Final Report Summary - HILDA (High Integrity Low Distortion Assembly)
Executive Summary:
HILDA is intended to promote and facilitate the use of Friction Stir Welding (FSW) in the joining of steel panels and components in assemblies required for the shipbuilding industry in Europe. Presently, this process is carried out using various techniques used in the fabrication and building of ships. These techniques involve high heat input to establish localised consolidation of the metals being welded together. The FSW process has many advantages over these processes, amongst these are low temperature input resulting in low distortion, simplified preparation, joining of dissimilar metals possible, improved design capability and high weld integrity.
Project Context and Objectives:
HILDA will deliver a cost effective, low distortion welding process for EU shipyards to allow them to maintain competitiveness and produce light, strong, more fuel-efficient vessels. The solid state technique will enable the modular construction of dimensionally accurate, high strength, corrosion resistant fabrications across the entire range of steels, enabling the introduction of stronger, tougher, corrosion resistant steels into the industry. HILDA will develop the fundamental metallurgical knowledge required to predict the complex phase changes and stress regimes present in welding steel. This will enable the proven, energy efficient, low hazard and environmentally benign technology of friction stir welding, widely deployed in aluminium construction for the aerospace and rail industries, to be transferred to steel shipbuilding. To achieve this breakthrough, HILDA will use real world data from FSW of high strength low alloy steels as an input to develop a computer model of the thermo-mechanical processing process at the heart of the solid state welding technique.
A combination of numerical and experimental techniques will be implemented to determine sensitivity over a wide range of process input variables and to understand and characterize these parameters in order to enhance the microstructure and mechanical properties while mitigating the effects of post weld distortion. The general aims of the program are summarized below:
• Understand and characterize the process quality in relation to metallurgical, corrosion and wear performance.
• Understand, trace and quantify relationships between intermetallic reactions (if any) and process parameters.
• Develop a predictive interrelationship between computational thermal transients / plastic deformations and particle dispersion characteristics, through an innovative analytical approach based on the continuous dynamic recrystallization phenomenon.
• Determine optimum process parameters via numerical and analytical sensitivity studies.
• Transfer knowledge of methods to the industrial and academic communities by providing design guidelines on FSW of steel over a range of thicknesses.
The primary objective of this work is to create a knowledge base on FSW that is applicable to FSW of steel ship structures with specific microstructural and mechanical properties that not only exceeds the limitations of conventional joining processes but accomplishes this in a cost advantageous manner with minimal distortion. By establishing a European wide collaborative effort that builds on the commercial, technical and academic merits of the carefully selected partnership, it is believed that the research outcomes will be highly attractive to the Eurozone ship building and ship repair industries.
Project Results:
Work Package 1: Scientific and Technical results
At the commencement of project HILDA, the state of the art for FSW of steel can be summarised as follows:
• FSW of steel had been demonstrated to be achievable in the laboratory but only in short lengths, often less than 1 m, as the tools used had a very short lifetime.
• The welding speeds attained were generally approx. 100 mm/min in 6 mm thick steel.
• Welds had only been produced in the butt and lap configurations.
• The FSW environment in steel, for example the temperatures and forces generated within the weld zone, was largely unknown.
• The FSW process parameter envelope, i.e. the combinations of tool rotational speed and tool traverse speed that will result in a good weld, was unquantified.
• The influence of process parameters on the weld microstructure was largely unknown.
• The influence of process parameters upon the mechanical properties of the weld was unknown.
• The effect of process parameters upon distortion in the welded fabrication was unquantified.
• The influence of process parameters upon tool life was largely unknown.
By the end of the HILDA project, significant advances had been made in the FSW of steel. Within work package 1, TWI has significantly expanded the process envelope for steel FSW and developed a greater understanding of the thermo-mechanical process involved in making welds. The influence of the process control parameters on the weld quality has been investigated jointly with the University of Strathclyde (UoS, work package 2), and the process shown to be capable of producing fabrications with low distortion and excellent mechanical properties though tool cost and life considerations currently make FSW more expensive than GMAW and laser welding.
The process advancements made during HILDA can be summarised as:
• 6 mm thick DH36 steel can be friction stir welded at speeds of up to 500 mm/min, significantly surpassing the targeted benchmark fusion welding speed of 350 mm/min. Preliminary work to further extend the welding traverse speeds has showed that DH36 steel can potentially be welded at up to 700 mm/min.
• FSW speeds of 300–450 mm/min produce a good balance of productivity, tool life and weld properties, making the process technically competitive with equivalent arc welding processes.
• The forces generated during FSW of 6 mm thick steel can be maintained below a level that will allow welds to be produced on existing equipment developed for the FSW of light metals.
• The spindle rotational speed and concentricity requirements for the FSW of steel are no greater than those used in many existing FSW machines for welding light metals (e.g. aluminium, magnesium).
• Welds can be produced in either position or force control. The ability to utilise position control, combined with the low tool rotational speed requirements, may permit the use of converted milling machines for FSW of DH36 steel, thus potentially reducing the initial capital costs required to adopt the process.
• The temperatures generated during steel FSW are lower than in fusion welding but considerably greater than the temperatures during aluminium FSW (Figure 1).
• The FSW process in steel appears tolerant of workpiece material defects and poor prior preparation.
• FSW of DH36 steel produces significant benefits in terms of reduced Health and Safety risks when compared with conventional fusion welding techniques.
During the weldability investigation, it quickly became apparent that the current drawbacks of the FSW process in steel are centred on the available tool technology. These are:
• There is a monopoly supplier of tools for steel, based outside the EU and thus the quality and security in the supply of tools cannot be assured.
• The FSW tools currently available for steel are expensive and short-lived, leading to a typical best case welding cost of approx. €77 / metre for the good quality FSW tools alone.
• The current tool material technology places limits on the design of the tools, thus restricts the weld geometries that can currently be achieved. In addition, these design limitations also make the process in steel more susceptible to root defects than is the case with other metals.
Moreover, three aspects of the technical progress made in steel FSW are of particular significance:
1. The generation of a body of knowledge regarding the interaction between the process parameters, the microstructures that these produce in the weld zone and the mechanical properties of the welds.
2. The increased tolerance of the process to defects; FSW is seen to produce welds with properties that exceed those of conventional fusion welds even in the presence of defects (reported more fully by UoS in work package 2).
3. The generation of data regarding the FSW environment in steel and the requirements on tool materials for welding steel.
The first aspect, the generation of knowledge by UoS on the influence of the process parameters on the weld microstructure / mechanical property relationship offers the potential to control the process for specific results. For example, it may be possible to tailor the processing conditions to produce welds that are exceptionally tough for use in cryogenic applications, or which have enhanced fatigue properties, or a more general balance of properties for non-specialist applications. The data generated also provide the underlying knowledge required for the modelling tasks undertaken by the University of Malta (work package 3) and Cenaero (work package 4) to enable the development by GeonX (work package 5) of the expert system (software) that will predict the properties and distortion associated with the FSW of steel.
The high defect tolerance of the process, for instance failure to remove surface oxides from the steel plates prior to welding or the presence of a root defect due to tool limitations, makes for an industrially robust welding technique and mitigates to an extent the current issues arising from suboptimum tool materials and designs. It will however require that end users and classification societies reconsider some fundamental assumptions regarding the presence of weld defects and approach approval regimes with a view to accepting “fitness for purpose” as a goal rather than freedom from defects. Lloyd’s Register (LR) is leading the work in this area, ensuring that an appropriate regulatory framework is in place to expedite the industrial application of FSW in Europe. The relevant report produced by LR, “Application of Friction Stir Welding of Steels for the Ship building Industry”, has been appended to this document.
The data generated on the welding environment, in particular the temperatures and forces experienced by the tool, will be invaluable in generating performance specifications for, and guiding the development of, next generation tool materials for steel FSW that improve both the technical performance and economic viability of the technology.
Work Package 2: Scientific and Technical results
As demonstrated by previous research, friction stir welding (FSW) of steel presents numerous advantages across many industrial sectors and especially for marine applications compared to conventional fusion welding techniques. However, the fundamental knowledge of the process on steel remained relatively limited prior to project HILDA, thus obstructing the industrial uptake of the process on steel. In the course of WP2, the University of Strathclyde (UoS) has carried out a large-scale microstructure and property evaluation of friction stir welded 6 mm thick low alloy steel grade DH36 plate, commonly used in the European shipbuilding industry. The in-depth experimental testing programmes examined butt welds produced by a wide range of process parameters through microstructural characterisation and mechanical testing, also focusing on the interactions between weld metallurgy and associated mechanical properties. The key findings of these programmes are outlined below.
A preliminary process parameter envelope has been developed in close collaboration with TWI by employing varying tool rotational and traverse speeds. The trialled welding traverse speeds commenced from the originally recommended by the FSW tool manufacturer (100 mm/min) and were gradually increased up to 500 mm/min. By substantially increasing the range of attainable traverse speeds, this development programme has concluded that high speed FSW of steel is feasible; new parameter sets have been established which deliver high integrity welds and enhance the process’s techno-economic competitiveness relative to fusion welding methods through a step change increase in the conventionally recognised welding speed. Expanding the process envelope even more should be made possible when various tooling issues (encountered during the project and discussed by TWI in deliverables Del 1.2 and 6.1) are appropriately addressed through new research.
