Metal Transfer During Gmaw With Thermal Pulsation Conem 2006

  • Uploaded by: Sérgio Barra
  • 0
  • 0
  • November 2019
  • PDF

This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA


Overview

Download & View Metal Transfer During Gmaw With Thermal Pulsation Conem 2006 as PDF for free.

More details

  • Words: 4,591
  • Pages: 10
IV C on g re s s o Na c i o na l d e E n ge n ha r ia M ecâ ni ca 22 a 25 d e A go s to 2 006 , R ec if e-P E

CHARACTERISTICS OF THE METAL TRANSFER DURING GMAW WITH THERMAL PULSATION Sérgio Rodrigues Barra1 Carlos Eduardo Aguiar Lima Rodrigues2 Augusto José de Almeida Buschinelli3 [email protected] 1

Integrated Center of Manufacture and Technology – SENAI Cimatec Av. Orlando Gomes, 1845 – Piatã. CEP: 41650-010. Salvador – Bahia - Brazil 2 Federal University of Uberândia – Laprosolda / UFU Campus Santa Mônica. Bloco 10. CEP: 38.400-902. Uberlândia – MG – Brazil 3 Federal University of Sanata Catarina – Labsolda / UFSC Campus Universitário da Trindade. Caixa postal – 476. CEP: 88040-900. Florianópolis – Santa Catarina – Brazil Resumo. This work evaluates the influence of thermal GMAW on the stability of the electric arc, the maintenance of the condition one drop per pulse (ODFP) and the weld pool stirring. Low carbon steel, Al-Mg alloy, AWS ER 70S-6 and AWS ER 5356 were adopted as base metal and consumable, respectively. Electronic power source, system of acquisition of welding signals (arc’s voltage, current and wire feeding speed) and synchronized shadowgraphy were used. The results show that, depending on the alloy and on the adopted operational package, the use of the thermal pulsation affects the ODFP condition, dephases the current wave and feeding electrode wave signals, vary cyclically the value of thearc’s length and increase the weld pool stirring, when compared to the conventional pulsation process. Palavras-chave: Thermal GMAW, thermal pulsation, electric arc, metal transfer 1. INTRODUCTION The use of the thermal pulsation in GMAW process was introduced in Brazil in the middle of the 90s (Dutra et al, 1995; Barra, 1998), despite literature already makes citation of this variant in the end of the 80s (Street, 1990). The process of thermal pulsation, taking into account the characteristics of the current pulse of the velocity of wire feeding, will be able to present, basically, two configurations in the operational package (Barra, 2003; Yamamoto et al, 1998; Dutra et al, 1995; Street, 1990). In the first wave form (see Fig. 1a) the cyclical change of the frequency of the current pulses and of the wire feeding speed is applied in a synchronized way, ranging from of 0,5 to 10 Hz, as a way to guarantee the maintenance of the arc’s length (Barra, 2003). The second form wave (Fig. 1b) is different from the previous one for the fact that the fixed wire feeding speed is kept in an intermediate level (Barra and Buschinelli, 2004; Barra, 2003; Yamamoto, 1998). Therefore, it is evident in Fig. 1 that thermal GMAW must be aligned to the advantages of welding using conventional pulsed GMAW with the advantages of the pulsed TIG welding (thermal TIG). However, the real advantages of the variables are still not well documented as much

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

Modulação do sinal da unidade de alimentação de arame Alta energia

Velocidade do arame

Velocidade arame

in the process (physical of the arc) as in the metallurgic field (phase of transformation and structural refining).

Sinal da unidade de alimentação de arame constante

Baixa energia

Corrente

Corrente

Tempo

Tempo

Impt Im pt

Imbt

Imbt

Impt - corrente média no pulso térmico Imbt - corrente média na base térmica

Modulação do sinal da unidade de pulso Im pt - corrente média no pulso térmico Im bt - corrente média na base térmica

Tempo

Tempo

(a)

(b)

