Chapter 3
Thermal, Metallurgical and Mechanical Phenomena in the Heat Affected Zone 1
As a general rule, welding operations modify the metallurgical structures and the local properties of the assembled parts deeply. In certain extreme cases, they can be at the origin of defects such as cracks, porosities or local embrittlement. Consequently, it is important to thoroughly understand the influence of various factors such as the thermal welding cycles and the chemical composition of the parts to be assembled.
In this chapter, we will successively examine:
– particular thermal phenomena associated with the operations of welding;
– the microstructural modifications and principal metallurgical consequences associated with these thermal phenomena;
– the modifications of the mechanical properties in the zone adjacent to the molten metal, i.e. in the heat affected zone (HAZ).
3.1. Thermal aspects related to welding
In spite of their great diversity, the welding processes have localized heat concentration as a common characteristic.
A welding operation can thus be described as a short passage of a small amount of matter at a very high temperature followed by a cooling, primarily by conduction in the adjacent parts: cooler adjacent metal, clamping tools, electrodes (in the case of spot welding). The thermal cycle at any point close to the welded zone thus represents the heat dissipation associated with the welding, and thus depends on variables related to the process (quantity of effective heat applied), on the material (thermal conductivity) and the configuration of the assembly. This localized heating can be carried out without displacement of the heat source (in the case of resistance or aluminothermic welding) or by displacement of the source relative to the part (in general arc and laser welding).
In the second case, if a constant rate of travel is considered, the thermal characteristics of a quasi-stationary mode, where the temperature redistribution around the source adopts a stable form in relation to time, can be determined.
A resolution of the simplified heat equation was proposed by Rosenthal [ROS 46] as early as 1935, followed up by Rykaline [RYK 61], Clyde and Adams [CLY 58]. The principal points are the following: if one imagines a specific heat source q moving at a constant speed along an axis x, the differential heat equation is written, in a series of coordinates (x,y,z) (see Figure 3.1):
with:
– θ: temperature;
– t: time;
– α: thermal diffusivity (m2 s−1) of material = λ/pC, with:
- λ: thermal conductibility (J m−1°K−1s−1),
- ρC: voluminal calorific capacity (J m−3°K−1).
A system of mobile coordinates can be defined related to the source by posing: ξ = x − vt.
In stationary mode 
, the preceding equation then becomes:
Figure 3.1. Definition of the system of co-ordinates where a heat source moves at a constant speed v
Equation [3.2] can be solved by considering two distinct situations (see Figure 3.2).
In the case of welding “thick”1 products, the dissipation of heat occurs primarily radially compared to the source. At any point, the change in temperature according to time and the distance is expressed:
with:
– r: outdistance point considered compared to the source of heat
– q: calorific contribution (J);
– v: rate of travel of source (m/s). The q/v term indicates linear energy or the quantity of heat introduced per unit of length of the welded joint;
− θ0: initial temperature.
In the case of welding thin products, the heat dissipation is negligible in depth. The isotherms are thus perpendicular to the surface of the parts. A simplified solution of equation [3.2] is written:
where d indicates the thickness of the welded parts.
Portevin and Séférian [POR 34] put forward a pictorial image, a thermal solid, to represent welding temperatures in spatial terms, expressed in equations such as [3.3] or [3.4]. Figure 3.3 presents the example of such a diagrammatic representation, axis Ox indicating the welding direction (x indicating distance as well as time), axis Oy the distance to the axis of the welded joint, and the axis Oz the temperature.
With a given distance from the axis (y1 fixed), the thermal cycle θ(t) is described by the intersection of the thermal solid with the plane parallel to (xOz) in y1. In the same way, at a given temperature θ1, the form of the isotherm is described by the intersection of the thermal solid with a plane parallel to (xOy) in θ1. It is precisely by using this last representation that the differences in temperature distribution during welding of thin or thick products can be appreciated (see Figure 3.4). It will thus be observed that the network of isotherms is much denser upstream of the passage of the heat source, which translates to the fact that the rise in temperature (heating phase) is faster than the descent (cooling phase).
Figure 3.3. 3D representation of welding temperature distribution in the case of a mobile source; from [EAS 83]
Because of the finite speed of heat propagation, all the fibers parallel with the welding axis do not simultaneously reach their maximum temperature. The plot of points corresponding to this maximum is the curve (m). Compared to the direction of displacement of the source of heat, all the points upstream of this curve are in the heating phase, and behind, in the cooling phase. The cooling and heating rates are correspondingly greater as the linear energy (q/v) is weaker or the products are thicker and the conductibility greater.
In the case of welding of thin products, the temperature distribution is more spread out in the longitudinal and transverse direction than in the case of thick materials. The difference derives from the possibility, in this last case, of heat dissipation throughout its depth.
Compared to materials with lower conductibility (e.g. austenitic stainless steel, λ ≅ 0.02 Wmm−1°C−1), the welding of material such as aluminum (λ ≅ 0.23 Wmm−1°C−1) leads to a widened isothermic distribution, albeit less extended in the longitudinal direction of the source displacement.
Figure 3.4. Example of temperature distribution during the welding of (a) thin or (b) thick sheet steel assembled under identical conditions; from [EAS 85]
At this stage, we must keep in mind that the analytical calculations presented rest on many assumptions (specific heat source, thermal properties independent of temperature, ignoring latent heat due to phase transformations and heat losses by convection or radiation, etc.). Although new digital methods (see for example [GRO 94, KOV 86, WET 95, ZAC 93]) allow for a more complete consideration of parameters, experience shows that this simplified approach makes it possible to solve the great majority of problems with satisfactory precision.
Figure 3.5. Diagrammatic representation of the thermal cycle in HAZ. Definition of the parameters θM, 
, 
We will now reconsider the thermal cycleθ (t) applicable at any point of the HAZ. Whatever the type of welding process (fixed source or mobile), it typically comprises (see Figure 3.5):
– a very fast heating phase, of which the mean velocity is typically from some 102°C/s (arc welding), rising to 104°C/s for the HDE processes such as laser welding [DEV 89];
– a passage at a maximum temperature θM. In general, the duration at this temperature is very short, and is all the more reduced the higher θM is;
– a more or less fast cooling phase. To simplify matters, we generally characterize the intensity of this cooling by time passing between two given temperatures:
- either between 800 and 500°C, the cooling parameter is then 
,
- or between 700 and 300°C (
). The relationship between this parameter and the previous one generally lies between 2 and 4.
The choice of these two criteria was guided in particular by the fact that the majority of metallurgical transformations occur in these two temperature ranges in the case of carbon-manganese steels.
Experience actually shows that the final metallurgical structure at any point of the HAZ depends almost entirely on the maximum temperature θM reached at this point, and on the cooling parameter as defined by 
. Certain thermal aspects concerning these two parameters will now be more fully developed.
3.1.1. Maximum temperature attained in the HAZ
It was seen that expressions like [3.3] and [3.4] described the temperature change according to time in the respective cases of welding thick or thin products. The maximum temperature is reached at a time tM such that:
Moreover, in order to avoid the singularity (θM →∞) when (r→0), the limiting condition is imposed: θM = θf, melting point, when r = R, radius of the molten metal deposit.
We are then led to the following expressions:
– maximum temperature attained in the case of thick sheet welding (two-dimensional flow of heat in a section perpendicular to the welding direction):
– maximum temperature attained in the case of thin sheet welding (unidimensional flow of heat in a section perpendicular to the welding direction):
Figure 3.6. Evolution of peak temperature according to the distance to the fusion line. Comparison of the analytically calculated temperatures and those recorded in experiments (submerged arc welding); from [DEV 87]
Thus, the maximum temperature in the HAZ varies in approximately inverse proportion to the distance r to the fusion line (i.e. in thin materials) or to its square (i.e. thick materials). In the case of plates, the maximum temperature reached does not depend on the product thickness.
In spite of their simplicity, these expressions generally give a good account of the maximum temperature evolution in the HAZ, as Figure 3.6 testifies. This presents a comparison of experimental data recorded by thermocouples of thermal cycles in the HAZ, with the results of an analytical calculation (welding of thick products in the case presented here).