High traverse speed FSW is expected to affect the weld quality. For this purpose, metallographic examination of the 2000 mm long welds was performed to characterise the evolved microstructure, thus offer an insight into the anticipated mechanical property behaviour of the weldments. This analysis has established that significant grain refinement of the original DH36 parent material microstructure is promoted during welding. Moreover, FSW of steel creates a complex metallurgical system which is highly dependent on the employed process parameters. The slow traverse speeds (100-200 mm/min) generate a microstructure of refined ferrite grains, the intermediate traverse speeds (250-400 mm/min) create an acicular bainitic ferrite-rich microstructure and the fast traverse speeds (450-500 mm/min) give rise to heterogeneous microstructures of acicular ferrites and acicular bainitic ferrites. Although the assessed welding parameters have produced no defects in the main body of the weld zone hence increasing the confidence in the quality of high speed FSW, two types of process related flaws have been detected in most welds. Lack of penetration initiating a weld root flaw and the top surface breaking flaws require tackling through new developments in the processing conditions and the FSW tool for steel.
Mechanical testing of the welds was conducted to assess the evolution of mechanical properties with increasing tool rotational and traverse speed and evaluate the impact of the complex metallurgical system in steel FSW. Transverse tensile testing has consistently demonstrated that all slow and intermediate welds have higher yield and tensile strength compared to the parent material regardless of minor surface breaking flaws (as characterised above). Most fast weld samples fractured in the weld zone or were classed as unstable, suggesting reduced tolerance to parameter variations; hence, optimisation of the high speed welding parameters (450-500 mm/min) is critical. The weld hardness is seen to increase with increasing traverse speed and this is attributed to the evolution of harder phases in the microstructure. Still, the majority of the examined welds showed hardness distribution within classification society rules. Notably, an improvement in impact toughness with increasing traverse speed has been recorded and this provides additional weight to the potential of high speed steel FSW.
Since the fatigue performance of welded components is of paramount importance in marine applications, this has been thoroughly assessed by producing new DH36 steel welds based on the findings in the previously outlined testing programmes. An extensive standard operating procedure document, specific to FSW, was drafted and executed in collaboration with LR and CMT to address the relevant knowledge gap for this process on low alloy steel and generating novel data for future assessment by classification societies. The weldments’ yield strength, geometry, metallurgical features and hardness distribution were recorded in support of the fatigue testing. The effect of varying welding parameters was also evaluated. Microstructural observations, hardness measurements and transverse tensile properties were in excellent agreement with the corresponding in the previous testing programmes. All examined steel grade DH36 friction stir welds exhibited excellent fatigue performance, better than relevant international recommendations for fusion welding and equal to high quality laser welding, irrespective of minor surface breaking flaws which were identified. The relation between weld flaws and fatigue performance has been established through metallographic examination, fracture surface analysis and additional fatigue testing with the critical defect removed. This has revealed that minor embedded flaws do not initiate cracks whilst surface breaking irregularities, i.e. lack of penetration or the indentations on the weld top surface by the tool shoulder are the critical factor for crack propagation. The latter is a serious challenge for future FSW process developments which, when addressed, is expected to upgrade the welds’ fatigue life dramatically.
In parallel, knowledge of the thermo-mechanical response of the examined material is equally important with regard to further refinement of process parameters and future improvements in tooling and welding machine specifications. To this end, thermo-mechanical deformation studies were implemented in partnership with Cenaero by hot compression testing in conditions, i.e. temperature and strain rate, which simulate the actual FSW process on a Gleeble thermo-mechanical simulator. This novel investigation has determined that the alloy’s flow stress is appreciably affected by the test temperature and strain rate, i.e. increasing with decreasing temperature and increasing strain rate. This significant finding relates to FSW of steel since it suggests that process parameters which produce regions of low temperature and high strain rate in a friction stir weld may promote the formation of embedded flaws due to inadequate material flow. This flow stress evolution is only opposed by the tests at 800oC, in which increased flow stress is recorded as compared to a lower test temperature for constant strain rate. This observation provides an insight into future process parameter optimisation work since it specifies that welding in the vicinity of 800oC is better avoided to reduce the forces exerted on the FSW tool for steel. The original material property data were employed by the University of Malta and Cenaero for validation of their predictive models.
The key technical objectives of WP2 were the expansion of the scientific knowledge on FSW of steel in support of the process’s future introduction in industry and the generation of data not previously available on steel FSW for the validation of the predictive local and global models formulated in WP3 and WP4. It is evident from the preceding discussion that both objectives have been successfully fulfilled; with respect to the former, the testing programmes conducted by UoS have advanced the understanding of the underlying phenomena during steel FSW and identified high speed welding parameter sets which result in an attractive combination of acceptable quality level and economically competitive processing conditions. The latter objective has been realised by producing a novel and wide-ranging experimental dataset on DH36 steel which has been supplied to UoM and Cenaero. Additionally, the scientific knowledge and the corresponding recorded datasets will favour the progress in process control regimes and tooling design and material, an area in which improvements are greatly desired. Finally, the noteworthy findings on the steel friction stir welds’ performance have formed the basis of extensive dissemination activities; these are summarised in Del 2.1 2.2 and 7.3.
Work Package 3: Scientific and Technical results
Work package 3 set out to develop and validate global thermo-elastoplastic FSW numerical models. These models were then used to investigate the influencing parameters giving rise to residual stresses and distortion. The numerical models are integrated with the local models and together provide an expert system platform that was developed in work package 5. The scheme of work undertaken within this work package followed 3 main tasks (detailed in the periodic report) and these were all fully achieved.
The University of Malta (UoM) has led work package 3 which was primarily focused on developing global finite element model for the prediction of residual stresses and distortion. The models were validated with experimental tests results. These included transient temperature and strain measurements, residuals stresses and distortion measurements and provided significant information towards the knowledge and understanding of the FSW process. These tests were in part conducted within work package 3 together with the work performed in work package 1 (TWI) and 2 (University of Strathclyde). Adequate calibration with heat loss and other influencing parameters was conducted leading to good correlation with experimental tests.
Three thermo-elastoplastic models were developed, ranging from relatively simple analytical algorithms that are able to define the welding contraction forces and angular distortion to a fully transient non-linear material properties thermo-elastoplastic numerical model capable of predicting the evolution of residual stresses and distortion. The latter requires significant computational time and consequently a hybrid model that takes into account all non-linearities and influences due to heat sinks and clamping boundary conditions was developed. This model provides an accurate representation of the residual stresses and distortion using less computational time. The development and validation of these models was presented in deliverable Del 3.1 and includes:
• Phase I models: Reduced solution methods based on elastoplastic analytical algorithms that identify FSW contraction forces and applied to elastic static room temperature material properties structural models.
• Phase II models: Hybrid models that take into account non-linear material properties but are solved in a two-step heating and cooling cycle in a static manner.
• Phase III models: Transient elastoplastic models that apply the developed thermal strains in a transient manner with non-linear material properties included.
The local – global modelling strategy adopted in this study and applied to FSW of DH36 steel is shown in Figure 2.
Following the development of the global models, work package 3 focused on using these models to understand and optimise the FSW process by identifying the influential parameters affecting residual stresses and distortion. A series of welding parameters and boundary conditions were considered, ranging from slow to fast traverse speed FSW. In this case, no thermocouple measurements were available and thus the local models developed in work package 4 (Cenaero) were used to identify the thermal transients, material flow, microstructural properties and TMAZ. All these parameters are required for the successful numerical modelling through the global models. The results of this parametric study were presented in deliverable Del 3.2. An analytical and numerical approach was adopted to investigate the following effects:
• Heat loss to machine bed
• Tool penetration depth
• Structural restraint and pressure developed by clamping
• Tool traverse and rotational speed
• Plate stiffness with respect to FSW
Figure 3 shows the typical predicted out-of-plane distortion for different heat loss parameters.
Within work package 3, UoM generated comprehensive global numerical models capable of predicting the residual stresses and distortion developed due to FSW. Various numerical modelling strategies were adopted, ranging from computationally efficient models to more robust non-linear thermos-elastoplastic models that were integrated with the local models. All tasks and deliverables were attained and reported. The main conclusions of the numerical parametric study include:
• The heat loss to the machine bed highly influences the development of thermal strains. Appropriate insulation or heat extraction can be developed for specific welding parameters leading to more efficient welds in terms of process, residual stresses and distortion.
• The tool shape and penetration depth are also crucial. Higher penetration depth leads to more stirring and shearing thereby more heat input, residual stresses and distortion. However, it is important that the tool is significantly plunged into the material to ensure the production of fully penetrated welds.
• The FSW process on steel requires significant clamping that in turn influences the evolution of residual stresses and distortion. For instance, free angular distortion is inhibited as a result of clamping. Furthermore the application of side clamping reduced the longitudinal residual stresses in a similar fashion to mechanical tensioning in conventional welding.
• A sensitivity analysis on the different welding parameters was also conducted. It was concluded that faster welds generally lead to less distortion; however, this is highly dependent on the structural stiffness of the plate and heat loss boundary conditions.
