You are on page 1of 6

18.20.5.

2010,RonovpodRadhotm,eskRepublika

TIG WELDING PROCESS EFFECT ON THE IRON ALUMINIDE Fe3Al Heinz NEUMANN a, Jaromr MORAVECb, Josef BRAD c
a b

TU v Liberci, Studentsk 2, 461 17 Liberec, esk republika, heinz.neumann@tul .cz TU v Liberci, Studentsk 2, 461 17 Liberec, esk republika, jaromr.moravec@tul .cz
c

TU v Liberci, Studentsk 2, 461 17 Liberec, esk republika, josef.bradac@tul .cz

Abstract Iron aluminides belong to difficult-to-weld materials. Low plasticity and high spatial stress values cause cracks during the welding process. For the analysis and development of welding procedures for these materials, numerical simulations in the Sysweld programme are used. This programme however does not make it possible to locate, on the basis of the computed data of spatial stress, places with high risk of cracks. For this reason the Technical University of Liberec developed a program called MOVYPRO, which is able to detect potential critical places prone to cracks, using the Sysweld program output data. The article deals with the analysis results of the Fe3Al butt welds of plates and presents the welding procedure. The results of the thermal analysis by SYSWELD programme are verified by checking the temperature time curves in the selected places of a weldment, as well as against the weld pool geometry. The results of the mechanical analysis of the spatial stress state are verified by determining the residual spatial stress, using a magneto elastic method. The article examines and compares the results of the mechanical analysis carried out for an elasto-plastic material model and for an elasto-viscoplastic material model. The article presents a summary of the required input parameters for the SYSWELD simulation programme. The article also describes the results of comparative analyses and gives recommendations for using the SYSWELD programme in interconnection with the MOVYPRO programme. The acquired results and relevant data help to increase the credibility of results of the simulation computation. 1. INTRODUCTION

Intermetalic phases on the basis of Fe-Al alloys are characterized by specific properties, for which, they are considered highly perspective, despite of their brittleness at normal temperature. With regard to the structure arrangement, the lower coefficients have diffusions which result in a lower creep rate and difficult recrystallisation. Creation of Al2O3 oxides layers leads to a very good corrosive resistance in oxidizing and sulphur environment at high temperatures. Iron aluminides weldability is a topic which is still not discussed very much in technical literature [1]. Iron aluminides belong to difficult to weld materials, because they are susceptible to cracks. The welding technology is however supposed to be used for the products final production. The application of simulation computations and experimental tests should result in creation of the appropriate welding procedures, so that the risk of cracks is minimized. These issues were dealt with by the Technical university of Liberec, mainly within the scope of the GAR 106/07/1288 grant project. The research resulted into creation of butt welds of plates and fillet and corner welds. The outcomes of the experiments served for the acquisition of basic input data for simulation computations in the Sysweld programme. The main goal was to set up a computation methodology to determine critical parameters of temperature and deformation effects of the TIG welding process.

18.20.5.2010,RonovpodRadhotm,eskRepublika

2.

EXPERIMENTS

Experimental work was carried out for iron aluminide Fe3Al type alloyed Cr and Ce (16.13% Al, 2.85% Cr, 0.06% Ce, 0.46% Mn a 0.04% C in percentage by weight). Semiproducts for sheet production were prepared by melting in a vacuum furnace and cast into ceramic shell moulds. Consequently, semiproducts were gradually hot-rolled to the desired final thickness. First experiments were focused on the monitoring of the parent metal during melting by pulse welding current, by TIG welding method. Beads were created on the material by remelting the parent metal without using any filler material. The melting was carried out without preheating (cooling in the air) or with preheating at 400C (after melting, the samples were put into a furnace at 400C/1 hour, than cooled in the air). Within the grant project, butt welds without any filler material were carried out by TIG method (141) in the shield protection of argon. Welded plates were clamped in four places and were situated on a heated desk during the welding process. The preheating temperature was settled to 200C on the basis of the results of simulation computations. Automated welding by direct current with direct polarity was used (U = 13.9 V, I = 114 A, vw = 0.375 mm.s-1). In the weldments centre and in its cross direction, length temperature curves were measured in four places. The welding process was monitored by Weld.Monitor 3.5. After welding, the welded plates were put into an electric furnace preheated to 700C. After two hours the furnace was turned off and the cooling process continued in a shut-off furnace. 3. SIMULATION COMPUTATIONS AND DETERMINATION OF CRITICAL PLACES