3.1.2. Cooling parameter in the HAZ
With the maximum temperature, the cooling speed, quantified by the parameter 
, is the most significant determining factor for the metallurgical structure of the HAZ. Expressions [3.3] and [3.4] make it possible to calculate:
– the cooling parameter associated with thin sheet welding:
(temperatures expressed in °C);
– the cooling parameter associated with thick sheet welding:
Readings of thermal cycles taken for welded joints [DEF 75, MUR 67] show that the cooling parameter varies very little within the same welded joint. In this way, it can be said that the various zones of a welded deposit can be characterized by a single cooling parameter value 
(or 
). The continuing evolution of the microstructure thus depends only on that of the maximum temperature reached locally, all the points undergoing a practically identical cooling rate.
It now remains to link the cooling parameter to the welding conditions in a more precise way: if welded joints are produced, starting from products of various thicknesses while varying the linear energy of welding, the variation of the cooling parameter associated with these conditions is presented in Figure 3.7.
Beyond a certain thickness, the cooling parameter does not vary. This critical thickness, called the thickness limit, is correspondingly higher the greater the welding energy is. From the point of view of welding, a product will thus be regarded as “thick” for a given energy of welding, if its thickness is higher than the thickness limit.
In reality, this concept of thickness limit represents the modification of the nature of heat dissipation in the welded joint: below the limit, the flow of heat is two-dimensional, with normal isotherms compared to the plane of the product. Beyond this, the flow is three-dimensional and the isotherms take on a cylindrical form. There exists a zone of transition between these two modes, where the isothermal form is affected by radiation to the lower part of the component.
In practice, it is important to define the welding linear energy (term q/v in the preceding expressions) encountered in this phenomenon. This is the product:
– of linear electrical energy Ee, defined from the welding parameters: for example, in arc welding, carried out with a voltage U and current intensity I, the expression can be written: 
, E being frequently expressed in kJ/cm;
– of a thermal efficiency coefficient ηP, intrinsic to the welding process used, which characterizes the relationship between the energy actually transferred to the part and electrical energy. Table 3.1 indicates generally accepted values.
Table 3.1. Thermal efficiency coefficient of various welding processes; from [CHR 65, CON 87, DEV 85, GRO 94, ION 97, NIL 75]
| Welding process | Thermal efficiency ηp |
|---|---|
| Submerged arc welding | 0.9 to 1 |
| Welding with coated electrode | 0.7 to 0.85 |
| MIG welding | 0.7 to 0.85 |
| TIG welding | 0.2 to 0.8 |
| Electron beam welding | 0.8 to 0.95 |
| Laser | 0.4 to 0.7 |
It will be noted that TIG welding’s efficiency is generally low and varies with many factors: intensity, DC or AC, shielding gas, etc. In the same way, the efficiency of laser welding is very variable according to the reflectivity of the part: low for copper or polished aluminum parts, the efficiency (and thus the beam penetration) can be increased by depositing a fine layer of graphite or zinc phosphate.
The linear welding energy is also the product of a coefficient ηG related to the geometry of the deposit and assembly. By definition, a flat deposit (heat diffusion at an angle of 180°) will be expressed as: ηG = 1. A T-shaped assembly (diffusion of heat at 270°) will be characterized by the coefficient 
. This approach can be generalized for more complex examples (preparation of X-shaped, V-shaped chamfers, etc.).
The effective linear welding energy is then written: 
.
This value, as well as the thickness of the welded products, makes it possible to define the thermal welding mode: in a remarkable study [BER 72] based on a theoretical and experimental approach, G. Bernard could specify the field of thick and thin products with respect to welding heat dissipation.
Figure 3.8. Graph determining the cooling parameter 
according to the welding conditions; from [ATS 80]
In the absence of pre-heating (θ0 = 20°C), a part will be regarded as thick based on the criterion 
if its thickness d (mm) is such that:
E being expressed in kJ/cm. In the case of the criterion 
, this condition would be written:
From such criteria, it is then possible to calculate the Δt parameter values according to expression [3.7] or [3.8].
To enable rapid estimations, IRSID [ATS 80] developed a forecast graph of the value of 
starting from the various parameters in the case of arc welding (see Figure 3.8). Thus, as an example, we can consider a TIG welded joint with an energy 
of 15 kJ/cm on a T-shaped assembly. Starting with electrical energy (point A), corrections related to the geometry of the assembly (point B) and to the efficiency of the process lead to an effective energy E (point C) of 5 kJ/cm. Under these conditions, the welding of 8 mm thick materials will lead to a value of 
of 4 seconds (D).
Another example (MIG welding at 20 kJ/cm) shows how it is possible to take account of the pre-heated temperature θ0 (here 200°C) by a correction of the thickness and effective energy.
As an indication, Table 3.2 gathers some typical values of cooling speeds observed for various welding processes, in the range of thicknesses used in these processes.
of various welding processes
3.2. Microstructural modifications in the HAZ: metallurgical consequences of the thermal cycles of welding
3.2.1. Transformations in the HAZ during heating
In a strict sense, the rapidity of the welding thermal cycles does not make it possible to use the balance diagrams to forecast the true nature of the various phases in the vicinity of the line of fusion. However, following the example of Easterling [EAS 83], this type of diagram is a suitable starting point to interpret the microstructural modifications qualitatively. To serve as an example, structural steels, which are the most widely used in welding applications, will be considered by examining the diagram (Fe-C) in parallel with the maximum temperatures reached in a welded joint (see Figure 3.9).
Let us consider for example the case of welding a 0.15% steel. As we approach the molten zone, the following zonal situations are to be found.
The base metal has not undergone a transformation phase in the heating, i.e. it has been heated to a temperature lower than the transformation point A1 (727°C). On balance, it is thus a microstructure made up of ferrite α (solid solution carbon insertion in iron, of a centered cubic structure) and of iron carbides Fe3C (or cementite) or more precisely of pearlite, a lamellar aggregate of ferrite and cementite.
We have a subcritical zone, where a phase shift has not yet occurred. However, when the temperature reached is sufficiently high (for example θ >600°C), certain phenomena such as tempering, the spheroidizing of the cementite lamellae, the recrystallization in the case of welding of strain hardened products, ageing, etc., can possibly occur [GRA 95].
We also have a partial transformation zone (or intercritical zone), between A1 and A3 (≈ 830°C for the composition considered). In this field ferrite coexists with a newly formed phase, austenite γ, a solid insertion solution of cubic centered face structure. It is starting from this zone (θ> A1) that the heat affected zone begins.
From temperature A3 up to approximately 1,495°C, the transformation into austenite is total. The new structure thus replaces and erases any trace of the former ferritic structure. It is within this zone that an austenitic grain enlargement occurs: this is of a very small size at temperatures slightly higher than A3, and can reach a few hundred microns for the highest temperatures. In this case, the term employed is coarse grain zone (for example, when the austenitic grain size exceeds a few tens of microns) and in general it is this zone which is the most likely to cause certain metallurgical problems, as will be seen later.
We have another zone partially in the liquid state, where a solid ferritic phase (δ) and a liquid phase coexist. Very small in size (and thus difficult to observe micrographically), this zone constitutes the connection between the base metal and the molten metal. On the balance diagram, this corresponds to the temperature interval between the beginning and the end of fusion (solidus-liquidus interval).
Certain phenomena (local decarburization by enrichment of the liquid in the solute, intergranular liquation by eutectic formation at low melting point in the case of impurities, etc. [GRA 95, HAL 84]) can intervene in this bond zone.
Finally, the molten metal often has a composition different from that of the base metal, because of the volatilization of certain elements, reactions with the surrounding medium, or of enrichment by external elements (filler material). The first molten metal nuclei are solidified by epitaxy (crystallographic coherence) on the grains of the HAZ.
