LIBRARY “— Michigan State L University PLACE iN RETURN BOX to remove this checkout from your record. TO AVOID FiNES return on or before date due. DATE DUE DATE DUE DATE DUE We. ML _Ji _fi.____i fl =J ll MSU is An Affirmative Action/E qua! Opportunity institution chS—pA CONTROL O? HIIIDIALS PDOCISSINO VIRIABLSS IN PfiODUOTIOl IBSISTLNOS SPOT IILDING BY Michael John larcgoulie A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Metallurgy, Mechanics, and Materials Science 1991 5“." E‘— <‘ r91 — F’- -4-/ odd 0/2:- ABSTRACT CONTROL OF MATERIALS PROCESSING VARIABLES IN PRODUCTION RESISTANCE SPOT WELDING 3? Michael John Xeragoulis How can high-volume manufacturers resistance spot weld sheet steels with maximum quality, repeatability and efficiency? What are the key variables of this process? What are the requirements for feedback control? These are a few of the technical problems relevant in today's automotive industry. Effective answers to these questions stem from the knowledge of spot welding metallurgy, and the physical and cultural environment of the shop floor. In the spot welding process, some of the classical variables are: material weldability, zinc coatings, metal fit, sheet thickness, electrode alignment, force, time, current density, cooling, and others. This work focused on ranking the known variables according to their relative influence on process control. Consequently, a number of variables were characterized in both laboratory and plant environments. The results show that from a process control standpoint, peak efficiency and quality occur when operating close to the expulsion limit. Surprisingly, when welding about the expulsion limit, the key variables were found to be in the equipment maintenance area, rather than in the material properties and coatings area. Thus, through analysis, the long list of key variables was reduced to the following four important maintenance variables: 1) Electrical resistance of the total circuit. 2) Cooling water efficiency. 3) Mechanical condition of the weld guns. 4) Operating current. A distant fifth variable was metal fit, but only when metal fit was extremely poor. Additionally, it was found that an electrode misalignment of less than 40% had little impact on quality, provided the process was operating at expulsion. And, the power factor was found to be an effective feedback signal for keeping the process centered at the expulsion limit. The results of the study, conducted in both laboratory and plant environments, are discussed at length in this dissertation. The recommendations were successfully implemented in a large automotive plant. f e e P C. be a». 7. Q. If at (4].. flinfaalfi Tclflu v: 7.2.. q. ACKNOWLEDGMENTS First of all I would like to recognize my wife Lynne for her unfailing support, forbearance, prayers and encouragement throughout this project. I would like to express appreciation to my academic advisors, including Professor Kali Mukherjee and the faculty and graduate students at Michigan State University. I would also like to thank Dr. Jerry Gould of the Edison Welding Institute, and Professor Thomas Eagar of the Massachusetts Institute of Technology. I am profoundly indebted to my steadfast supporters at General Motors, including Michael Camp, James Armstrong, Richard Conley, Gary Sweeley, Dave Stevenson, Gerald Wood, Robert Angeli, John Stark, John Headley, Jerry Graul, Wayne Grandy, Robert FitzPatrick, Dave Purchase, James Gilbert, Bruce Kelly, Greg Nagel, and many, many others. Thanks to Medar, Inc. and Physical Acoustics Corp. for their willing assistance with the feedback controls. And a very special thanks to the enthusiastic members of the United Auto Workers, Locals 652 and 1618, who willingly sacrificed their time and creative energy to make this project a success. Without the help of each of the above persons, I could not have completed this work. This dissertation is therefore dedicated to all of them, toward the continuous improvement of the American manufacturing industry. Many thanks also to Dr. Parwaiz A. A. Khan, my friend and colleague, for his guidance and assistance in proofreading this manuscript. iii f—"C LIST OF LIST OF TABLE OF CONTENTS FIGURES TABLES 1. INTRODUCTION 1-1 1-2 1-3 Overview Spot Welding Variables Project Summary 2. LITERATURE REVIEW 2-1 2-2 2-3 2-4 NNN l \lO‘UI 2-8 Weld Microstructure Contact Resistance Heat Flow Fundamentals Current and Time - the Weld Lobe 2-4-1 Dynamics of Heat Generation 2-4-2 Dynamics of Heat Flow 2-4-3 Material Property Limitations Weld Force and Weld Pressure Fundamentals Mathematical Modeling Process Monitoring and Feedback Techniques 2-7-1 The Classical Dynamic Resistance Curve 2-7-2 A Theory for the Fundamental Changes at Expulsion A Novel Technique for In-Situ Measurement of Temperature and Voltage 3. EXPERIMENTAL STRATEGY AND SETUP 3-1 UUOJUUUN l QQGMéUN E hhbb III bUNH Goals and Strategy Laboratory Apparatus Production Apparatus Weld Microstructure Study Procedure Contact Resistance Measurement Procedure Weld Lobe Procedure Electrode Misalignment Study Materials and Electrodes Used in Plant Study BORATORY RESULTS AND DISCUSSION Weld Microstructure and Hardness Contact Resistance Data Weld Lobe Data Misalignment and Force Effects 4-4-1 Misalignment Effect 4-4-2 Force Effect iv Page vi xii UBOP’H 10 15 17 20 21 26 29 31 32 42 43 44 51 51 52 56 56 57 57 58 62 63 63 69 71 75 78 79 dlrfllduil. 4-5 Process Monitoring and Feedback Techniques 4-5-1 Power Factor Drop (PFD) 4-5-2 Acoustic Emission (AE) 4-5-3 Loop Efficiency U|U|UIUIUIUI'U LANT RESULTS AND DISCUSSION -1 Floor-based Research Strategy -2 Process Centering -3 Process Desensitization -4 Remaining Variables -5 Setup Guide -6 Process Monitoring and Feedback Techniques 5-6-1 Power Factor Drop (PFD) 6. CONCLUSIONS 7. RECOMMENDATIONS FOR FUTURE WORK REFERENCES APPENDICES APPENDIX APPENDIX APPENDIX APPENDIX APPENDIX APPENDIX General Motors Weld Lobe Procedure MOS-247 Mathematical Solution to the Common Area of Winking Circles Rewards and Benefits from Implementation of Findings in General Motors Plants A Technical Summary of the Resistance Spot Welding Process 0.1 The Physical Process D.2 The Electrical Process D.3 The Metallurgical Process D.4 The Impact of the Cultural Manufacturing Environment Elements of Process Control in Manufacturing Conclusion About Applied Research Strategies 81 82 90 93 99 99 103 105 106 110 110 110 117 119 120 127 142 152 170 170 171 179 180 184 186 Figure 1.1 LIST OF FIGURES Fifty four resistance spot welding process variables, arranged according to the "Four M's" of manufacturing processes. A normal spot weld in SAE 1005 DQAK cold rolled steel, 0.030 inch (0.75 mm). Microstructure shows a cast nugget formation, (Sample no 323). Photomicrograph of nugget edge, showing a rotating freezing direction. The upper sheet is cold rolled SAE 1005, the lower sheet is normalized SAE 1005. Schematic of resistances in a spot weld. Experimental model showing the onset of sheet separation with time. Figures c) and d) represent greater weld time, (Ref 20). a) Seam welding transition speed from Yamamoto and Okuda's experimental model, using bare steel. b) Measured internal peak temperature profiles for the two modes, (Ref 21). Typical Weld Lobe (Appendix A). A one-dimensional temperature distribution model for spot welding, a) thin sheet model, b) thick sheet model, (Ref 24). Typical nugget formation resulting from a 3:1 thickness ratio. This is a production weld of 0.040 inch (1 mm) to 0.120 inch (3 mm) drawing quality steels. Both sheets were hot-dipped galvanized. Nominal production parameters were 10 ka for 20 cycles at 600 pounds (261 kg). Notice how the nugget center lies at the mid-point of the stackup, rather than at the weld interface. vi Page 8 11 14 19 23 24 Greenwood's calculated temperature distributions after various weld times. a) 0.25 sec, b) 0.1 sec, c) 0.6 sec, (Ref 12). Schematic representation of dynamic feedback monitoring. Load cell displacement sensor to detect nugget expansion during the weld, (Ref 33). Nugget expansion profile as the weld develops, as a function of weld time, in cycles. The etched cross-sections (left side) show nugget development. The oscillograph data (right side) show vertical electrode movement during the weld. Notice the abrupt collapse at expulsion, letter 1, (Ref 33). Schematic illustration of acoustic emission sensors on weld gun, (Ref 34). Schematic signature of acoustic emission during spot welding, (Ref 34). Classical dynamic resistance curve, (Ref 35) Welding electrodes containing ultrasonic transducers in a pitch-catch configuration, (Ref 37). Schematic of the setup used for in-situ measurement of temperature and voltage, using half weld instrumentation, (Ref 52). Schematic of beadless microthermocouple attachment used for in-situ measurement, (Ref 53). In-situ temperature fields in full-size weld nuggets. The temperature field was measured at the peak of nugget size by the half-weld technique, (Ref 53). Convection in liquid nugget. Steady state convective heat transfer and heat conduction at a boundary surface, measured by the half-weld technique, (Ref 53). Convection swirling pattern proposed by Alcini, (Ref 53). vii 33 34 36 37 38 39 40 41 46 47 48 49 50 Photographs of laboratory spot welder. a) overall setup, b) closeup view of welder. Photograph of laboratory welder, showing weld tip arrangement. The lower electrode is stationary. Notice the acoustic emission sensor on the upper electrode shank. Electrical schematic of laboratory welder, with data acquisition computer. Sketch of gun translation device used for misalignment study. Photograph showing gun translation device installed on upper platon. Holders used to prevent sample rotation during misalignment study. Coupon weld of hardened SAE 1080 steel, welded to cold rolled SAE 1005, a) sketch, b) photomicrograph. Tukon hardness profiles of sample from Figure 4.1. Coupon weld of heat treated SAE 1005, welded to cold rolled SAE 1005. a) sketch, b) photomicrograph. Tukon hardness profiles of sample from Figure 4.3. Preweld contact resistance versus pressure. Weld lobe number 1 - a normal weld lobe for bare steel, under the conditions stated in Chapter 3. Electrode pressure I 10.2 ksi, (70 MPa). Force = 500 lbs, (217 kg). Photomacrograph showing the result of moderate expulsion. This is a laboratory weld, of 0.030 inch (0.75 mm) SAE 1005 uncoated steel. Parameters were 8.3 ka for 14 cycles at 500 pounds (217 kg). Notice the electrode indentation, and the expulsion "whisker" extruded from the weld interface, (Sample no 544). viii 53 54 55 59 60 61 64 65 67 68 70 72 73 Hot weld showing over-indentation. The material and parameters were identical to Figure 4.7, except that current was 15.8 ka. The electrodes stuck to this sample and left behind pieces of copper on the surface, (Sample no 333). Schematic of the experimentally misaligned condition. Weld lobe number 2 - 40% misaligned electrodes. Notice the leftward lobe shift caused by the misalignment. Still, lobes 1 and 2 share roughly an 80% overlap, when the upper limit is taken to the sticking/indentation point. Data from lobes 1 and 2, showing average current density as the abscissa rather than current. Force = 500 lb (217 kg) in both lobes. These data show the significant effect of changing the average pressure at the weld. These data approximate the magnitude of lobe shift that would have been measured by a 2x increase in weld force, for aligned electrodes. Dynamic power factor for a laboratory weld on SAE 1005 bare steel, 0.030 inches (0.75 mm). This weld did not have expulsion. The largest one-cycle drop in power factor near the end of the weld (when expulsion normally occurs), was 0.85%. Theoretical relationship between power factor and R/X . Tangents along short sections of this cu e may be used to linearly approximate this relationship. In addition, since X is largely constant during welding, it eprains why the power factor curve mirrors the resistance curve during welding. Dynamic power factor for a laboratory weld on SAE 1005 bare steel, 0.030 inches (0.75 mm). This weld had visible expulsion. The largest one-cycle drop in power factor near the end of the weld is 6.02%. Logic flow diagram for automatic feedback stepping by the power factor drop (PFD) technique. Correlation data between PFD signal and visual expulsion. These 32 data points were from 32 consecutive welds made under the conditions stated in Chapter 3. ix 74 76 77 80 83 86 91 92 4.17 The measured relationship between primary current and percent heat, for the laboratory welder at tap l on bare steel. There are 455 separate data points shown; each point repre- sents one cycle at the indicated heat. The tolerance band is calculated for the 95% confi- dence limits of the individual points. Joule's Law would imply a parabolic relationship; in reality however, the data do not agree with Joule's Law. The empirical relationship between percent heat and primary current may be expressed for the data of Figure 4.17: K = I /(%H), where n = 1.43 and K = 5.82 (+/- 0.49). Actual production data using the PFD feedback control. The material was hot-dipped galvanized steel 0.024 inch (0.6 mm) welded to 0.079 inch (2 mm). The weld tool was of the press/fixture type, with a large secondary loop. Notice how the stepper stepped both up and down, in order to maintain an average of 22% expulsion. a) individual data, b) histogram. Actual production data using the PFD feedback control. The materials were identical to those in Figure 5.1. The weld tool was also similar. Notice how active the stepper was for the first 125 welds. Because the initial heat setting was too low, there was little expulsion. Later the system balanced, for a total of about 15% "expulsion welds". a) individual data, b) histogram. Standard, simplified stepper curves deployed in the plant for non-feedback applications. Actual production data showing PFD feedback out of control. The material was hot-dipped galvanized steel 0.024 inch (0.6 mm) welded to itself. The stepper was initially disturbed by frequent false PFD's, until the operator intervened at about 275 welds. a) individual data, b) histogram. Model of "winking" circles. In the "winking" model of electrode misalignment, the electrodes are allowed to shift out of alignment by distance m, while the axes remain parallel. 95 96 112 113 114 116 82 B3 The quarter area A(m)/4 is 1/4 of A(m), the contact area at misalignment m. The summation of rectangular elements by calculus. The remaining contact area of misaligned electrodes. Schematic of resistance spot welding (Ref 62). Waveforms in resistance welding (Ref 63). Typical single phase SCR wiring diagram for resistance welding (Ref 63). xi B4 BS B6 DZ DS D7 Table LIST OF TABLES Page Example of cycle by cycle output produced by 88 the weld control. Each line represents one cycle during the weld. Example of end of weld summary output produced 89 by the weld control. Each line represents one weld. Floor setup guide that was refined and used 102 during the plant study. These are good empirical parameters for process control. xii 1. INTRODUCTION 1-1 Overview Resistance spot welding is the most popular method of joining sheet metal. This opinion is based upon the number of welds made per year and the annual material consumption of the major users of the spot weld process. The automotive industry for example, is by far the largest tonnage user of steel. It is estimated that the auto industry alone produces 90 billion spot welds per year, based upon a conservative average of 3000 spots per vehicle and a worldwide annual production rate of 30 million vehicles. Some of the reasons for the popular success of this welding process are the following: a) Relatively high energy density and rapid welding. b) No filler metal or shielding gas. c) Low heat input and low distortion. d) Relatively low cost of equipment and operation. e) Simplicity and maintainability of equipment. f) Highly automatable and flexible. Unfortunately, as with all high volume manufacturing processes, efficiency plays a major role in the cost of production. In spot welding, one of the main efficiency issues is weld quality. Weld quality is defined primarily in terms of the average bond diameter at the weld interface. Weld quality is important because it affects manufacturing costs in many ways. For example, since only acceptable parts can be used, weld quality directly affects rework, scrap, downtime, inventory costs, and the proliferation of redundant welds. However, if it should become possible to improve the control of weld quality on a broad scale, engineers would in time develop more streamlined products and processes that require less material, fewer welds, less rework, and less scrap to produce comparable vehicles. Significant cost savings could be achieved through the elimination of redundant operations, which would allow for less tooling, fewer operations, less floorspace, less manpower and less administration. Thus greater efficiency will be enjoyed by manufacturers who learn to optimize and control their weld processes. As a general rule, process control is the key to greater efficiency in manufacturing. But process control always requires adequate knowledge of the process. Process knowledge is only adequate once the key variables and their critical interactions are known. 