Analysis of mechanism of plasma and spatter in CO2 laser welding of galvanized steel

Analysis of mechanism of plasma and spatter in CO2 laser welding of galvanized steel

Optics & Laser Technology 31 (1999) 119±126 www.elsevier.com/locate/optlastec Analysis of mechanism of plasma and spatter in CO2 laser welding of ga...

206KB Sizes 4 Downloads 30 Views

Optics & Laser Technology 31 (1999) 119±126

www.elsevier.com/locate/optlastec

Analysis of mechanism of plasma and spatter in CO2 laser welding of galvanized steel Hyunsung Park 1, Sehun Rhee* Department of Mechanical Engineering, Hanyang University, Seoul, South Korea Received 7 October 1998; received in revised form 5 January 1999; accepted 3 February 1999

Abstract In laser welding, quality, reproducibility and formability are required. That is the great problem in the automation of the laser welding process. Therefore, construction of an on-line process monitoring system of high accuracy is requested. The light which is emitted from plasma and spatter in laser welding was detected by photo-diodes. It was found that the light intensity depends on welding speed, laser power and ¯ow rate of assist gas. The relationship between the plasma and spatter and the bead shape, and the mechanism of plasma and spatter were analyzed for on-line laser weld monitoring systems. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: CO2 laser welding; Plasma; Spatter

1. Introduction During high speed CO2 laser welding, defects may occur due to the state of the cross section of the specimen, gap, ¯ow rate of the assist gas, travel speed and laser power. On the production line, a slight alteration of the welding condition produces many defects and it is very dicult for operators to detect the defects with the naked eye. The defects are monitored in real time, in order to prevent continuous occurrence of defects, reduce the loss of material and guarantee good quality. Also, the shape of the weld bead is the main factor in deciding the strength of the workpiece. Therefore, development of a monitoring system in laser welding which detects weld defects and bead shape is required. There are many methods for laser monitoring systems, such as the methods using acoustic emission [1], optical signal and image processing. The method using * Corresponding author. Present address: 17 Haengdang-dong, Sungdong-gu, Seoul 133-791, South Korea. Tel.: +82-2-2290-0438; fax: +82-2-2299-6039. E-mail addresses: [email protected] (H. Park), [email protected] (S. Rhee) 1 Tel.: +82-2-2290-0432; fax: +82-2-2299-6039.

optical signal is the main object of research in monitoring systems. Chen [2] showed the relationship between weld quality and signal strength by using ultraviolet and infrared signal. Also, he presented the signal change due to welding speed, laser power and assist gas. Beyer [3,4] researched the e€ect of assist gas and analyzed sound and plasma signals during laser welding. He also presented the relationship between full penetration of the weld bead and the plasma signal of the top and bottom of the plates. Miyamoto [5±7] detected the pit using plasma signal and he used the photo-diode, which has a maximum peak wavelength of 950 nm. He considered variation of the signal by placing the sensors in di€erent angles. Recently, Farson [8] presented the relationship between optical and acoustic emissions using the ARMA (autoregressive moving average) model. The fact that mechanism of the laser plasma and spatter varied according to welding conditions can be used as data in weld quality estimation. The mechanism of plasma of galvanized steel in the detection of weld quality hasn't been analyzed completely and, especially, the relationship between spatter and weld quality has not yet been reported. Therefore, there are

0030-3992/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 0 3 0 - 3 9 9 2 ( 9 9 ) 0 0 0 1 3 - 4

120

H. Park, S. Rhee / Optics & Laser Technology 31 (1999) 119±126

many problems in monitoring the quality of laserwelded automotive parts. The objective of this paper is to describe the mechanism of the plasma and spatter, as varied by welding conditions (travel speed and laser power), and the resulting change in the bead shape of galvinized steel by signal processing.

2. Theoretical background of measurement It is known that plasma and keyhole are important factors when laser energy is transferred to the workpiece. Also the weight of the spatter plays an important role in weld quality. When the pressure of laser plasma is over 104 Pa and it is in local thermo-dynamic equilibrium (LTE), the radiation of plasma and spatter in LTE state is similar to the black body radiation [6]. The peak wavelength in this form of radiation is described in Wein's displacement law. lmax T ˆ 2897 …mm K†

…1†

It is known that the band of wavelength of the laser plasma is 190 to 400 nm in CO2 laser welding [9]. Thus, we should choose the ultraviolet (UV) detector for plasma and infrared (IR) detector for spatter. The sensor for monitoring is photo-diode. The current of the photo-diode varies according to the intensity of the light within the response range. The measured current is converted to voltage, and is ampli®ed to the voltage of acceptable range for data acquisition.