At this stage, it is necessary to note the following points:
– the transformations and temperature ranges evoked above correspond to a state of balance. In reality, the brevity of the thermal cycles means that phase transformations and structural homogenity intervene at temperatures sometimes considerably higher than these balanced temperatures;
– the particular case considered here (alloy Fe-0.15% C) presents a situation where phase shifts occur during heating. In the case of other materials (for example: aluminum alloys, copper, ferritic stainless steels, etc.), the thermal welding cycle intervenes in a single-phase area. A grain enlargement will then be found throughout the cycle, without possibility of refinement by phase shift;
– starting from expression [3.5] or [3.6] and the knowledge of the phase transformation temperatures, it is possible to predict the HAZ width in various conditions of welding. The calculations, confirmed by observations of welded joints, show that this can vary from a few hundreds of microns (for example: low power densities in laser welding [ION 97] and resistance welding [CHA 74]) to 15 mm in electroslag welding [DEB 82, HAR 80].
In addition to these microstructural modifications, welding operations have a marked influence on the precipitates possibly present in the base metal. Indeed, it is known that modern steels draw their mechanical properties (resistance, ductility, toughness) from an optimized combination of hardening, by the nature of the microstructure, the grain and precipitation refinement. This balance is significantly modified by the thermal welding cycles.
Generally, the dissolution of various particles (carbides, nitrides, sulfides, oxides) can be quantified, starting from the variation of their solubility product with the temperature, described in this form (case of diluted solutions) [RIS 74]:
MaNb ⇔ aM + bN, a and b defining the stoichiometry of the initial compounds:
ΔG 0 indicates the free energy of the reaction, T the absolute temperature, R the constant of perfect gases (1,987 cal. K−1 moi−1). Table 3.3 and Figure 3.10 present the solubility product of some compounds, from which the temperature at which they are totally soluble can be calculated.
Table 3.3. Expression of the solubility product of various compounds in austenite; concentrations in % by weight, T in °K
| Type of precipitate |  |
| NbN | 4.04 − 10230/T |
| VN | 3.02 − 7840/T |
| AlN | 1.79 − 7184/T |
| TiN | 4.35 − 14890/T |
| TiC | 5.33 − 10475/T |
| NbC | 2.26 − 6770/T |
| MnS | 2.93 − 9020/T |
| Al2O3 | 20.43 − 125986/T |
| SiO2 | 5.10 − 44801/T |
| MnO | −5.71 − 24262/T |
| Ti2O3 | 16.18 − 104180/T |
The higher the temperature reached locally in the HAZ, the greater the likelihood of precipitates decomposing. In decreasing order, the stability of these compounds is as follows: oxides, nitrides, sulfides, carbides.
For common compositions of structural steels, decomposition of carbides in the HAZ will be observed around 1,100-1,150°C (VC, TiC) [HRI 95], nitrides around 1,150 to 1,300°C, sulfides being generally dissociated around 1,100 to 1,200°C. Taking into account their very strong stability, oxide inclusions resulting from the steel making in the liquid state (alumina, silicates, etc.) are not affected by the welding cycle. Being very few, and relatively large in size (for example: several tenths to a few microns), these oxides generally do not play a role at the time of the heating phase in the HAZ. It is not the same for nitrides or carbo-nitrides. Indeed, it is known that these numerous, fine and dispersed particles slow down the growth of austenitic grain since the joint displacement this achieves is accompanied by a local increase in enthalpy.
Maximum size of grains dmax, in the presence of a population of precipitates of size r and whose voluminal fraction is VF, can be expressed by the Zener relation [ZEN 48]:
.
In general, an increased austenitic grain size in the HAZ leads thereafter to a coarser ferritic structure. To limit this phenomenon, it is possible to show that low additions of titanium (an element which forms, as was seen previously, very stable nitrides at high temperature) slow down the growth of the austenitic grain effectively (see Figure 3.11). It should however be stressed that this result is obtained only if these nitrides are relatively fine (a few tens of nanometers), i.e. primarily if they are formed after solidification during steelmaking and not in the form of coarser precipitates (micrometric particles, formed in the molten metal).
Figure 3.11. Influence of maximum temperature on the austenitic grain size in HAZ of steels containing various types of precipitates; from [KAN 76]
3.2.2. Transformations in the HAZ during cooling
In the case of structural steels, the microstructures observed in the HAZ of welded joints correspond to transformed products of austenite. It is in fact difficult to study these metallurgical transformations starting from traditional CCT diagrams (continuous cooling transformation). Indeed, these are generally established starting from austenitic treatments at a relatively low temperature (900-950°C) and for long durations, i.e. a situation exactly opposite to that encountered in the HAZ. In order to consider more representative conditions, the metallurgical phenomena starting from thermal heating and cooling cycles very close to those met in welding are studied. These tests are generally carried out on specific thermal simulators (Gleeble and Smitweld simulators, etc.). For a given material (see Figure 3.12) and austenitization conditions (temperature generally higher than 1,200°C, so as to simulate the HAZ with coarse grains), the transformations of the austenite according to the cooling speed using the CCT diagram in welding conditions or more simply of the evolution of hardening the final microstructure with temperature can be gathered.
Figure 3.12. (a) Example of CCT diagram in welding conditions (θM = 1,300°C); (b) corresponding graph (hardness-criterion of cooling Δt). Steel C = 0.18%, Mn = 1.4%, Si = 0.4%
According to the cooling speed, various types of microstructures are liable to be formed.
For very high speeds (lower than the critical quenching rate VCM, which corresponds to a plateau on the hardness-criterion cooling graph or HV-Δt), martensite appears. Having a tetragonal crystallographic structure, this supersaturated solid solution of carbon in iron is formed by a shearing mechanism without atomic diffusion. Identical in composition to that of the mother austenite, martensite has a hardness HVM which depends virtually only on the percentage of carbon in the steel alone, according to the expression: HVM=283+930(%C) [IRS 77]2.
At intermediate speeds, bainitic structures appear. Formed starting from metastable austenite at temperatures too low for the carbon to be able to diffuse at the grain boundaries, the bainite is characterized by the very fast growth of ferrite needles by shearing and a precipitation of carbides. Without returning to the complex debate relating to the description of bainitic structures (see for example [BHA 97, HEH 72, HON 95, OHM 74]), we will simply recall that faster cooling speeds lead to the formation of the harder lower bainite structures (carbide precipitation in the shape of thin platelets within the ferrite laths), whereas slower speeds are associated with higher bainite formation (carbides rejected from the ferrite lath joints, sometimes in the form of quasi-continuous cementite lamellae [CON 92]). In this last case, the local carbon content can be such that the formation of compounds between the laths occurs where martensite and austenite coexist (M-A compounds), the latter being stabilized because the transformation temperature Ms is lowered below room temperature [AND 65].
Slower cooling speeds in welding correspond to the precipitation of proeutectoïd ferrite at the old austenitic grain boundaries and the evolution towards a ferritoperlitic structure. It is also in this range of cooling speeds that some compounds (vanadium and niobium nitrides or carbo-nitrides, etc.), dissolved at the time of the heating phase, can re-precipitate in a more or less complete form.
Various more or less complex formulae have been proposed in order to describe the evolution of HVmax hardness in the large grained HAZ of C-Mn steels according to their composition and the speed of cooling. We can for example quote that of Yurioka [YUR 87], applicable to steels of less than 0.3%C, 5%Ni, 1%Cr without B (
expressed in s, content in %):
with:
These expressions reveal the concept of carbon equivalent (CE, a function of the level of C, M, Mo, etc.), which will be subject of further discussion in connection with cold cracking. At this stage, it must be retained that it is a factor quantifying the capacity for hardening. At a given cooling speed, limiting hardness amounts to limiting the percentage of carbon equivalent.
As has been previously stressed, the microstructure formed at any point of the HAZ depends on the temperature reached locally. Figure 3.13 illustrates, in the case of a 32C1 steel, the influence of the temperature on the transformation of austenite. If cooling is sufficiently fast (for example, 
s), all the zones where the temperature exceeds 1,000°C are martensitic. In the case of a slower cooling (for example, 
s), the part of the HAZ heated to around 1,000°C has a bainitic structure, whereas the coarse grained part of the HAZ (1,100 to 1,200°C) is martensitic. This microstructural difference explains why the majority of the metallurgical problems (toughness, sensitivity to cracking in the presence of hydrogen) occur primarily in the coarse grained part of the HAZ, heated to a higher temperature.