1-2 The Spot Welding Variables In manufacturing, processing variables may be conveniently listed in four categories, commonly known as the four M's: Material, Machine, Method, and Maintenance. The four M's shall now be defined in brief detail. 1) Material - Material variables are those variables which are physically carried to the weld process with the material to be welded. 2) Method - Method variables are the weld parameters and the weld requirements. These variables are largely elective and may stem from product design, tool design or the floor setup engineer. 3) Machine - Machine variables are variables due strictly to machine design and build. 4) Maintenance — The maintenance variables are associated with how and when maintenance is performed. The applicable variables from spot welding are listed in Figure 1.1. These spot welding variables were identified through input from many people, from both academic and plant backgrounds. Seeing this long list of variables, one might be tempted to conclude that the process is difficult to control, or even that it is uncontrollable. In this study, as many of these variables as possible were evaluated, in order to rank and isolate the critical variables. 1-3 Project Summary In this work a number of variables were characterized and tested for their relative influence. Testing was done in both laboratory and plant environments. Then the process setups were refined to minimize the impact of day to day variation. Peak efficiency and weld quality occurred consistently when operating as close to the expulsion limit as possible. When production weld equipment was calibrated 1 2 3 4 MATERIAL METHOD MACHINE MAINTENANCE Material Properties Weld Specifications Electrical Electrical Electrical Resistivity Strength Current Balance Actual Loop Resistance Thermal Conductivity Diameter Power Factor Actual Loop Area Heat Capacity Expulsion Designed Loop Area Density Indentation Loop Magnetism Magnetism Location Dsgn. Loop Resistance Flange Width Supply Voltage Turns Ratio Microstructural Factors Weld Parameters Mechanical Mechanical Cracking Current Gun Design Mechanical Tool Wear Grain Growth Force Force Balance Lubrication Hardness Weld Time Cooling System Cooling Water Flow Heat Aiiected Zone Squeeze Time Misalignment Misalignment inclusions Electrode Selection Phase Changes Liquatlon Cracking Shape Factors Die Practice Thickness Metal Fit Thickness Ratio Coating Damage Flatness Flange Condition Metal Fit Surface Factors Roughness Oil Oxide Zinc Paint Cloth. Paper Misc. Contaminants Figure 1.1 Fifty four resistance spot welding process variables, arranged according to the "Four M's" of manufacturing processes. 1 2 3 4 MATERIAL METHOD MACHINE MAINTENANCE Material Properties Weld Specifications Electrical Electrical Electrical Resistivity Strength Current Balance Actual Loop Resistance Thermal Conductivity Diameter Power Factor Actual Loop Area Heat Capacity Expulsion Designed Loop Area Density Indentation Loop Magnetism Magnetism Location Dsgn. Loop Resistance) Flange Width Supply Voltage Turns Ratio Microstructural Factors Weld Parameters Mechanical Mechanical Cracking Current Gun Design Mechanical Tool Wear Grain Growth Force Force Balance Lubrication Hardness Weld Time Cooling System Cooling Water Flow Heat Affected Zone Squeeze Time Misalignment Misalignment inclusions Electrode Selection Phase Changes Liquation Cracking Shape Factors Die Practice Thickness Metal Fit Thickness Ratio Coating Damage Flatness Flange Condition Metal Fit Surface Factors Roughness Oil Oxide Zinc Paint Cloth, Paper Misc. Contaminants Figure 1.1 Fifty four resistance spot welding process variables, arranged according to the "Four M's" of manufacturing processes. to operate this way, the list of important variables was reduced to just four basic maintenance items: 1) Electrical resistance of the total circuit. 2) Cooling water efficiency. 3) Mechanical condition of the weld guns. 4) Operating current. A distant fifth key variable was metal fit, but only when metal fit was extremely poor. The remarkable aspect of these conclusions is that so many of the classic variables shown in Figure 1.1 ended up being minor variables in the shop floor environment. It turned out that proper maintenance is the key process variable on the shop floor. And fortunately, of the four M's, maintenance is the preferred main variable since it is the most controllable. This dissertation summarizes a laboratory and plant study of the variables of resistance spot welding. The recommendations of this work were successfully implemented in a large automotive plant. The tangible result was the achievement of perfect weld quality during several extended production runs of over 8 million welds each. Data and analysis are included to explain the physical and metallurgical process of production resistance spot welding. 2. LITERATURE REVIEW This section is to acknowledge the research of others who may have had some bearing on this study. The reviews are organized topically, beginning logically with the weld microstructure, which is the most complete measure of weld quality and is therefore the best way to define process requirements. Then, in order to understand how spot weld microstructures develop, contact resistance and heat flow are reviewed. Next, weld parameters are discussed in terms of process requirements, using a measurement tool called a weld lobe. A review of mathematical models follows. These models attempt to predict weld formation. Several classical feedback control schemes are then presented. The chapter concludes with an explanation of a significant new in-situ measurement technique for gathering voltage and temperature weld data in a laboratory environment. 2-1 Weld Microstructure The microstructure of a spot weld resembles a rapidly cooled metal casting. 0n the one hand, it seems strange that spot welding is not officially classified as a fusion welding process by the Resistance Welder Manufacturer's Association (1). This is probably due to the fact that surface melting ideally does not occur in resistance welding. On the other hand, metallographic analysis reveals that melting does indeed take place in the interior "nugget" of the spot weld (Figure 2.1). Rapid cooling of the sheets by the copper electrodes causes martensitic transformation of the nugget. Surrounding the martensitic nugget are successive bands of thermally altered microstructure. The entire altered region surrounding the weld nugget is called the heat affected zone (HA2). The microstructure of a spot weld in mild steel has been classified by Kim (2). Solidification within the nugget follows the mechanism of cellular growth. This mechanism applies to all rap‘ily solidifying alloys (3-5). In short, solidification occurs by epitaxial growth of favorably oriented grains at the solid-liquid boundary. The fastest growing grains are those oriented with growth direction parallel to the maximum thermal gradient. These grains quickly dominate and cut off all slower, misoriented grains. Even though solidification is rapid, there is still opportunity for the segregation of low melting eutectics such as FeS to the center of the weld, causing "hot cracking" or porosity at the centerline. Segregation is greatest when the freezing direction remains in a straight line from the nugget wall to the nugget center, since unidirectional freezing gives the largest prior austenite grain size. If the freezing direction is gradually rotated, such as occurs near the edges of a spot weld (Figure 2.2), the favored grain orientation will also .E o c O. o Figure 2.1 A normal spot weld in SAE 1005 DQAK cold rolled steel, 0.030 inch (0.75 mm). Microstructure shows a cast nugget formation, (Sample no 323) . .. gr"? 0.610} " .025 mm Figure 2.2 Photomicrograph of nugget edge, showing a rotating freezing direction. The upper sheet is cold rolled SAE 1005, the lower sheet is normalized SAE 1005. 10 rotate, trapping in segregates as new grains cut off old grains. Excessive segregation in high strength low alloy (HSLA) steels has been shown to cause a kind of hot cracking known as hold-time sensitivity (6,7). However, core defects in spot welds generally do not pose a weld strength problem for at least two reasons. First, since structural loads are transmitted through the sheet, they are focused around the soft perimeter (HAZ) of the weld. Second, the freezing direction places the outer nugget in residual compression, which tends to inhibit fatigue growth of core cracks, since cracks do not penetrate the compressive ring. Hot welds, with loss of metal and internal voids from expulsion, have been tested and found to be structurally sound for the most part (8,9). 2-2 Contact Resistance For the purpose of this study, contact resistance is defined as the total resistance of the weld before current is passed. In Figure 2.3, it is the sum of the resistance before welding has taken place. This resistance seems to be 5 to 30 times greater than resistance after one or two weld cycles. Contact resistance can be significant because it provides extra joulean heating early in the weld sequence. The largest part of contact resistance seems to be due to surface film and asperities touching at the interfaces R2, 11 BULK COPPER R1 ALLOY 43.2 R3 BULK STEEL R4 R5 BULK STEEL 436 R7 BULK COPPER ALLOY Figure 2.3 Schematic of resistances in a spot weld. 12 R4, and R6. This is in contrast with the bulk resistances R1, R3, R5, and R7 which are quite low in the cold preweld condition. R3 and R5 are estimated to be 6 micro-ohms per sheet for 304 stainless steel, based upon a cylindrical volume 0.030 inches (0.75 mm) thick and 0.25 inches (6.35 mm) in diameter, using a material resistivity of 25 micro ohm-cm: 1 R= P" [2.1] a 0.075 R=25 2 [2-2 i 0.63 f. 4 [2.3] R = 6 micro ohm Therefore, bulk resistance is seen to be one to two orders of magnitude smaller than the total contact resistance. Interfacial resistance is made up of many parallel microasperity paths (lo-12). These cause the data to show significant variation since surface roughness and cleanliness are generally not repeatable, especially in an industrial environment. There is however an inherent relationship of contact resistance with pressure since the current can only pass through areas of actual asperity contact. Pressure influences the amount of micro contact area by crushing the surfaces together. 13 Contact resistance is most influential when weld time is short, according to a study of half cycle welds made by condenser discharge. Nakata, et al (13) showed in 1975 that contact resistance has two components: surface film resistance and asperity resistance (known as spreading resistance in physics). Although surface film resistance is one or two orders of magnitude larger than asperity resistance, its heat contribution is small because the film breaks down in only 1/16 cycle (~1 ms). Therefore the main source of heat in contact resistance is due to asperities. Nakane and Torii (14) also concluded in 1973 that contact resistance hardly affects spot welding. They dispelled the earlier "linear” theory that contact resistance generates local hot spots which in turn generate more heat and grow rapidly larger. They found that contact resistance disappeared within 1/4 cycle after heating begins, and that heat flow quickly erased local hot spots. Then they modeled the current flow path, factoring in contact area at the weld interface and the sheet separation around the contact area. Their conclusions were: the contact area limits the diameter of the current path and this diameter does not change appreciably during the course of welding before expulsion, but grows abruptly when expulsion occurs. The increase in weld area at expulsion lowers current density, which discourages further nugget growth (Figure 2.4). 14 MN 2 ' i 2 . - 2 2 2 ‘ ;/ ¢ \\ () Figure 2.4 M \l/ 2 KO 2 W Experimental model showing the onset of sheet separation with time. Figures c) and d) represent greater weld time, (Ref 20). 15 An interesting experimental study of the flow of heat across metallic interfaces under pressure was done by Weills and Ryder in 1949 (15). They concluded that thermal resistance at the interface is analogous with electrical spreading resistance. As such, it decreases with increasing temperature and pressure, or by the inclusion of oil or a soft metal plating. Using high speed cinematography of half welds, Satoh, et al concluded in 1970 (16,17): "the contribution of the contact surface is mainly to provide the place for sheet separation to occur, rather than the place where contact resistance prevails (since contact resistance exists for such a short time)." In 1990, Kim and Eagar (18) concluded that, "the contact area at the faying interface has a greater effect on nugget growth than does the contact resistance. It is found that the ease of spot welding of bare steel is due to the small contact size, not due to high resistance. The heat generation rate at the electrode interface is about double that of the faying interface when welding galvanized steel due to the large contact area at the faying surface created by the molten zinc." 2-3 Heat Plow Fundamentals The heat flow out of spot welds through the electrodes is a key parameter. Heat flow is expectedly high during the later stages of weld development. Some contributing factors are: nugget temperature, sheet thickness, copper electrodes, 16 and water cooling, to name a few. For welding below expulsion, the peak amount of heat flow may be approximated by the electrical energy input. This is especially true late in the weld schedule when the weld approaches thermal balance (heat in = heat out), and nugget growth stops. Thus if the input energy density were constant for a series of welds, the final penetration of those welds should be defined and stabilized by the ability of the electrodes to draw heat away from the weld. Neglecting heat flow in the plane of the sheets, all heat leaving the weld must cross the contact boundary between the electrode and the sheet. Thus, a stable, well controlled process would have low thermal (and electrical) resistance at the electrode-sheet interface. In 1990, Calva and Eagar published an experimental and numerical analysis of thermal contact conductance, specifically as it relates to spot welding thin, galvanized sheet (19). They cited the difficulty of producing a repeatable thermal balance due to the combined effects of coating variation and material thickness. This thermal instability was causing unstable penetration. They concluded that improved weldability would result by increasing the contact resistance at the faying interface and decreasing it at the exterior interface. The results agree with an earlier work by Eagar, et. al. regarding galvanized high strength low alloy steels (20). 17 An insightful comparison of short and long weld times was done by Yamamoto and Okuda in 1978 (21). They studied resistance seam welding of bare steel, but their results provide an interesting view of spot welding as well. To gain understanding from their arguments, it is helpful to consider for the moment that spot welding could be a special case of seam welding, where travel speed equals zero. They showed that similar to spot welding, low speed seam welding is controlled by heat conduction. However, high speed seam welding is controlled by contact resistance, since the weld time is too short for heat conduction to dominate. The influence of sheet thickness on transition speed between modes was both calculated and measured, (see Figure 2.5). From a control standpoint, a process governed by heat conduction is more stable than a process governed by contact resistance. This is because the thermal properties of commercial grade steel are more repeatable than its surface properties. 2-4 Current and Time - the Weld Lobe The weld lobe of Figure 2.6 is a research tool commonly used to study the weld variables current and time. A weld lobe is a plot of iso-diameter lines on rectangular coordinates of weld time (ordinate) and weld current (abscissa). To understand how current and time operate, we 18 10 - . E I (u) CONTACT . E . RESISTANCE iii TYPE 0. i- .. U) ‘2’ 5 ~ . 9 g L (I) HEAT . . cououcnon . _ TYPE q a) l l l l l l l l l l 0 1.0 2.0 THICKNESS (mm) SURFACE I/ l WELD INTER- - ——————————————————— FACE b) SURFACE ———> TEMPERATURE Figure 2.5 a) Seam welding transition speed from Yamamoto and Okuda's experimental model, using bare steel. b) Measured internal peak temperature profiles for the two modes, (Ref 21). 19 r—NOMINAL NUGGET CURVE -—4 MAX. CURRENT LINE PRODUCES EXPULSICN _1 ON 30TH FIRST AND SECOND COUPON WELDS MIN.CURRENT LINE (PRODUCES MIN. NUGGETS) wntitIME (CYCLES) I T I i I CURRENT (KILO AMPS) Figure 2.6 Typical weld lobe (Appendix A). 