Figure 1. The GMAW variant. In (a) wave I - synchronized modulation of energy and wire feed speed and in (b) wave II - modulation only in the energy. Notice the behavior of the Arc’s length (ll0). Additionally, the recent insertion of this variable as a new option of welding process creates a literature of nonsense in the standard form that represents it. Specifications as GMAW doubly pulsed, GMAW with thermal pulsation, GMAW pulsed with thermal pulsation, thermal GMAW or even interpulse, have been adopted by different authors (Barra, 2003; Yamamoto, 1998; Dutra et al, 1995; Street, 1990). Detailed information about the operational and metallurgic characteristics of the thermal pulsation, as well as proposals for standardization of the intrinsic parameters of the variable, are presented by Barra (2003). 2. EXPERIMENTAL PROCEDURE As the research aims to evaluate the possible advantages and limitations of the thermal pulsation, in relation to the conventional pulsation to the stability of the electric arc area and to the weld pool stirring, two forms of thermal pulsation waves were adopted, having one conventional pulsation wave as reference (see Fig. 2). The first form of thermal pulsation wave (Fig. 2b), called wave I, is characterized by the synchronized modulation of the frequency of current pulsation and of the wire feeding speed. The second form of thermal pulsation adapted wave (Fig. 2c), called wave II, presents only modulation in the signal of current pulsation, keeping the feeding value of the wire in an intermediate level. The supervision of the characteristics of the thermal pulsation waves was done with the immediate register of the source output signals (current – I, arc’s voltage – U and wire feeding speed - va ) and the high speed filming of the arc area behavior. In the first phase, oscillograms, containing signals of I, U and va, were obtained, via A/D converter, for all the conditions tested (see Fig. 2). In addition, this phase aimed to generate a database containing the behavior of the conventional pulsation waves and the waves I and II in different experimental conditions. The second phase, using the shadowgraphy technique, aimed to evaluate possible differences in the arc’s length, in the existence of instability in the ODFP conditions, and in the interfaces (passages) of the thermal pulse (pt) to the thermal base (bt) and vice-versa. Four packages of thermal pulsation were chosen (with values set at intervals of average current Im and va) with the objective to evaluate the two forms of waves proposed (wave I and wave II), in the welding of different alloys (low carbon steel and Al-Mg). The Table 1 relates the packages used

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

300

250

250

200

200

150

100

300

Thermal base

200

150

100

300

600

100

0 0

900

150

50

0

0

250

500

750

1000

0

30

30

30

Tensão (V)

40

Tensão (V)

40

20

20

0

0 600

900

250

500

750

0

1000

Velocidade arame (m/min)

Velocidade arame (m/min) 600

900

500

1000

750

1000

10

5

5

0

0 300

250

Tempo (ms)

10

0

750

20

Tempo (ms)

5

1000

0 0

Tempo (ms) 10

750

10

10

10

500

Tempo (ms)

40

300

250

Tempo (ms)

Tempo (ms)

Tensão (V)

250

50

50

Velocidade arame (m/min)

Thermal pulse

Corrente (A)

300

Corrente (A)

Corrente (A)

in the evaluation of the thermal pulsation possible effects. More detailed information about the deposition conditions can be got in Barra (2003).

0

250

500

750

1000

0

250

500

Tempo (ms)

Tempo (ms)

Tempo (ms)

(a)

(b)

(c)

Figure 2. Oscillograms containing the synchronized signals of I, U and va (Al-Mg alloy). Where: (a) conventional pulse, (b) wave I e (c) wave II. Table 1. The packages used in the evaluation of the stability of the thermal pulsation process. Wave

Alloy Low carbon steel

Wave I Al-Mg Low carbon steel Wave II Al-Mg

Thermal pulse

Thermal base

va-pt = 7 m/min e Impt = 212 A

va-bt = 3 m/min e Imbt = 119 A

vs = 25 cm/min, DBCP = 18 mm e Imt = 188 A va-pt = 7 m/min e Impt = 117 A

va-bt = 4 m/min e Imbt = 67 A

vs = 25 cm/min, DBCP = 18 mm e Imt = 94 A va-pt = 5 m/min e Impt = 212 A

va-bt = 5 m/min e Imbt = 119 A

vs = 25 cm/min, DBCP = 18 mm e Imt = 188 A va-pt = 5,5 m/min e Impt = 117 A

va-bt = 5,5 m/min e Imbt = 67 A

vs = 25 cm/min, DBCP = 18 mm e Imt = 94 A

Where: DBCP represents the distance between the contact tube to the piece and Impt, Imbt, va-pt and va-bt are the average current and the wire feeding speed in the thermal pulse and in the thermal base, respectively.