Figure 3.13. Influence of maximum temperature in the HAZ on a CCT diagram in the welding of a steel 0.3% C, 0.8% Mn, 0.2% Si, 0.3% Cr
3.2.3. Case of multipass welding
Beyond a certain thickness, the product assembly generally involves a number of successive deposits (welding involving several passes, or multipass). Except in the last deposit, all the points of the HAZ undergo, during this operation, a complex succession of reheating cycles [DEV 87]. In reality, if a point is located in the HAZ in the immediate vicinity of given pass N, the progressive distance of the deposits means that their metallurgical influence quickly becomes negligible (maximum temperature lower than 500°C) beyond pass N + 2. Moreover, it is understood that austenitization at very high temperature erases the metallurgical influence of the preceding cycles at lower temperature; therefore, the metallurgically effective thermal history in a multipass HAZ can be summarized with some characteristic combinations, comprising at most three thermal peaks. Thus, the three following zones can be distinguished (see Figure 3.14).
Figure 3.14. Diagrammatic illustration of the thermal cycles and the microstructures encountered in multipass welding; from [MAT 94]
A zone with coarse grains is unaffected by the subsequent passes, because it is heated to a temperature lower than Ac1 (zone A in the figure, or SRCGHAZ: subcritically reheated coarse grain heat affected zone). In general, the microstructure and the mechanical properties of this zone are relatively close to those of a zone with coarse grains in a rough state (CGHAZ, zone D).
A zone with coarse grains heated in the intercritical field, between Ac1 and Ac3 (ICCGHAZ: intercritical reheated coarse grain heat affected zone, zone B). The transformation (γ→ α) occurs in particular in the zones of easy diffusion, such as the former austenitic grain boundaries, the zones richest in carbon (carbides, M-A compounds) or the segregated zones. The austenite which is formed locally is increasingly carbon enriched the lower the temperature is. With cooling, the austenitic sites can be transformed into martensite or to austenite (M-A compounds) according to a mechanism described in the preceding section. As many studies have highlighted the poor character of these compounds with respect to toughness [IKA 80, LAM 99, MAT 94], it is generally considered that the zone heated in the intercritical field is, with the coarse grained zone, the zone of least toughness in multipass welded joints. However, the brittleness of this zone all but disappears with a reheating to a sufficient tempering temperature (for example: 350–400°C) related to the deposit of a subsequent welding pass or a stress relieving heat treatment.
When structural steels must provide guarantees of toughness in multipass welded joints, manufacturers and steelmakers particularly endeavor to minimize the proportion of local brittle zones (LBZ, coarse grain or intercritical zones [HRI 95, TOY 86]) in the vicinity of the molten metal by controling the conditions of welding and the chemical composition.
Finally, a zone with fine grains (FGHAZ, fine grain heat affected zone) is encountered, associated with a temperature peak slightly higher than Ac3. The very fine microstructures associated with this type of treatment correspond to structures resulting from normalization treatments and present excellent mechanical properties. A wise choice of welding conditions, so as to maximize this zone with fine grains and to favor the transformation of M-A compounds, makes it possible to guarantee good toughness properties of multipass joints [IKE 95, KAW nd, KOS 81, LIN 89].
To conclude, it should be noted that certain software, based on a thermal and metallurgical modeling, have been developed in order to forecast the risk of brittle zones appearing in multipass HAZs and their effects on mechanical properties [DEV 91, KAP 98, REE 94].
3.2.4. Cold cracking
The concept of weldability has often been confused with and reduced to sensitivity to cold cracking, which gives some indication of the importance of this phenomenon.
During arc welding, the molten metal is likely to dissolve hydrogen derived from the decomposition of water present (ambient air, moisture, etc.), or from that of hydrogenated matter (greases, etc.) and possibly from hydrogen in the welding gas; the dissolved quantity can be high. Results published by Granjon [GRA 95] show that, during structural steel welding, the quantity of diffusible hydrogen present in 100 grams of molten metal can reach, on average:
– 15 cm3 in welding with electrodes with a rutile coating,
– 9 cm3 in welding with powdered flux,
– 7 cm3 in welding with electrodes with a basic coating,
– 4 cm3 in MIG welding.
During solidification, cooling and any possible allotropic transformations, the solubility of hydrogen decreases appreciably. This fact could be responsible for the formation of porosities during solidification and will generate thereafter a diffusion of gas towards the atmosphere and, if this is insufficient (fast cooling), a supersaturation of the solid metal with in situ formation of H2 molecules. The gas thus produced gathers in the spaces caused by structural defects and decohesions between metal and inclusions. In these places, the gas pressure rises with the diffusion of hydrogen and can reach elevated levels, which generates considerable localized stresses likely to combine with shrinkage stresses. If the structural state of the metal is not sufficiently strong under these conditions, a crack is likely to develop.
This precise situation is encountered during the welding of structural steels likely to undergo a (γ ⇔α) transformation; it leads to cold cracking.
Indeed, in a joint:
– the transformation in molten metal (when it is susceptible to a (γ →α) transformation, which is generally the case) starts before that of the HAZ because, if it cools like the latter, it is lower in carbon and its Ms point is higher. Any hydrogen contained is thus released, partially towards the atmosphere and partially towards the HAZ (as long as the latter is in an austenitic state);
– the (γ →α) transformation in the HAZ, by the effect of heat gradients and the differences in hardenability caused by the enlargement of the austenite grains, progresses from the coolest parts (close to the base metal) towards the limit of the molten metal.
Consequently, hydrogen is trapped in the most overheated part of the HAZ, whereas its solubility decreases as a result of cooling and especially from the passage to the α state. If this part is transformed into martensite (not very plastic, even brittle), it can fissure under the action of stresses caused by the localized hydrogen pressures and the shrinkage stresses. Thus, cold cracking will occur.
To assess the risks of cold cracking, it is necessary to carry out tests for cracking under the welding conditions considered. If it is easy to reproduce thermal cycles and the hydrogen loading (it is sufficient to apply identical welding conditions), it is on the other hand very difficult to reconstitute the pattern of shrinkage stresses. For this the tests employed will be conventional. The most common and most significant is the implant test defined by French standards NF A 89-100 and NF A 03-185 and by the recommended procedure IIS/IIW-447-73. A notched cylindrical test-piece is implanted in a thick plate. The test weld is carried out under the selected conditions then, upon the solidification of the molten metal, the implant is subjected to traction (see Figure 3.15).
Figure 3.15. a) Example of implant with a circular notch; b) implant subjected to traction after deposit of a weld bead
Micrographic sections taken after cooling and waiting time allow the detection of possible cracks. While simultaneously varying the welding conditions (energy and thickness of the support plate) and the load applied to the implant, we have the necessary information to plot a cracking curve which distinguishes, within the tensile effort/cooling parameter coordinates, the fields where cracking occurs and where it is avoided (see Figure 3.16). Such a curve is characterized by the steel, the welding process and the hydrogen content under consideration. This procedure has made it possible to qualify the metallurgical weldability of many steels with respect to the risk of cold cracking and to define minimal energies to employ when welding them without risk. It should however be noted that the influence of hydrogen content is foremost; it is often enough to lower this to make cold cracking disappear, as shown in Figure 3.17 for a simple C-Mn steel. It is then this factor that the welder will have to consider as a priority.
However, if lowering the hydrogen content is impossible or insufficient, the welder can still intervene by carrying out a careful pre-heating and possibly a post-heating. These two operations (which we should not regard as being heat treatments) consist of modifying the thermal evolution of a weld by heating up the parts which constitute the assembly at a higher temperature than room temperature (this is preheating) and possibly by maintaining this heat, after welding, for a given time (this is post-heating).