3. Experiment

Fig. 1. Spectral response of UV photo-diode.

spatter ejected from the weld pool is similar to the temperature of the weld pool, therefore we chose the infrared range photo-diode, the wavelength is 0.7±1.7 nm (1700±4100 K). The ampli®ers were made for ampli®cation of the signals. Speci®cations of the sensors are listed in Table 1 and the spectral responses of photo-diodes are plotted in Figs. 1 and 2. Fig. 3 shows a schematic diagram of the signal measurement system. Sensors are attached to the moving part of the CO2 laser welding machine, to measure the signal of plasma and spatter. The sensors are placed in di€erent positions. There are two sensors for the plasma signal; the low aiming angle sensor (UV1) is for plasma plume and the high aiming angle sensor

Two types of photo-diodes were used in this experiment; ultraviolet and infrared range. The spectral response range of the ultraviolet photo-diode is 260 to 400 nm and of the infrared range photo-diode is 700 to 1700 nm. Spatter is the molten droplets ejected from the weld zone. It is generally known that the temperature band of the weld pool is from the melting temperature (about 1770 K) to the boiling temperature (about 3000 K). It is assumed that the temperature of Table 1 Speci®cations of the sensors

Type Spectral response range Peak sensitivity wavelength (lp) Dark current (maximum) Photo sensitivity (at lp)

UV photo-diode

IR photo-diode

GaAsP 260±400 nm 370 nm 50 pA 60 mA/W

InGaAs PIN 700±1700 nm 1550 nm 1.5 nA 0.95 A/W

Fig. 2. Spectral response of IR photo-diode.

H. Park, S. Rhee / Optics & Laser Technology 31 (1999) 119±126

121

Fig. 3. Schematic diagram of the laser monitoring system.

Table 2 Aiming angles and distances from welding focus to sensors

UV1 UV2 IR

Angle (8)

Distance (mm)

10 73 6

310 260 310

Flow rates of the assist gas (He) were 85 ft3/h (leading nozzle) and 40 ft3/h (tail nozzle). The focus of the laser beam is at the bottom of the specimen. The experiment is carried out 5 times for reliability. Chemical compositions of the test plate are listed in Table 4.

4. Results and discussion (UV2) is for plasma in the keyhole. Aiming angles and distances of the sensors are listed in Table 2. The data acquisition board used in order to receive the measured data through a PC is a 12 bit resolution and the sampling rate is 1000 samples/s. The power of the CO2 laser welding machine is 6 kW and the speci®cations are listed in Table 3. The specimen was galvanized steel of 1.5 mm thickness which is used in car bodies. Bean-on-plate welding was carried out at a speed range of 3 to 9 m/min at 6 kW. The power range was 3 to 6 kW at 3 m/min.

4.1. Signals of plasma and spatter Fig. 4 shows the acquired signals detected during laser welding at 6 kW, 6 m/min. The high voltage region of plasma plume signal (UV1) is shown. Approximately 0.5 s later, the signal settles down to a steady state. The signal of spatter (IR) also has a high initial voltage region. It is thought that the abnormality of the initial plasma generation and the transient range in the laser head transferal in¯uences this phenomenon.

Table 3 Speci®cations of the CO2 laser welding machine Laser power

TEM mode

Beam diameter (mm)

Focal length (mm)

Focus spot size (mm)

Focus depth (mm)

Focus number

Assist gas ¯ow angle

CW 6 kW

TEM01

30

350

0.31

3.67

11.67

208

Table 4 Chemical compositions of test plates

Galvanized

C (wt%)

Si (wt%)

Mn (wt%)

P (wt%)

S (wt%)

Fe (wt%)

Zn (g/m2)

0.0032

0.002

0.065

0.009

0.008

bal.