2.1. Procedure for filming in high speed (shadowgraphy) a) Formation of the image (shade) For the filming (visualization) of the metallic transference process, it is necessary to reduce the light generated in the arc’s area by means of a selective filtering that allows only the light in the

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

same wave length of the laser. As the welding torch is located between the laser source and the lens of the camera, a record of the shade generated by the wire, the drop and the weld pool will be made (see Fig. 3 and 4). Details about the synchronization of the I and U signals with the "synchronized shadowgraphy" image technique may be obtained in Vilarinho et al. (2000).

(a)

(b)

(c)

Figure 3. Details of the testing bench apparatus of laser shadowgraphy: (a) laser source and optic set; (b) Form of deposition and (c) camera and monitorial set.

(a)

(b)

Figure 4. Shade generated by laser beam (a) and metal transfer picture in arc region (b). b) Measuring and synchronization of the signals of the camera’s voltage with the signals of I, U and va For the precise determination of which image (picture) would accurately correspond to one definite point in the waves of arc’s voltage and current, a software to capture and synchronize the signals was used (Vilarinho et al., 2000). After the synchronization is done, the program generates oscillograms of the I, U values with their respective numbers in the pictures, Fig. 5. Thus, the picture (digital image) corresponds to one definite wave point, which will be identified by its written number. The conditions used in the signals survey were: Acquisition in 12 bit and 10 kHz; time of acquisition 5 s, camera shutter release tax of 2000 pictures and exposure of 1/20000. 3. RESULTS AND DISCUSSION 3.1. Variation in the value of the arc’s length (ll0) The use of the conventional pulse, following the basic premises of ODFP, Txa = Txf (equality between the fusion tax and the feeding tax) and a value of base current, necessary to the maintenance of the arc, does not induce the significant variation in the arc’s length (ll0) in none of the studied alloys. In the form of wave I, the value of l0 does not present, in general, a significant variation between the phases of thermal pulse (pt) and thermal base (bt), making a steady arc’s region. This

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

characteristic is related mainly to the imposition of the synchronized modulation between the signals of average current (Im) and the wire feed speed (va), to the related phases (see left side of Fig. 6). However, because of the differences normally presented in the dynamics (inertia) of the signals of current (dI/dt) and in the wire feeding speed (dva/dt), a small variation of l0 in the interfaces pt/bt and bt/pt may occur. Oscillograms with the synchronized signals (full line – I and U and dotted line – image)

Referring picture * (red dotted line)

300

I (A ) e S inal C âmera

250

200

10

150

100

0

2

4

6

8

10

12

14

16

18

20

22

24

26

50

0

2

4

6

8

10

12

T e m p o [s ]

x 10

-3

18

45

U (V) e Sinal C âmera

40

0

2

4

6

8

10

12

14

16

18

20

22

24

26

35

30

20

25

20

15

10 0

2

4

6

T e m p o [s ]

8

10

12 x 10

-3

24 Comment: ODFP condition is seen and that the detachment is carried through in the final third of the current pulse (image 20). * The acquired pictures are turned of 90° in the counter clockwise direction.

Figure 5. Example of shadowgraphy technique. The establishment of the ODFP condition is seen jointly with weld pool undulation (Al-Mg alloy). As the inertia in the wire feeding (dva/dt) is greater than in the current signal (dI/dt), a small instability in the initial instants of each phase (pt and bt) is seen. At the beginning of the thermal base phase, there is a slight increase in the projection of the wire (ll), because of the fast change in the value of the average current (interface between Impt and Imbt), while the variation in the va value is relatively slower (interface between va-pt and va-bt). In the case of the interface from bt/pt, in the beginning of the thermal pulse phase, the shown mechanism is inverted. In these interfaces, the signal (oscillogram) of va presents a peak in the initial instants of each phase, as a result of the inertia of the wire feeding system used (see Fig. 2(b)). Good to mention that the intensity of this form of instability in the transition between the phases (variation in l0) will be related to the intensity of the instant loss of condition Txf = Txa, which causes a variation in the projection of the wire in relation to the contact tube (variation in stickout l). Because of the apparent stability presented in the arc’s region, the wave I does not show limits in its used field and, because of the maintenance in l0 (U ≈ constant), alterations in the superficial aspect (scales), in the weld bead geometry (penetration, reinforcement and width), in the microstructure of the fusion zone, and in the interval of one thermal period (Tt = tbt + tpt), will be generated, in principle, only by signals modulation of the Im and va.

