Abstract
The structureborne acoustic emission signals during resistance spot welding on 2024 aluminium alloy were detected in real time and analysed to find the characteristics of signals corresponding to the physical phase of welding process. The curve fitting models were developed based on the acoustic emission count and positive peak of nugget nucleation event. These mathematical models were used to predict the tensile–shear strength of spot weld. The results showed that the physical phases of welding process can be characterised by the acoustic emission signals detected during the resistance spot welding process. The acoustic emission count and the positive peak of nugget nucleation event of resistance spot welding on 2024 aluminium alloy have good relevance to the tensile–shear strength of spot weld, and these correlations were fitted to comply with some functional relationships, which can be used to realise the prediction of the strength of spot weld. It can be concluded that the prediction of weld strength based on the acoustic emission count of nugget nucleation event has good performance, but the prediction of weld strength based on the positive peak of nugget nucleation event is more susceptible to the expulsions and has poor performance when there are expulsions.
Keywords
Introduction
Resistance spot welding (RSW) is widely used in the automotive body assembly process. The quality of spot weld depends primarily on the size of the nugget, bearing strength of the spot weld, expulsion, process stability and other factors.1 Because the nugget formation is invisible during the RSW process, monitoring and improving the quality of spot welds are a continuing process in RSW research.2, 3 These researches range over many subjects such as the contact resistance or dynamic resistance measurement in the RSW process,4, 5 the impact evaluation of expulsion on welding process,6– 8 the monitoring for electrode invalidation during welding9 and the monitoring for sonic emission (SE) during welding.10 The prediction or assessment for the welding quality can be realised based on the above researches.
Cho and other researchers monitored the process variables in the primary circuit of the welding machine and obtained the variation of the dynamic resistance across electrodes. In order to test the reliability of such a system, an artificial intelligence algorithm to estimate the weld quality using the primary dynamic resistance was proposed.5 Primoz and other researchers investigated the estimation of the spot weld strength on the basis of the study of the SE signals detected during the RSW on zinc coated steels and concluded that the instrumentation required by sonic sound signal detecting is simpler and easier to use.10
The researchers proposed to predict or assess the spot weld quality of RSW based on the mathematical modelling. Óscar and other researchers set the quality level of a RSW joint by ultrasonic non-destructive testing and then developed an artificial neural network capable of reliably predicting the quality level of RSW joints from three welding parameters: welding time, welding current and electrode force.11 Luo and other researchers developed the non-linear models by the method of non-linear multiple orthogonal regression assembling design and realised the prediction of the quality of spot welds.12 Pouranvari and other researchers proposed a failure mechanism to describe both interfacial and pullout failure modes based on the effects of welding current, welding time, electrode pressure and holding time on the weld nugget size. In the light of this mechanism, an analytical model was proposed to predict failure mode and to estimate minimum nugget diameter (critical diameter).13 Besides, Zhou et al. and Marashi et al. also developed the analytical models between the weld failure and the area attribute of nugget or spot weld to predict or assess the weld quality.14, 15
The numerical computation method was introduced to study the quality of RSW joints. Yang et al. developed an integrated computational model by considering microstructural heterogeneity and residual stress of spot welds. The resistance spot weld process model was used to predict residual stress distribution, which was mapped into the local mechanical model.16 Ma and Murakawa simulated the RSW process by finite element method program developed with the coupling of the electrical, thermal and mechanical fields. The results showed that the nugget size and its formation process predicted by finite element method agree well with the experimental results.17 Eisazadeh et al. also developed a mechanical–electrical–thermal coupled model in a finite element analysis environment. Via simulating this process, the phenomenon of nugget formation and the effects of process parameters on this phenomenon were studied. Using this analysis, shape and size of weld nuggets were computed and validated by comparing them with experimental results.18
The objective of the present work is to introduce a new testing method, which monitors the structureborne acoustic emission (AE) signals of the physical phases and physical phenomenon in real time during RSW process, to obtain the characteristic information about the nugget formation and then predict the quality of spot welds.
Experimental
The microphone is used as the sensor to detect the SE signals. Because there is no contact between the sensor and the electrode or workpiece, the air is used as the transmission medium, as shown in Fig. 1a. Therefore, there are some losses for the SE signals detected in RSW. Especially, the microphone is not able to detect the SE signals from the solid–solid phase transformation and cracking. Therefore, there is a certain deficiency to study the nugget quality based on the SE signals.
a sonic emission signals sensing; b AE signals sensing
The AE signals can be detected by the piezoelectric sensor mounted on the electrodes or workpiece, which is structureborne sensing as shown in Fig. 1b. Because the electrodes or workpiece are used as transmission medium, the losses of AE signals of nugget nucleation are less. The sensor is sensitive enough to detect the AE signals from the solid–solid phase transformation and cracking.