20 must examine their dependent variables, heat generation and heat flow. Heat generation and heat flow result from the passage of current. They produce a final temperature distribution which is of key importance in producing quality and stability in the process. The weld lobe simply records the distribution of final weld diameters. As such, the lobe indirectly tracks where the favorable temperature distributions lie, within the realm of current and time. 2-4-1 Dynamics of Heat Generation Because of Joule's Law, current should be a parabolic contributor to the power needed to weld. Specifically: H=I-R [2.4]. Weld time, on the other hand, linearly affects energy input as simply: Energy = Power x Time [ 2.5 ]. It is worth considering how the energy supplied to make a weld is used. First, let us remember that (approximately) 100% of the electrical energy used is released as heat. Local resistance and current density determine where the heat is released. Heat is released at all points having 21 finite resistance and current density, according to Equation 2.4. Initially the hot spots are at the interfaces R2, R4, and R6 (Figure 2.3), since they have the highest resistance. Over time the contact resistance is quickly broken down, and the resistance "flame" for the completion of the weld shifts toward the bulk, which by this time has heated somewhat and become more resistive. As the bulk heats it becomes even more resistive and heats at an ever faster rate in a sort of "thermal runaway". But to prevent an explosive melt down and loss of control, heat flow comes into play. 2-4-2 Dynamics of Heat Flow Provided the energy input rate is less than the maximum ability of the electrodes to dissipate this energy, the weld will continue heating until a quasi-steady state is reached. This condition is described as: [2.6 I, (Q = (Q rms)in rms out where Qrms is the root mean square of energy density with respect to time. If energy input is above this critical output rate, "thermal runaway" will occur, resulting in explosive welding and poor quality (22). In contrast with thermal runaway, quasi-steady state is desired from a process control point of view. Here is the 22 scenario leading up to the quasi-steady state: 1) As heat is liberated within the weld, it begins to flow conductively toward the water-cooled electrodes. Heat flow causes the surface of the weld touching the electrodes to be the coolest, most temperature-controlled part of the weld. Heat flow in the plane of the sheet is largely negligible, due to the lack of a nearby heat sink in the sheet direction. Lateral heat flow is especially small with thin sheets. 2) The region farthest from the electrodes (R4 in Figure 2.3) attains the highest temperature, since it is cooled the least. In addition, peak temperatures occur well after the interface has melted and R4 equals zero. Thus when heat flow is given time to stabilize, it is heat flow rather than resistance which determines the hottest location in the weld. This peak temperature zone becomes the center of the nugget casting. The nugget center is located halfway between the electrodes when the electrode geometry, material, and cooling are the same on both sides of the weld (Figure 2.7). It is observed that even when welding sheets of dissimilar thickness, the nugget center favors the midpoint between the electrodes, rather than the weld interface (Figure 2.8). (This midpoint principle may not be observed when dissimilar metals are welded). 3) The size of the nugget melt is restricted by the balance of heat generation and heat flow, provided there is enough weld time to achieve the quasi-steady state condition of Equation 2.6. Any additional time will not appreciably increase the weld pool size. Additional current however, will increase the pool size slightly, since the quasi-equilibrium will require this extra heat to be dissipated. The extra dissipation can only occur if the thermal gradient becomes steeper by the presence of a thicker weld pool and a thinner weld skin. The thermal gradient shown in Figure 2.7 may be approximated from the weld microstructure. For example, if the weld penetration in two sheets of 0.030 inch (0.75 mm) mild steel were 60% as in Figure 2.7, the thermal gradient could be estimated, assuming a melting 23 —> TEMPERATURE 'Nmmo T(meit) Figure 2.7 A one-dimensional temperature distribution model for spot welding, a) thin sheet model, b) thick sheet model, (Ref 24). Figure 2.8 .E o v Q 0 Typical nugget formation resulting from a 3:1 thickness ratio. This is a production weld of 0.040 inch (1 mm) to 0.120 inch (3 mm) drawing quality steels. Both sheets were hot-dipped galvanized. Nominal production parameters were 10 ka for 20 cycles at 600 pounds (261 kg). Notice how the nugget center lies at the mid-point of the stackup, rather than at the weld interface. poin‘ tran, woul grad weld dis: Prod Surf PTO: equi 25 point of 1525 C, and the outer weld surface at the transformation temperature (727 C), the thermal gradient would be: dT T2 - T1 - = [ 2.7 ] dX t?(1.' P) T1 = low temperature boundary T2 = high temperature boundary t = sheet thickness p = weld penetration 1525 - 727 C C -— = 66,500 -— [ 2.8 1 0.03 (1 - 0.6) in in The maintenance of an aggressive yet stable thermal gradient during the final (quasi-steady state) cycles of the weld has great influence on the final temperature distribution and penetration of the weld. The goal is to produce melting at the interface, while keeping the outer surface as cool as possible. This goal dictates certain process requirements commonly seen in production welding equipment. These are at least: 1) Copper electrodes of high thermal conductivity. 2) Sufficient electrode force for good thermal coupling with the workpieces. 3) An electrode cooling system for repetitive welding applications. 26 2-4-3 Material Property Limitations As seen in the previous example, the stock thickness has bearing upon the thermal gradient needed to preserve the desired temperatures at the two critical boundaries, (the outer surface and the near edge of the weld pool). This is unfortunate when welding thinner stock, because the thermal gradient eventually becomes limited by workpiece conductivity, rather than heat sink efficiency. When this happens, the heat cannot escape as fast as it is generated, and therefore the quasi-steady state cannot be attained, except when current density is reduced. Therefore quasi-steady state can only occur with current density below that required for welding. The problem with spot welding without enough current density is that sufficient melting does not occur, and diffusion bonding becomes the welding :mechanism. Diffusion bonding is rather sensitive to ‘variation in surface cleanliness, since contamination is a diffusion barrier. In this scenario, weld strength could be unpredictable . To reughly estimate the quasi-steady state thermal gradient limitation, we shall use the Fourier Law for one dimensional steady-state conductive heat transfer. Q=-k-— [2.9] 27 Where, Q = heat flow towards the electrodes [power/area], k = thermal conductivity of the sheet material [power/length/C], dT/dx = thermal gradient [C/length] Rearranging for thermal gradient, dT Q ‘— = T ‘— [ 2.10 ] dx k For mild steel, we will use a thermal conductivity of k = 0.7544 W/[in-C], (0.0297 W / [mm-C1), and assume that it is constant in the weld skin temperature range: 727 - 1525 C. As an example, the power density of welding might be roughly: v Q = I'- [ 2.11 1 A 1 amp volt Q = 10,000- -———————— 0.098 in? E 2'12 1 watts Q = 102,000 2 [ 2'13 1 in Solving Equation 2.10, dT 102,000 -=-————- [ 2141 dx 0.7544 dT C - =“135,000 - dx in Ati theI 35011; 165% mi ThUS Dre: theI or t intE Th1E HowE 28 which is the peak thermal gradient to be found in this material at this power level. The minimum "skin" of solid material between the nugget and electrode may be readily predicted by assuming a linear thermal gradient: Q =__ [2.16 k I At our power level and temperature range (727 - 1525 C), the theoretical minimum skin thickness to be found in the sample would be: 1525 - 727 in 135000 c [ 2.17 1 x2 - x1 = 0.0059 in [ 2.18 1 In practice, it is seen that when sheet thickness is less than about 5 times the theoretical minimum skin 'thickness, some tendency toward diffusion bonding begins. 'Thus, in terms of process control and weld quality, it is preferable to weld thicker sheet. It is proposed that ‘thermal instability occurs when welding thin sheets because (of the presence of a thermal barrier at the electrode-sheet .interface, which restricts heat flow below the input rate. {This means Qin > Qout and thermal runaway is predicted. However in practice, a setup operator will compensate and 29 reduce current to a visually acceptable level. It is this lower current which shows the tendency toward diffusion bonding. To conclude this discussion of the thermal aspect of spot welding, the goal of the process is to produce a particular thermal profile within the sheets which will permit melting and bonding at the sheet interface, while keeping the external surfaces as cool as possible. The actual thermal profile depends upon two dynamic phenomena: 1) Heat generation, produced by electrical resistance to current flow. 2) Heat flow, dictated by the characteristics of the material (provided the time of welding is sufficiently long). Thus, current and time are expected to be main variables affecting the thermal profile of the weld. 2-5 Weld Force and Weld Pressure Fundamentals Consider the scenario of the shop floor tradesman who reduces force in order to increase weld diameter and 3penetration. Although not a sanctioned method of process icontrol, reducing weld force does often increase the apparent heat of resistance welds. The warming effect of lower force has been explained as follows: 1) Contact resistance increases due to less asperity-crushing, which in turn generates more heat. L_um for: the As c' Pres app] toge fleas aver “Eld C01-r 30 2) Weld area decreases at all interfaces due to less elastic/plastic yielding of the bulk, especially when using ball-faced electrodes. The reduced area increases heat due to higher current density. Remember that local heat is proportional to local current squared (Equation 2.4). 3) As contact electrical resistance increases per item 1 above, contact heat flow resistance will also increase (15). Resistance to heat flow at the interfaces R2 and R6 (Figure 2.3) will cause more heat to be retained inside the weld. Weld pressure in this study is defined as the weld force (delivered by the two opposing electrodes) divided by the electrode contact area. P=- [2.19] weld pressure electrode force electrode contact area 3"‘1'0 II II II .As defined, P is really an average pressure, since the true pressure varies across the face of the electrode due to the applied mechanics considerations of clamping two sheets 'together between flat or ball-faced tips (14). Nonetheless, P is of value in process study because of ease of measurement, and because of the implied correlation between average and true pressure. Experimentally, weld pressure was varied rather than weld force, by changing the contact area. The above correlation with contact resistance is documented in Section 4-4. mat) The} the I (24} E! m m 1 . . , m Gre. EXP ion the nth 31 2-6 Mathematical Modeling Ando and Nakamura contributed to the one dimensional mathematical understanding of spot welding in 1957 (23). They developed the heat time constant approach, which led to the law of thermal similarity introduced by Okuda in 1973 (24): "For the case when the plate thickness and the diameter of the electrodes are magnified by n times, if we also change the current density by l/n times (which2 is current by n times), and heating time by n , the new temperature distribution becomes similar to the original one." Okuda's law of thermal similarity is based purely upon a one dimensional consideration of bulk properties and does not even consider the influence of the interfaces. Consequently, actual welding begins to deviate from this rule when welding thin sheets, because there is less bulk to mask the interfacial effects, and because steeper thermal gradients are required for thin sheets (23,25). Practical welding of thick plates also deviates from this law since there is fanning of the current path and heat loss into the plane of the plate (26). An important early numerical model was done by Greenwood in 1961 (26) . Using a heat conduction model, he explained the classical understanding of heat flow and jcn11ean heating within the weld. He showed that local thermal spikes are soon leveled out by conduction and that "the temperature (distribution) tends to an equilibrium deS deti the 32 form" which "bears no resemblance to the heat production pattern" (Figure 2.9). Although he neglected surface resistance, the latent heat of fusion, and the variation of material properties with temperature, his results still correlate with the actual temperature distribution in completed spot welds. The fact that such a simple model still does a good job predicting final temperatures, even with its sweeping assumptions, indicates that perhaps contact resistance is not unduly influential in the later stages of welding, once a quasi steady-state of heat generation and heat flow is established. Newer numerical models have included more exact data on material properties and interface resistance, but Greenwood's predictions about the thermal distribution for completed welds still stand (27-32). 2-7 Process Monitoring and Feedback Techniques There have been longstanding efforts to develop adaptive welding systems by various means. An ideal system would detect process variation and automatically compensate as shown schematically in Figure 2.10. In practice, these systems are usually designed to terminate the weld when the desired nugget size has been reached or expulsion has been detected. Ideally, one would h0pe to "guarantee" quality as the welds are made. Unfortunately, many of these techniques r. l ._ A LIL VAUWZVE LaueEEAe .0 e.n( ,. /..,, , , /, . [Ii 2, .. ”WI .. ,,/ / ./ Fig} I - 'c m i \ m . . / Electrode edge F .I' l ' i000'c uoa'c l200'c I300'c’ - ' OTC g //// ififlt ' '3 i /// II- 333;: g / 6001: I m _ - 2°»: E ‘6 Figure 2.9 Greenwood's calculated temperature distributions after various weld times. a) 0.025 sec, b) 0.1 sec, c) 0.6 sec, (Ref 26). 34 INPUT OUTPUT + . PROCESS FEEDBACK «— Figure 2.10 Schematic representation of dynamic feedback monitoring. we ac re Tc di 1: de 2 :1 35 work under laboratory conditions, but are not broadly accepted by manufacturing plants. One reason they are rejected is that they often require sensors on the weld gun. To a plant person, sensors are seen as unreliable in the dirty and often destructive environment of the secondary loop. Of the feedback controls which have been developed to date, Stiebel, et al (33) have used a mechanical expansion sensor on the gun to detect nugget expansion. The sensor and the signal obtained are shown in Figures 2.11 and 2.12. Havens (34) tried acoustic emission detection using piezo-electric sensors on the electrode holder. Acoustic sensors and resulting signals are shown in Figures 2.13 and 2.14. Dickinson, et al. (35) tried dynamic resistance monitoring using voltage and current leads on the electrode holders (Figure 2.15). Nagel and Lee (36) terminated the weld at the resistance change which occurs at expulsion, as per Figure 2.15. Burbank and Taylor (37) developed ultrasonic through-transmission feedback using piezo—electric crystals inside the electrode cooling cavity, according to Figure 2.16. Although each method is ingenious and technically very different from the others, they all share the handicap of requiring sensors on the gun. All have shown some success in laboratory testing, yet none have achieved any significant market penetration, due to cost, practicality, or robustness in the plant environment. 36 F J //W Figure 2.11 Load cell displacement sensor to detect nugget expansion during the weld, (Ref 33). 37 Figure 2.12 Nugget expansion profile as the weld develops, as a function of weld time, in cycles. The etched cross-sections (left side) show nugget development. The oscillograph data (right side) show vertical electrode movement during the weld. Notice the abrupt collapse at expulsion, letter i, (Ref 33). 38 ACOUSTIC EMISSION TRANSDUCER Am) 3 T0 CONTROLLER a «; WELDING ’7 ELECTRODES fiéééééééééa LOGIC , , CIRCUIT ACOUSTIC EMISSION TRANSDUCER Schematic illustration of acoustic emission Figure 2 . 13 sensors on weld gun, (Ref 34). 39 EXPULSION Iu o :1 t .I a. 2 < .J < z 2 m H“ I] “I “ll llllllllllltlimllt in Illll mm ,TIME ELECTRODE I POST WELD ELECTRODE CLAMPDOWN START NUGGET DELAY :giszEGl-D UNCLAMP FORMATION WELD TIME Figure 2.14 Schematic signature of acoustic emission during spot welding, (Ref 34). 40 name menus: mam ween eaoeTII meme ssoumcu com rIIIsr ‘ _, [ suture SURFACE ”WWI » ' .