These procedures:
– slow down the joint cooling which is applied with a cooling medium — the base metal — no longer at room temperature but maintained at the temperature of preheating. If the base metal is not too hardenable, this slowing down modifies the (γ → α) transformation by allowing a certain proportion of bainite to form before the martensitic transformation; the metal of the HAZ is less hardened and thus less hard but especially less brittle. In addition, starting from the pre-heating temperature just as starting from the temperature of post-heating, the final cooling is very slow since it corresponds to all of the pre-heated and possibly post-heated area. The end of the martensitic transformation thus occurs very slowly in the presence of a hydrogen quantity which gradually decreases since it diffuses into the atmosphere, outside the joint. The relatively long duration of post-heating and between the latter and the ambient temperature constitute a tempering operation to which the already formed martensite is subjected, causing a further decrease in its brittleness;
– increase the HAZ depth and decrease the temperature gradients. The dilation gradients are in this way decreased and, consequently, the possibilities of shrinkage and thus significant stresses are reduced.
All these effects have the same beneficial result; they decrease or eliminate the risk of cold cracking.
To assess a priori the risk of cold cracking, on a practical level, a particular parameter called carbon equivalent (Ceq) has been used on a regular basis, calculated on the chemical composition of the parent metal. Many formulae have been proposed, the best known being that employed since 1940 by the International Institute of Welding [DEA 40]:
With a constant Ceq, metallurgical weldability would be constant and the lower the Ceq, the lower the risk of cold cracking. Such a method of assessing this risk does not appear very satisfactory, indeed:
– Ceq does not take into account the hydrogen content of the molten metal, the importance of which has just been discussed (certain formulae have been proposed which remedy this oversight);
– Ceq does not take into account the cooling conditions, which are a determining factor concerning the proportion of martensite formed;
– Ceq regrettably confuses the capacity for hardening (and thus the risking of martensite brittleness) which depends on the percentage of carbon and secondly hardenability (and thus the risk of martensite formation) which depends on the content of alloying elements.
Ceq can therefore only be used to make comparisons within very limited fields.
Currently, some NF EN standards define limiting values for this Ceq. As an example, and to highlight the improvements in weldability achieved thanks to the developments in iron and steel processes, the maximum guaranteed value can be cited for steels meeting standard NF EN 10113 and delivered in a normalized state or after thermomechanical treatment in the form of flat products with a thickness ranging between 16 and 40 mm.
Lastly, it is important to stress that the development of many iron and steel products during the last few decades has been guided mainly by the desire to decrease the sensitivity to cold cracking. Consequently, the concept of metallurgical weldability has played a crucial role in the evolution of structural steel composition.
3.2.5. Lamellar tearing
It will be noted later that welding leads to the formation of stresses and significant deformations. A possible consequence of this phenomenon is lamellar tearing, which corresponds to a particular type of crack developing parallel to the bond zone of the welded joints and perpendicular to the grain of flat-rolled products (see Figure 3.18). Its appearance corresponds to an insufficient ductility of the welded material in the direction of its thickness, caused by the presence of inclusions aligned in the direction of rolling.
Examination of the ruptures primarily reveals a characteristic step-like break corresponding to decohesion between the inclusions (essentially sulfides) lengthened by rolling and the matrix. This type of cracking can be avoided:
– by the joint design, avoiding procedures which lead to high levels of stress in the short transverse direction as much as possible (geometry, preliminary buttering, limitation of the elastic limit for the molten metal, etc.);
– by the choice of the metallurgical quality of the steel, which must have a sufficient ductility in its thickness. This characteristic is checked on cylindrical tensile test-specimens taken in this direction, and on which striction (Z%) is measured after rupture. So we talk about “Z… steels” in connection with products with guaranteed properties in their thickness, which are obtained by limitating the inclusion content (in particular expressed by the sulfur behavior) and/or controling the morphology of these inclusions (treatment in production by addition of calcium, rare earths (cerium, etc.) [OLE 81], etc.).
3.3. Influence of the thermal cycles on the mechanical properties of the HAZ
Important modifications in microstructures caused by welding are naturally accompanied by significant changes in the mechanical properties compared to the base metal. These will be examined with regard to localized tensile, hardness or toughness properties.
3.3.1. Modifications of the mechanical properties of hardness or traction in the HAZ
The modifications of mechanical properties in the HAZ (mechanical resistance, or more simply hardness [GRU 81]) must obviously be understood in comparison with the base metal, whose properties can result from various combinations (thermomechanical composition/treatments). Thus, it is possible to enumerate among the big groupings of structural steels, for example the C-Mn steels of ferritopearlitic structure obtained by normalization treatments, high strength steels (HSLA) combining refinement of the ferritic grain and fine hardening precipitation, steels with high or very high strength (ferritobainitic, or bainitic with very low carbon (ULCB)), dual-phase steels (ferrite-martensite), quenched and tempered steels, etc.
In general, the thermal cycles associated with the coarse grain zone lead to a localized increase in hardness compared to that of the base metal, particularly in the case of a fast cooling after welding. The purpose of this mechanical heterogenity is often to displace the rupture site towards the base metal in the case of smooth tensile or bending tests across the welded joints. In the case of mechanical tests comprising a notch located in the HAZ, the presence of the less hard adjoining base metal contributes to the development of a plastic zone and to a rise in apparent toughness [DEV 86a]. In certain extreme cases (very large difference in hardness between the welded zone and the base metal, very narrow welded zone, as in the case of welds by laser or electron beam), a marked deviation of the rupture can be noted towards the least hard zones [DEV 86b, GOL 76].
More generally, this illustrates the possible difference between intrinsic mechanical behavior in a given zone and that obtained in the presence of the adjoining zones with different behavior patterns, possibly in the presence of a mechanical defect. The whole of these rather complex questions, the interaction of the mechanical properties in the various parts of the welded joints, is known as the matching of welded joints (see for example [LON 94, TOY 92]).
In certain cases (base metal with high yield strength obtained by thorough thermomechanical treatment), the thermal cycles can involve the formation of softened zones compared to the base metal. Softening appears at a temperature corresponding to Ac1 or Ac3 [DEV 83, ITO 00] and increases the higher the energy is (for example: aluminothermic welding of rails). Very often, the triaxiality of the stresses and the localized consolidation make it possible to avoid an excessive concentration of deformations within the softened zone. It is advisable to limit the amplitude of softening (difference in hardness between the base metal and the HAZ) as well as the width of the softened zone [DEV 84], so as to avoid rupture or a premature wear by concentration of the deformation in this zone. Finally, it should be noted that the presence of this softened zone can sometimes be associated with a cracking due to localized hydrogen (stress oriented hydrogen induced cracking, or SOHIC) [TAK 95].
3.3.2. Toughness properties of the HAZ
The mechanical properties in the HAZ are the subject to increasingly severe demands from users and of construction codes, with the aim of guaranteeing the viability of the welded joints with respect to brittle fracture. The toughness of the HAZ can clearly be assessed by direct removal of notched samples from within welded joints. However, this method presents many difficulties (delicate positioning of the notch within very narrow zones with high microstructural gradients, residual stresses, etc.) which leads in practice to an excessive spread in the toughness results. This is why the tenacity of these zones is often assessed on the basis of experimental on test-samples that have undergone simulated thermal treatments representative of those encountered in the HAZ.
On the basis of such simulation tests, Figure 3.19 presents the influence of the cooling speed in the coarse grain HAZ on the transition temperature to the level 28 Joules of impact strength. In parallel, it includes the corresponding evolution of hardness in the HAZ.
Figure 3.19. Influence of the cooling speed on impact strength and hardness in the coarse grain HAZ; steel C = 0.18%, Mn =1.4%, Si = 0.4%; from [BER 75]
Looking at the fastest cooling speeds, three successive phases will be noted.
An optimal toughness in the HAZ corresponding to the end of the martensite area. The martensites obtained in welding are frequently self-tempered laths, characterized by the presence of carbide sticks aligned according to the direction <111> of the slats. Indeed, in steels with a low percentage of carbon, the temperature of the end of martensitic transformation MF is so high that it allows a certain tempering of martensite even during the welding cycle. The presence of lower bainite in these predominately martensitic HAZs is rather frequent. The toughness of the HAZ structures obtained from these fast cooling speeds is thus similar to that of tempered martensites and the mixed martensite-bainite structures obtained by traditional heat treatments [HEI 75, OHM 74].
