness reaches a maximum) and in empirical descriptions of the subsequent phase transformations. Model-based relationships between the indices inform the features of the weldability charts that are constructed. To whom correspondence should be addressed Published under licence by Ltd

3 IOP Conf. Series: Materials Science and Engineering 3 (0) 00 doi:0.088/ x/3//00 Figure. Empirical chart showing carbon equivalent ranges for various steels (bars) and typical ranges of weld energy (energy flux) for various processes (boxes). SA = submerged arc, MMA = manual metal arc. Note that weld energy is measured in MJm - and MJm - for thin (full penetration) and thick (partial penetration) welding, respectively. There is no relation in this chart between the axes.. Modelling of weldability.. Weld thermal cycles The thermal cycle T (r,t) at a point in a fully penetrating weld HAZ (e.g. a laser weld) is [5]: Aq (, ) T0 exp / vd (4ct) T r t 4 r at () where T 0 is initial (or preheat) temperature, A is fraction of incident energy absorbed by the weld, q is weld power, v is welding speed, d is plate thickness, r is lateral distance from the energy source, and for steels λ is thermal conductivity (4 Wm - K), ρ is density (7800 kgm -3 ), c is specific heat capacity (40 Jkg - K), and a is thermal diffusivity (.5 x 0-6 m s - ) []. Provided that the peak temperature lies above about 000 o C (i.e. the point of interest lies in the HAZ close to the weld fusion line), equation () may be differentiated and solved to obtain the cooling time between 800 and 500 o C in the HAZ, Δt, for a two dimensional (thin plate) state of heat flow: Aq t vd 4c ()

4 IOP Conf. Series: Materials Science and Engineering 3 (0) 00 doi:0.088/ x/3//00 where T 0 T0 (3) Corresponding equations may be derived in a similar manner to describe heat flow in the HAZ produced by a surface point energy source (e.g. manual metal arc, metal inert gas, tungsten inert gas or submerged arc welding), in which r becomes the radial distance from the energy source in a threedimensional (thick plate) temperature field: Aq r T r, ) 0 exp v t 4 at T( t Aq t v (4) (5) where T T. 0 (6) Equations ( and 5) indicate that the quantities Aq/(vd) (thin plate welding) and Aq/v (thick plate welding), which are measures of the weld energy input, characterise the cooling rate in the HAZ within the austenite transformation range. These are used as the process index... Phase transformations in the HAZ Diffusion-controlled transformation of austenite to a transformation product i is modelled using the form of the Johnson-Mehl equation relevant to nucleation on grain edges [6]: V i = exp-{0.69.(t/t i 50 ) } (7) where V i is the volume fraction of i formed after a time t, and t i 50 is the time required for half the transformation to occur. Figure illustrates microstructural development as a function of cooling time in steels and shows the times required for a given amount of transformation. The transformation of bainite is thus described by: V b = exp-{0.69.(t/t b 50 ) } (8) 3

5 IOP Conf. Series: Materials Science and Engineering 3 (0) 00 doi:0.088/ x/3//00 where V b is the volume fraction of bainite formed after a time t, and t b 50 is the time required for half the transformation to occur. The term t b 50 is not always clearly defined in published data [7], and so is expressed as a function of the time resulting in a bainite-free microstructure, t b 0, and the time for a ferrite-free microstructure, t f 0 ; quantities that can be obtained easily from continuous-coolingtransformation (CCT) diagrams: t b 50 = exp(ln(t b 0.t f 0 ) /). (9) Any remaining austenite is assumed to transform to martensite. Microstructural transformation is then characterised by: V m = exp-{0.69.(t/t m 50 ) } (0) where V m is the volume fraction of martensite formed for cooling with a characteristic time t, and t m 50 is the characteristic cooling time that results in a microstructure containing 50% martensite. Figure. Chart showing model-based variation of microstructure and hardness with weld energy..3. Hardness of austenite transformation products A simple carbon equivalent C eq is used here, and justified later: C eq = C + Mn/ + Si/4 () where element symbols refer to composition (wt%). Cooling times that result in the formation of microstructures containing 50% martensite t m 50, 0% ferrite t f 0, and 0% bainite t b 0 as a function of C eq may then be obtained by regression analysis of published data for constructional steels [7]: t m 50 = exp(7.74 C eq -.96) 4