45/45

122

H. Park, S. Rhee / Optics & Laser Technology 31 (1999) 119±126

Fig. 4. Output signals of sensors (6 kW, 6 m/min).

Fig. 6. FFT of output signals of sensors (6 kW, 6 m/min).

The detected signal, when the travel speed is reduced to 5 m/min, is plotted in Fig. 5. As the travel speed decreases, UV2 and IR signals increase, but UV1 decreases. Particularly, the ac component of UV2 signal increases at 5 m/min. Figs. 6 and 7 show the fast Fourier transfrom (FFT) of the signals. Generally, the amplitude of the signals at 5 m/min are higher than the signals at 6 m/min. It is also shown that the ac component of signals increase at 5 m/min. Leong [10] suggested that oil contamination is the reason for the high frequency ac component. However, the cleanliness of the surface was guaranteed during the experiment.

Miyamoto [7] and Beyer [11] suggested that a hot spot is produced in the keyhole near the opening in CO2 laser welding and the hot spot moves up and down periodically. When the hot spot in the keyhole moves upward to the top surface of the workpiece, it expands and the plasma signal decreases. The increase of the ac component of the signal is due to this phenomenon. The results of FFT show that the frequency of the signal is below 50 Hz. The cut-o€ frequencies of the sensors are approximately 100 MHz and the frequency response of the ampli®er is 20 kHz. We can verify the propriety of the sampling rate. It is possible to use a 1

Fig. 5. Output signals of sensors (6 kW, 5 m/min).

Fig. 7. FFT of output signals of sensors (6 kW, 5 m/min).

H. Park, S. Rhee / Optics & Laser Technology 31 (1999) 119±126

123

Fig. 10. Penetration depth according to travel speed.

Fig. 8. Output signals of sensors (6 kW, 3 m/min).

kHz sampling rate with only a dc component of the signal for the detection of weld defects. Fig. 8 shows the signals at 3 m/min. We can ®nd that the dc level of the signal decreases. 4.2. E€ect of travel speed Fig. 9 shows the average voltage of the signals and photographs of bead cross sections according to the

travel speed. Beyer [4] suggested that the plasma signals increase as the heat input increases. However, the signals of UV2 and IR increase within the travel speeds of 9 to 4 m/min and fall abruptly at 3 m/min. The reason is that the signals increase as the heat input increases until 4 m/min, when the keyhole is opened and plasma leaks out from the keyhole at 3 m/ min. We can see in the bead cross section photographs of Fig. 9 that the penetration depth increases as travel speed decreases and full penetration occurs at 3 m/ min. Penetration depth and bead width according to travel speed are plotted in Figs. 10 and 11. As the speed decreases and, thus, heat input increases, the penetration depth also increases. Also, the width of the

Fig. 9. Signal of sensors according to travel speed (6 kW).

124

H. Park, S. Rhee / Optics & Laser Technology 31 (1999) 119±126

4.4. Distribution of signal and weight factor

Fig. 11. Bead width according to travel speed.

bead widens during partial penetration, and when full penetration is reached, the bead width narrows. The pattern is similar to that of the UV2 and IR signals. Although the heat input increases as the travel speed decreases, the UV1 signal shows a pattern of overall decrease. The reason can be found in the analysis of the signal in Fig. 12. If we do not use the assist gas, the UV1 signal will increase as travel speed decreases as did the UV2 and IR signals. Under the condition that the ¯ow rate of assist gas is supplied uniformly, if the volume of the gas per unit length increases as travel speed decreases, much dispersion of plasma occurs. The plasma plume is easily a€ected by assist gas, which is the reason for the reverse pattern of the UV1 signal.

When a correlation coecient is found in order to analyze the relationship between each signal, the correlation coecients of all combinations of the signals are under 0.5. That means signals are independent of each other and indicate the state of welding individually. Thus, the monitoring system needs an integrated defects inference algorithm using 3 signals of sensors. Fig. 16 shows the distribution of the signals according to travel speed. Each data shows the average of the signals from the sensors during welding and the circular symbol shows the average of the signals after 5 repetitions under identical conditions. The distribution of the signals indicates the uniformity of the signals in the experiment. UV2 signals have almost