An abrupt drop in toughness is then observed when the cooling speed decreases. This is related to the evolution of lower bainite towards the higher bainite, whose slat sizes or bainitic packages offer less cleavage resistance.
For the slowest cooling speeds (ferrite-carbide structures), a stabilization of toughness is finally observed, possibly after a slight fall in the temperature of transition. As already seen (see section 3.2.2), this cooling speed field corresponds to ferrite extension and certain authors [OTT 80] put forward a connection between the degradation of toughness and the thickness of the ferrite proeutectoïd joints appearing at the old austenitic grain boundaries.
At this stage, it should be noted that simultaneous requirements concerning toughness and resistance to cracking can sometimes appear contradictory: indeed, the guarantee of a minimum toughness can be dependent on the limitation of the cooling parameter (for example:
). On the other hand, the desire to avoid cold cracking can result in the requirement for a minimum cooling speed, in order to avoid too strong a martensitic proportion (generally 80%), that is to say: 
. It can thus be seen that the simultaneous respect of these two criteria can lead to the concept of a metallurgical weldability field between the values Δt1 crit and Δt2 crit.
The influence of various elements present in C-Mn and micro-alloyed steels on toughness in the coarse grain HAZ can be schematized as follows.
Figure 3.20. Influence of the percentage of carbon on the transition temperature TK28J in the HAZ structure (martensite + lower bainite); from [DEV 88]
Carbon decreases the toughness of the HAZs equally in the case of fast and slow cooling speeds. Figure 3.20 thus shows the unfavorable influence of carbon on the transition temperature of impact strength for HAZs of optimal microstructure (martensite-bainite), therefore cooled relatively quickly.
In the case of the slowest coolings (for example, when 
s), carbon and nitrogen (the latter having an even more marked influence) have a detrimental influence on toughness [DEV 88]. Nitrogen’s detrimental effect exists both in as-welded and stress relieved states.
It is also interesting to limit silicon a certain extent, an element which leads to the formation of M-A compounds decreasing the toughness of the coarse grain HAZ and the reheated HAZ in the intercritical field [TAI 95].
Within certain limits (P < 20 × 103%), phosphorus seems to have rather little influence on the HAZ toughness in the case of high energy welding, but has a significantly detrimental effect in the case of welding at low energy levels, in particular after stress relief [GRO 86].
As has been observed previously, controlled additions of titanium slow down the growth of the austenitic grain. This beneficial action on toughness is also reinforced by the fixing of part of the free nitrogen. If moreover the aim is to avoid coarse nitride precipitation in the liquid state, the titanium and nitrogen content must be such that the product of their activity remains lower than the product of solubility in the liquid state. Various pieces of research have shown that effective additions of titanium must be carried out in steels with a low nitrogen content (<5 × 10−3%) and must be themselves fairly low (Ti ≈ 10 × 10−3%) [BAK 97, DEV 88, YAO 90].
Recently developed steels, comprising complex titanium oxides, also show very good toughness properties in the HAZ. In this case, these thermally very stable oxides can be used as sites of ferritic germination with cooling in the HAZ, even at high temperature (1,400°C, for example), and involve the formation of a fine, disorientated acicular ferrite structure.
Boron is a very effective element with which to increase hardenability and to limit the formation of proeutectoïd ferrite or Widmanstätten ferrite. It is however advisable to limit the B content to 10-15 ppm to avoid an embrittlement by precipitation of boro-carbides or the formation of M-A compounds [RAM 86].
The gammagene property of nickel significantly lowers the transformation temperatures in cooling and refines the grain size. From a content of 0.3 to 0.5%, it improves toughness in the HAZ as well as in the base metal.
In the case of slow cooling (for example when 
s) or at the time of a later tempering, the dispersoid elements such as niobium and vanadium can reprecipitate and lead to a certain embrittlement. For a given grade of steel, it should however be noted that this possibly unfavorable influence is largely compensated by the fact that the presence of these elements is generally associated with a significant reduction in the percentage of carbon.
3.3.3. Residual stresses associated with welding
The origin of residual welding stresses is related on the localization of the heat source and the variations of the mechanical properties of materials according to the temperature. Two sources of residual stresses can be distinguished:
– residual stresses, purely thermal in origin, result from the following fact: subjected to a rise in temperature Δθ, an element of the heated central zone should dilate to the degree ε = α Δθ, α indicating the linear dilation coefficient of the material. Actually, this expansion is very limited since the neighboring cold parts play a restraining role. The element is thus subjected to a compression from them. The flow limit being very weak at high temperature, all the deformations that appear correspond to plastic deformations. With cooling, the situation is reversed: the heated central zone cannot retract freely, and is put under pressure by the neighboring zones. After cooling, the heated zones (molten metal and HAZ) will be subjected to residual tensile stresses, the neighboring zones being, for reasons of balance, in compression. In the molten metal, the stresses thus created can be very high, in the order of material yield strength at room temperature. Figure 3.21 presents a typical example of longitudinal distribution or transverse stresses (compared to the direction of welding) within a welded joint;
– residual stresses associated with allotropic transformations on cooling: in the case of construction steels, the transformation of austenite is always accompanied by a more or less marked expansion. If this occurs at relatively low temperature (temperable materials, fast cooling, etc.), the associated deformation will no longer be plastic, but elastic. The expansion, opposed by the neighboring zones, results in the formation of residual compressive stresses.
Superimposed on the thermal stresses, these can decrease the level of tensile stresses, and even lead to compression stresses in the HAZ [MAS 80].
Figure 3.21. Typical example of longitudinal distribution and transverse residual stresses within a welded joint, from [MAC 77]
The residual stresses resulting from welding are superimposed on the in-service stresses, and for this reason play a role in various phenomena or properties: cold cracking, fatigue strength, rupture, stress corrosion, etc.
3.3.4. Influence of residual stress relieving heat treatments in the HAZ
In certain cases, the residual stresses of welding can seriously reduce the service performance of assemblies, and it is necessary to lower their level by stress relieving treatment. The principle aspect of this operation consists of transforming the residual elastic stresses into plastic deformations. In general, this is carried out by a post-welding heat treatment at a temperature such that the elastic limit of material is sufficiently reduced to obtain this transformation. This type of treatment thus comprises:
– a slow heating to a temperature where relief is sufficiently significant;
– a maintenance at this temperature (for example: to around 530 to 630°C for C-Mn and micro-alloyed steels and, 670 to 710°C for Cr-Mo steels, etc.);
– a sufficiently slow cooling so as not to introduce new residual stresses.
We will restrict ourselves here to pointing out some general metallurgical consequences of this type of treatment, by recalling that the precise choice of conditions depends on many factors (type of material, thickness, and of course any regulations in place [BER 78]).
In general, the tempering associated with the relief treatment has a beneficial effect on quenched microstructures, although this is less favorable when the initial structures in the HAZ are ferrite-carbide. A favorable effect caused by the restoration of possible strain hardened-aged zones during welding, or after cold working should also be noted.
As has been seen, the precipitation of micro-alloying elements (Nb, V) is generally incomplete during coolings in the HAZ. The complementary precipitation which can occur during the stress relieving heat treatment will generally have a detrimental influence on toughness. Thus, the metallurgical effect of the treatment will often result from a conflict between a mechanism of matrix softening (favorable) and a more or less significant precipitation (unfavorable).
The tempering treatment can sometimes have a post-heating role, allowing the diffusion of hydrogen out of the welded joints, lowering the levels of hardness, and thus reducing the risk of cold cracking (see section 3.2.4).
In the case of low alloyed steels, the additions of Mn, Cr, Ni can lead to a certain deterioration of tenacity for a stress relief treatment followed by a slow cooling, contrary to the additions of Mo [BLO 80]. This phenomenon is more accentuated the coarser the grain involved and the greater the temper. To avoid the problems of reversible temper brittleness [BLO 88], it is important to limit the impurity content, which can be achieved by, for example, using the Bruscato parameter: 
= (10P + 4Sn + 5Sb + As)/100, lower than a value of 10 for demanding specifications [BER 97].