The models developed in this work package were embedded to the local models (Cenaero) in the expert system implemented in work package 5 (GeonX), hence providing design engineers with a platform to analyse and optimise FSW for different plate dimensions and configurations, together with the investigation of influencing parameters.
Dissemination was also fundamental towards the successful completion of this project. The main contributions from the University of Malta are outlined in Del 7.3.
Work Package 4: Scientific and Technical results
Work package 4 (Cenaero) is dedicated to the microstructural and thermo-mechanical modelling of FSW of DH36 steel plates in a butt joint configuration. The numerical model is used to obtain material flow, metallurgical characteristics and process parameters influencing the microstructure of the welded joint.
The local thermo-mechanical phenomena which occur during the welding stage of the process are described by a thermo-fluid model. This local model is based on incompressible flow formulation and is used to simulate the velocity and temperature distributions induced by the stirring action of the rotating tool between the plates being welded. The temperature distribution and material flow are computed by solving the continuity, momentum and energy equations for incompressible flow with appropriate contact / friction conditions. These contact / friction conditions assist in modelling the friction between the rotating FSW tool and steel workpieces which produces frictional shear stresses and frictional heat.
The thermo-fluid model is based on a rigid visco-plastic approximation and is described by an Eulerian formulation. In this model, the material is analysed as a viscous fluid flowing across an Eulerian mesh and interacting with the rotating FSW tool. The type of Eulerian formulation used depends on the tool’s axisymmetric properties. The friction and boundary conditions are adapted to the type of Eulerian description depending on process parameters and tool geometrical features.
In the numerical model, the Norton Hoff visco-plastic law which is proportional to the effective viscosity is proposed to model the behaviour of the steel. The identification of the viscosity is attained from the evolution of flow stress as a function of temperature and strain rate through data generated on a Gleeble machine (Figure 4) as provided by work package 2 (University of Strathclyde, UoS). The evolution of thermal properties has also been determined experimentally by UoS as a function of temperature.
Material properties have been identified on a range of temperatures and strain rates which closely match the conditions occurring during FSW of DH36 steel. For each test temperature of the experimental plan, the material consistency and strain rate sensitivity are determined by inverse analysis as represented in Figure 5.
A 3D numerical analysis of material and thermal flow during FSW of steel grade DH36 in the case of an axisymmetric tool was produced based on the alloy's visco-plastic behaviour. Temperature distribution examples are represented in Figure 6.
An initial numerical sensitivity analysis showed the influence of friction coefficient on the mechanical and thermal response. The workpiece velocity is strongly related to the friction coefficient value, particularly in the thermo-mechanical affected zone (TMAZ). The velocity tends to be the same as the tool velocity by increasing the friction coefficient value and as plastic deformation is becoming higher than friction heat flux.
The thermo-fluid model provides the metallurgical model with the velocity and temperature distributions in order to predict grain size evolution and phase fraction. The prediction of the grain size evolution is addressed by JMAK model which takes into account all stages of the discontinuous dynamic recrystallization. To predict the phase fraction, a user interface enables the use of several existing phase transformation models from thermo-fluid model solutions. A streamlines algorithm is performed after the thermo-fluid model in order to determine temperature, strain rate and strain cycles. These cycles are extracted along computed particle path lines during FSW.
A numerical sensitivity analysis has been undertaken to evaluate the influence of process parameters such as tool traverse and rotational speed on the FSW process. The above two process parameters are the most significant in FSW of steel and control the generated heat, yielding and cooling rate within the weld zone. A wide range of process parameters has been explored using the thermo-fluid and the microstructural evolution models (parameters classified in slow, intermediate and fast groups).
For a fixed rotational speed, an increase in the traverse speed reduces the extent of the heat affected zone (HAZ). The reduction is more visible on the friction stir welded workpiece’s bottom surface which is at a greater distant from the frictional heat source.
The peak value on the maximum temperature map is reached in the vicinity of the top surface at the advancing side for all three weld groups. The increase of this peak follows the increase of the rotational speed. The lower boundary of the maximum temperature map is affected by the tool traverse speed. The lowest value is seen in the fast group. Thus, traverse and rotational speed contribute to the expansion of the temperature gradient in the HAZ. The extent of the TMAZ is far more pronounced in the slow group and this extent reduces with increasing traverse speed.
The influence of process parameters is minor in terms of maximum phase fractions. The fraction distribution of martensite and recovered ferrite is similar for the fast and intermediate group. A larger area of the HAZ is affected by phase transformation for these groups. On the contrary, phase change takes place mainly on the advancing side for the slow group.
The dynamically recrystallized zone is attenuated by accelerating the process parameters as displayed in Figure 7. For the slow group, the grain refinement covers the entire TMAZ. By increasing the rotational and traverse speed, the dynamically recrystallized zone is seen as concentrated on the top surface and the advancing side. The slow group shows more homogeneous grain size and more extensive grain refinement than the other two groups.
This analysis concludes that the slow welds present the most homogeneous and refined microstructure. Increasing the traverse speed leads to more heterogeneity in the microstructure and asymmetry between advancing and retreating side.
Work Package 5: Scientific and Technical results
The objective of work package 5 (GeonX) was to combine the knowledge acquired in work packages 1-4 to design an expert system intended for use in an industrial setting. The system will output the evolution of distortions during and after FSW, with the final aim of minimising expensive post-process corrections by an optimisation of the process parameters.
The expert system involves development of the FSW analysis module within GeonX’s software Virfac® in order to facilitate the setting up of the local model for the thermo-fluid analysis, and the subsequent global transient thermo-mechanical analysis with the welding energy evaluated by the local model. The two-step analysis scheme is represented in Figure 8.
The FSW analysis in Virfac is carried out in three stages:
1. Thermo-fluid analysis on a local model in order to calculate the power generated by the tool
2. Thermo-mechanical analysis on the local model
3. Inserting the results from the local model into the global model using the inherent strain method
For the thermo-fluid analysis, a visco-plastic material constitutive law is used. Four types of material constitutive laws are interfaced in Virfac to represent those available in Morfeo. They are:
1. Newtonian fluid
2. Norton-Hoff
3. Bingham
4. Herschel-Bulkley
Preparation of the FSW analysis is performed by defining the FSW tool (Figure 9) followed by the definition and extraction of the local model (Figure 10).
The thermal dissipation calculated by the thermo-fluid analysis is used to define the heat source for the transient thermo-mechanical analysis. Several heat source shapes are available in Virfac: cylindrical, ellipsoidal, conical, double-ellipsoidal, as well as a superposition of several heat sources. The welding process parameters, i.e. trajectory, welding speed, and heat source are defined using Virfac® Welding designer as shown in Figure 11.
The thermo-mechanical analysis may be carried out on the full component or, in the case of very large geometries, on sub-components. The plastic strain map on the sub-component is then extruded to the full component. A linear elastic calculation is then performed which results in the final distortions and residual stresses present in the full component.
The approach described above couples a thermo-fluid analysis with a thermo-mechanical analysis in order to predict the residual stresses and distortion that develop during the FSW process.
Work Package 6: Scientific and Technical results
Friction stir welding of steel comparator study
The objective of work package 6 was to undertake a techno-economic comparison between FSW and submerged arc welding (SAW) in order to provide a body of data on the real world benefits of steel FSW over SAW for ship construction. Technical difficulties with poor quality FSW tooling supplied to the project by a monopolistic, non-EU supplier meant that the initial aim of constructing two large comparator panels, one welded by FSW and another by SAW, could not be achieved. To overcome this issue, it was agreed with the Project Officer that the comparator study would be based upon technical and economic data already obtained during earlier stages of the project, together with further data provided by TWI, CMT and NAP. A report on the technical challenges encountered with the FSW tool for steel supplemented this study, since work undertaken by CMT had already determined that the performance and cost of tooling was the most significant factor in evaluating the economic benefits of the process.
NAP conducted a detailed study of the economic basis for using FSW, building upon the earlier work performed by CMT. This assessed the direct replacement of SAW by FSW to manufacture stiffened panels according to existing designs and working practices. The analysis was based upon technical and production rate data on the two processes provided by TWI and UoS which were then incorporated into NAP’s economic model. A second study considered the benefits accruing from changing the panel design and production process to accommodate the particular advantages offered by FSW. This redesigned panel was agreed with Lloyd’s Register (LR) as being a technically viable solution with at least equivalent performance to conventional panels.
The results of these studies were captured by the generation of a production sensitivity response analysis of the FSW tool cost and service life, reproduced in Figure 12. This analysis uses realistic tool cost and life data generated during the HILDA project but makes no allowance for future tool cost reductions or any assessment of the added value that FSW can bring to either the manufacturing process or the subsequent service life of the ship produced. Based upon these considerations alone, the use of FSW for ship production actually increased the cost of construction by 12%; this is clearly an unacceptable state of affairs and one due entirely to the poor performance of the tools supplied to the consortium.
The current state of the art FSW tools for steel are a commercially available hybrid material tool of wear resistant pcBN particulates in a toughened WRe binder. The cost of these tools varies with exchange rates but is approx. €5.000,00 per tool (in 2014). The life of the tool also varies but during the course of the HILDA project, this was found to average approx. 30 m of weld. The sensitivity analysis indicates that for FSW to compete directly with SAW, with no allowance made for other factors, either:
• The tool life must improve to surpass 105 m of weld per tool at the current cost of € 5.000,00 per tool, or
• The tool cost must reduce to €1.250,00 if the existing tool life of 30 m of weld is not improved.