‘IIII/V’W’V", ’Illiflp III”, " ’///////////I////, "///,////,' Cs RESISTANCE -. Till—O Figure 2.15 Classical dynamic resistance curve, (Ref 35) . ‘ m I I I l SPEI BEA 1: ‘A‘ ' COI L 41 EL ECT RODE 4 I“ TRANSMITTING CRYSTAL I I l I v i 'I t I l i l 1 I ' 3 sea» PATH sf . . RECEIVING CRYSTAL WATER-PROOF CONNECTOR (- COOLANT CAVITY L— Figure 2.16 Welding electrodes containing ultrasonic transducers in a pitch-catch configuration, (Ref 37). 42 2-7-1 The Classical Dynamic Resistance Curve Perhaps the most common feedback control method found in the literature is dynamic resistance (10,35,36,38-45). These systems usually employ voltage and current sensors in the secondary loop. V/I represents the measured resistance, taken at a point each half-cycle when dI/dt = 0, in order to eliminate inductive effects. The classic dynamic resistance curve is shown in Figure 2.15. The curve has been characterized in terms of six key stages (35,36). Stages I and II include clamping of the workpieces by the electrodes and the passage of the first 1-2 cycles of current. During this time resistance is unstable and decreases rapidly to a minimum value which signifies the end of stage II. During stage III, resistance begins to rise due to bulk heating. Stage IV and V represent nugget growth, where resistance begins to taper off due to the increasing area at the weld interface. Stage ‘V is also accompanied by some electrode indentation. Stage ‘VI begins with expulsion of molten metal from the weld interface. During this time there is a rapid decrease in :resistance due to an abrupt increase in weld area as the enclosed nugget ruptures. Resistance continues to fall aafter expulsion, due to additional electrode indentation. This curve has been used in a variety of ways to provide meaningful feedback and control. Some adjust the cur rat cha exp Con Cur So the Si 2 Var 00c the no:- 43 current each cycle during stage III in order to control the rate of resistance change (41). Others Observe the rate change during stage V (43). Still others terminate upon expulsion detection at the abrupt resistance drop (36). 2-7-2 A Theory for the Fundamental Changes at Expulsion The existence of the resistance drop at expulsion (shown in Figure 2.15), is well documented (35,36,41-46). The reason for the R-drop at expulsion has been given by Lee and Nagel: "Because of the sudden spread of the molten metal, the current is no longer restricted to the hot region. This sudden change in the conduction mode is typified by a sudden drop in the resistance. Expulsion also signifies the maximum nugget size for that electrode size, geometry, and force. For all practical purposes the nugget growth process terminates at expulsion" (46). Therefore, an abrupt increase in the faying interface contact area occurs at expulsion. When this happens, current density falls, which is why the nugget growth halts. So we see that expulsion can have a stabilizing effect upon the process since nugget size variation is less. The nugget size will be roughly the same even when input parameters vary, provided there is still enough energy for expulsion to occur. Of course, over welding severely could permit thermal runaway even after expulsion, but there should normally be a considerable current range in between tin ele eXp res tru be tee 6x1 44 expulsion and unacceptably hot welds. This range will be discussed further in Section 4-4-4. Consistent with the fundamental resistance change at expulsion, the electrodes experience abrupt motion at this time (33,47-49). Figure 2.12 shows the classic pattern for electrode movement. The key here is the abrupt drop at expulsion. In addition, the acoustic emission signal at expulsion has also been used to terminate the weld as seen from Figure 2.14 (8,34,50,51). Unfortunately, this method did not prove very reliable in the present study, as discussed in Section 4-5-2. 2-8 A Novel Technique for In-Situ Measurement of Temperature and Voltage Measurement of data is an abiding problem in scientific research. Once in a while, a researcher will develop a truly clever and insightful way to collect data. One has to be impressed with the great effort some have given toward collection of hard-to-get data. This section mentions an interesting data collection scheme that has been used in the study of resistance welding. Alcini (52,53) extensively developed the half weld technique, whereby electrodes were cut in half along their axis and arranged to make a weld on the sheet edge (Figure 45 2.17). This setup used a quartz "window" to contain the melt and allowed for wirebonding of micro thermocouples on an axial plane of the weld (Figure 2.18). Using an elaborate procedure, he was able to measure voltage and temperature from within the weld with significant accuracy. Some of his results are shown in Figures 2.19 through 2.21. 46 Optical , I ’ ’ Measuring —> ’ Equipment Half Weld with Weld Zone Containment Data Acquisition and I I I | I Digital Conversion [ c:::: _ = D Computer for Data Processing Figure 2.17 Schematic of the setup used for in-situ measurement of temperature and voltage, using half weld instrumentation, (Ref 52). Pm All? Aka 47 Schemes Circmt ofthc Readies Micro-Thermocouple I i I i Pure P: 1 j P: .1395 Rh l Spar Weld Elecvode ‘- Sheet: Being Welded . , Weld Nugget Spar Weld Electrode Axis of Symmetry A Beadicss Micro-We Attached to an Area (Approximting a Point) on the Half Weld Surface Figure 2.18 Schematic of headless microthermocouple attachment used for in-situ measurement, (Ref 53). 48 / r¢—-—L0mm->l Cycle 20 Electrode 700 °C + —— T ——f 1500 °C Sheet 1525+. ‘EW\ Fayt'ng Interface—I» 1472 2 153 + ”EBJ; Figure 2.19 In-situ temperature fields in full-size weld nuggets. The temperature field was measured at the peak of nugget size by the half-weld technique, (Ref 53). (I) EIecrrode Sheet f Liquid Nugget + A. Liquid Nugget To Solid Sheet I Boundary (2) ‘ \\ :flfi \ ? (1540°C) ? —'T—— 6 \\ 2 \ 2 / é Tw(1505°C) ”—8.— ----8 Figure 2.20 Liquid Nugget \ Solid Sheet 7WIJI \\\\\\\\\\\\\\\\ \\\\\\\\ Secfion AAA Convection in liquid nugget. Steady state convective heat transfer and heat conduction at a boundary surface, measured by the half-weld technique, (Ref 53). 50 \ Electrode / Sheet f \ Figure 2.21 Convection swirling pattern proposed by Alcini, (Ref 53). 3. EXPERIMENTAL STRATEGY AND SETUP 3-1 Goals and Strategy The goal of this study was to understand the spot weld process from the aspect of key variables, and then learn to control the process in spite of the many variables existing in the manufacturing environment. To achieve this goal it was necessary to conduct experiments to filter out key variables from less significant variables. Then, with the list of key variables and optimum setup conditions in hand, the next phase would be to implement a process control plan with measured performance feedback. The strategy used to achieve the goals may be summarized as follows: 1) Obtain management support to sustain project resources. 2) Develop laboratory facilities. 3) Conduct scientific studies, both in lab and plant. 4) Develop training courses. 5) Assemble trained preventive maintenance teams. 6) Bring every weld process under control (1600 guns). 7) Enhance program based on learned improvements. 8) Implement monitoring and preventive maintenance plans. 51 WE fe 52 3-2 Laboratory Apparatus The laboratory welder consisted of a single gun spot welder as shown in Figures 3.1 and 3.2. The welder was custom manufactured as a laboratory model by the Resistance Welder Corporation of Bay City, Michigan. The important features were: 1) An air cylinder to extend and pressurize the electrodes. 2) A gun translation stage and sample holders (used for the electrode misalignment studies). 3) A fixture transformer driven by a 480 volt 60 Hz power line. The transformer was rated: 50 kVA at 440 volts, secondary voltage 3.2 - 5.0 volts, Girton Industries Model A 454. 4) A Legend weld control by Medar Corporation of Farmington Hills, Michigan (software program number 7181, revision 7). 5) An IBM personal computer used for data acquisition, as illustrated in Figure 3.3. 6) A Robotron toroidal current meter, model WS-20 from Robotron Corporation of Southfield, Michigan. 7) A digital low resistance ohmmeter, model 247001-11 by Biddle Instruments was used for static resistance readings of clamped sheets and machine circuits. A 10 amp current is supplied by the meter to take readings directly in micro-ohms. 8) A hydraulic force gage, range 0-1000 pounds (0-434 kg). Figure 3.1 53 Photographs of laboratory spot welder. a) overall setup, b) closeup view of welder. 54 Figure 3.2 Photograph of laboratory welder, showing weld tip arrangement. The lower electrode is stationary. Notice the acoustic emission sensor on the upper electrode shank. 55 0 0 ./ j I 7 D —I Liam I an . I Z . IBM 035%,?“ PERSONAL (digital) DATA COMPUTER -—-> FIRING CURRENT SIGNAL WORKPIECE SENSING LEADS TOROID SCR pa), \ 2 I. V ’/ SECONDARY 3' ' [xi-J LOOP PRIMARY LOOP 3 [ WELD TRANSFORMER INPUT POWER INDYmm 60 hz AC 200 amp breaker OUTPUT POWER avmm 60 hz AC 15 RA Figure 3.3 Electrical schematic of laboratory welder, with data acquisition computer. g2 9E 56 3—3 Production Apparatus The primary apparatus used for production data was a hard automation fixture-type welder with a daily production rate of approximately 1000 automobile hoods per day. The weld control and data acquisition system were identical to the laboratory system shown in Figure 3.3. The material welded was 0.024 inch (0.6 mm) hot dipped galvanized steel welded to 0.080 inch (2.0 mm) hot dipped galvanized steel. 3-4 Weld Microstructure Study Procedure In order to study weld microstructure, samples were welded of the following steels: 1) Cold Rolled SAE 1005 welded to Quenched SAE 1080 Spring Steel. 2) Cold Rolled SAE 1005 welded to Quenched SAE 1005. These welds were sectioned in the transverse direction along the weld centerline. They were metallurgically polished and etched. Photomicrographs and microhardness (Tukon) profiles were made. 57 3-5 Contact Resistance Measurement Procedure To measure contact resistance before welding, the following procedure was used in order to create nearly ideal conditions: 1) Acetone-washed stainless steel coupons 0.030 x 1.5 x 5 inches (0.75 x 38 x 127 mm). 2) A-nose class II copper electrodes, 0.250 inch (6.35 mm) diameter face. 3) Misalignment eliminated by visual alignment of electrodes. 4) Tips made parallel by clamping lightly on a rotating file. 5) Tip to tip resistance (with no coupon) was 20 - 53 micro ohms. 6) Tips were allowed to impact the coupons as in actual welding. 7) Measurements taken 1 minute after clamping, to stabilize. 8) Five data points were collected at each pressure level. 9) The Biddle micro ohmmeter was used measure resistance. 3-6 Weld Lobe Procedure The importance of current and time was confirmed using the laboratory welder. Data were gathered to plot a weld lobe of 0.030 inch (0.75 mm) bare steel. A weld lobe is a plot of iso-diameter lines on rectangular coordinates of weld time (ordinate) and weld current (abscissa). The 58 procedure used was MUS-247 (General Motors), Appendix A, with the following modification. In this study, the lobe was plotted to the outer bounds of acceptable weld quality. This means minimum nugget diameter on the low current end, and electrode sticking or over-indentation on the high current end. These two extremes span the domain of acceptable manufacturing. Experimental parameters were: Force = 500 pounds (217 kg) Electrode Diameter = 0.25 inches (6.35 mm) Minimum Weld Diameter = 0.158 inches (4.0 mm) Nominal Weld Diameter = 0.197 inches (5.0 mm) # Preconditioning welds = 100 Maximum Weld Indentation = 50% Coupon Size = 1.5 x 5 x 0.030 inches (38 x 127 x 0.75 mm) Material = Low carbon uncoated steel, SAE 1005 DQAK Electrode Material RWMA Class II copper, (1/2% Cr) Electrode Geometry Pointed "A" nose. 3-7 Electrode Misalignment Study A simple gun translation device was added to the weld cylinder at its base in order to measure the effect of electrode misalignment. This device is shown in Figures 3.4 and 3.5. As the screw is tightened, the entire cylinder and the upper electrode are shifted 0.050 inches (1.27 mm) per revolution. During the contact resistance study involving electrode misalignment, a sample holder was used in order to prevent sheet rotation for highly misaligned conditions. This sample holder, shown in Figure 3.6, was made of 59 CYUNDER MOUNTING FOOT O / Q 195 0M.0000 MIRI¢Y£W.A&MW§ /00«0n0u..w0 0 .0»..0.o.w.00000 4- éfifihfifibdi RPDVPSVIPPP. 0. .00000 .00 $000.. 0.0 000000. 000 /.0000 00¢..«000w00 000000. 6&YHIotevedaéfiw 0 0 ...0 000000 \\\\\ l\\V\\ AWR§%.¢ 20o \\— 900” ”130000 30 300 00 .000 Izw0fiflziw.éxv 0!“:00033v‘0 /r 0fbfi0s0 J. J \ PARALLEL GUIDES O DOWN HOLD- PLATE x0000006/ STATIONARY ANCHOR PLATE TRAVEL SCREW FOR PRECISE MISALIGNMENT CONTROL (0.050“ TRAVEL PER TURN - 1.25mm/TURN). Sketch of gun translation device used for misalignment study. Figure 3.4 6O Figure 3.5 Photograph showing gun translation device installed on upper platon. 61 Figure 3.6 Holders used to prevent sample rotation during misalignment study. 62 non-magnetic stainless steel bar stock 0.25 inches thick by 2 inches wide (6.4 mm x 51 mm), and strips of fabric-reinforced rubber (conveyor belt material). 3-8 Materials and Electrodes Used in Plant study A practical setup guide was developed from the plant study. This guide was intended to identify robust weld parameters for a full range of common automotive applications. The nominal plant material was drawing quality aluminum killed (DQAK) steel in the thickness range 0.020 - 0.120 inches (0.5 - 3.0 mm), in any welded combination of bare or galvanized. In addition, four common electrode configurations were represented as follows: Case A: Two opposing ball-nose caps of 5/16 inch (7.94 mm) radius, with a small flat at the very tip. Case B: Two opposing A-nose caps, 0.250 inch (6.35 mm) face diameter. Case C: One ball nose cap and one flat cap. Case D: One A-nose cap and one flat cap. 4. LABORATORY RESULTS AND DISCUSSION Laboratory data were gathered concerning the following topics: 1) Ield microstructure and hardness. 2) Contact resistance. 3) weld lobes. 4) Effect of mechanical considerations on the weld lobe. 5) Process nonitoring and feedback control. These topics were also discussed in chapter 2, where the work of other researchers was acknowledged. 4-1 nicrostructure and Hardness Samples were welded of the following steel materials: 1) Cold Rolled SAE 1005 welded to Quenched SAE 1080 Spring Steel 2) Cold Rolled SAE 1005 welded to Quenched SAE 1005 Micrographs and Tukon micro hardness profiles were made (Figures 4.1 - 4.4). Hardness data are displayed in Rockwell C and B scales for convenience. Items of note in Figures 4.1 and 4.2 are: 1) Greater nugget penetration into spring steel (1080) due to lower electrical and thermal conductivity. 2) Nugget hardness is nearly constant, indicating a well-mixed melt from a carbon solution standpoint. 3) The HA2 of CR 1005 begins at the edge of the nugget at 26 Re hardness. It smoothly tapers off to 46 Rb. 63 64 QUENCHED IRE-HARDENED TEMPERED BASE SPRING STEEL ZONE ZONE mfiL COLD-ROLLED HA2 BASE SAE 1005 STEEL (soft) ugh-r: so a) V REGION OF PHOTOMICROGRAPH Figure 4.1 Coupon weld of hardened SAE 1080 steel welded to cold rolled SAE 1005, a) sketch, b) photomicrograph . 65 Base Nugget < HAZ —> Metal 6.0 35 _. ._ 69Rc c D 5.0 _ 66F! 668C I 5 £9 iI _. \ § 4.0 -1 C .. U) :3 3.0 . 2 C) s E I 2.0 « Mb scan 57“” 47nb 46Rb -0.02 0.00 0.02 0.04 0.06 0.08 0.10 Position, (inches from nugget edge) I 081005 + OUENCHED SPRING STEEL Figure 4.2 Tukon hardness profiles of sample from Ffiignirwe 44.11. 66 4) The near HA2 of quenched spring steel consists of a wide re-hardened zone with full hardness equal to nugget hardness. The far HA2 consists of a tempered zone which appears to bottom out at 52 Rc. Outside the HA2, the base material is fully hard at 66 RC. 5) The boundaries of the prior austenite growth cells are plainly visible inside the nugget. Items of note in Figures 4.3 and 4.4 are: 1) Nugget penetration is roughly equal in both materials. 2) Maximum hardness is only 26 Rc due to the low carbon content. Thus even the untempered martensitic nugget would still be quite tough from a brittle fracture standpoint. 3) Hardness gradually fades to 50 Rb for CR 1005, 87 Rb for Quenched 1005. 4) Base microstructure in CR 1005 is coldworked ferrite with an imperceptible amount of pearlite and carbides. 5) Base microstructure in quenched 1005 is grain refined equiaxed ferrite with imperceptible amounts of carbon products such as martensite, pearlite and carbides. 