In certain cases (low alloy steels, alloys containing nickel, etc.), it is observed that stress relief heat treatments lead to an intergranular cracking in the HAZ, known as reheat cracking [DHO 87, DHO 92].
This phenomenon occurs in the presence of significant residual stresses, accentuated particularly by localized stresses caused by the geometry, in coarse grain zones weakened by intergranular precipitations (AlN, Mo, Nb, V carbides) and segregated impurities (S, P, As, Sn, Sb, etc.) within these same joints. The appearance of such cracks is generally interpreted from a precipitation hardening of the grain interior during the post-welding treatment, which limits all stress relief or plastic deformation within the grains, by concentrating them on the less resistant grain boundaries. The prevention of reheat cracking thus involves the adoption of less sensitive materials, but also by use of suitable welding conditions: low resistance molten metal, buttering, increased contribution of heat and/or pre-heating, use of the temper bead method, i.e. narrow passes to achieve a refinement of the preceding ones, etc.
3.4. Bibliography
[AND 65] ANDREWS K.W., “Empirical formulae for the calculation of transformation temperature”, Journal of the Iron and Steel Institute, no. 7, July 1965.
[ATS 80] ATS/OTUA, “Conseils pour le soudage des aciers de construction métallique et chaudronnée à limite d’élasticité garantie ≤ 420 N/mm2”, Cahiers de l’ATS/OTUA, 1980.
[BAK 97] BAKER T.N., Titanium Technology in Microalloyed Steels, Institute of Materials Publications, 1997.
[BER 72] BERNARD G., PRUDHOMME M., “Compléments à l’étude des phénomènes thermiques dans les joints soudés”, Revue de Métallurgie, July-August 1972.
[BER 75] BERNARD G., “A view-point on the weldability of C-Mn and microalloyed structural steels”, Microalloying Symposium 75, Washington, October 1975.
[BER 78] BERNARD G., Opportunité des traitements thermiques de relaxation dans les constructions soudées, Doc. IIS/IIW no. X.913-78, 1978.
[BER 97] BERANGER G., HENRY G., LABBE G., SOULIGNAC P., Les aciers spéciaux, Lavoisier, 1997.
[BHA 92] BHADESHIA H.K.D.H., Bainite in Steels, Institute of Materials, 1992.
[BLO 80] BLONDEAU R., “Les aciers faiblement alliés soudables, influences des éléments d’addition”, Soudage et Techniques Connexes, January-February 1980.
[BLO 88] BLONDEAU R., “La fragilité de revenu réversible, facteurs et remèdes”, Traitement thermique, 220, 1988.
[CHA 74] CHANDEL R.S., GARBER S., “Mechanical and metallurgical aspects of spot-welded joints in heat-treated low-carbon mild steel sheet”, Metals Technology, September 1974.
[CHR 65] CHRISTENSEN N., DAVIES V. DE L., GJERMUNDSEN K., “Distribution of temperatures in arc welding”, British Welding Journal, February 1965.
[CLY 58] CLYDE M., ADAMS C.M., “Cooling rates and peak temperatures in fusion welding”, Welding Research Supplement, May 1958.
[CON 87] CONNOR L.P., Welding Handbook, volume 1, American Welding Society, 1987.
[CON 92] CONSTANT A., HENRY G., CHARBONNIER J.C., Principes de base des traitements thermomécaniques et thermochimiques des aciers, PYC Edition, 1992.
[DEA 40] DEARDEN J., O’NEILL, “A guide to the selection and welding of low alloy structural steels”, Trans. Int. Welding, 1940.
[DEB 82] DEBROY T., SZEKELY J., EAGAR T.W., “Temperatures profiles, the size of the Heat-Affected Zone and dilution in electroslag welding”, Materials Science and Engineering, no. 56, 1982.
[DEF 75] DEFOURNY J., BRAGARD A., “Caractérisation des cycles de température en soudage bout à bout à l’arc submergé de tôles d’aciers au moyen de deux grandeurs du champ thermique”, Revue Belge de la Soudure, no. 3, Lastijdschrift, 1975.
[DEV 83] DEVILLERS L., KAPLAN D., MOUSSY F., Soudabilité des aciers trempés et revenus obtenus par trempe directe, Final research report, CECA no. 7210 KA/308, 1983.
[DEV 84] DEVILLERS L., KAPLAN D., MOUSSY F., Prévision et comportement de la zone adoucie des aciers trempés-revenus, IRSID Report no. RE 1137, September 1984.
[DEV 85] DEVILLERS L., KAPLAN D., Propriétés d’emploi d’assemblages obtenus par faisceau d’électrons des tôles de forte épaisseur en acier de construction, Final research report CECA no. 7210 KA/310, IRSID Report RE 1252, December 1985.
[DEV 86a] DEVILLERS L., KAPLAN D., MAAS E., “Aspects mécaniques de l’évaluation de la ténacité d’éprouvettes métallurgiquement hétérogènes”, Revue de Métallurgie, CIT, April 1986.
[DEV 86b] DEVILLERS L., KAPLAN D., Interpretation of deviated fracture of electron beam steel welds in notch bending tests, IRSID Report no. RE 1319, 1986.
[DEV 87] DEVILLERS L., KAPLAN D., JANSEN J.P., “Simulation thermique des zones affectées par la chaleur en soudage multipasse des aciers de construction”, Soudage et Techniques connexes, March-April 1987.
[DEV 88] DEVILLERS L., KAPLAN D., JANSEN J.P., “Aspects thermiques et métallurgiques de la ténacité des ZAC”, Revue de Métallurgie, CIT, March 1988.
[DEV 89] DEVILLERS L., KAPLAN D., Aspects métallurgiques de l’assemblage des tôles de faible épaisseur par LASER CO2, Final research report CECA no. 7210 KA/314, IRSID Report RCA 89333, October 1989.
[DEV 91] DEVILLERS L., KAPLAN D., TESTARD P., An approach for predicting microstrucrures and toughness properties in HAZ of multipass welds, IRSID Report 91316, March 1991.
[DHO 87] DHOOGE A., VINCKIER A., “Reheat cracking — a review of recent studies”, Int. J. Pres. Ves. and Piping, 27, 1987.
[DHO 92] DHOOGE A., VINCKIER A., “La fissuration au réchauffage- revue des études récentes (1984-1990)”, Le Soudage dans le Monde, vol. 30, no. 3-4, 1992.
[EAS 83] EASTERLING K., Introduction to the Physical Metallurgy of Welding, Butterworths, 1983.
[GOL 76] GOLDAK J., BIBBY M., “The toughness of EB welds in HSLA steels”, Welding of HSLA Structural Steels Conference, Rome, ASME Publications, 1976.
[GRA 95] GRANJON H., Bases métallurgiques du soudage, Publications du soudage et ses applications, 2nd edition, 1995.
[GRO 86] GRONG O., AKSELSEN O.M., “HAZ toughness of microalloyed steels for offshore”, Metal Construction, September 1986.
[GRO 94] GRONG O., Metallurgical Modelling of Welding, Institute of Materials, 1994.
[GRU 81] GRUMBACH M., Les essais de dureté, Collection IRSID-OTUA, 1981.
[HAL 84] HALE G.E., NUTTING J., “Overheating of low-alloy steels”, International Metals Review, vol. 29, no. 4, 1984.
[HEH 72] HEHEMANN R.F., KINSMAN K.R., AARONSON H.I., “A debate on the bainite reaction”, Metallurgical Transactions, vol. 3, May 1972.
[HEI 75] HEISER F.A., DE FRIES R.S., “Temper embrittlement in steel for thick wall gun tubes”, Journal of Metals, January 1975.
[HON 95] HONEYCOMBE R.W.K., BHADESHIA H.K.D.H., Steels, Microstructure and Properties, Edward Arnold, 1995.
[HRI 95] HRIVNAK I., “Weldability of modern steel materials”, ISIJ International, vol. 35, 1995.
[IKA 80] IKAWA H., OSHIGE H., TANOUE T., Effect of M-A constituent on HAZ toughness on a high strength steel, Document IIS-IIW, no. IX-1156-80, 1980.