The use of FSW for steel is clearly an immature technology when compared with the existing arc based welding technologies that have been in use for over 100 years, therefore considerable scope for cost reduction and process improvement exists. FSW does introduce a range of other benefits to the shipbuilding sector, particularly in terms of improved weld properties, reduced Health and Safety risks and enhanced environmental performance. Placing an economic value on these aspects of FSW is difficult, especially when trying to establish the monetary value to the manufacturing shipyard rather than the operator of the vessel. Equally, it is difficult to establish a true monetary cost for the elimination of ill health and injury. NAP produced data on these benefits as summarised in Table 1. Comprehensive and compiled data for total EU-wide costs related to death and sickness arising from welding-related causes were not available, hence the comparisons were made on data from the USA which has a similar technology and economic base to the EU.
Factoring in these societal-economic benefits of FSW, in particular the potential for substantial reduction in the cost associated with long-term sickness increases the competitiveness of FSW when compared with traditional fusion welding techniques.
The potential for tool cost reduction exists; presently, there is a monopoly supplier of tools for steel and thus no competition with respect to price. From data available in the public domain regarding the tool manufacturing process, it is estimated that the unit cost of producing the current generation of pcBN/WRe tools is in the region of €1.000,00 – €1.350,00. Therefore, the difference between this cost and the current selling price represents an allowance for recovery of the substantial development activities and a profit element.
The work carried out in work package 6 confirmed that the major cost driver in steel FSW for shipbuilding applications is the price and lifetime of the tool used. To be competitive as a direct replacement for arc welding of stiffened panels, either the service life of the existing tool technology tool needs to be extended to 105 m of weld, or the unit price of the tool needs to be reduced in the order of €1,250. It is envisaged that both of these improvements are technically feasible as the relevant technology matures; however, the current situation of a single tooling supplier may act against a price reduction.
The entry of new FSW tool providers into the market is expected to have a major impact upon the economics of the process. As the benefits of steel FSW become more apparent and the operating environment of the tools becomes better understood, the attractiveness of producing FSW tools for steel increases. To date, some companies have considered starting development programmes for steel FSW tools but chose not to do so until they are more certain that a market exists for the product. The results being generated in HILDA and by another FP7 project (MOSAIC) are demonstrating that there are real benefits to be achieved from steel FSW and thus establishing a market demand for improved, lower cost tools. New entrants to the FSW tool market will eliminate the monopolistic supply that currently exists and may introduce different tool materials and design concepts that could bring about the required step change in tool cost and performance. In the meantime, a significant output of project HILDA is the development of a “Guidance for approval” document by LR titled “Application of Friction Stir Welding of Steels for the Ship building Industry” and published by the Lloyd’s Register Technical Association (session 2015-16); this report is appended to the present project final report.
Work Package 7: Scientific and Technical results
The aim of the work carried out by CMT in work package 7 is the dissemination and exploitation of the project’s foregrounds and more specifically, ensuring the flow of information and requirements into the project from external end users. The same applies for the outgoing information on the consortium’s achievements and results; CMT is responsible for collecting the industry’s (mainly shipyards) requirements. In this case, the industrial end users as well as the academic world were supplied with information through workshops, brochures, the HILDA website and social media content, conferences, papers and direct contacts.
Apart from the tasks of information management and dissemination, the HILDA consortium delivered an economic assessment of the FSW process and the expert system. Finally, the use of results and a “business plan” document were presented to demonstrate the ongoing developments and the project’s achievements and to state the identified research gaps on the subject.
The major activities during the 1st period were focused on identifying the shipyards’ requirements in order to achieve the best possible project results and maximise its exploitation potential. A user group (HILDA IAG) was established; this user group intends to provide a platform for discussion between the consortium and the wider industrial community. Except for industry partners, the IAG embraces research institutions and universities to discuss the academic results and research findings from HILDA.
The dissemination work performed during the project has substantially raised the awareness and interest of the major shipyards on the process. The potential to weld dissimilar materials, steels of very high strength and other characteristics meets the demand of the shipbuilding industrial society to produce maritime assets for rough and extreme conditions and strict owner requirements. This dissemination work ensured the initiation of a network of contacts beyond HILDA. Currently, the consortium is aiming for further research work, especially with focus on the FSW tool for steel. The German shipyards, organised in a national working group on ship production technologies, have instructed CMT to proceed with the FSW tool development work.
Furthermore, it was acknowledged that the end users (shipyards) will only recognise the techno-economic potential of FSW if the process is competitive to the state of the art joining technologies. To this end, an initial techno-economic assessment was performed by generating information to allow shipbuilding industries and other users to evaluate the advantages and disadvantages of the FSW process against established fusion welding methods. This assessment was based on the direct investment costs for a new assembly line comprising butt welds (as examined in HILDA) without taking into account possible economic enhancements in the life-cycle of a ship or environmental and safety benefits. The aim is to provide a realistic techno-economic comparison of the potential benefits of steel FSW in the context of industrial applications, therefore relevant to potential end users.
The main efforts in the 2nd period concentrated on establishing solutions to the tooling issue which hampered the project. Therefore, workshops and meetings with FSW tool manufacturers and material experts were organised. Moreover, the process’s economic assessment was updated towards a tool cost / service life response surface indicating a target envelope of production cost for the shipbuilding industry (Figure 12).
Except for HILDA’s primary focus on the shipbuilding industry and specifically the manufacturing of stiffened panels, other industries and applications were screened towards the possible use of steel FSW. Additionally, the developed expert system and FSW knowledge hub are of broad interest among all manufacturing companies in various transport modes.
By the end of project HILDA, FSW has been further developed towards the use in the maritime industry. Achieving welding traverse speeds comparable to fusion welding techniques and in accordance with production standards, steel FSW can compete with these state of the art welding methods. The process parameter envelope generated within HILDA in conjunction with mechanical testing and recommendations by class provide the basis for its potential use in industrial applications. However, the main bottleneck for introduction in a wide range of application from an economic point of view is the FSW tool. At this stage of development, FSW can only be economically viable in applications with potential high lifecycle costs, e.g. for maintenance and repair; underwater piping, offshore structures and highly loaded ship structure are some examples.
The risk of tool failure and the lack of alternative sources of tooling were stated in the proposal for this project. This emerged as a major obstacle for the experimental work during HILDA as well as for any follow-up applications. Availability of a reliable tool material would not only allow Europe to reap the technical, health and safety and economic benefits that can be obtained from FSW in steel but could also promote advances in other areas, for instance the development of high temperature, wear resistant bearings for power generation, aircraft engines and machining applications. However, the capability to weld extremely high strength steels, steels with special properties, and the dissimilar joining of steel to aluminium is a strong need within the maritime industry and this was stated during the workshops and working group meetings organised by the consortium.
Moreover, the possibility of incorporating an auxiliary energy source for pre-heating the steel along the weld line prior to FSW would appear to offer a potential advantage, both economic and technical, that ought to be further investigated. Such an approach may allow European companies that have already made significant investment in automated laser welding systems for their panel lines to adopt the FSW process at relatively low extra cost.
Project HILDA: Next steps for friction stir welding of steel
A significant conclusion from the above experimental programmes has been that microstructural examination, fatigue and tensile testing, and hardness and impact toughness measurements suggest acceptable (at a minimum) mechanical properties of the fast welds. Therefore, a step change improvement in the currently employed welding traverse speeds has been confirmed, but there is considerable scope for further examination of high speed welding. This would render more weight to the argument for introduction of steel FSW in industrial applications and primarily generate the required datasets which will trigger relevant classification society guidelines and specifications (in the context of the marine sector). One more fundamental research question is the assessment of more grades of steel such as high strength and poor weldability (in fusion welding), e.g. super bainitic steels.
The FSW tool for steel with the associated matters of unreliable performance, high costs and short operational life has emerged during HILDA as the main limiting factor in the economic competitiveness of the process. An obvious requirement for wider introduction of steel FSW in large-scale manufacturing is the development of a cost-effective and durable tool which will consistently deliver high integrity welds; this is a matter broadly accepted as critical in the uptake of the process. Equally, it is worth exploring the pre-heat assisted (e.g. through induction heating) FSW which can be utilised to soften the steel ahead of the tool, hence exposing the latter to much reduced forces. Moreover, the current tool technology for steel only allows for welding in butt and lap configuration. Critically, the future introduction of FSW in shipbuilding will not only dependent upon cost related issues but also on the process’s capability to weld more geometries, predominantly fillet welding. In view of the present tooling manufacturing methods and materials, this will require a step change advancement in tool technology.
Potential Impact:
The research output and dissemination activities performed through project HILDA have already had substantial impact on the international academic and research community. The development of world-class, beyond the state of the art knowledge and expertise on an innovative and very promising joining process has raised the status of all partners involved in HILDA. The links between partners and to industry will enable members of the consortium to participate successfully in future research projects.
Moreover, there is considerable potential for uptake of the FSW process on steel by shipbuilding and other industries once the tooling-related issues that have emerged during HILDA are addressed satisfactorily.
The project has had negligible socio-economic impact or wider societal implications to this date.