6) HA2 in both materials consists of a partially transformed far HA2 band and a fully transformed near HA2 band. 7) Remains of prior austenite growth cell boundaries are visible inside nugget. Fortunately, microstructural discrepancies due to material variation (of approved materials) are rarely seen in practice. Therefore, common microstructural discrepancies are usually dependent upon welding conditions. This is good news since, by controlling weld conditions we ‘will also control the microstructure. Thus as welding 67 OUENCHED HAZ use SAE 1005 STEEL METAL COLD-ROLLED HA2 ass SAE 1005 STEEL METAL Gm“) «do a) V REGION OF PHOTOMICROGRAPH Figure 4.3 Coupon weld of heat treated SAE 1005 welded to cold rolled SAE 1005. a) sketch, b) photomicrograph. rteu‘e - IE-uU\CCCF\ “auZCEQI 68 Nugget |<= HAZ —————> 32:; 6.0 .32 C I D 5.0 L- g . LT. \ § 4.0 - 5 _. 8 UJ 3.0 ~ 2 D _ E I 2.0 6 599° 51Rb 5080 -0.02 0.00 0.02 0.04 0.06 Position, (inches from nugget edge) I 061005 + OUENCHED1005 Figure 4.4 Tukon hardness profiles of sample from Figure 4.3. 69 fabricators, our destiny is in our own hands rather than someone else's. If we can control our process then we will be quite reasonably assured of the narrow boundaries of microstructural variation. While it remains possible to generate microstructural discrepancies such as cracks and porosity by over welding, even the existence of these discrepancies does not necessarily constitute a structural problem. If the discontinuity is away from the load-bearing edges of the weld, it generally does not weaken the joint. The tolerance of spot welds for metallurgical unsoundness in the weld core is due to compressive cooling stresses around the weld which tend to prevent crack growth. That microstructure is not a significant factor in controlled spot welding of low carbon steel indicates how well-suited the material is for the process. 4-2 Contact Resistance Figure 4.5 demonstrates the inherent pressure / resistance relationship for 304 stainless steel. The results show two significant conclusions: 1) Contact resistance is a significant function of electrode pressure, especially below 8 ksi. 2) Random variation decreases with increasing pressure, as seen in the range and standard deviation data. 70 i w _{ l E i c 1.0 d I 9 i E ‘ CYUNDER AIR PRESSURE (shown in psi above each point) i "' 08 a i E - \ i - i 30 40 i (”5' 0.6 4: N ? <2: . 5° 70 80 l- 0 4 4 90 c0 ' (T) d a I g 0.2 —4 \\ I _ \\b____qyz/"6M“\«rr”’6\\\\sy_..—e I 0.0 T i T T i T i i i i i i I f “I 0 4 8 12 16 20 24 28 CALCULATED ELECTRODE PRESSURE, (ksi) (for 0.25 inch diameter electrode. 6.36 mm) l AVERAGE OF 5 PTS 0 STANDARD DEVIATION Figure 4.5 Preweld contact resistance versus pressure. O O ('1 71 The pronounced random variation indicates that operating a stable, production welding process would require either stabilization of contact resistance or minimization of its influence upon weld quality. Although contact resistance varies widely at the start of spot welding, it is desirable to minimize the impact of this variation upon the final temperature distribution. Later in this study we will address the issue of contact resistance sensitivity. 4-3 leld Lobe Data The importance of current and time was confirmed using the laboratory welder. Using the lobe procedure outlined in Section 3-6, a weld lobe was generated for 0.030 inch (0.75 mm) bare steel. The following observations can be readily made from Figure 4.6: 1) The area enclosed by points A, B, C and D is the window of acceptable weld quality for this material under the conditions listed. Within this envelope, moving toward the right produces hotter, larger welds. 2) The curve connecting points A and B is the iso-diameter boundary for a weld diameter of 0.158 inches (4.0 mm). Welds along this boundary typically have a microstructure as in Figure 2.1. 3) Points E and F define the expulsion limit. Welds to the right of this curve consistently expel molten metal as shown in Figure 4.7. 4) The sticking / indentation limit is shown with endpoints C and D. A typical "hot” microstructure along this line is shown in Figure 4.8. /:UC-r\~ar\ RIF. CH\\ UQ¢—|F 72 24 B F C 14 - 6 12 ~ £2 g — OER o 10 ‘ “NOE” INDENTATION 2 SIZE INDENTATION AND/OR " 25:: ZONE sncmNo EB ZONE v 8 -—4 g — k cmamoi avmsmn 4 -« «0mm) umn mmunwmmnm I I I I I I I 4 B 12 20 WELD CURRENT (ka) Figure 4.6 Weld lobe number 1 - a normal weld lobe for bare steel, under the conditions stated in Chapter 3. Electrode pressure a 10.2 ksi, Force = 500 lbs, (217 kg). (70 MPa). Figure 4.7 73 .E o v Q o Photomacrograph showing the result of moderate expulsion. This is a laboratory weld, of 0.030 inch (0.75 mm) SAE 1005 uncoated steel. Parameters were 8.3 ka for 14 cycles at 500 pounds (217 kg). Notice the electrode indentation, and the expulsion "whisker" extruded from the weld interface, (sample no 544). 74 E o v Q o Figure 4.8 Hot weld showing over-indentation. The material and parameters were identical to Figure 4.7, except that current was 15.8 ka. The electrodes stuck to this sample and left behind pieces of copper on the surface, (sample no 333) . 75 5) From Figure 4.6 it is apparent that current is a more influential parameter than time. This conclusion is consistent with Joule's Law, Equation 2.4 4-4 Misalignment and Force Effects In order to simulate plant conditions concerning weld pressure in the laboratory, it was necessary to consider both electrode force and misalignment, since misalignment influences contact area as discussed in Appendix B. A simple modification to the weld lobe procedure provided adequate data to assess the relative importance of misalignment and force. The gun translation device, which was added to the welder, was shown in Figure 3.4. It provided a controlled means of inducing electrode misalignment. After translating the upper electrode 40% (0.10 inches, 2.54 mm) away from the axis of the lower electrode as shown in Figure 4.9, a new weld lobe was generated, using the same procedure. The results are shown in Figure 4.10. In comparing lobe 2 with lobe 1, the following conclusions may be drawn: 1) All of the vertical curves are shifted left, by 100 to 1000 amperes. 2) About 80% of lobe 1 is common to both lobes. Only the rightmost 20% of lobe 1 is not shared by lobe 2. This means that if the process is initially set to the expulsion limit according to lobe 1, and then the electrodes should become misaligned by 40%, then the weld will still be acceptable according to lobe 2. Conversely, if the process begins with misaligned tips and is set to operate near the low current limit of lobe 2, then by 76 19 Electrode m Workplace R l Workplace Electrode m - 0.100 inches. (2.54 mm) 2R - 0.250 Inches. (6.36 mm) m I TR 40% misalignment [9 Figure 4.9 Schematic of the experimentally misaligned condition. 77 B' F‘ c: ‘— 14 —‘ W ........ 6‘: N 10 a 1‘ g -= '. '-. .‘, 8 _‘ 1. i E - 9 I: 6 -« A v ------ \ {-1 x E’ o' 4 _ LOBE SHIFT DUE TO A 40% ELECTRODE MISALIGNMENT I I I I I I I I I 4 a 12 16 20 24 WELD CURRENT (ka) MISAIJGNED ALIGNED ELECTRODES ELECTRODES (Lan #2) (1.er #1) Figure 4.10 Weld lobe number 2 - 40% misaligned electrodes. Notice the leftward lobe shift caused by the misalignment. Still, lobes 1 and 2 share roughly an 80% overlap, when the upper limit is taken to the sticking/indentation point. 78 aligning the electrodes, the process will begin producing undersize welds. However, in either lobe 1 or 2, if the process is initialized at the expulsion limit then, regardless whether the electrodes become more or less aligned, the weld size should remain adequate. 4-4-1 Eisalignsent Effect Ordinarily, under laboratory conditions, misalignment would be seen as a major variable. This is because within the spot weld research community, the upper current limit of the weld lobe has traditionally been the expulsion limit (Appendix A). Treating expulsion as the upper limit, lobes 1 and 2 do not exhibit enough overlap before and after misalignment. Thus, under the no expulsion rule, misalignment would not pass as a minor variable. However, since misalignment is difficult to control in manufacturing, we learned from this study how to desensitize the process, by disregarding the no expulsion rule. Misalignment has been shown to be a minor variable, when it is less than 40% of the electrode diameter. For on the one hand, misalignment increases local pressure, which tends to reduce visual weld heat. On the other hand, misalignment also increases current density by an equal amount. These two changes nearly cancel each other, and the resulting lobe shift is relatively small. The trick which allows misalignment to be treated as a minor variable is to set the process to operate near the expulsion limit. In 79 order to center the process at expulsion, we must redefine the upper limit of the weld lobe to include the hot welds, out to the point of over-indentation and/or electrode sticking. Then if some misalignment should occur, the process will still make acceptable welds, as per Figure 4.7, albeit with more expulsion. 4-4-2 Force Effect A 40% misalignment reduces electrode contact area by 50% (for the mathematical solution to the common area of winking circles, see Appendix 8). Since halving the welding area essentially doubles current density and pressure, the data of Figure 4.10 was replotted in Figure 4.11 in terms of average current density. Even though mechanical and electrical considerations imposed by the sheets have an impact on actual pressure and current density, these mechanics considerations are ignored here. Under the assumption that these considerations are negligible, Figure 4.11 simulates the effect of doubling the weld force. Force has been shown to exhibit a strong influence upon the size and position of the weld lobe. As force increases, the weld lobe increases in width and moves to the right (compare lobes l and 3). Thus one may conclude that force is an unavoidable major variable. However, force is a very acceptable key variable from a practical point of view, A 12 4 3. Z a) '. 3 1 £2 _ 5. 5. o I I 5‘ 1° ‘ E 5 N 1 : . z I I 0 T '. : / 9 a =. LU 8 ‘ 35 ‘5 / 2 2 '2 a F: K 3 1” e 4 ........... /1. A' E' D' d _—> > 4 _ 1.an SHIFT DUE TO A 2x CHANGE IN PRESSURE AT THE ELECTRODES I I I I I I I I T I I I I I r 80 160 2M0 3K) 400 ‘ED' 9%) SM) 720 AVERAGE CURRENT DENSITY THROUGH THE WELD (kafln " 2) .......... ELECTRODE PRESSURE ELECTRODE PRESSURE I 10.2 RSI I- 20.4 ksl CONWMHVWEA CONWMNVWEA I 0.049 in"2 = 0.025 in"2 (LOSE #1) (LOBE #3) (data from lobe #2) Figure 4.11 Data from lobes 1 and 2, showing average current density as the abscissa rather than current. Force - 500 lb (217 kg) in both lobes. These data show the significant effect of changing the average pressure at the weld. These data approximate the magnitude of lobe shift that would have been measured by a 2x increase in weld force, for aligned electrodes. 81 since it can be easily set, monitored, and maintained at the process. Remember, the stated goal in this study was to establish weld parameters which define a stable, robust process with as few key variables as possible. 4-5 Process Honitoring and Feedback Techniques As a matter of common practice in industry, current steppers are used to compensate for tip wear according to a programmed boost schedule. Steppers are quite reliable. However, even with their use, there can still be a need for compensation of other types of process variation. Conventional steppers do not adjust for many significant variables such as current, tip alignment, cable wear, material and coating. One of the goals of this study was to identify static parametric regimes where the deleterious impact of process variables is minimal. Assuming that this can be done, there may still be practical situations where the magnitude of process variation exceeds that allowed by even the most robust static control scheme. In that case, one must either minimize the Offending process variation, or compensate dynamically, so as to remain inside of a "moving" weld lobe. 82 4-5-1 Power Factor Drop (PFD) Certain weld controls can empirically measure the power factor each half cycle. In this study a Medar Legend (tm) weld control was used with a data acquisition system, as described in Section 3. Power factor data for bare low carbon steel are shown in Figure 4.12. On first inspection, the power factor curve appears to resemble the dynamic resistance curve shown in Figure 2.15. The reason for this resemblance lies in the definition of power factor. Most commonly, power factor is defined as: P.F. = COS“) [ 4.1 1 where Q is the angle between voltage and current. This equation is true for sinusoidal waveforms. But in resistance welding, part of the sinusoidal wave is removed in order to regulate the weld current. In any case, a more appropriate definition is used to correlate power factor and resistance (54). That is: P.F. = 4.2 2 2 [ ] where R a total circuit resistance, and XL = inductive reactance of the circuit. Since XL is mainly dependent upon 83 74% .2 72% m g —1 70% -* |—- 0 _. g - m ... I; 66% — O ... n- 6... 1 62% — 60% Figure 4.12 I I I I I I I I I I I I I 2 3 4 5 8 7 8 9 10 11 12 13 14 WELD CYCLE (60 hz-ac) (AT 80% HEAT) I NEGATIVE HALF-CYCLE + POSITIVE HALF-CYCLE Dynamic power factor for a laboratory weld on SAE 1005 bare steel, 0.030 inches (0.75 mm). This weld did not have expulsion. The largest one-cycle drop in power factor near the end of the weld (when expulsion normally occurs), was 0.85%. 84 geometric and magnetic effects in the transformer and secondary loop which should normally not change during welding, it may be treated as a constant of the weld setup. Therefore, the most dynamic term in Equation 4.2 is resistance, and more specifically, the resistance of the weld. Figure 4.13 shows Equation 4.2 for the practical range of resistance spot welding. And from Figure 4.12 we see that the laboratory welder operated in the range 61% to 72% power factor. Over small ranges such as this, a tangent on the curve in Figure 4.13 indicates only a small amount of nonlinearity between power factor and resistance. In addition, by comparing Figures 4.12 and 4.14, notice how expulsion produces an instantaneous drop in power factor. There are two significant conclusions to state here: 1) As long as impedance is constant, power factor can be used to indirectly measure small resistance changes. This is important because of the ease with which power factor is measured, without the use of secondary leads or sensors. 2) The power factor drop (PFD) at expulsion essentially measures the resistance drop of Nagel and Lee (36), and therefore may be used as expulsion detection signal. The second conclusion is very significant because it improves the economic opportunity for feedback control. By having the weld control provide power factor data as a by-product of its own internal algorithms (54), it is essentially ”free data". This data stream may be processed after each weld. Current adjustments can be made 85 90% 80% 8 5 70% < H. E 60% 3 O D. 50% 40% 0.5 Figure 4.13 ID 15 2D R/xL Theoretical relationship between power factor and R/xL ratio. Tangents along short sections of this curve may be used to linearly approximate this relationship. In addition, since X is largely constant during welding, it explgins why the power factor curve mirrors the resistance curve during welding. 8t5 74% -‘ j 72%-J I A O 70% ~ /s/.\' l.— 0 m E 68% J a.“ // cr ~ , / ”3" 66* _ \\/ PFD -6.02% O -1 O- ¢g%.e - ......................... g ..- a ‘i i 60%! I I I 1 T 1* I 1 1 I I I r i 1 2 3 4 5 6 7 8 9 1O 11 12 13 14 WELD CYCLE (60 hz-ac) UVISOHIHEWR) I NEEUVUVEWTALF4TWCLE -+ FNDSHTVEITALFJTWCLE Figure 4.14 Dynamic power factor for a laboratory weld on SAE 1005 bare steel, 0.030 inches (0.75 mm). This weld had visible expulsion. The largest one-cycle drop in power factor near the end of the weld is 6.02%. 87 automatically, keeping the process operating as closely as possible to the expulsion limit. Fortunately, as discussed in Section 4-4, probably the most robust way to operate the process is at the expulsion limit, since it is close to the center of the weld lobe. Whereas a conventional stepper blindly compensates for tip wear alone, a dynamic stepper could compensate for any variable, including tip wear. The data presented in this report show that this system works, both in the laboratory and plant environments. Its practical limitations and advantages are explained in Section 5-6-1. The power factor signal is processed for feedback as shown in Figure 3.3. First, power factor is internally measured and stored for each half cycle Of current. These numbers are available for data collection at the printer port after each weld is complete. Table 4.1 shows an example of this postweld output. Next, the control identifies the largest decrease in power factor from one cycle to the next. Each polarity is considered separately, and the larger of the two results is reported as the PFD (power factor drop). PFD is also directly available at the printer port after each weld as shown in Table 4.2. Next, the PFD is compared to a threshold (say 1%). If PFD is equal to or greater than the threshold, the control concludes that expulsion has occurred, as in Figure 4.14. Likewise, PFD less than the threshold means that expulsion Table 4.1 Example of cycle by cycle output produced by the weld control. Each line represents one 88 cycle during the weld. CYCLE VOLT AMPS PF+ PF- ifiEAT 1 475 113 6442 6551 40 2 474 104 6619 6818 40 3 474 104 6755 6920 40 4 474 103 6848 6963 40 5 474 100 6898 6987 40 6 475 107 6885 6415 40 7 474 115 6320 6294 40 8 474 109 6278 6321 40 89 Table 4.2 Example of end of weld summary output produced by the weld control. Each line represents one weld. WELD AEPS VOLTSLEHEAT PFD 63 476 80 00.85 66 476 80 06.02 66 476 80 05.57 64 476 80 01.07 64 475 80 00.98 64 475 80 00.93 64 475 80 01.31 64 476 80 00.92 0040101pr6! 