The materials used in the experiment were 2024 aluminium alloys. The alloy plates had a thickness of 2 mm. The specification of samples was 25×100 mm. Two specimens were welded overlay by 25 mm lengths, as shown in Fig. 2. The RSW machine used in the experiment was a YR-A05CM2 ac device with a silicon controlled rectifier welding source. The materials of electrodes were Cu–Cr–Zr alloys, and the experimental measurements were made with new and old electrodes. The diameter of head–face for the new electrode was 6 mm. The head–face of the old electrode has been wearing away during use, and the diameter of the head–face was 8 mm. Specimens of the 2024 aluminium alloy were processed by chemical cleaning and drying before the welding experiment.
Specimen specification
The AE signals in the welding process were monitored in real time. The harmonic frequency of piezoelectric sensor used in the experiment was 150 kHz, which was mounted on the wall of the electrode as shown in Fig. 1b. The AE signals detected during the welding process were transferred to the computer for further processing and analysis.
Results and discussion
Characteristics of AE signals in RSW
Figure 3 shows the structureborne AE signals detected in a typical process of RSW on 2024 aluminium alloy. It can be found that the stage features of the welding process including the electrode loading event (arrow 1 in Fig. 3), nugget nucleation event (arrow 2 in Fig. 3) and electrode unloading event (arrow 3 in Fig. 3) are distinguished from the AE signals detected in the welding process. As the cracks are generated at the end of the solid–solid phase transformation, the cracking event is detected, which is indicated by the arrow 4 in Fig. 3.
Acoustic emission signals detected during RSW on aluminium alloy
The nugget nucleation event is composed of several AE subevents. The AE signals of nugget nucleation event pointed as arrow 2 in Fig. 3 are exacted as shown in Fig. 4. It can be found that the whole process of nugget formation corresponds to the AE signals of the nugget nucleation event. The area A shown in Fig. 4 presents several primary events of the nugget nucleation event in the AE signals, which corresponds to primary effects from the instantaneous current impulses affecting the materials, the materials melting, the nugget nucleation and growth. Therefore, the waveform characteristics contain a wealth of information about the quality of nugget formation, and we can make in-depth research of the quality characteristics of nugget formation by the AE signals detected in the RSW process.
Acoustic emission signals of nugget nucleation event
In order to study the quality of the nugget by the AE signals, two characteristic parameters of AE event including the AE count and the positive peak are introduced to evaluate nugget formation. A subevent is exacted from the AE event of nugget nucleation as shown in Fig. 5. It can be seen that the subevent is composed of a lot of impulses. We set a threshold value (shown in Fig. 5a) for the impulses of the AE subevent; the total count of the impulses with the peak value of greater than the threshold value is called the AE count of an AE subevent. A nugget nucleation event of AE signals as shown in Fig. 4 includes several AE subevents. The total count of the AE counts included in all the subevents of nugget nucleation is called the AE count of the nugget nucleation event. There is a positive maximum value in every AE subevent of nugget nucleation event such as the M value or the M′ value in Fig. 5b. The highest one is called the positive peak of nugget nucleation event. The AE signals released in the welding process are a manifestation of the energy release as the welding current affecting the materials to form the nugget. These two characteristic parameters are used to characterise the energy feature of the nugget formation in the RSW process and further to assess the nugget quality.
a AE count; b positive peak of nugget nucleation event
Curve fitting analysis
The spot weld strength depends largely on the nugget dimension for the RSW. But in light of the influence of the nugget microstructure, expulsions and welding defects, the spot weld strength is not entirely due to the nugget dimension. For the reason described above, the tensile–shear strength of spot weld and the characteristic parameters of AE signals are non-linearly related. In order to collect more experimental data, the quantity of experiment is about 150 times. The welding current ranges from 16 000 to 26 000 A, the electrode force ranges from 0·08 to 0·2 MPa and the current duration ranges from 0·12 to 0·20 s. The AE count and positive peak of the nugget nucleation event of the welding experiment are collected in real time, and the tensile–shear strength of the spot weld is measured by mechanical properties testing.
Figure 6 presents the relationship between the tensile–shear strength of the spot weld and the AE count of the nugget nucleation event. It can be found that the tensile–shear strength of the spot welds welded by the new and the old electrodes shows a similar relationship with the AE count. However, because the diameter of the new electrode is smaller, the nuggets growing up in the RSW process are smaller too and the bearing capacity of the spot welds is on the low side. The data distribution in Fig. 6 can be fitted using the following logarithm equation
Relationship between tensile–shear strength of weld and AE count of nugget nucleation event
Relationship between tensile–shear strength of weld and positive peak of nugget nucleation event
Prediction of spot weld strength
The fitting models above can be used to predict the tensile–shear strength of the spot weld according to the characteristic parameters of AE signals. It can be concluded from the scatter plot in Figs. 6 and 7 that the fitting models above are suitable for the prediction as the AE count or the positive peak is at a relatively low level. That is to say, the AE count or the positive peak at a high level does not necessarily fit above functional relations with the spot weld strength.