Simulations were computed in the Sysweld programme and the acquired data from the experiments were used as input parameters to define and describe the heat source for thermal-metallurgical analysis. So called double ellipse model was used for the computations. Thermal-metallurgical and mechanical analyses by the Sysweld programme provide many interesting and useful data, but do not give solutions for places susceptible to cracks. Credibility of simulation computations depend on many factors. The results of thermal metallurgical analysis are verified by examining the weld pools shape and size, by the temperature field parameters and the metallographic analysis. According to the scratch pattern and weld pools measurements, a 3D computation model is designed. Verification of the results of mechanical analysis is a problematic issue. The computed values of deformations are compared with the measured values without difficulties. What is however not easy, is the examination of computed values of the spatial stress. Within the grant project, a possibility of an application of a magneto-elastic method for the determination of spatial stress was tested. For the assessment of different variants of technological parameters, it would be very beneficial; in case of difficult to weld materials, to detect critical places in the weldment, where cracks can occur. The Sysweld programme is unfortunately not able to solve this problem. Another question is according to which parameter or quantity such a place can be detected. It was finally decided that the critical places will be determined by comparing the computed values of spatial stress according to HMH theory, set by the Sysweld programme and the real strength limit values at a given temperature. Following the simulation computation, critical places were determined in a programme which was created in the Technical University in Liberec [2]. The results of the analysis are presented in a table, which shows a number of critical places for individual computation times. The critical places for individual computation times can be then determined by the Sysweld programme. The Sysweld programme can also retroactively determine the computed stress curves in individual node points in time dependence and compare these data with critical stress values in a graph. If we include a temperature curve in the graph, we can assess in which area and at which temperature the computed stresses are bigger than the critical values.

18.20.5.2010,RonovpodRadhotm,eskRepublika

The complexity of a credible determination of state of stress by the Sysweld simulation programme results from the fact that the computed stress values depend on the used model describing the materials behaviour. At present two types of constitutive models of material are used for the computation of residua stresses during and after welding. The first one is an elasto-plastic model with an isotropic hardening, and the second one is elasto-plastic model with a kinematic hardening. However, during the real welding process, stress relaxation takes place because of the high temperatures. In fact, the residual stress values are lower than indicated in the case of both elasto-plastic material models. An elasto-plastic model cannot take into account the visco-plastic processes, which occur by stress relaxation. In the framework of this submission we compared the computations for an elasto-plastic model with a kinematic hardening against an elastoviscoplastic model of the materials behaviour. 4. RESULTS OF SIMULATIONS AND COMPARATIVE ANALYSES

When welding plates by but welds using the method described in section 2 we managed to create welds without cracks. Fig. 1. shows the scratch pattern of a weld. The fusion line is not clear, which makes the determination of welding presents a geometry basic more input for difficult. simulation Nonetheless, the welding pool geometry computations.

Fig. 1. Weld macrostructure butt weld made without any filler material Fig. 2. shows the result of an analysis for the node point No. 5579, in which a critical place was detected during computation, using an elasto-plastic (EP) model of the materials behaviour. It is a point which is situated on the fusion line, in the centre of the welds length and the maximum temperature is equal to 1500C. It was proved that the detection of critical places corresponds to the moments, when the temperature is high, always reaching over 900C [3]. By simulation computations practically all input data for temperatures over 900C are determined by approximation or estimation and this, of course, leads to a question about the level of credibility of data which we receive from simulation computations. Even though critical places were detected, no cracks were found in the weld and that is why an elasto-viscoplastic model of the materials behaviour (EVP) was used for other computations, including also the relaxation behaviour. As it can be observed from Fig. 2., the computed stress values lowered considerably and the place is not detected as critical any more.