List of Websites:
http://www.hilda-europe.eu/
HILDA is intended to promote and facilitate the use of Friction Stir Welding (FSW) in the joining of steel panels and components in assemblies required for the shipbuilding industry in Europe. Presently, this process is carried out using various techniques used in the fabrication and building of ships. These techniques involve high heat input to establish localised consolidation of the metals being welded together. The FSW process has many advantages over these processes, amongst these are low temperature input resulting in low distortion, simplified preparation, joining of dissimilar metals possible, improved design capability and high weld integrity.
Project Context and Objectives:
HILDA will deliver a cost effective, low distortion welding process for EU shipyards to allow them to maintain competitiveness and produce light, strong, more fuel-efficient vessels. The solid state technique will enable the modular construction of dimensionally accurate, high strength, corrosion resistant fabrications across the entire range of steels, enabling the introduction of stronger, tougher, corrosion resistant steels into the industry. HILDA will develop the fundamental metallurgical knowledge required to predict the complex phase changes and stress regimes present in welding steel. This will enable the proven, energy efficient, low hazard and environmentally benign technology of friction stir welding, widely deployed in aluminium construction for the aerospace and rail industries, to be transferred to steel shipbuilding. To achieve this breakthrough, HILDA will use real world data from FSW of high strength low alloy steels as an input to develop a computer model of the thermo-mechanical processing process at the heart of the solid state welding technique.
A combination of numerical and experimental techniques will be implemented to determine sensitivity over a wide range of process input variables and to understand and characterize these parameters in order to enhance the microstructure and mechanical properties while mitigating the effects of post weld distortion. The general aims of the program are summarized below:
• Understand and characterize the process quality in relation to metallurgical, corrosion and wear performance.
• Understand, trace and quantify relationships between intermetallic reactions (if any) and process parameters.
• Develop a predictive interrelationship between computational thermal transients / plastic deformations and particle dispersion characteristics, through an innovative analytical approach based on the continuous dynamic recrystallization phenomenon.
• Determine optimum process parameters via numerical and analytical sensitivity studies.
• Transfer knowledge of methods to the industrial and academic communities by providing design guidelines on FSW of steel over a range of thicknesses.
The primary objective of this work is to create a knowledge base on FSW that is applicable to FSW of steel ship structures with specific microstructural and mechanical properties that not only exceeds the limitations of conventional joining processes but accomplishes this in a cost advantageous manner with minimal distortion. By establishing a European wide collaborative effort that builds on the commercial, technical and academic merits of the carefully selected partnership, it is believed that the research outcomes will be highly attractive to the Eurozone ship building and ship repair industries.
Project Results:
Work Package 1: Scientific and Technical results
At the commencement of project HILDA, the state of the art for FSW of steel can be summarised as follows:
• FSW of steel had been demonstrated to be achievable in the laboratory but only in short lengths, often less than 1 m, as the tools used had a very short lifetime.
• The welding speeds attained were generally approx. 100 mm/min in 6 mm thick steel.
• Welds had only been produced in the butt and lap configurations.
• The FSW environment in steel, for example the temperatures and forces generated within the weld zone, was largely unknown.
• The FSW process parameter envelope, i.e. the combinations of tool rotational speed and tool traverse speed that will result in a good weld, was unquantified.
• The influence of process parameters on the weld microstructure was largely unknown.
• The influence of process parameters upon the mechanical properties of the weld was unknown.
• The effect of process parameters upon distortion in the welded fabrication was unquantified.
• The influence of process parameters upon tool life was largely unknown.
By the end of the HILDA project, significant advances had been made in the FSW of steel. Within work package 1, TWI has significantly expanded the process envelope for steel FSW and developed a greater understanding of the thermo-mechanical process involved in making welds. The influence of the process control parameters on the weld quality has been investigated jointly with the University of Strathclyde (UoS, work package 2), and the process shown to be capable of producing fabrications with low distortion and excellent mechanical properties though tool cost and life considerations currently make FSW more expensive than GMAW and laser welding.
The process advancements made during HILDA can be summarised as:
• 6 mm thick DH36 steel can be friction stir welded at speeds of up to 500 mm/min, significantly surpassing the targeted benchmark fusion welding speed of 350 mm/min. Preliminary work to further extend the welding traverse speeds has showed that DH36 steel can potentially be welded at up to 700 mm/min.
• FSW speeds of 300–450 mm/min produce a good balance of productivity, tool life and weld properties, making the process technically competitive with equivalent arc welding processes.
• The forces generated during FSW of 6 mm thick steel can be maintained below a level that will allow welds to be produced on existing equipment developed for the FSW of light metals.
• The spindle rotational speed and concentricity requirements for the FSW of steel are no greater than those used in many existing FSW machines for welding light metals (e.g. aluminium, magnesium).
• Welds can be produced in either position or force control. The ability to utilise position control, combined with the low tool rotational speed requirements, may permit the use of converted milling machines for FSW of DH36 steel, thus potentially reducing the initial capital costs required to adopt the process.
• The temperatures generated during steel FSW are lower than in fusion welding but considerably greater than the temperatures during aluminium FSW (Figure 1).
• The FSW process in steel appears tolerant of workpiece material defects and poor prior preparation.
• FSW of DH36 steel produces significant benefits in terms of reduced Health and Safety risks when compared with conventional fusion welding techniques.
During the weldability investigation, it quickly became apparent that the current drawbacks of the FSW process in steel are centred on the available tool technology. These are:
• There is a monopoly supplier of tools for steel, based outside the EU and thus the quality and security in the supply of tools cannot be assured.
• The FSW tools currently available for steel are expensive and short-lived, leading to a typical best case welding cost of approx. €77 / metre for the good quality FSW tools alone.
• The current tool material technology places limits on the design of the tools, thus restricts the weld geometries that can currently be achieved. In addition, these design limitations also make the process in steel more susceptible to root defects than is the case with other metals.
Moreover, three aspects of the technical progress made in steel FSW are of particular significance:
1. The generation of a body of knowledge regarding the interaction between the process parameters, the microstructures that these produce in the weld zone and the mechanical properties of the welds.
2. The increased tolerance of the process to defects; FSW is seen to produce welds with properties that exceed those of conventional fusion welds even in the presence of defects (reported more fully by UoS in work package 2).
3. The generation of data regarding the FSW environment in steel and the requirements on tool materials for welding steel.
The first aspect, the generation of knowledge by UoS on the influence of the process parameters on the weld microstructure / mechanical property relationship offers the potential to control the process for specific results. For example, it may be possible to tailor the processing conditions to produce welds that are exceptionally tough for use in cryogenic applications, or which have enhanced fatigue properties, or a more general balance of properties for non-specialist applications. The data generated also provide the underlying knowledge required for the modelling tasks undertaken by the University of Malta (work package 3) and Cenaero (work package 4) to enable the development by GeonX (work package 5) of the expert system (software) that will predict the properties and distortion associated with the FSW of steel.
The high defect tolerance of the process, for instance failure to remove surface oxides from the steel plates prior to welding or the presence of a root defect due to tool limitations, makes for an industrially robust welding technique and mitigates to an extent the current issues arising from suboptimum tool materials and designs. It will however require that end users and classification societies reconsider some fundamental assumptions regarding the presence of weld defects and approach approval regimes with a view to accepting “fitness for purpose” as a goal rather than freedom from defects. Lloyd’s Register (LR) is leading the work in this area, ensuring that an appropriate regulatory framework is in place to expedite the industrial application of FSW in Europe. The relevant report produced by LR, “Application of Friction Stir Welding of Steels for the Ship building Industry”, has been appended to this document.
The data generated on the welding environment, in particular the temperatures and forces experienced by the tool, will be invaluable in generating performance specifications for, and guiding the development of, next generation tool materials for steel FSW that improve both the technical performance and economic viability of the technology.
Work Package 2: Scientific and Technical results
As demonstrated by previous research, friction stir welding (FSW) of steel presents numerous advantages across many industrial sectors and especially for marine applications compared to conventional fusion welding techniques. However, the fundamental knowledge of the process on steel remained relatively limited prior to project HILDA, thus obstructing the industrial uptake of the process on steel. In the course of WP2, the University of Strathclyde (UoS) has carried out a large-scale microstructure and property evaluation of friction stir welded 6 mm thick low alloy steel grade DH36 plate, commonly used in the European shipbuilding industry. The in-depth experimental testing programmes examined butt welds produced by a wide range of process parameters through microstructural characterisation and mechanical testing, also focusing on the interactions between weld metallurgy and associated mechanical properties. The key findings of these programmes are outlined below.
A preliminary process parameter envelope has been developed in close collaboration with TWI by employing varying tool rotational and traverse speeds. The trialled welding traverse speeds commenced from the originally recommended by the FSW tool manufacturer (100 mm/min) and were gradually increased up to 500 mm/min. By substantially increasing the range of attainable traverse speeds, this development programme has concluded that high speed FSW of steel is feasible; new parameter sets have been established which deliver high integrity welds and enhance the process’s techno-economic competitiveness relative to fusion welding methods through a step change increase in the conventionally recognised welding speed. Expanding the process envelope even more should be made possible when various tooling issues (encountered during the project and discussed by TWI in deliverables Del 1.2 and 6.1) are appropriately addressed through new research.