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

Behavior of l0 – Wave I (a) Thermal base (low carbon steel) A

Behavior of l0 – Wave II (b) Thermal base (low carbon steel)

B

C

A

C

A

B

(c) Thermal base (Al-Mg alloy) A

(d) Thermal base (Al-Mg alloy)

B

300

D

E

250

Corrente (A)

C

C

B

F B

A

C

200

Illustrative oscillogram of the wave of current (I) (localization of the images)

150

100

50 300

600

900

1200

1500

Tempo (ms) 40

40

Oscillogram of U – Wave II

Oscillogram of U – Wave I 30

Tensão (V)

Tensão (V)

30

20

20

10

10

0 200

0 300

600

900

1200

1500

Tempo (ms)

40

Al-Mg alloy

400

500

600

700

800

Tempo (ms)

Al-Mg alloy

30

Tensão (V)

30

Tensão (V)

300

40

20

20

10

10

0 300

0 400

600

800

1000

1200

1400

450

1600

D

E

E

750

900

1050

Low carbon steel (b) Thermal pulse (low carbon steel)

F

D

F

D

(c) Thermal pulse (Al-Mg alloy) D

600

Tempo (ms)

Low Carbon steel (a) Thermal pulse (low carbon steel) Tempo (ms)

E

F

(d) Thermal pulse (liga Al-Mg alloy) E

F

Figure 6. Behavior of the arc’s length (ll0) according to the material and to the form of pulsation wave. For the form of wave II, the variation in l0 occurs during all thermal period (Tt), generating instability in the arc’s region. As the wire feeding speed is kept constantly and the current signal varies periodically, between Impt and Imbt, there will be predominantly loss in Txf =Txa condition.

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

Thus, for the thermal base phase, as TxfTxa, l0 will increase, causing the reduction in l and making the fusing of the wire with the contact tube possible, see right side of Fig. 5. The experimental results demonstrated that the adjustment in wave II is more complex compared to the form of wave I and that its range of operation – adopted relations for Imbt x tbt and Impt x tpt – will be limited, on one side, by the possibility of short circuit (ll0 = 0) and, on the other, by the fusing of the wire with the contact tube (ll= 0). Generically, for that it does not occur, the value of tbt will have to be reduced and always lower than the value of tpt and, for the used value for tpt, the concern will be to prevent the fusing of the wire with the contact tube (extreme value of l0). As wave II is characterized by the cyclical change in the l0 value, the instability presented in the arc’s region will be the function of the combination of the effects generated by the loss in the Txf =Txa and by the variation in the value of the arc’s voltage (cyclical variation in U). Therefore, alterations in the way of metal transferring, in the superficial aspect (scales), in the weld bead geometry (penetration, reinforcement and width) and in the microstructure of the fusion zone, during a thermal period (Tt), will be originated by the variation, complexity and combination of the values of Im, U and l (Joule effect). 3.2. Loss of the condition of one drop transferred per pulse (ODFP) In the form of wave I, in the implemented experimental conditions, it was observed that the possibility of loss of ODFP condition is most likely to occur at the beginning of the thermal base phase. This characteristic demonstrates a connection with the "thermal inertia" left by the thermal pulse phase. Probably, the wire initiates the bt hot and, as a consequence, it promotes conditions for the release of more than one drop, in the initial pulses, until the phase (bt) starts the regime (see Fig. 7). At this point, it is important to emphasize that this effect was seen only in the most resistant material (low carbon steel). In the less resistant material, Al-Mg alloy, ODFP was kept and the only noticed variation was the occurrence of a small increase in the diameter of the drop in the tip of the wire in the beginning of bt. In the form of wave II, the cyclical variation in the value of the wire projection (ll), caused by the imposition of a fixed value of va and of the modulation in Im, causes a variation in the part of heat generated in the wire, by Joule effect (see Eq. 1). In this condition, it is expected that the oscillation of the heat produced by Joule effect influences the instability of the ODFP condition, mainly in the welding of materials with high resistivity. Txf = α.I+β β.ll.I2