[IKE 95] IKEUCHI K., LIAO J., TANABE H., MATSUDA F., “Effect of temper-bead thermal cycle on toughness of weld ICCGHAZ of low alloy steel SQV-2A”, ISIJ International, vol. 35, 1995.
[INO 98] INOUE K., ISIJ International, vol. 38, no. 9, 1998.
[ION 97] ION J.C., “Modelling of LASER welding of Carbon-Manganese steels”, in H.C. Cerjak (ed.), Mathematical Modelling of Weld Phenomena 3, Institute of Materials, 1997.
[IRS 77] IRSID, “Courbes dureté-paramètre de refroidissement en conditions de soudage”, collection established at the IRSID, 1977.
[ITO 00] ITO R., SHIGA C., KAWAGUCHI Y., HAYASHI T., SHINOHARA H., TORIZUKA S., Microstructural analysis of softening region on weld HAZ of ultra-fine grained steels, Document Institut International de la Soudure no. IX-1968-00, 2000.
[KAN 76] KANAZAWA S., “Improvement of weld fusion zone toughness by fine TiN”, Transactions ISIJ, vol. 16, 1976.
[KAP 98] KAPLAN D., STUREL T., “Couplage d’un modèle métallurgique de soudage et de critères locaux de rupture pour prévoir la ténacité des joints soudés multipasses”, 9èmes Journées AFIAP, Paris, October 1998.
[KAW nd] KAWABATA F., AMANO K., TOYODA M., MINAMI F., Tempering effect by succeeding weld passes on multi-layered HAZ toughness, doc. Kawasaki Steel, no date.
[KOS 81] KOSO M., MIURA M., OHMORI Y., “Microstructure and toughness of weld HAZ in 785 MNm−2 steel”, Metals Technology, December 1981.
[KOU 86] KOU S., WANG Y.H., “Computer simulation of convection in moving arc weld pools”, Met. Trans. A, vol. 17A, December 1986.
[LAM 99] LAMBERT A., DRILLET J., GOURGUES A.F., STUREL T., PINEAU A., Microstructure of M-A constituent in HAZs of HSLA steel welds in relation with toughness properties, IRSID report, no. 99 21, 1999.
[LIN 89] LIN Y., ABKEN M.G., BOWKER J.T., “Mechanical and microstructural analysis of multi-pass welded HAZ”, Proceedings of Recent Trends in Welding Science and Technology, ASM International, Gatlinburg, 1989.
[LON 94] LONGAYGUE X., KAPLAN D., MAURICKX T., “Etude paramétrique de l’influence des facteurs d’hétérogénéité mécanique sur le comportement en service des assemblages soudés”, Revue de Métallurgie, CIT, June 1994.
[MAK 77] MACHERAUCH E., WOHLFAHRT H., “Different sources of residual stress as a result of welding”, in Residual Stresses in Welded Construction and their Effects, The Welding Institute Publications, London, 1977.
[MAR 80] MARQUET F., DEFOURNY J., MARIQUE C., MATHY H., BRAGARD A., Recherches sur les connaissances de base en vue d’améliorer le soudage des aciers sous haut apport calorifique, Rapport final de recherche CECA no. 7210 LA/303, 1980.
[MAS 80] MASUBUCHI K., Analysis of Welded Structures, Pergamon Press, 1980.
[MAT 94] MATSUDA F., FUKADA Y., OKADA H., SHIGA C., IKEUCHI K., HORII Y., SHIWAKU T., SUZUKI S., Review of Mechanical and Metallurgical Investigations of M-A constituent in Japan, Doc. IIS-IIW no. IX-1782-94, 1994.
[MIE 92] MIETTINEN J., “Mathematical simulation of interdendritic solidification of low-alloyed and stainless steels”, Acta Polytechnica Scandinavica, Chemical Technology and Metallurgy Series, no. 207, 1992.
[MUR 67] MURRY G., PRUDHOMME M., CONSTANT A., “Etude des transformations dans la zone thermiquement affectée lors du soudage et son intérêt pour l’estimation de la dureté sous cordon”, Revue de Métallurgie, 64, no. 6, June 1967.
[NIL 75] NILES R.W., JACKSON C.E., “Weld thermal efficiency of the GTAW process”, Welding Journal, Welding Research Supplement, January 1975.
[OHM 74] OHMORI Y., OHTANI H., KUNITAKE T., “Tempering of the bainite/martensite duplex structure in a low-carbon low-alloy steel”, Met. Sci., vol. 8, 1974.
[OKA 92] OKAGUCHI S., ISIJ International, vol. 32, no. 3, 1992.
[OLE 81] OLETTE M., GATELLIER C., “Action d’une addition de calcium, de magnésium ou des éléments des terres rares sur la propreté inclusionnaire des aciers”, Revue de Métallurgie, CIT, December 1981.
[OTT 80] OTTERBERG R., SANDSTRÖM R., SANDBERG A., “Influence of Widmanstätten ferrite on mechanical properties of microalloyed steels”, Metals Technology, October 1980.
[POR 34] PORTEVIN A., SEFERIAN D., XIe Congrès de l’Acétylène et de la Soudure Autogène, Rome, vol.3, 1934.
[RAM 86] RAMBERG M., AKSELSEN O.M., GRONG O., “Embrittlement phenomena in the intercritical HAZ of boron-containing steels”, Advances in Welding Science and Technology, Publications de l’ASM, Gatlinburg, 1986.
[REE 94] REED R.C., BHADESHIA H.K.D.H., “A simple model for multipass steel welds”, Acta Metall. Mater., vol. 42, no. 11, 1994.
[RIS 74] RIST A., ANCEY-MORET M.F., GATTELLIER C., RIBOUD P.V., “Equilibres thermodynamiques dans l’élaboration de la fonte et de l’acier”, Techniques de l’Ingénieur, M 1733, 1974.
[ROS 46] ROSENTHAL D., “The theory of moving sources of heat and its applications to metals treatments”, Transactions ASME, vol. 68, 849, 1946.
[RYK 61] RYKALINE N.N., “Calcul des processus thermiques de soudage”, Soudage et techniques connexes, no. 1-2, 1961.
[TAI 95] TAILLARD R., VERRIER P., MAURICKX T., FOCT J., “Effect of silicon on CGHAZ toughness and microstructure of microalloyed steels”, Met. Trans. A, vol. 26, February 1995.
[TAK 95] TAKAHASHI A., OGAWA H., “Influence of softened heat-affected zone on stress oriented hydrogen induced cracking of a high strength line pipe steel”, ISIJ International, vol. 35, 1995.
[TOY 86] TOYODA M., Fracture toughness evaluation of steel welds (Review), Osaka University Document, July 1986.
[TOY 92] TOYODA M. (ed.), Proceedings of the Workshop Mismatching and its Control, Osaka University, Tokyo, July 1992.
[WEI 95] WEISS D., SCHMIDT J., FRANZ U., “A model of temperature distribution and weld pool deformation during arc welding”, Mathematical Modelling of Weld Phenomena 2, Institute of Materials, 1995.
[YAO 90] YAO S., KOÇAK M., Influence of titanium on HAZ microstucture and toughness of offshore steel welds: literature review — part 1, Report GKSS, no. 90/E/39, 1990.
[YUR 87] YURIOKA N., OKUMURA M., KASUYA T., COTTON H.J.U., “Prediction of HAZ hardness of transformable steels”, Metal Construction, April 1987.
[ZAC 93] ZACHARIA T., DAVID S.A., “Heat and fluid flow in welding”, Mathematical Modelling of Weld Phenomena, Institute of Materials, 1993.
[ZEN 48] ZENER C., Transactions ASME, no. 45, 1948.
1 Chapter written by Dominique KAPLAN and Guy MURRY.
1 The notions of thin or thick products will be quantified more precisely later.
2 Other similar expressions have been proposed: HVM = 293+812(%C) (Satoh), 305+802(%C) (Duren) or HVM= 294+884[%C(1-0.3 (%C))2] (Yurioka), etc.