High traverse speed FSW is expected to affect the weld quality. For this purpose, metallographic examination of the 2000 mm long welds was performed to characterise the evolved microstructure, thus offer an insight into the anticipated mechanical property behaviour of the weldments. This analysis has established that significant grain refinement of the original DH36 parent material microstructure is promoted during welding. Moreover, FSW of steel creates a complex metallurgical system which is highly dependent on the employed process parameters. The slow traverse speeds (100-200 mm/min) generate a microstructure of refined ferrite grains, the intermediate traverse speeds (250-400 mm/min) create an acicular bainitic ferrite-rich microstructure and the fast traverse speeds (450-500 mm/min) give rise to heterogeneous microstructures of acicular ferrites and acicular bainitic ferrites. Although the assessed welding parameters have produced no defects in the main body of the weld zone hence increasing the confidence in the quality of high speed FSW, two types of process related flaws have been detected in most welds. Lack of penetration initiating a weld root flaw and the top surface breaking flaws require tackling through new developments in the processing conditions and the FSW tool for steel.
Mechanical testing of the welds was conducted to assess the evolution of mechanical properties with increasing tool rotational and traverse speed and evaluate the impact of the complex metallurgical system in steel FSW. Transverse tensile testing has consistently demonstrated that all slow and intermediate welds have higher yield and tensile strength compared to the parent material regardless of minor surface breaking flaws (as characterised above). Most fast weld samples fractured in the weld zone or were classed as unstable, suggesting reduced tolerance to parameter variations; hence, optimisation of the high speed welding parameters (450-500 mm/min) is critical. The weld hardness is seen to increase with increasing traverse speed and this is attributed to the evolution of harder phases in the microstructure. Still, the majority of the examined welds showed hardness distribution within classification society rules. Notably, an improvement in impact toughness with increasing traverse speed has been recorded and this provides additional weight to the potential of high speed steel FSW.
Since the fatigue performance of welded components is of paramount importance in marine applications, this has been thoroughly assessed by producing new DH36 steel welds based on the findings in the previously outlined testing programmes. An extensive standard operating procedure document, specific to FSW, was drafted and executed in collaboration with LR and CMT to address the relevant knowledge gap for this process on low alloy steel and generating novel data for future assessment by classification societies. The weldments’ yield strength, geometry, metallurgical features and hardness distribution were recorded in support of the fatigue testing. The effect of varying welding parameters was also evaluated. Microstructural observations, hardness measurements and transverse tensile properties were in excellent agreement with the corresponding in the previous testing programmes. All examined steel grade DH36 friction stir welds exhibited excellent fatigue performance, better than relevant international recommendations for fusion welding and equal to high quality laser welding, irrespective of minor surface breaking flaws which were identified. The relation between weld flaws and fatigue performance has been established through metallographic examination, fracture surface analysis and additional fatigue testing with the critical defect removed. This has revealed that minor embedded flaws do not initiate cracks whilst surface breaking irregularities, i.e. lack of penetration or the indentations on the weld top surface by the tool shoulder are the critical factor for crack propagation. The latter is a serious challenge for future FSW process developments which, when addressed, is expected to upgrade the welds’ fatigue life dramatically.
In parallel, knowledge of the thermo-mechanical response of the examined material is equally important with regard to further refinement of process parameters and future improvements in tooling and welding machine specifications. To this end, thermo-mechanical deformation studies were implemented in partnership with Cenaero by hot compression testing in conditions, i.e. temperature and strain rate, which simulate the actual FSW process on a Gleeble thermo-mechanical simulator. This novel investigation has determined that the alloy’s flow stress is appreciably affected by the test temperature and strain rate, i.e. increasing with decreasing temperature and increasing strain rate. This significant finding relates to FSW of steel since it suggests that process parameters which produce regions of low temperature and high strain rate in a friction stir weld may promote the formation of embedded flaws due to inadequate material flow. This flow stress evolution is only opposed by the tests at 800oC, in which increased flow stress is recorded as compared to a lower test temperature for constant strain rate. This observation provides an insight into future process parameter optimisation work since it specifies that welding in the vicinity of 800oC is better avoided to reduce the forces exerted on the FSW tool for steel. The original material property data were employed by the University of Malta and Cenaero for validation of their predictive models.
The key technical objectives of WP2 were the expansion of the scientific knowledge on FSW of steel in support of the process’s future introduction in industry and the generation of data not previously available on steel FSW for the validation of the predictive local and global models formulated in WP3 and WP4. It is evident from the preceding discussion that both objectives have been successfully fulfilled; with respect to the former, the testing programmes conducted by UoS have advanced the understanding of the underlying phenomena during steel FSW and identified high speed welding parameter sets which result in an attractive combination of acceptable quality level and economically competitive processing conditions. The latter objective has been realised by producing a novel and wide-ranging experimental dataset on DH36 steel which has been supplied to UoM and Cenaero. Additionally, the scientific knowledge and the corresponding recorded datasets will favour the progress in process control regimes and tooling design and material, an area in which improvements are greatly desired. Finally, the noteworthy findings on the steel friction stir welds’ performance have formed the basis of extensive dissemination activities; these are summarised in Del 2.1 2.2 and 7.3.
Work Package 3: Scientific and Technical results
Work package 3 set out to develop and validate global thermo-elastoplastic FSW numerical models. These models were then used to investigate the influencing parameters giving rise to residual stresses and distortion. The numerical models are integrated with the local models and together provide an expert system platform that was developed in work package 5. The scheme of work undertaken within this work package followed 3 main tasks (detailed in the periodic report) and these were all fully achieved.
The University of Malta (UoM) has led work package 3 which was primarily focused on developing global finite element model for the prediction of residual stresses and distortion. The models were validated with experimental tests results. These included transient temperature and strain measurements, residuals stresses and distortion measurements and provided significant information towards the knowledge and understanding of the FSW process. These tests were in part conducted within work package 3 together with the work performed in work package 1 (TWI) and 2 (University of Strathclyde). Adequate calibration with heat loss and other influencing parameters was conducted leading to good correlation with experimental tests.
Three thermo-elastoplastic models were developed, ranging from relatively simple analytical algorithms that are able to define the welding contraction forces and angular distortion to a fully transient non-linear material properties thermo-elastoplastic numerical model capable of predicting the evolution of residual stresses and distortion. The latter requires significant computational time and consequently a hybrid model that takes into account all non-linearities and influences due to heat sinks and clamping boundary conditions was developed. This model provides an accurate representation of the residual stresses and distortion using less computational time. The development and validation of these models was presented in deliverable Del 3.1 and includes:
• Phase I models: Reduced solution methods based on elastoplastic analytical algorithms that identify FSW contraction forces and applied to elastic static room temperature material properties structural models.
• Phase II models: Hybrid models that take into account non-linear material properties but are solved in a two-step heating and cooling cycle in a static manner.
• Phase III models: Transient elastoplastic models that apply the developed thermal strains in a transient manner with non-linear material properties included.
The local – global modelling strategy adopted in this study and applied to FSW of DH36 steel is shown in Figure 2.
Following the development of the global models, work package 3 focused on using these models to understand and optimise the FSW process by identifying the influential parameters affecting residual stresses and distortion. A series of welding parameters and boundary conditions were considered, ranging from slow to fast traverse speed FSW. In this case, no thermocouple measurements were available and thus the local models developed in work package 4 (Cenaero) were used to identify the thermal transients, material flow, microstructural properties and TMAZ. All these parameters are required for the successful numerical modelling through the global models. The results of this parametric study were presented in deliverable Del 3.2. An analytical and numerical approach was adopted to investigate the following effects:
• Heat loss to machine bed
• Tool penetration depth
• Structural restraint and pressure developed by clamping
• Tool traverse and rotational speed
• Plate stiffness with respect to FSW
Figure 3 shows the typical predicted out-of-plane distortion for different heat loss parameters.
Within work package 3, UoM generated comprehensive global numerical models capable of predicting the residual stresses and distortion developed due to FSW. Various numerical modelling strategies were adopted, ranging from computationally efficient models to more robust non-linear thermos-elastoplastic models that were integrated with the local models. All tasks and deliverables were attained and reported. The main conclusions of the numerical parametric study include:
• The heat loss to the machine bed highly influences the development of thermal strains. Appropriate insulation or heat extraction can be developed for specific welding parameters leading to more efficient welds in terms of process, residual stresses and distortion.
• The tool shape and penetration depth are also crucial. Higher penetration depth leads to more stirring and shearing thereby more heat input, residual stresses and distortion. However, it is important that the tool is significantly plunged into the material to ensure the production of fully penetrated welds.
• The FSW process on steel requires significant clamping that in turn influences the evolution of residual stresses and distortion. For instance, free angular distortion is inhibited as a result of clamping. Furthermore the application of side clamping reduced the longitudinal residual stresses in a similar fashion to mechanical tensioning in conventional welding.
• A sensitivity analysis on the different welding parameters was also conducted. It was concluded that faster welds generally lead to less distortion; however, this is highly dependent on the structural stiffness of the plate and heat loss boundary conditions.
The models developed in this work package were embedded to the local models (Cenaero) in the expert system implemented in work package 5 (GeonX), hence providing design engineers with a platform to analyse and optimise FSW for different plate dimensions and configurations, together with the investigation of influencing parameters.