90 did not occur, as in Figure 4.12. For signal processing, there are counters which keep a running tally of expulsion welds and no expulsion welds. The counters are programmed to trigger a 1% heat change at a preset number of welds. If there are not enough expulsion welds the current is automatically increased. Conversely, if there are too many expulsion welds the current is decreased. After every automatic current change, both counters are reset and the tallying process restarts at the new heat setting. The flow diagram for this process is shown in Figure 4.15. Laboratory correlation between PFD and visual expulsion was quite good when welding bare steel (Figure 4.16). Laboratory data for galvanized steel were not gathered however, plant data for galvanized steel is provided in Section 5-6-1. 4-5-2 Acoustic Emission (AE) In this study, acoustic emission (AE) was considered in parallel along with PFD monitoring and visual expulsion monitoring. The PAC - 2250 system was attached to the lab welder for evaluation. It was set up and calibrated by the manufacturer, Physical Acoustics Corp. In initial testing, good correlation between acoustic emission energy and visual expulsion could not be obtained and the concept was dropped in favor of PFD monitoring, for the following reasons: 91 WELD “ms YES EXPULSION N0 DETECTED ? INCREMENT THE EXPULSION COUNTER DECREASE HEAT INCREASE HEAT BY 1% Fa! ALL BY 1% FOR ALL SUCCESSNE SUCCESSIVE WELDS WELDS Figure 4.15 Logic flow diagram for automatic feedback stepping by the power factor drop (PFD) technique. 7% 92 xx X X x A 6% "‘ C) u. 9-, 5%>~ (L O a: X X (3 4%:— a: 2 () 3%.J 1< u. E 2% - 3 O EXPULSION THRESHOLD 0- 1% — , . I 1... . 1‘ . 0% n I ‘1 I I I ' 6 7 8 9 10 CURRENT (ka) . =IWDVMHBLE )(=\N§BLE EXPULSION EXPULSION Figure 4.16 Correlation data between PFD signal and visual expulsion. These 32 data points were from 32 consecutive welds made under the conditions stated in Chapter 3. 93 1) PFD showed good statistical correlation with expulsion as per Figure 4.16. 2) PFD measurement is "free" data since it needs no sensors in the secondary loop, and adds no fundamental cost to the welding equipment. While acoustic emission remains an interesting topic for resistance welding, it did not appear to be a viable solution to the feedback problem at this time in light of newer, competing technology. 4-5-3 Loop Efficiency As a result of this study a simple method of automatically detecting significant process change was invented. Since there are measures of energy efficiency in the basic electrical data already resident in weld controls, a control can be inexpensively programmed to monitor the energy efficiency of the welding circuit. The controls used for data acquisition were altered to detect wear on the equipment, when that wear resulted in a change in the electrical load. To understand this method, consider the welder for a moment simply as a current supply, programmable in increments of % heat, % current or amps. One could measure the relationship between the programmed (or expected) output and the actual output. In a well-behaved process, as the user calls for more output, the machine should respond by producing more current. 94 In the % heat system, as a consequence of Joule's Law, Equation 2.4, current should follow heat according to the formula: I = K-%H [4.3] In this expression, I is either primary or secondary current, %H is the programmed value of percent heat, and K is a proportionality constant which lumps together the total load (inductive and resistive). K has been referred to as the K-factor. Rearranging for the K-factor, K=— [4.4] shown with the proportioning exponent n, which compensates for nonlinearity. Ideally, n = 2 and Joule's Law is obeyed. However, actual measured n values in this study ranged from 1.3 to 1.8 (see Figures 4.17 and 4.18). It is postulated that n values are not 2 because of measurement inaccuracies in the control, especially concerning how I(rms) and %H are determined. In any case, K-factor is a process constant which reflects the empirical electrical efficiency of the welding circuit. It is useful for process control since it can be used to detect changes in the welder due to fundamental process changes, while remaining relatively insensitive to 95 _.s O o 80- 60* 40‘ 20‘ PRIMARY CURRENT, amps (output) I = 0.644 (%H) + 16.38 amps (+/- 3.16) O 0 Figure 4.17 I I T I T 20 40 I 60 % HEAT (input) 80 The measured relationship between primary current and percent heat, for the laboratory welder at tap 1 on bare steel. There are 455 100 separate data points shown: each point repre- sents one cycle at the indicated heat. The tolerance band is calculated for the 95% confi- dence limits of the individual points. Joule's Law would imply a parabolic relationship; in reality however, the data do not agree with Joule's Law. 96 10 3 .. K = 5.62 (+/- 0.49) 32‘ 5 , ‘6' q BI 35 6-‘ ‘5 w,,; "‘ .~ *5 d E 8 5 4 — ‘6' a: ”T _1 x 2 .1 _ r1=‘L43 O I I I I I T I I 0 20 40 60 80 100 % HEAT (input) Figure 4.18 The empirical relationship between percent heat and primary current may benexpressed for the data of Figure 4.17: K = I /(%H), where n = 1.43 and K = 5.82 (+/- 0.49). 97 parametric changes. In other words, a change in K-factor indicates that the process has changed in a significant way, such as cable wear, or a change in the transformer turns ratio. Simply changing the weld schedule or stepper schedule will not alter the K-factor. Similarly, in % current systems, actual current should follow % current according to the formula: I %I = 100- [ 4.5 1 I(max) where Imax is the maximum current available when firing the SCR's at the full 180 degree conduction angle (minus a power factor or voltage compensation angle, if any). I and Imax can be either primary or secondary current. Equation 4.5 can be rearranged as was Equation 4.3 yielding, I I(maX) c = —— - [ 4.6 1. %I 100 C has been named the C-factor. Because the C-factor is a new concept, as of this writing its inventors Britton and Doede (55) had not yet tested its linearity. However being inherently simpler and more elegant than the K factor, if C-factor becomes a popular monitoring tool, it may steer the industry toward the %I programming convention. The C-factor will first need to be tested for linearity with respect to loop wear. 98 In the amperage programming convention, loop efficiency monitoring can be in done a variety of ways. K-factor can be used for a %H comparison (Equation 4.4), or C-factor for %I (Equation 4.6). It is important to note that in the amperage programming convention the system automatically compensates for loop wear. It becomes, by definition, a constant amperage power source, as discussed in Appendix D. Because electrical efficiency is a product of controllable variables, a gradual change in inefficiency calls for preventive maintenance rather than feedback and automatic adjustment. Blind compensation for a maintenance need leads to further wear and loss of efficiency. It may best serve the user's interests to be passively alerted of the need for repair, so that the maintenance can be performed during a period of scheduled downtime. Methods like this are called passive monitoring rather than feedback monitoring because they do not compensate for changes they detect. If this particular system performed automatic compensation, it would be similar to the constant current system described in Appendix D. This technique is called loop efficiency monitoring because most of the wear on electrical components occurs in the (secondary) loop of the welding circuit. 5. PLANT RESULTS AND DISCUSSION In addition to the variables studied in chapter 4, there is a host of additional variables imposed by the plant environment (see Figure 1.1). These variables range in their significance, and a complete study of each of them would require years of testing. In order to test the practical implications of the results obtained thus far, a change in research strategy was imposed at this point in the study. It was decided to take the conclusions of Chapters 2 and 4 to the factory floor in order to begin a pilot implementation study. 5-1 Floor-based Research Strategy Conducting research on the factory floor involves coordinating activities among the many people who are connected with the processes. Communication is essential, both with those directly involved with the equipment, such as operators and maintenance personnel, and those more distantly involved, such as engineers and managers. To have a successful project with meaningful, fruitful results requires that all persons who can affect the research results be sufficiently aware and supportive so they may provide necessary assistance at whatever hour, day or night. Thus good, frequent, and friendly communications are 99 100 extremely important to preserve rapport with the people on the floor. Because floor people are notoriously results oriented, it is also necessary to reward them for their efforts. Rewards vary from personal recognition to improved line performance, uptime, and part quality. See Appendix C for examples. First, weld improvement teams were assembled to evaluate and recondition the plant equipment. Armed with the laboratory results that key variables were probably current, weld time, and force, we began to measure and record these variables on each weld gun. Knowing that electrode alignment was only a minor variable, we inferred that metal fit was also a minor variable (56). We used an infra-red viewer to detect hot areas in the tooling, and a micro-ohmmeter to measure the resistance of the secondary loops. Soon, we began to notice certain trends in the information: 1) Up to 50% of secondary current cables were found to be substandard by either the ohmmeter or the infra-red viewer. 2) Up to 30% of cooling water lines were found to be non-functional by the infra-red viewer. 3) Up to 25% of weld guns were in need of rebuilding, due to worn bushings or cylinders. 4) Weld current, time, and force varied widely throughout the plant. There seemed to be no single strategy for weld setup based upon either the material to be welded or the electrodes being used. 101 Because of the strength of these four trends, we began to refer to them as the four key variables. They were the root cause behind the majority of discrepant welds in the plant. Notice how all four key variables are maintenance related. In order to address the above problem concerning parameter selection (trend number 4), a set of weld parameters was taken from the General Motors Resistance Welding Handbook (57) and adapted for easy floor interpretation on the shop floor (see Table 5.1). In setting up standard weld parameters without extensive lab testing there was a risk that the parameters would not be optimum. However, because weld lobes for low carbon steel are so large (as shown in Section 4-3 and 4-4), the opportunity to minimize process variation outweighed the risk that the actual parameters might not be exactly optimum for some applications. So as a first approximation, Table 5.1 values were implemented on all 1600 weld guns in the plant. The decision to use standard parameters ended up helping the project in several important ways: 1) The bookkeeping and training were simplified, since only simple weld schedules were used, (eg. no current sloping or pulsing). 2) Once the reconditioning work was complete and the weld quality verified, if a later weld was not holding, the focus was not on the weld parameters themselves, but on the root cause for the discrepancy. People accepted the weld parameters and did not change them arbitrarily. 102 h<._u wZO ((0 “021‘ ”20 O #5”. wZO S<0 Ungraaan wZO 0 n go 8024 03.. 6.26 $02.33 on: c M§._ 33.967713 g ”dam MM.“ Hfiau £83126 6:6 xoowozsi 625m; 32.5.me .26 ES. 68:2 2a 8.83. .r ... E> n. Floor setup guide that was refined and used during the plant study. These are good empirical parameters for process control. Table 5.1 103 3) Because the parameters were derived from an official company handbook they were politically benign, and needed only a little "salesmanship" to gain acceptance plant-wide. During the setup stage, whenever the standard parameters were found to produce discrepant welds, the standard was adjusted to reflect the latest information. It soon became obvious that the initial weld current for new electrodes was related to the electrode geometry. Empirically, this relationship is shown in the caption under Table 5.1. It is postulated that contact area is the most significant shape factor in assigning the starting current for a given electrode geometry. This is because contact area has a direct bearing on the average current density of the weld. Other factors which may be sensitive to initial electrode geometry are the distribution of current density and pressure across the electrode face. Therefore, a flat faced electrode will require more current than a ball faced electrode, even if the initial contact areas are similar (14,21,58). 5-2 Process Centering Process centering has to do with adjusting a process to operate in the center of its parametric envelope. Ideally, a centered process uses parameters such that the measured average output equals the specification mean. In the case of spot welding, the measured output is nugget diameter, and 104 a centered process operates at the center of its weld lobe, above the minimum diameter and below the electrode sticking point. A few of the 1600 plant guns reconditioned still needed slight adaptation beyond the Table 5.1 requirements. This need surfaced during the first few production runs. These setups were finessed in order to achieve the required weld size for the entire electrode life. Usually only initial current and the stepper schedule were adjusted. In our plant, electrodes are changed as a group. Therefore electrode life is considered to be the number of production shifts between tip changes. This practice is in contrast with a popular research technique of measuring the maximum number of welds obtainable, which has been shown to be a statistical variable (59). Thus it was not necessary to obtain every last weld from the electrodes, but only to make it safely to the next tip change interval. Typically this production interval is from one to four shifts (from 1000 to 10,000 welds). One welding machine required a special deviation that is worth mentioning. Some sections of this machine ran at an extremely high production rate of 800 welds per hour. The air-cooled welding cables were heating up and oxidizing, which caused significant current losses within a few days' time. To cure the problem, the weld time was shorted from 16 cycles to 10 cycles in order to lower the electrical duty 105 cycle, (the percent of time when the circuits are actually passing weld current) from 5.9% to 3.7%. In exchange for a shorter weld time, the starting weld current was raised from 9.5 ka to 10.0 ka. The material welded in this application was HSLA hot-dip galvanized sheet steel, approximately 0.030 inches (0.75 mm) thick. Efforts to further optimize weld setups still exist in many areas of the plant. Nonetheless Table 5.1 served as a very good first approximation for the vast majority of plant applications. At present, continuing studies focus mainly on electrode shapes, electrode materials, and preventive maintenance systems rather than on process fundamentals, as was done in this study. 5-3 Process Desensitization As a rule of thumb, operating near expulsion decreases process sensitivity and decreases the number of key variables. Running with the welds hot makes resistance spot welding easier to control because there are fewer sources of critical variation. When using this strategy, one usually only needs to monitor cable wear, water flow, force, and current. Variation in material, fitup, thickness, coating, dirt, voltage, water temperature, alignment, etc. can be largely ignored. 106 However, a price for the extra robustness of running hotly is shorter electrode life. Intuitively, the hotter the welds, the worse the electrode life will be. The upper limit to weld heat is defined by the following factors: 1) The minimum electrode life required to operate economically, which is measured conveniently in even multiples of the number of parts in a standard production interval (eg. the number of shifts). 2) The appearance requirements (if any) for the weld. Hot welds generally leave an indentation mark from the electrodes when significant expulsion occurs. 