(1)

Where: I - current of welding; l - projection of the wire in relation to the contact tube (stickout); α - constant characterizing anode reactions and presents relation with the type of gas, electrode and polarity; β - constant associated with the resistive heating of the electrode.

The experimental observations demonstrated that in the form of wave II, because of the cyclical variation in l value, for materials with large electric resistance, in the thermal base as well as in the thermal pulse, the appearance of three distinct regions of the drops detachment may occur, or either:  Region 1 (high values of l0 and U and low value of l) In the end of the thermal pulse and in the beginning of the thermal base, where l presents the lower value, the drop will be detached with an irregular form and with bigger diameter than the diameter of the electrode (wire), after the occurrence of more than a current pulse. For details see points A and F, in the right side of the Fig. 8 and Fig. 9;

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

Region 2 (intermediate values of l0, U e l) In the intermediate part of the thermal pulse and the thermal base, the drop will be detached with a regular form and equal diameter to the one of the electrode, after the occurrence of a current pulse (ODFP maintenance). For details see points B and E, in the right side of Fig. 8;  Region 3 (low values of l0, U and high value of l) In the final part of the thermal base and in the initial of the thermal pulse, more than one drop will be detached with regular forms, but with different diameters, after the imposition of one current pulse. For details see points C and F, in the right side of Fig. 8. 

In the welding of the Al-Mg alloy, using the form of wave II, it was not verified the loss of ODFP condition, during all the thermal period (Tt). It was only noticed the variation in the form and the diameter of the drop in the interface pt/bt, and a tendency for the neck in thermal pulse phase to get longer (see left side of Fig. 8). Possibility of loss of ODFP condition (Wave I - Beginning of the thermal base (region A)) (a) Low carbon steel 275

269

279

(b) Al-Mg alloy 2195

2186

350

2197

250

F in al do p ulso térmico

Fin al do pu lso térmico

P rimeiro pu lso da base térmica

300

P rimeiro pu lso da base térmica

200

250

150

200

S inal d a câmera

150

100

100 230

240

250

260

270

2160

2170

2180

2190

2200

280

50

50

0.115

0.125

0.135

0.145

1.08

1.085

1.09

1.095

1.1

Figure 7. Effect of wave I on ODFP. In (a) loss of ODFP (high wire electric resistivity) and (b) ODFP maintenance (low wire electric resistivity). Possibility of loss of ODFP condition during Tt – Wave II (a) Thermal base – Al-Mg alloy B

A

C

A

(a) Thermal base – Low carbon steel B

400

250

D A

B

E

F

350

C

D

F

E

A

B

C

300

Corrente (A)

Corrente (A)

200

C

150

100

250 200 150 100

50 50 0

0 200

300

400

500

600

700

800

300

900

D

(a) Thermal pulse – Al-Mg alloy E

450

600

750

900

1050

Tempo (ms)

Tempo (ms)

F

D

(b) Thermal pulse – Low carbon steel E

F

Figure 8. Instability in ODFP condition, caused by the variation in l and resistivity (material type).