Dissemination was also fundamental towards the successful completion of this project. The main contributions from the University of Malta are outlined in Del 7.3.
Work Package 4: Scientific and Technical results
Work package 4 (Cenaero) is dedicated to the microstructural and thermo-mechanical modelling of FSW of DH36 steel plates in a butt joint configuration. The numerical model is used to obtain material flow, metallurgical characteristics and process parameters influencing the microstructure of the welded joint.
The local thermo-mechanical phenomena which occur during the welding stage of the process are described by a thermo-fluid model. This local model is based on incompressible flow formulation and is used to simulate the velocity and temperature distributions induced by the stirring action of the rotating tool between the plates being welded. The temperature distribution and material flow are computed by solving the continuity, momentum and energy equations for incompressible flow with appropriate contact / friction conditions. These contact / friction conditions assist in modelling the friction between the rotating FSW tool and steel workpieces which produces frictional shear stresses and frictional heat.
The thermo-fluid model is based on a rigid visco-plastic approximation and is described by an Eulerian formulation. In this model, the material is analysed as a viscous fluid flowing across an Eulerian mesh and interacting with the rotating FSW tool. The type of Eulerian formulation used depends on the tool’s axisymmetric properties. The friction and boundary conditions are adapted to the type of Eulerian description depending on process parameters and tool geometrical features.
In the numerical model, the Norton Hoff visco-plastic law which is proportional to the effective viscosity is proposed to model the behaviour of the steel. The identification of the viscosity is attained from the evolution of flow stress as a function of temperature and strain rate through data generated on a Gleeble machine (Figure 4) as provided by work package 2 (University of Strathclyde, UoS). The evolution of thermal properties has also been determined experimentally by UoS as a function of temperature.
Material properties have been identified on a range of temperatures and strain rates which closely match the conditions occurring during FSW of DH36 steel. For each test temperature of the experimental plan, the material consistency and strain rate sensitivity are determined by inverse analysis as represented in Figure 5.
A 3D numerical analysis of material and thermal flow during FSW of steel grade DH36 in the case of an axisymmetric tool was produced based on the alloy's visco-plastic behaviour. Temperature distribution examples are represented in Figure 6.
An initial numerical sensitivity analysis showed the influence of friction coefficient on the mechanical and thermal response. The workpiece velocity is strongly related to the friction coefficient value, particularly in the thermo-mechanical affected zone (TMAZ). The velocity tends to be the same as the tool velocity by increasing the friction coefficient value and as plastic deformation is becoming higher than friction heat flux.
The thermo-fluid model provides the metallurgical model with the velocity and temperature distributions in order to predict grain size evolution and phase fraction. The prediction of the grain size evolution is addressed by JMAK model which takes into account all stages of the discontinuous dynamic recrystallization. To predict the phase fraction, a user interface enables the use of several existing phase transformation models from thermo-fluid model solutions. A streamlines algorithm is performed after the thermo-fluid model in order to determine temperature, strain rate and strain cycles. These cycles are extracted along computed particle path lines during FSW.
A numerical sensitivity analysis has been undertaken to evaluate the influence of process parameters such as tool traverse and rotational speed on the FSW process. The above two process parameters are the most significant in FSW of steel and control the generated heat, yielding and cooling rate within the weld zone. A wide range of process parameters has been explored using the thermo-fluid and the microstructural evolution models (parameters classified in slow, intermediate and fast groups).
For a fixed rotational speed, an increase in the traverse speed reduces the extent of the heat affected zone (HAZ). The reduction is more visible on the friction stir welded workpiece’s bottom surface which is at a greater distant from the frictional heat source.
The peak value on the maximum temperature map is reached in the vicinity of the top surface at the advancing side for all three weld groups. The increase of this peak follows the increase of the rotational speed. The lower boundary of the maximum temperature map is affected by the tool traverse speed. The lowest value is seen in the fast group. Thus, traverse and rotational speed contribute to the expansion of the temperature gradient in the HAZ. The extent of the TMAZ is far more pronounced in the slow group and this extent reduces with increasing traverse speed.
The influence of process parameters is minor in terms of maximum phase fractions. The fraction distribution of martensite and recovered ferrite is similar for the fast and intermediate group. A larger area of the HAZ is affected by phase transformation for these groups. On the contrary, phase change takes place mainly on the advancing side for the slow group.
The dynamically recrystallized zone is attenuated by accelerating the process parameters as displayed in Figure 7. For the slow group, the grain refinement covers the entire TMAZ. By increasing the rotational and traverse speed, the dynamically recrystallized zone is seen as concentrated on the top surface and the advancing side. The slow group shows more homogeneous grain size and more extensive grain refinement than the other two groups.
This analysis concludes that the slow welds present the most homogeneous and refined microstructure. Increasing the traverse speed leads to more heterogeneity in the microstructure and asymmetry between advancing and retreating side.
Work Package 5: Scientific and Technical results
The objective of work package 5 (GeonX) was to combine the knowledge acquired in work packages 1-4 to design an expert system intended for use in an industrial setting. The system will output the evolution of distortions during and after FSW, with the final aim of minimising expensive post-process corrections by an optimisation of the process parameters.
The expert system involves development of the FSW analysis module within GeonX’s software Virfac® in order to facilitate the setting up of the local model for the thermo-fluid analysis, and the subsequent global transient thermo-mechanical analysis with the welding energy evaluated by the local model. The two-step analysis scheme is represented in Figure 8.
The FSW analysis in Virfac is carried out in three stages:
1. Thermo-fluid analysis on a local model in order to calculate the power generated by the tool
2. Thermo-mechanical analysis on the local model
3. Inserting the results from the local model into the global model using the inherent strain method
For the thermo-fluid analysis, a visco-plastic material constitutive law is used. Four types of material constitutive laws are interfaced in Virfac to represent those available in Morfeo. They are:
1. Newtonian fluid
2. Norton-Hoff
3. Bingham
4. Herschel-Bulkley
Preparation of the FSW analysis is performed by defining the FSW tool (Figure 9) followed by the definition and extraction of the local model (Figure 10).
The thermal dissipation calculated by the thermo-fluid analysis is used to define the heat source for the transient thermo-mechanical analysis. Several heat source shapes are available in Virfac: cylindrical, ellipsoidal, conical, double-ellipsoidal, as well as a superposition of several heat sources. The welding process parameters, i.e. trajectory, welding speed, and heat source are defined using Virfac® Welding designer as shown in Figure 11.
The thermo-mechanical analysis may be carried out on the full component or, in the case of very large geometries, on sub-components. The plastic strain map on the sub-component is then extruded to the full component. A linear elastic calculation is then performed which results in the final distortions and residual stresses present in the full component.
The approach described above couples a thermo-fluid analysis with a thermo-mechanical analysis in order to predict the residual stresses and distortion that develop during the FSW process.
Work Package 6: Scientific and Technical results
Friction stir welding of steel comparator study
The objective of work package 6 was to undertake a techno-economic comparison between FSW and submerged arc welding (SAW) in order to provide a body of data on the real world benefits of steel FSW over SAW for ship construction. Technical difficulties with poor quality FSW tooling supplied to the project by a monopolistic, non-EU supplier meant that the initial aim of constructing two large comparator panels, one welded by FSW and another by SAW, could not be achieved. To overcome this issue, it was agreed with the Project Officer that the comparator study would be based upon technical and economic data already obtained during earlier stages of the project, together with further data provided by TWI, CMT and NAP. A report on the technical challenges encountered with the FSW tool for steel supplemented this study, since work undertaken by CMT had already determined that the performance and cost of tooling was the most significant factor in evaluating the economic benefits of the process.
NAP conducted a detailed study of the economic basis for using FSW, building upon the earlier work performed by CMT. This assessed the direct replacement of SAW by FSW to manufacture stiffened panels according to existing designs and working practices. The analysis was based upon technical and production rate data on the two processes provided by TWI and UoS which were then incorporated into NAP’s economic model. A second study considered the benefits accruing from changing the panel design and production process to accommodate the particular advantages offered by FSW. This redesigned panel was agreed with Lloyd’s Register (LR) as being a technically viable solution with at least equivalent performance to conventional panels.
The results of these studies were captured by the generation of a production sensitivity response analysis of the FSW tool cost and service life, reproduced in Figure 12. This analysis uses realistic tool cost and life data generated during the HILDA project but makes no allowance for future tool cost reductions or any assessment of the added value that FSW can bring to either the manufacturing process or the subsequent service life of the ship produced. Based upon these considerations alone, the use of FSW for ship production actually increased the cost of construction by 12%; this is clearly an unacceptable state of affairs and one due entirely to the poor performance of the tools supplied to the consortium.
The current state of the art FSW tools for steel are a commercially available hybrid material tool of wear resistant pcBN particulates in a toughened WRe binder. The cost of these tools varies with exchange rates but is approx. €5.000,00 per tool (in 2014). The life of the tool also varies but during the course of the HILDA project, this was found to average approx. 30 m of weld. The sensitivity analysis indicates that for FSW to compete directly with SAW, with no allowance made for other factors, either:
• The tool life must improve to surpass 105 m of weld per tool at the current cost of € 5.000,00 per tool, or
• The tool cost must reduce to €1.250,00 if the existing tool life of 30 m of weld is not improved.