Equation (2) is a mathematical model for the tensile–shear strength of spot weld based on the AE count of the nugget nucleation event. The relationship between the measured strength and the strength estimated by the equation (2) is shown in Fig. 8. Equation (4) is a mathematical model for the tensile–shear strength of spot weld based on the positive peak of the nugget nucleation event. The relationship between the measured strength and the strength estimated by equation (4) is shown in Fig. 9. Tables 1 and 2 respectively show the statistical parameters of the fitting lines in Figs. 8 and 9.
Relationship between measured strength and strength estimated by fitting model based on AE count of nugget nucleation event
Relationship between measured strength and strength estimated by fitting model based on positive peak of nugget nucleation event
Statistical parameters of fitting line in Fig. 8
Statistical parameters of fitting line in Fig. 9
The slope value of the fitting line in Fig. 8 is 1·0559, which is closer to the constant 1 than the slope value 0·8828 of the fitting line in Fig. 9. The root square R2 of the fitting line in Fig. 8 is 0·9079, which is closer to the constant 1 than the R2 of 0·7057 of the fitting line in Fig. 9. It can be found from the data contrast in Tables 1 and 2 that the mathematical model based on the AE count has a good performance in the prediction of the tensile–shear strength of the spot weld, and the mathematical model based on the positive peak has a poor performance because of the greater predictive error.
For the prediction of spot weld strength based on the positive peak of the nugget nucleation event, the expulsion is an important factor to influence the accuracy. Figure 10 shows the whole profile of AE signals detected in real time during the RSW process with expulsions. The area B in Fig. 10, which is enlarged to display in Fig. 11, is the primary events of the nugget nucleation event in the AE signals. It can be seen from Fig. 11 that the nugget nucleation event includes several primary subevents, which correspond to the different effects as the welding current was affected during the process of nugget formation. Two primary subevents included in the area C in Fig. 11 are enlarged to display in Fig. 12. The expulsion takes place in the second subevent in Fig. 12 (or in the area C in Fig. 11). The instantaneous release of energy as the expulsion taking place will affect the positive peak greatly. It can be found from Fig. 12 that the positive peaks of the subevent fluctuate according to the expulsion impact. Especially, the strongest energy release of expulsion induces the largest positive peak M, which is larger than the M′ or M″ corresponding to the other subevents as the expulsion does not take place. The expulsion induces certain losses of molten metal in the nugget during melting and restrains the nugget growth, which reduces the bearing capacity of the spot weld. Therefore, although there is greater energy release during the nugget formation, the nugget formation process with expulsion corresponds to a spot weld with lower bearing capacity. The factors above make a negative effect to the prediction of the tensile–shear strength by the mathematical model based on the positive peak of the nugget nucleation event.
Whole profile of AE signals detected in real time during RSW process with expulsions
Main event of nugget nucleation event in AE signals detected in real time during RSW process with expulsions
Impact of expulsions on positive peak of nugget nucleation event in main event
Conclusions
The physical phases and physical phenomenon including the electrode loading, nugget nucleation, electrode unloading, expulsion and cracking can be characterised by the structureborne AE signals detected in the RSW process. The AE count and the positive peak of the nugget nucleation event are optional parameters to characterise the quality of nugget formation.
The AE count and the positive peak of the AE signals detected in the RSW on the 2024 aluminium alloy show a better relevance to the tensile–shear strength of the spot weld. The tensile–shear strength of the spot weld and the AE count of the nugget nucleation event are more in line with the logarithm relationship, and the tensile–shear strength of spot weld and the positive peak of the nugget nucleation event are more in line with the polynomial relationship.
The AE count and the positive peak of the AE signals detected during the RSW on the 2024 aluminium alloy can be used to predict the tensile–shear strength of spot weld based on the fitting models. The prediction of tensile–shear strength based on the AE count of nugget nucleation event has good performance. However, the prediction of tensile–shear strength based on the positive peak of nugget nucleation event is more susceptible to the expulsions, and the difference between the measured and estimated value will be greater when there are expulsions.
Footnotes
Acknowledgements
This work is supported by the Natural Science Foundation Project of Chongqing Science and Technology Commission of China (grant no. cstcjjA50013), the fund of the State Key Laboratory of Solidification Processing in NWPU (grant no. SKLSP201204) and the Major Project of Outstanding Achievements Conversion of Chongqing Colleges (grant no. KJZH11215).