18.20.5.2010,RonovpodRadhotm,eskRepublika

1600 1400 Stress (MPa), Temperature (C) 1200 1000 800 600 400 200 0 0 50 100 150 200 250 Time (s) 300 350 400 450 Temperature Critical stress Calculated stress - model EVP Calculated stress - model EP

Fig. 2. Stress analysis during welding process for node point No. 5579 Fig. 3. shows a difference between the mechanical analysiss results and the results completed by an analysis of critical places; it demonstrates the stress distribution in a cross section of the plate and in a zoomed detail of the weld area we can see the distribution of critical places (white crosses); the node point No. 5579 is marked on the melting boundary. It is evident that critical places are not to be found in places with the highest stress. A classical approach to deal with the results of a mechanical analysis does not offer a survey on the possible risk of defects occurring in the weld.

Fig. 3. Stress distribution in plates cross section and the critical places in the same section Fig. 4. shows the weld macrostructure, where only remelting on the plate surface took place. One sample was chosen for a testing simulation computation and a comparative

18.20.5.2010,RonovpodRadhotm,eskRepublika

analysis, because during these experiments cracks would always occur. The figure shows changes in the structure pattern of the basic material, in the area of melting boundary and on the lower side of the plate. An interesting finding is that in area with a distinctive thermal and deformation effect, the grain becomes more refined. The remelting was carried out by the TIG method in the shield gas He+25%Ar, by an impulse current (Iimp=250 A, Ib=12.5 A, vw=0.18 m.min-1, the pulse period and basic current of 0.01s). Notwithstanding the considerable changes of the base structure, the exact determination of fusion line for the specification of the weld pool geometry for simulation computation remains very difficult. Fig. 4. Weld macrostructure - bead made on plate without filler metal Fig. 5. shows the results of a stress analysis for the node point No. 9023 on the fusion line, where a critical place was once more detected. Stress computations were again carried out for the aforementioned models of material. Computed stress values for the EVP model are lower than in case of the EP model, however a critical place occurs during one computation time.
1600 1400 Stress (MPa), Temperature (C) 1200 1000 800 600 400 200 0 0 20 40 60 80 100 Time (s) 120 140 160 180 200 Temperature Critical stress Calculated stress - model EVP Calculated stress - model EP

Fig. 5. Stress analysis during welding process for node point No. 9023 5. CONCLUSION

The executed experiments and simulation computations proved that while using the Sysweld simulation programme it is possible to verify the influence of technological parameters on the inner stress state of the welded parts during a welding process. By help of the developed programme and a methodological procedure it is possible to determine critical places susceptible to cracks. The obtained results show that computed stress values depend, as expected, on the used computed model of materials behaviour, mainly in the area of high temperatures, i.e. in the area where there is a higher risk of cracks, according to the results of performed analyses. To increase the credibility of the simulation computation for iron aluminides, it will be necessary to ensure the determination of temperature dependences of the input parameters for computation (thermal physical and mechanical material properties), as well as for the range of temperature above 800C, i.e. for a temperature

18.20.5.2010,RonovpodRadhotm,eskRepublika

range, for which critical places are firstly detected for the given material. Another improvement of simulation computations should be reached by including the relaxation processes during welding and also during the subsequent heat treatment. The requirements for quantity and quality of input data for simulation computations thus become more exigent. The analysis of a welding process by simulation computations is beneficial for the preparation of complex weld constructions as well as for solving problems concerning the welding of difficult to weld materials. However, the simulation computations in the Sysweld programme still do not make it possible to determine critical places in the welded details, nor during welding, nor during the cooling process. It would be beneficial to upgrade the programme by this possibility. It would also be efficient to verify other criteria for the assessment of critical places, for example the exhaustion of plasticity supplies. ACKNOWLEDGEMENTS The research was the part of the projects 106/07/1288 financed by the Grant Agency of the Czech Republic and MSM 4674788501 financed by the Ministry of Education of the Czech Republic. LITERATURE
[1] David SA, Santanella ML. In: Stoloff NS, Sikka VK, editors. Physical metalurgy and rocessing of intermetallic compounds. Chapman & Hall; 1996. p. 65575. [2] Moravec J.: Analza kritickch mst svaovacho procesu na zklad simulac v programu SYSWELD. /Disertan prce/. Liberec 2008. ISBN 978-80-7372-315-6. [3] NEUMANN, H.: Analza meznch teplotnch a deformanch ink na aluminidy eleza. In Zvranie 2009, konference: 4. 6. 11. 2009. Tatransk Lomnica, Hotel Urn, Slovensk republika, [CD-ROM]. 9 s. ISBN 978-80-89296-09-5.

You might also like