The use of FSW for steel is clearly an immature technology when compared with the existing arc based welding technologies that have been in use for over 100 years, therefore considerable scope for cost reduction and process improvement exists. FSW does introduce a range of other benefits to the shipbuilding sector, particularly in terms of improved weld properties, reduced Health and Safety risks and enhanced environmental performance. Placing an economic value on these aspects of FSW is difficult, especially when trying to establish the monetary value to the manufacturing shipyard rather than the operator of the vessel. Equally, it is difficult to establish a true monetary cost for the elimination of ill health and injury. NAP produced data on these benefits as summarised in Table 1. Comprehensive and compiled data for total EU-wide costs related to death and sickness arising from welding-related causes were not available, hence the comparisons were made on data from the USA which has a similar technology and economic base to the EU.
Factoring in these societal-economic benefits of FSW, in particular the potential for substantial reduction in the cost associated with long-term sickness increases the competitiveness of FSW when compared with traditional fusion welding techniques.
The potential for tool cost reduction exists; presently, there is a monopoly supplier of tools for steel and thus no competition with respect to price. From data available in the public domain regarding the tool manufacturing process, it is estimated that the unit cost of producing the current generation of pcBN/WRe tools is in the region of €1.000,00 – €1.350,00. Therefore, the difference between this cost and the current selling price represents an allowance for recovery of the substantial development activities and a profit element.
The work carried out in work package 6 confirmed that the major cost driver in steel FSW for shipbuilding applications is the price and lifetime of the tool used. To be competitive as a direct replacement for arc welding of stiffened panels, either the service life of the existing tool technology tool needs to be extended to 105 m of weld, or the unit price of the tool needs to be reduced in the order of €1,250. It is envisaged that both of these improvements are technically feasible as the relevant technology matures; however, the current situation of a single tooling supplier may act against a price reduction.
The entry of new FSW tool providers into the market is expected to have a major impact upon the economics of the process. As the benefits of steel FSW become more apparent and the operating environment of the tools becomes better understood, the attractiveness of producing FSW tools for steel increases. To date, some companies have considered starting development programmes for steel FSW tools but chose not to do so until they are more certain that a market exists for the product. The results being generated in HILDA and by another FP7 project (MOSAIC) are demonstrating that there are real benefits to be achieved from steel FSW and thus establishing a market demand for improved, lower cost tools. New entrants to the FSW tool market will eliminate the monopolistic supply that currently exists and may introduce different tool materials and design concepts that could bring about the required step change in tool cost and performance. In the meantime, a significant output of project HILDA is the development of a “Guidance for approval” document by LR titled “Application of Friction Stir Welding of Steels for the Ship building Industry” and published by the Lloyd’s Register Technical Association (session 2015-16); this report is appended to the present project final report.
Work Package 7: Scientific and Technical results
The aim of the work carried out by CMT in work package 7 is the dissemination and exploitation of the project’s foregrounds and more specifically, ensuring the flow of information and requirements into the project from external end users. The same applies for the outgoing information on the consortium’s achievements and results; CMT is responsible for collecting the industry’s (mainly shipyards) requirements. In this case, the industrial end users as well as the academic world were supplied with information through workshops, brochures, the HILDA website and social media content, conferences, papers and direct contacts.
Apart from the tasks of information management and dissemination, the HILDA consortium delivered an economic assessment of the FSW process and the expert system. Finally, the use of results and a “business plan” document were presented to demonstrate the ongoing developments and the project’s achievements and to state the identified research gaps on the subject.
The major activities during the 1st period were focused on identifying the shipyards’ requirements in order to achieve the best possible project results and maximise its exploitation potential. A user group (HILDA IAG) was established; this user group intends to provide a platform for discussion between the consortium and the wider industrial community. Except for industry partners, the IAG embraces research institutions and universities to discuss the academic results and research findings from HILDA.
The dissemination work performed during the project has substantially raised the awareness and interest of the major shipyards on the process. The potential to weld dissimilar materials, steels of very high strength and other characteristics meets the demand of the shipbuilding industrial society to produce maritime assets for rough and extreme conditions and strict owner requirements. This dissemination work ensured the initiation of a network of contacts beyond HILDA. Currently, the consortium is aiming for further research work, especially with focus on the FSW tool for steel. The German shipyards, organised in a national working group on ship production technologies, have instructed CMT to proceed with the FSW tool development work.
Furthermore, it was acknowledged that the end users (shipyards) will only recognise the techno-economic potential of FSW if the process is competitive to the state of the art joining technologies. To this end, an initial techno-economic assessment was performed by generating information to allow shipbuilding industries and other users to evaluate the advantages and disadvantages of the FSW process against established fusion welding methods. This assessment was based on the direct investment costs for a new assembly line comprising butt welds (as examined in HILDA) without taking into account possible economic enhancements in the life-cycle of a ship or environmental and safety benefits. The aim is to provide a realistic techno-economic comparison of the potential benefits of steel FSW in the context of industrial applications, therefore relevant to potential end users.
The main efforts in the 2nd period concentrated on establishing solutions to the tooling issue which hampered the project. Therefore, workshops and meetings with FSW tool manufacturers and material experts were organised. Moreover, the process’s economic assessment was updated towards a tool cost / service life response surface indicating a target envelope of production cost for the shipbuilding industry (Figure 12).
Except for HILDA’s primary focus on the shipbuilding industry and specifically the manufacturing of stiffened panels, other industries and applications were screened towards the possible use of steel FSW. Additionally, the developed expert system and FSW knowledge hub are of broad interest among all manufacturing companies in various transport modes.
By the end of project HILDA, FSW has been further developed towards the use in the maritime industry. Achieving welding traverse speeds comparable to fusion welding techniques and in accordance with production standards, steel FSW can compete with these state of the art welding methods. The process parameter envelope generated within HILDA in conjunction with mechanical testing and recommendations by class provide the basis for its potential use in industrial applications. However, the main bottleneck for introduction in a wide range of application from an economic point of view is the FSW tool. At this stage of development, FSW can only be economically viable in applications with potential high lifecycle costs, e.g. for maintenance and repair; underwater piping, offshore structures and highly loaded ship structure are some examples.
The risk of tool failure and the lack of alternative sources of tooling were stated in the proposal for this project. This emerged as a major obstacle for the experimental work during HILDA as well as for any follow-up applications. Availability of a reliable tool material would not only allow Europe to reap the technical, health and safety and economic benefits that can be obtained from FSW in steel but could also promote advances in other areas, for instance the development of high temperature, wear resistant bearings for power generation, aircraft engines and machining applications. However, the capability to weld extremely high strength steels, steels with special properties, and the dissimilar joining of steel to aluminium is a strong need within the maritime industry and this was stated during the workshops and working group meetings organised by the consortium.
Moreover, the possibility of incorporating an auxiliary energy source for pre-heating the steel along the weld line prior to FSW would appear to offer a potential advantage, both economic and technical, that ought to be further investigated. Such an approach may allow European companies that have already made significant investment in automated laser welding systems for their panel lines to adopt the FSW process at relatively low extra cost.
Project HILDA: Next steps for friction stir welding of steel
A significant conclusion from the above experimental programmes has been that microstructural examination, fatigue and tensile testing, and hardness and impact toughness measurements suggest acceptable (at a minimum) mechanical properties of the fast welds. Therefore, a step change improvement in the currently employed welding traverse speeds has been confirmed, but there is considerable scope for further examination of high speed welding. This would render more weight to the argument for introduction of steel FSW in industrial applications and primarily generate the required datasets which will trigger relevant classification society guidelines and specifications (in the context of the marine sector). One more fundamental research question is the assessment of more grades of steel such as high strength and poor weldability (in fusion welding), e.g. super bainitic steels.
The FSW tool for steel with the associated matters of unreliable performance, high costs and short operational life has emerged during HILDA as the main limiting factor in the economic competitiveness of the process. An obvious requirement for wider introduction of steel FSW in large-scale manufacturing is the development of a cost-effective and durable tool which will consistently deliver high integrity welds; this is a matter broadly accepted as critical in the uptake of the process. Equally, it is worth exploring the pre-heat assisted (e.g. through induction heating) FSW which can be utilised to soften the steel ahead of the tool, hence exposing the latter to much reduced forces. Moreover, the current tool technology for steel only allows for welding in butt and lap configuration. Critically, the future introduction of FSW in shipbuilding will not only dependent upon cost related issues but also on the process’s capability to weld more geometries, predominantly fillet welding. In view of the present tooling manufacturing methods and materials, this will require a step change advancement in tool technology.
Potential Impact:
The research output and dissemination activities performed through project HILDA have already had substantial impact on the international academic and research community. The development of world-class, beyond the state of the art knowledge and expertise on an innovative and very promising joining process has raised the status of all partners involved in HILDA. The links between partners and to industry will enable members of the consortium to participate successfully in future research projects.
Moreover, there is considerable potential for uptake of the FSW process on steel by shipbuilding and other industries once the tooling-related issues that have emerged during HILDA are addressed satisfactorily.
The project has had negligible socio-economic impact or wider societal implications to this date.
List of Websites:
http://www.hilda-europe.eu/