3) The absolute upper limit of weld heat is electrode sticking, when the electrode bonds to the workpiece. Sticking is totally unacceptable, since it halts the process and sometimes leads to equipment and part damage. Another price for running hot is dirtier equipment, which will require deslagging and washing on a scheduled basis. Although some have claimed that running hot wastes electricity (60), the automotive community has regarded electricity as a minor cost of spot welding, since each weld uses only 0.0005 kw-hr of energy as described in Appendix D. 5-4 Remaining Variables The question was raised by plant personnel about what to do once the variables were optimized if the process still failed at some future point to produce acceptable weld quality. (In all of the 1600 guns included in this study, less than 1% failed to yield acceptable weld quality after 107 optimization). This section focuses on this group of exceptional welds. In almost every case there was an obvious and direct assignable cause to the problem. Most of the exceptional welds were in extremely poor metal fit areas, where the electrodes and workpieces would not seat, even under full electrode force. When the clamped workpieces either do not contact each other or do not contact one of the electrodes, the current path through the workpiece becomes disturbed and current density will be low. In this case the gun must crush the sheets together in order to establish good thermal and electrical contact for welding. Over welding and using a pulsation weld schedule has worked with some success, although electrode life suffers accordingly. A better solution to metal fit problems is to eliminate them ahead of time, in the metal forming operations. Occasionally electrodes become coated with organic buildup from oil, paper, or cloth from the manufacturing environment. This type of coating acts as an insulator and will suddenly cause weld current to drop to zero in the middle of a production run. There are several practical strategies for dealing with this annoying problem: 1) Lightly file the contamination off of the electrodes, taking care not to change the face diameter, which by this point in the stepper is at a particular size which is appropriate for this number of welds. 2) Replace the electrode with a used one that is worn to about the same diameter. 108 3) Program a minimum weld current limit into the weld control. The limit will halt the machine when minimum current is not detected. This "low current fault" condition effectively alerts personnel and prevents the manufacture of any discrepant welds due to this problem. Other common causes of low current during welding are: 1) An open condition in one of the primary or secondary circuit connections. 2) Any source of extra resistance in the secondary loop such as a worn cable or loose connection. 3) An increase in the inductive reactance of the secondary loop due to a change in loop area or a change in the amount of magnetic material in the secondary loop. 4) A large drop in supply voltage. 5) A change in the weld schedule. These considerations are discussed in Appendix D. As observed from Figures 4.6 and 4.10, operating at or above the expulsion limit reduces the susceptibility to critical loss of current density. This susceptibility is due to variation in pressure or weld area. In fact, over welding in this way reduces the influence of many minor variables, such as metal thickness and coating (see Figure 1.1 for an extended list of variables). Making use of the full width of the weld lobe as defined in Section 4-3, one can center the process and operate the furthest from quality problems. This is done by setting the current near the middle of the weld lobe, at the expulsion limit. 109 While operating at expulsion, the degree of process robustness enjoyed depends upon: 1) The width of the weld lobe, 2) the chosen setup conditions, and 3) the range of variation encountered among the major and minor variables. Although welding hotly increases the system's tolerance for variation, excessive heat produces negative results, such as over—indentation, poor electrode life, and worst of all, electrode sticking. Electrode sticking is most often observed with new electrodes. The system with the widest weld lobe will show the least tendency toward electrode sticking and the greatest insensitivity to normal process variation. To increase the lobe width of a process one might consider changing electrode shape, diameter, material, or force. The shape of the electrode is expected to affect lobe width because of its effect on current profile (58). The electrode material is expected to affect lobe width because of its effect on heat flow out of the workpiece and because of alloying, pitting and contamination (61). The electrode force was shown to affect lobe width in Figures 4.6 and 4.11. The reasons for this force - lobe width relationship are complex, including a relationship between force and thermal and electrical coupling at the electrode-sheet boundary, and the effect mechanical deformation has upon the contact area and the current path (14,17,18). 110 5-5 Setup Guide The parameters used to set up weld guns during the plant implementation study are shown in Table 5.1. The parameters were tested and refined empirically on 1600 production weld guns. In practice, the tabular values place the process very near to the expulsion limit, at a force where the process is robust and well-behaved. To complete the setup, we only fine-tuned the current to the expulsion limit and measured weld size destructively to demonstrate that the minimum bond diameter requirement had been met. The column in Table 5.1 labeled "design maximum current at 81% heat" was used to test transformer capacity. The transformer must be suitably overdesigned to provide head room for the stepper to compensate for electrode wear over thousands of welds. 5-6 Process Monitoring and Feedback Techniques 5-6-1 Power Factor Drop (PFD) Under plant conditions, production tests of the PFD stepper described in Section 4-5-1 were run using galvanized steel (0.024 inch [0.6 mm] welded to 0.079 inch [2.0 mm]). These results are shown in Figures 5.1 and 5.2. From these data one may conclude the following: 111 1) When in automatic feedback mode, the weld heat will step either up or down to balance the amount of expulsion obtained. 2) There is sometimes a rapid rate of stepping initially to compensate for incorrect starting current. An example of this heat correction is shown in Figure 5.2a. Once the current is adjusted to the expulsion limit, the current steps are more randomly up and down, with only a long term upward trend. 3) The feedback process is in control. Figure 5.1b shows a classic pattern of control for this application. About 78% of the welds exhibit PFD less than the expulsion threshold. Ideally, we expected approximately 75% Of the welds to be below expulsion, since the counter limits were set to 6 no-expulsion welds and only 2 expulsion welds, (a rat-o of 3:1). Figure 5.2b shows a higher percent of welds below expulsion (85%) due to the initial warm up adjustment during the first 150 welds (Figure 5.2a). 4) The long term stepper trend may be approximated by a straight line. This means that generic stepper gradients might be characteristically linear throughout electrode life. This trend would be a worthwhile subject for future study, perhaps using a similar data collection scheme. However, just knowing that the process naturally may want to step at a linear rate throughout the electrode life helped to simplify our plant's stepper schedules for non-feedback applications. As a result, in most cases they were shortened down to one or two steps, as characterized in Figure 5.3. Certain interesting problems were identified whenever the feedback system failed. The biggest problem was false PFD's. A false PFD tricks the control into reading the weld as hotter than it is. This condition was initially prevalent when welding with galvanized steels. There is a false PFD associated with expulsion of the zinc coating early in the weld schedule. If the control reads this as the PFD for iron expulsion, it will incorrectly turn down the 96 STEP HEAT PFD96 FREQUENCY OF OCCURENCE 9) Figure 5.1 112 40% 30% . 20% _ 10% d M 0% F%JF g I l I .‘-!i L ‘FL 2% ~ i . 3% « 4% 1 1 1 1 1 s 1 1 0 2a) 1“” 6“) mm NUMBER OF WELDS 2496 ”0:330" ‘ d‘nsornnu. gfifixgfixg ; Hm-1ns POWER FACTOR DROP Actual production data using the PFD feedback control. The material was hot-dipped galvanized steel 0.024 inch (0.6 mm) welded to 0.079 inch (2 mm). The weld tool was of the press/fixture type, with a large secondary loop. Notice how the stepper stepped both up and down, in order to maintain an average of 22% expulsion. a)individual data, b) histogram. 113 um I'- D 30% 1 I l DJ 20% 1 'u— (D #3 1ms~gfififig -———- 0% 1s 2 A l ‘ ' he i. I ' l E 2% - i O. 3* —4 4* I T 1 T I I— I I T T 0 an «m an an 1d» a) NUMBER OF WELDS a EXPULSION 2“ I THRESHOLD ; PH)-1J% I.” (D 2 [LI I D C) 8 EXPULSION “- WELDS o 15% OF TOTAL > >. C) 2 [U 8 [U C l .. . .. .. '1 1 _ .A fi'.‘_ 014' 1% 2% 396 4% b) POWER FACTOR DROP Figure 5.2 Actual production data using the PFD feedback control. The materials were identical to those in Figure 5.1. The weld tool was also similar. Notice how active the stepper was for the first 125 welds. Because the initial heat setting was too low, there was little expulsion. Later the system balanced, for a total of about 15% "expulsion welds". a)individual data, b) histogram. 114 60% 9363‘»)? 9‘“ / 66V 55 .\° 40% s / y I- Lfio“ Q3190 ff: 0 , 110* E 9‘ p3\<>“$ IE 1655 pNND U) 20% f \ fl ‘68 9' 503 E/L 4&3“. ‘“‘“#5.120 ?,,,ssa 1 ipPLfimh / g/ 0% 4 1 0 2 4 6 NUMBER OF WELDS, (in thousands) Figure 5.3 Standard, simplified stepper curves deployed in the plant for non-feedback applications. 115 heat in order to try to eliminate the perceived expulsion. Unfortunately, false PFD's can lead to cold welds. Through trial and error, we found in the plant that PFD's at zinc expulsion could be filtered out with pre-pulse welding and initial data blanking. This way, the control is set to look for PFD only after the zinc has expelled. Other false PFD's can arise from dirty, arcing ground blocks, from loose, arcing secondary cables, or from poor water flow. False PFD's must eliminated or filtered, otherwise the noise will be interpreted as expulsion. Another area of difficulty for PFD feedback control is welding very thin galvanized steel to itself. In the case of Figure 5.4, the material as specified is too thin to allow bulk resistance and heat flow to stabilize as discussed in Section 2-4-3. Therefore, even without feedback control this material will fluctuate in and out of expulsion. On light gauge steel, the window for quasi steady—state welding is much smaller and harder to hold. In practice, when the steel is thinner than 0.028 inches (0.7 mm), the process is subject to excessive variation due to coating variables. There was insufficient time in the present work to resolve this problem. Consequently, it remains for future study. The problematic data are shown in Figure 5.4. 96 STEP HEAT ES PFD % FREQUENCY OF OCCURENCE b) Figure 5.4 116 40% 30% j MANUALDOOSTTO RESTORE wELD QUALITY 20% 1 \ 10% '1 H .1. __,, T; ‘I .1 IV WWII 111 II n W 1% 1 II I I II III/I III IIIV’VIII III III 2... I I 3% T \ 4% .4 \ 5% _ I I FALSE PFD's DUE TO zmc EXPULSION MHWSMHIB 6% I I I I I I I 0 1a) 2a) 3m) 1“» NUMBER OF WELDS 24% I EXPULSION 4 THRESHOLD 20% 1 60916013106 PFD ' "1" « “sum 3 emmsnw :N%Osnnn. ; “sum 16% A 21%OFTOTAL I, §E;:F:cario\s «\p PROCEJLRE FOR DETERMINBC THE .t CLJ.1.D;LIT‘.'QF Bud) STEEL MATERL-XLS carbon mild steels, medium strength steeis, high Strength steels and high low allov steels. The procedure is used to determine the praCtical limits of selcaoiiity by the reSistahce spot welding method, utilizmg weldability lobes in comunCtion With data lrorn physical oestrUCtive test methoqs. Details are contained herein for determining the selection of test conditions. developing interpreting and evaluating “this specilication establishes the requirements and procedures used to determine 1.». elcaoility by the reSistance spot welding method for coated and uncoated lovi- Strength for ir.5trurr.entation, and Several formuias have been develooeo to obtain the various process parameters l. 2 equipment weicability lobes. and constants that take into account the metal thicmess, yield strength and 17.6 ’ 3 coating of the test material. Single pomt. Single phase reSistance sp0t welder with air operated cylinder. capable oi developing at least 2,556 pounds of electroce force. is capable of delivering Asgarxus Equipment Required. that ahc \ 't;\. .~\ transformer required weld current at the electrode tips for the speCiiic machine control Mocel 273 Dullers Current Analyzer or equivalent for measuring weld Strain Gauge the 2.i.l 2.1.2 -\ smtable melding secondary Circuit and throat opening. 2.l.3 .-\ suitable welding control With phase shilt .ieat Type 55 timer lunCtions. ’ 2.2 tiectrOdes ii 4 MA Class ll Zirconium copper l'ti'vi :1 materials. sizc 3 Lj/D‘irlo o.d.) and Size 3 i7, S—in o.d.) closed Shanks with ”A" nose or bailnose ::.::s cressed to the specii'ied lne oallnose cap when used must be specxlied in paragraph 2.3.5. The eleCtrooes require a cooling water flow rate of tip contaCt diameter for the test material. dressed to a truncated cone wit‘. a 45 degree angle utilizing the tip dresser 0.5 gal/min With cooling tubes set INc-lflCii from ir.side surface of electroce ShanK (Figure l). instruments 39C4-3ix’ and Uaytronic \iooel 327-3 or equivalent for measuring 2.3 2.3.1 cycles and secondary current. 2.3.2 Lebow Transducer, Model Conditioner/indicator, electrode force. brooxs flow meter, Model .\0. l358bClblCES or equivalent to measure water flow through each electrooe. 2.3.3 128 APPENDIX A - 2 Che»rclet-Portiac-Canaca Group Sheet 2 of 15 MER‘Efai Motors Corporation August 3, 1984 Advance-d Manufacturing Engineering ‘Rev. 1 (7/18/85) MOS-2147 SPECIFICATIC\S .-\.\D PROCEDLRE FOR DETERMINING THE ' “w LLLJAEILIT ‘1 JF BQO\ STEEL MATERiALS -1;oarst'.51cont'd.) 2.3.4 0131 Caliper - \Zitutoyo .‘tiodei 505629 or equivalent to measure weld nugget diameters. T\) \e‘ .5 Eiectrooe Tip Dresser - Tipalcy \o. T-361AC-B or equivalent. 't.ateriai Requirements A minimum of 36 square feet of test material is required. Material must be clean, rust-free and flat. J» eid Lobe 1nterpretatio”. .—\ weldability lobe is a means of graphically expressing the numerous combinations of weld current and weld time which will produce satisfactory welds with a specific set of conditions. ‘11 eid current values are plotted on the "X" axis and weld time values on the "‘1" 3X15 lFigure 5). The left most boundary of the lobe defines the combinations of w eid current and weld time which will produce minimum acceptable nugget diameters. Points to the left of this curve will produce nugget diameters less than the minimum speCified. The right most boundary of the lobe defines the flashing pomt. Points to the right of this curve results in weldments with excessive heat energy evidenced by heavy expulsion, eleCtrooe indentation, and/or electroce sticking. Points between the two curves result in welds of satisfaCtory quality. The distance between these two boundaries at a given weld time is referred to as the "weld range." A wide weld range is very deSLrauie and indicative that the material would be tolerant to changes of the manufacturing process variables such as tip geometry and voltage fluctuations. For convenient reference there is a third curve (shown as a dashed line) which indicates the currents and weld times which will proauce a nominal diameter weld nugget. A point on this line may be seleCted for a nominal weld schedule. Procedure 5.1 beleCting of Process ConStants 5.1.1 Tip Contact Diameter Measure the test material thickness accurately with a micrometer and insert the thickness value in the formula (Table 1) to determine the tip electrode contact diameter required. 5.1.2 Nominal Nugget Diameter From the formula (Table I) the nominal nugget diameter is equal to 0.86 times the tip contact diameter. 5.1.3 Minimum Nugget Diameter by formula (Table l) the minimum nugget diameter is equal to 0.69 times the tip contaCt diameter. 5.1.4 Force The electrOde tip force is calculated by the formula shown in Table 1. 129 APPENDIX A - 3 Charmer-Par.mac-Canada Croup Sheet 3 of 15 Generai '..0tors Corporation August 3, 1984 Advanced Manufacturing Engineering “Rev. 1 (7/18/85) MUS-2’47 \j‘ e \lt e 59"‘IF1C.~\TIO.\S AND PROCEQLRE FOR DETERMINING THE A h- \9 5.1.6 5.1.8 Setup of \ 5.4..i 5.2.2 5.2.3 at ELDAoILIT \ CF 501.1) SIEEL \m‘iERIALS weld Time The weld time requirements are obtained from the formula shown in Table 11. Obverning Material when spot welding the test material to mild steel, determine the force reqUired 1from formula) for each material and use the weld schedule that is commensurate with the lower electrode force. w eid Looes Required ReSiStance spot weld tubes are required for the following conditions: \11 elding the test material to itself. Welding the test material to a nominal 6.59mm (0.635 inch) mild steel 1Fb\.Slo-5E or F'bftiS lei-51;). welding the ten material to a nominal 1.9lmm 10.075 inch) mild Steel 1Fb.