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

Low carbon steel Loss of ODFP in thermal pulse phase (Wave II) End of first pulse

Begin of second pulse

End of second pulse

Detachment still in base current

Figure 9. Drop presenting great diameter and detachment after multiple current pulses (detachment absence after some pulses - loss of ODFP). 3.3. Weld pool stirring The use of the conventional pulsation produces a constant agitation in the weld pool. For the thermal pulsation, it is noticed the occurrence of different degrees of stirring in the pool, between the pt and bt phases. It is necessary to evaluate "how" and "how much" the thermal pulsation may affect the level of agitation of the weld pool, once this instability in the liquid metal may generate differences in the geometry of the weld bead and in the microstructure in the fusion zone. Figure 10 illustrates, for wave I, the difference generated in the level of agitation of the weld pool. Notice that the instability of the pool will be bigger in the thermal pulse (see the undulation difference of the pool). Stirring in weld pool – Wave I (Al-Mg alloy) Thermal base (region B)

Middle of tp

Interface of tp with tb

Middle of tp

Interface of tp with tb

Middle of tb

End of tb

Thermal pulse (Region E)

Middle of tb

End of tb

Undulation in Ondulação weld pool na poça

Figure 10. Agitation difference in weld pool (superior pictures – low agitation in thermal base; inferior pictures – high agitation in thermal pulse). In the form of wave I, because of the modulation in the values of Im and va, it is supposed that many factors contributed for the difference in the weld pool stirring, that is, the turbulence in the pool will vary according to the change in the values of the electromagnetic force (Fem), the force due to superficial tension gradient (Fγ), the force due to convection in the liquid flow (buoyancy Fb), pressure of the arc (Fpa), and frequency of detachment of the drops, between the pt and bt phases. For the form of wave II, the operating mechanisms demonstrated a bigger complexity. Besides the described variation above, there will be, still, the effect produced by the cyclical change in the values of the welding voltage (U), of the projection of the wire (ll) and of the distance of detachment of the drop (variation in the length of the arc – l0). 4. CONCLUSION In the adopted experimental conditions, the results make you infer that:

IV Congresso Nacional de Engenharia Mecânica, 22 a 25 de Agosto 2006, Recife-PE

a) The maintenance in the value of the arc’s length (ll0) will depend on the type of adopted thermal wave; b) In the form of wave I, the loss of ODFP condition may occur in the beginning of the thermal base (combination of "thermal inertia" and resistivity). For the form of wave II, the ODFP maintenance is more complex because of the cyclical variation in l0; c) In comparison to the conventional pulse, the use of thermal GMAW modifies the level of agitation in the weld pool. 5. ACKNOWLEDGEMENT Professor Américo Scotti of Federal University of Uberlândia (UFU) is thanked for fruitful discussions. CNPq and CAPES for financial support. The welding teams of welding laboratories of UFSC and UFU are gratefully acknowledged for helping with welding experiments. 6. REFERENCES Barra, S. R. e Buschinelli, A. J., 2004, “MIG/MAG Térmico: Efeito da Distância Entre Pulsos, Corrente Média Total e Desnível Térmico”, Anais do III Congresso Nacional de Engenharia Mecânica, Belém, Brasil. Barra, S. R., 1998, “Influência dos Procedimentos de Soldagem sobre a Resistência à Cavitação de Depósitos Obtidos com a Utilização de Arames Tubulares de Aços Inoxidáveis Ligados ao Cobalto”, Dissertação de Mestrado – Programa de Pós-Graduação em Engenharia Mecânica, UFSC, Florianópolis, Brasil, pp. 1-132. Barra, S. R., 2003, “Influência do Processo MIG/MAG Térmico Sobre a Microestrutura e a Geometria da Zona Fundida”, Tese de Doutorado – Programa de Pós-Graduação em Engenharia Mecânica, UFSC, Florianópolis, Brasil, pp. 1-209. Dutra, J. et al., 1995, “O Processo MIG/MAG Pulsado com Pulsação Térmica”, Anais do XXI ENTS, Caxias do Sul, Brasil, pp. 889-902. Street, J. A., 1990, “Pulsed Arc Welding”, Abington Publishing Special Report, Cambridge. England, pp. 1-57. Vilarinho et al., 2000, “Development of an Experimental Technique for Studying Metal Transfer in Welding: Synchronized Shadowgraphy”, The Int. Journal for Joining of Materials, Vol. 12, No. 1, JOM, Denmark, pp.1-12. Yamamoto, H. et al., 1998, “MIG Welding of Aluminum – Process and Power Source”, IIW Doc. XII-1433-98. 5. RESPONSIBILITY FOR THE INFORMATION The authors are the only responsible for the printed material included in this paper.

Related Documents


More Documents from "Daisy"