4.3. E€ect of laser power Fig. 13 shows the signal change and bead cross sections according to laser power. Full penetration occurs at 4, 5 and 6 kW and partial penetration occurs at 3 and 3.5 kW. In partial penetration, the amount of plasma and spatter and bead size increases as laser power increases. However, the maximum signal of the UV2 and IR occur at 5 kW, in full penetration. This is because the volume of plasma and spatter is dependent on heat input below 5 kW, but over 5 kW it is dependent on the size of the keyhole, where a large amount of plasma and spatter leak out. Penetration depth and bead width according to laser power are shown in Figs. 14 and 15. The pattern of penetration depth is similar to the pattern of UV2 and IR signals and bead width increases as the laser power increases. The relationship between the bead shape and the plasma and spatter signal shows certain patterns. Thus, the signals can be applied to the regression analysis and the arti®cial neural network for the estimation of bead shape.

Fig. 12. Signal of sensors according to travel speed (no gas, 6 kW).

H. Park, S. Rhee / Optics & Laser Technology 31 (1999) 119±126

125

Fig. 13. Signal of sensors according to laser power (3 m/min).

Fig. 14. Penetration depth according to laser power.

Fig. 15. Bead width according to laser power.

uniform voltage, but UV1 have scattered data. The signal, which has a low degree of variance, is able to indicate the accuracy of welding. Thus we can make an index, which shows the accuracy of the signals. This index can be used as weight factors for weld quality monitoring systems. The standard deviation is calculated to show the distribution of each signal, in order to observe the degree of variance. However, if the distances of the sensors are changed, the standard deviation of the signals will also change. For normalization, the standard deviation is divided by the mean value of the signals. The normalized standard deviations are 0.109 (UV1), 0.03 (UV2) and 0.058 (IR). Therefore, in monitoring defects, the weight factors of the sensors are 0.15 (UV1), 0.55 (UV2) and 0.3 (IR).

126

H. Park, S. Rhee / Optics & Laser Technology 31 (1999) 119±126

5. Conclusions The volume of plasma in the keyhole and spatter increase with the increase of heat input and signals abruptly decrease at full penetration. Plasma plume signal decreases with the increase of heat input because of the e€ect of assist gas. The size of the keyhole a€ects the signal and bead shape. The relationship between the bead shape and the plasma and spatter signal have certain patterns. Thus, the signals can be used as data for estimation of bead shape.

References

Fig. 16. Signal distribution of sensors according to travel speed.

[1] Farson D, Hillsley K, Sames J, Young R. Frequency-time characteristics of air-borne signals from laser welds. Journal of Laser Applications 1996;1:33±42. [2] Chen HB, Li L, Brook®eld DJ, Williams K, Steen WM. Laser process monitoring with dual wavelength optical sensors. In: ICALEO '91, 1991. p. 113±22. [3] Gatzweiler W, Maischner D, Beyer E. On-line diagnostics of process-control in welding with CO2 lasers. In: High power CO2 laser system and applications, SPIE 1020, 1988. p. 142±8. [4] Maischner D, Drenker A, Seidel B, Abels P, Beyer E. Process control during laser beam welding. In: ICALEO '91, 1991. p. 150±5. [5] Mori K, Sakamoto H, Miyamoto I. Detection of weld defects in tailored blanks. Journal of the Japan Welding Society 1996;4:689±93. [6] Miyamoto I, Kamimuki K, Maruo H, Mori K, Sakamoto M. In-process monitoring in laser welding of automotive parts. In: ICALEO '93, 1993. p. 413±24. [7] Miyamoto I, Mori K. Development of in-process monitoring system for laser welding. In: ICALEO '95, 1995. p. 759±67. [8] Farson D, Ali A, Sang Y. Relationship of optical and acoustic emission to laser weld penetration. Welding Journal 1988;4:142s±8s. [9] Ono M, Nakada K, Kosuge S. An investigation on CO2 laserinduced plasma. Journal of the Japan Welding Society 1992;2:239±45. [10] Leong KH, Hunter BV. Characteristics of monitor for laser welding. In: Proceeding of Automotive Laser Applications Workshop, 1997. p. 26±35. [11] Beyer E, Behler K, Herziger G. Plasma absorption e€ects in welding with CO2 lasers. In: High power CO2 laser system and applications, ECO1 SPIE 1020, 1988. p. 84±95.