6 IOP Conf. Series: Materials Science and Engineering 3 (0) 00 doi:0.088/ x/3//00 () t 0 f = exp(9.954 C eq ) (3) t 0 b = exp(6.99 C eq +.453). (4) The volume fractions of the austenite transformation products martensite, bainite and a ferrite-pearlite mixture, V m, V b and V fp respectively become (equation (7)): V m = exp{ln(0.5).(t/t m 50 ) } V b = exp{ln(0.5).(t/t b 50 ) } - V m V fp = - (V m + V b ). (5) (6) (7) Regression analysis of published data for welds in constructional steels [7] gives the following equations for the average hardness of martensite, H m, bainite, H b and ferrite-pearlite, H fp, in terms of the carbon equivalent used here: H m = C eq H b = C eq H fp = C eq, from which the average maximum HAZ hardness, H max, is calculated using the rule of mixtures: H max = V m H m + V b H b + V fp H fp. (8) (9) (0) () 3. Results and discussion The chart in figure is constructed using axes of the material index C eq (on a linear scale) and the process index weld energy (on a logarithmic scale). It shows empirical composition ranges for steels and ranges of weld energy for various processes, and can be used as a scoping tool in a preliminary assessment of candidate materials and processes. However, no link is yet made between the indices. The cooling curves commonly superimposed on CCT diagrams for steels are often labelled with empirically-determined volume fractions of transformation products. Figure shows model predictions of the volume fractions of transformation products resulting from cooling curves that are characterised by different values of t. The variation of hardness with cooling time for a given steel, which may be calculated for a range of steel compositions using the methods described, is included schematically. The chart summarises the volume fractions of austenite transformation products based on the CCT diagram. It may be reconstructed for a different steel composition by using the models described. The variation of phase transformation products and maximum HAZ hardness may be assessed as a function of weld energy using this chart. Figure 3 is constructed using the same axes as figure, scaled on the material axis for constructional steels, and shows model-based predictions of microstructural development in the HAZ after welding. The solid line bounds the region of martensite formation. Broken contours to the left of the boundary refer to conditions that result in a given volume fraction of martensite in the HAZ. Contours 5

7 IOP Conf. Series: Materials Science and Engineering 3 (0) 00 doi:0.088/ x/3//00 describing given volume fractions of bainite and a ferrite-pearlite mixture (not shown) may be constructed in a similar manner. By using a weldability criterion based on HAZ microstructure, e.g. 50% martensite, candidate combinations of welding parameters and steel composition may be assessed. By using equations () and (5) the weld energy scale may be modified to take into account preheating to a given temperature. The chart uses a material index and a process index to present solid state transformations in the HAZ of constructional steels, and is similar to the Schaeffler diagram, which shows weld metal microstructure in stainless steels following liquid-solid phase transformations using two material indices. Figure 3. Chart showing model-based microstructural development in the heat affected zone of constructional steel welds. The axes are the process index (weld energy) and the material index (C eq ). The solid boundary shows the lower limit of martensite formation. Broken contours indicate a given volume fraction of martensite. Figure 4 shows contours of constant maximum HAZ hardness, created using the model described for compositions relevant to constructional steels, and plotted using the axes of figure. Literature measurements of HAZ hardness adjacent to the fusion line (where the highest value is normally found) obtained from welds made using the manual metal arc, metal inert gas, submerged arc and laser processes [4, 8-0] are included. Experimental data are seen to agree well with the model-based hardness contours. By using a weldability criterion based on a maximum permitted hardness, such as that of 350 HV used in many branches of construction, an initial estimate of the welding parameters for safe welding of a steel of given composition may be obtained from figure 3. Alternatively the weldability of a range of steels within a given range of weld energy may be assessed rapidly. Despite its simplicity, the material index C eq appears to be a reliable measure of hardenability in constructional steels (model-based hardness contours lie in correct relation to experimental data on the y-axis of the chart). Similarly, the process index - weld energy input - appears to characterise the thermal cycles experienced in the HAZ well (experimental data and model predictions lie correctly on the chart x-axis). 6

8 IOP Conf. Series: Materials Science and Engineering 3 (0) 00 doi:0.088/ x/3//00 Figure 4. Chart showing hardness development in the heat affected zone of constructional steel welds. The axes are the process index (weld energy) and the material index (C eq ). Literature data for maximum heat affected zone hardness in welds produced by laser beam (thin plate), metal inert gas, tungsten inert gas and submerged arc (thick plate) processes are shown together with contours of hardness calculated using the model. 4. Conclusions The mathematical model used is a good predictor of the maximum hardness developed in the HAZ of constructional steels for a range of welding processes and steels. The charts are a user-friendly graphical display of the effects of variations in characteristic material and process indices on hardness and hardenability in constructional steels. The methods used provide a starting point to understand and model welding of steels currently under development, such as ausferritic and carbide-free grades. 5. References [] Ashby M F 0 Materials Selection in Mechanical Design (Oxford: Butterworth-Heinemann) [] Ion J C, Ashby M F and Easterling K E 984 Acta Met [3] Ion J C, Shercliff H R and Ashby M F 99 Acta Met. Mater [4] Ion J C, Salminen A S and Sun Z 996 Weld. J. 75 5s [5] Rosenthal D 946 Trans. ASME [6] Johnson W A and Mehl R F 939 Trans. Am. Inst. Min. Met. Eng. Iron Steel Div [7] Inagaki M and Sekiguchi H 960 Trans. Nat. Inst. Met. Japan 0 [8] Albright C E, Hsu C and Lund R O 99 Fatigue strength of laser-welded lap joints Proc. Laser Materials Processing Conf. ICALEO 90 eds S L Ream, F Dausinger and T Fujioka (Orlando: Laser Institute of America) pp [9] Moore P L, Howse D S and Wallach E R 003 Microstructure and properties of autogenous high-power Nd:YAG laser welds in C-Mn steels Proc. 6th Int. Conf. Trends in Welding Research eds S A David, T DebRoy, J C Lippold, H B Smartt and J M Vitek, (Pine Mountain, GA, USA, 5-9 April 00) (Materials Park: ASM International) pp [0] Ion J C 005 Laser Processing of Engineering Materials (Oxford: Elsevier) 7