\.s l6-5E or Fb.\'.S 16-5U). Minimum Spot Spacmg Select minimum spot spacmg for the metal thickness and welding combinations as shown in Table 111 Equipment Electrode Tip Tuning Reduce the electrode tip force sufficiently to allow a file (parallel in thiCKness, smooth or dead smooth, double cut) between tips to be rotated through a plane perpendicular to the center line of electrooes. This operation will create parallelism between the electrooe faces and adjust the tip diameters. Check contact diameter by measuring the carbon imprint per Section 5.2.4. Tip Alignment Check by carbon imprint (Section 5.2.4). Both eleCtrOdes should indicate the required tip diameter if in alignment. The metal panel is removed from between the carbon to determine alignment. Tip Dressing If the tip diameter is larger than required, use the electrode tip dresser at low electrode force to dress the tips to the proper diameter by rotating the dressing tool between electrode tips. Measure carbon imprint to verify correct diameter. 130 APPENDIX A - 4 t:hevroiet-Pontiac-Canada group Sheet '4 of 15 uenerai ~\.10tors Corporation August 3, 1964 Advanced Manufacturing Engineering *Rev. 1 (7118/85) MUS-247 SPECIFICATIONS AND PROOEDLRE FOR DETERMINING THE w ELSAEJILI". \ OF BODY STEEL MATERIALS 5. Procedure ’cont'd.) 5.2.4 Carbon 1mprint Place a sheet of carbon paper over a sheet of plain paper with carbon Side down. Fold the combination in the center over a metal panel. Place this combination between the eleCtrodes. Make the carbon imprint at the elECtrode force required for the weld test (Figure 6). The carbon imprint indicates the eleCtrode contact area. \ieasure to verify the proper tip diameter. 5.3 Sequence of Tests 5.3.1 Conditioning weius After the e1ectrooes are aligned, tuned and dressed, one hundred conditioning spot welds are made at a weld schedule that would proouce a nominal Size weld nugget at the nominal weld time (per formula Table 1). This stabilizes weld results and reduces the scatter of ten points. For galvanize coated metal the weld current should be adjusted from some lower heat value that does not result in sticking, graduaily increasmg the heat after several welds to obtain a brassing condition on the test coupons. 5.3.2 11010 Time SensitiVity Test weld tests are initially conduCted on the test material to determine whether an) embrittlement of the weld zone occurs due to the rapid quenching (60 cycles hold time). Hold time sensitive materials exhibits Shear-type weld failures when peel tested. At five cycles of hold time, spot weld together three sets of coupons (Figure 14A) at the maximum weld time determined (per formula Table 11) for the test material. Peel test the second weld Per Figure 46 to determine the heat setting required to obtain the minimum weld nugget diameter (per formula Table 1). Next, read)ust the hold time to 60 cycles. At the same heat setting that proouced the minimum size weld nugget at five cycles of hold time, spot weld together three more sets of coupons and examine the weld zone. If shear type weld failures occur, the test material is considered to be hold time sensitive. Slightly higher heat settings (3 percent maximum increase) may be used at 60 cycles hold time to obtain the proper nugget size. if good full size minimum nuggets are developed, the material is considered not to be bold time sensitive. Chevrolet-Pen tiac-Canada Group Cenerai .‘viotors Corporation 131 APPENDIX A - 5 Sheet 5 of 15 August 3, 1984 Advanced .\ianufaCturing Engineering *Rev. 1 (7/18/85) MOS-247 SPECIFICATIONS AND PROCEDLRE FOR DETERMINING THE w ELDAEILIT) OF DODY STEEL MATERIALS \A e Procedure 1cont'd‘.) 5.3.3 ‘4 eld Lobe Generation 5.3.3.1 5.3.3.3 5.3.3.4 A minimum Nugget Diameter Boundary with the weld time set at maximum per formula, adiust the current to obtain the minimum nugget size per formula and peel test three coupons to verify (Figure 5, Point 1). Repeat this procedure for Points 2, 3 and 4, for intermediate, nominal and minimum weld times, respeCtively (Table 11). Connect these points to establish the minimum weld nugget boundary of the weld lope. .\ominal Nugget Diameter Curve with the weld time reset to maximum as above, adiust the weld current to obtain the nominal nugget Size (per formula) and peel test three sets of welded samples to verify the nominal weld nugget sizes (Figure 5, Point 5). Repeat this procedure for points 6, 7 and S of the weld lobe. Adjusting the weld times as per 5.3.3.1. Connect these points with a dashed line to establish the nominal nugget diameter curve of the lobe. Tensile Test and Chisel Samples Tensile and chisel samples are made at the nominal nugget diameter and only at the nominal weld time. Five tensile samples are made (Figure 2) first, followed by five chisel samples. The sequence of spot weld placement on the chisel samples are A, B and C per Figure 3. These semiples are required to verify the weld quality at the nominal schedule (SeCtion 6.6). Expulsion Boundary To obtain Pomt 9 (Figure 5) increase the weld current slowly at maximum weld time to obtain a flashing condition on the second spot weld of the peel test coupon. Make only one weld sample at the flashing condition to determine this point. Repeat this procedure to obtain Points 10, 11 and 12 of the weld lobe. Adjusting the weld times as Per 5.3.3.1. ConneCt the points to obtain the expulsion boundary of the weld lobe. Connect all upper and lower points to complete the weld lobe. 132 APPENDIX A - 6 Chevrolet-Pontiac-Canada Group Sheet 6 of 15 General .Motors Corporation August 3, 1984 Advanced Manufacturing Engineering *Rev. 1 (7/18/85) MISS-247 SPECIFICATIONS AND PROCEDL’RE FOR DETERMINING THE 1 ELJAEIL1T\ OF BOD‘t STEEL MATERIALS Procedure (cont'd.) 5.3.4 Procedure Variations Due to material processing, carbon content, nitrogen and alloying material differences, variations in procedure may be required. Difficulties may arise in obtaining the desired points on the weld lobe. it may be necessary to increase the mimimum weld time by several cycles above that specified per formula to obtain weld nuggets. Different interpretations of the high side of the weld lobe may be necessary because of the material characteristics. Flashing may occur prematurely and would indicate a narrow weld range. However, a moderate amount of flashing may be tolerable and acceptable. In this case. excessive flashing and/or sticking of the electrode to the test material may be taken as the high side of the lobe. For galvanized steel, heavy brassing may be the criteria for determining the high side of the lobe if flashing is not obtained before excessive indentation. .Mimimum nugget diameters may be difficult to obtain. The weld nuggets formed may be oval. For this situation the average diameter should be used. The average diameter may be obtained by measuring the major and minor diameter and computing the arithmetic average. Reasons for Rejection 6.1 6.2 6.3 6.4 Hold Time Sensitive Material that is found to be hold time senSitive to itself is re)ected and terminates any further teSting. brittle Welds If during peel tests, brittle or erratic weld nuggets are obtained, the material is rejected and terminates further testing. Enough data shall be taken to substantiate this condition. Laminated Base Metal If during peel tests of the spot welds, lamination of the base material is observed, the material is rejected. Narrow V4 eld Range Minimum weld range requirements at the nominal weld times are listed below for the various metal thicknesses: 2000 amperes for 0.041 inch and larger. 1800 amperes for thicknesses less than 0.041 inch. Materials with less than the minimum weld range requirements for the specific metal thicknesses is caused for rejection. 133 APPENDIX A - 7 Chevroiet-Pontiac-Cahada Group Sheet 7 of 15 General Motors Corporation August 3, 1984 AC\3T‘1CBC Manufacturing Engineering *Rev. 1 (7/18/85) MES-247 SPECIFICATIONS AND PROCEDLRE FOR DETERMINING THE 11 ELDAEILIT) OF bODt STEEL MATERIALS Reasons for Rigecuon (cont'd.) (I e 6.5 w eld Lope POSition on Graph weld iooes that do not overlap sufficiently with those previously developed for approved sources may also be rejected. 6.6 Chisel TeSt Samples The material mav also be rejected if two or more of the 15 sp0t welds on the Chisel samples snear at the interface. LSSwider:k(0621Cl 134 APPENDIX A - 8 Chevrolet-Pontiac-Canada Group Sheet 8 of 15 General Motors Corporation August 3, 1984 Advanced Manufacturing Engineering ’Rev. 1 (7/18/85) MOS-247 SPECIFICATIONS AND PROCEDLRE FOR DETERMINING THE w ELJABILITV OF BOD) STEEL MATERIALS TABLE I PROCESS CC‘NST’ANTS FOR WELD/ABILITY LOBE DEVELOPMENT t = ‘iietal thickness (inches). ys : \ ield strength of steel (K51). d = Contact tip diameter (inches). A = Contact tip area (square inches). n : \ominai nugget diamEter (itic:.esi‘. m : J..ir.imun: nugget diameter (inches). P = w elding pressure (psi). F = \\ eld force (pounds). , = d2 4 .007 1.65 O = 1.65! -.CC7 n = .86 1.65t - .007 = .860 m = .69 1.65t - .007 = .690 A = 1.296t -.00555 P = 60) s . 10200 F -.- AP NOTE: For all materials less than 0.04l-incti thick, use the following process constants: 500 pounds Force 0.25" Tip Contact Diameter 0.20" Nominal Nugget Diameter 0.16" Minimum Nugget Diameter 135 APPENDIX A - 9 Chevrolet-Pentiac-Canada Group Sheet 9 of 15 General Motors Corporation August 3, 1984 Advanced Manufacturing Engineering 'Rev. 1 (7/18/85) \«iDS-247 SPECIFICATIONS AND PROCEDL'RE FOR DETERMINING THE w ELDAEILITV OF EODY STEEL MATERIALS TABLE II SINGLE PL' LSE bare Galvanize \.1h. "41.)". : .LS9F 41.2 \lin. “.T. = ."09F + 6.2 Noni/4.1. :.SI3F +1.9 Nom.\X.T. =.013F+6.9 1nt.w.T. :.317F .2,5 lnt.\4.T. :.017F +7.5 Max/1431. : .0221: + 3.1 Max. \11.T. : .OZZF + 8.1 DL'AL PL'LSE - 4 CV. COOL (For Thicmesses Above .089”) Eare Galvanize Min. “.1. : 3371: + 0.9 .\iin. 1LT. = .007F + 5.9 Nom. 11.1. = 0091‘ . 1.2 Nom. “.T. = .CC9F + 6.2 \iax.‘w.T. :.011F. 1.5 \iax.\4.1. :.011Fo6.5 bare Schedule - bare to bare and bare to Galvanize W hen bare is Governing. Galvanize Schedule - Galvanize to Galvanize and bare to Galvanize when Galvanize is Governing. Also one side Zinc at Interface. Dual Pulse - .089-ln. thick and over. NOT E: Governing material is determined by the force required, whichever is lower. NOTE: In general the lower force determined by formulas for materials would result in being the governing material. However, for all materials less than 0.041-ln. thick—use 500 pounds. 136 APPENDIX A - 10 Chevrolet-Pontiac-Canada Group Sheet l0 of 15 General .Motors Corporation August 3, 1984 Advanced NianufaCturing Engineering ’Rev. 1 (7/18/85) MUS-247 SPECIFICATIONS AND PROCEOLRE FOR DETERMINIING THE 4 ELDABILIT) OF BOD‘t STEEL MATERIALS TABLE III Required Spot Spacing -- Peel Test Coupons Thickness \iinimum Spot Spacing -- Inches lr-cnes Bare/Bare Galvanize1Bare Galvanize/Galvahize .03; 172 518 374 .335 5/8 3/4 7/8 .041 3, 4 7/8 1 ”347 7/8 I I-II'S .35 1-1/8 1-1/4 1-3/8 .367 1-1;4 1-378 l-l/2 .075 1-3/8 1-1/2 1-3/4 .689 1-578 1-314 2 .135 1—3/4 2 2-1/4 .123 2 2-1/9 I-I/Z NC‘TLt’ l400-pounds force or above--use size 3 shanks and electr0de caps (.\'.\i\ Z 6146). Metal tniCknesses above 0.089 in. requires dual pulse weld schedules. Tensile test coupon size for metal thicknesses above 0.089 in. should be 2" x 6". ’ For eleCtrooe tip contact diameters above 0.290 ih.«use 3.le 60:6 Caps—Reworkedu45 degree taper. 137 APPENDIX A - 11 Chevrolet-Pontiac-Canada Group Sheet 11 of 15 General .Motors Corporation August 3, 1984 Advanced Manufacturing Engineering *Rev. 1 (7/18/85) MOS-247 SPECIFICATIONS AND PROCEDLRE FOR DETERMINING THE 11 ELDABILIT) CF BODY STEEL MATERIALS 1/8"- IN)" V‘ 0 Q‘\\ ‘3 I ) NOTE: Top of water deflection tube to be . I do? at 45-degree angle. 1 I ‘\'~\ 3:) - {L \\\ \x. 1 03- 3.“ \ \ :\\\ .51: FIGLRE l - Correct W ater deflection Tube Installation for Closed Shank EleCtrode 138 APPENDIX A - 12 Chevrolet-Pontiac-Canada Group Sheet 12 of 15 General Motors Corporation August 3, 1984 Advanced Manufacturing Engineering ‘Rev. 1 (7/18/85) MUS-247 SPECIFICATIONS AND PROCEDURE FOR DETERMINING THE \4 ELDABILITY OF BODY STEEL MATERIALS /——- DIRECTION OF GRAIN 1“)“ gnu THICKNESS OVER AP L) LEV-T" ( A B .035"-.088" 1-1/2 INCHES 1.3-E I. .089"-.120" 2 INCHES 6 I\:-E SPOT \‘ ELDED COUPONS FOR TENSILE TEST FIGURE 2 DRIVE CHISEL BETWEEN PIECES UNTIL WELD NUGGETS ARE INDICATED I IR ECTION OF GRAIN Coupon Size - 1-1/2 x 5 INCHES SPOT WELDED COUPONS - CHISEL TESTED FIGURE 3 it) Hi In 139 APPENDIX A - 13 Chevrolet-Pontiac-Canada Group Sheet 13 of 15 General Motors Corporation August 3, 1984 Advanced Manufacturing Engineering 'Rev. 1 (7/18/85) MOS-247 SPECIFICATIONS AND PROCEDURE FOR DETERMINING THE )1 ELOABILITN OFBOD\ STEEL MATERIALS A II k II 5 Ll-r-SPSCTHED-W 0., \ / 1 SPOT sncnic T 4 snow [31 ‘u—c/ DIRECTION OF GRA.N SPOT WELDED COUPONS FOR PEEL TEST FIGURE 4A VICE GRIPS SECOND SPOT WELD PEEL FORCE IRST SPOT WELD / V a BENCH VlSE g % SPOT WELDED COUPONS - PEEL TESTED FIGURE 48 140 APPENDIX A - 14 Chevrolet-Pontiac-Canada Group Sheet 1’4 of I5 General Motors Corporation August 3, 191% Advanced Manufacwring Engineering ‘Rev. I (7/18/85) MOS-2&7 SPECIFICATIONS AND PROCEDLRE FOR DETERMINING THE \I ELDABILIT Y OF BOD\ STEEL MATERIALS -——NOM I NAL NUGGET CUR OE _J L, 5/ / MAX. CURRENT LINE 1 PROOUCES EXPULS'CN U 6 1 ON BOTH FIRST AND “I \ seco~o coupon wsws '3‘ - 7 11 - :3 -v- - — - — — — —- -- NOMINAL WELD TIME "' .i a :3 MIN. CURRENT ;' 9 LINE (PRODUCES 3 " MIN. NUGGETSI 12 —1 “'i V II I I I I I CURRENT (K I LO AMPS) TYPICAL VELD LOSE FIGURE 5 141 APPENDIX A - 15 Chevrolet-Pontiac-Canaoa Group Sheet IS of IS GeneraI Motors Corporation August 3. I98a onanced ManuIaCturing Engineering *Rev. I (7/18/85) MOS-2’47 SPECIFICATIONS AINjD PROCEDURE FOR DETERMINING THE \L ELDABILIT\ OF BODY STEEL MATERIALS h r: r: - . FOLDED PAPER UPP-R EL-CIROOE CARBON STEEL PANEL fiLOWER ELECTRODE ELECTRODE CAR EN IMPRINT FIGURE 6 APPENDIX 8 Mathematical Solution to the Common Area of Winking circles 142 APPENDIX B - 1 Mathematical Solution to the Common Area of Winking Circles The contact area of misaligned spot weld electrodes can be represented by "winking" circles, as shown in Figure 8.1. The common shaded area A(m), is of interest in determining current density and pressure density in spot welding. The assumptions of this area approximation are: 1) Aligned electrodes are co-axial, flat and perpendicularly faced. 2) Misaligned electrodes are only misaligned in a radial direction, as in Figure 8.2. Therefore the electrodes axes are still parallel, although they are no longer co-axial. 3) Misalignment, m, is defined as the distance between the axes. 4) Radius, R, is defined as 1/2 the face diameter of one of the electrodes. 5) A(m) is the contact area A, at misalignment m. 6) III-(1;) is the fractional area A, at fractional misalignment m (the fractional terms are dimensionless). Consider the quadrant area A(m)/4, as in Figure 8.3. The area of the quadrant at misalignment m, may be determined using rectangular elements as shown in Figure 8.4. The solution follows as shown in Equations 8.1 through 8.13. A plot of Equation 8.13 is shown in Figure 8.5. From the graph we see that the percent lost area due to misalignment is always greater than the percent 143 APPENDIX B - 2 + 2H m R - radius of (flat-faced) electrode m - displacement of electrode centers A(m) - common electrode area of mlsallgned electrodes Figure 8.1 Model of "winking" circles. 144 APPENDIX B - 3 (L Electrode Figure 8.2 Workpiece R l Workplace Electrode m - 0.100 Inches. (2.54 mm) Q 2R - 0.250 Inches. (6.36 mm) m R - 40% misalignment N In the "winking" model of electrode misalignment, the electrodes are allowed to shift out of alignment by distance m, while the axes remain parallel. 145 APPENDIX B - 4 Figure 8.3 The quarter area A(m)/4 is 1/4 of A(m), the contact area at misalignment m. 146 APPENDIX B - 5 > ’- Figure 8.4 The summation of rectangular elements by calculus. 147 APPENDIX B - 6 100% % REMAINING AREA, A(m) \j\ 80% \ \\ 60% \ 40% \ 20% \‘ 0% \ 0% 20% 40% 60% 80% 100% MISALIGNMENT (% of diameter), fin" Figure 8.5 The remaining contact area of misaligned electrodes. 148 APPENDIX B - 7 misalignment, (% Lost Area = 100 - % Remaining Area). For example, at 40% misalignment the lost contact area is 50%. For electrodes 0.25 inches (6.35 mm) in diameter, this represents a misalignment of 0.10 inches (2.5 mm). This scenario is used in Section 4-4 of the main report. 149 APPENDIX B - 8 A x(max) -== I h(x) dx [ 8. 4 O x(max) A = 4-J. h(x) dx [ 8. 0 where m h = 2 2 - - [ 8. R - x 2 and 2 x(max) = 2 [m] [ 8. R - _ 2 x(max) 2 2 x(max) Am = 2 x ‘[ 1R - x dx - 2 nil! dx 0 0 [ 8. x(max) 2 2 2 x Am = 2 x-JR - x + 2-R -asin - - 2 m-x [ 8. R 0 150 APPENDIX B-9 + 2-R -asin 2 Am = 2‘R -asin [B.7] [8.8] [13.9] 151 APPENDIX B - 10 The partial misalignment, m , is defined: m m a — , m = 2-R-m [3.10] 2-R And the remaining area in dimensionless terms, is defined: _ _ A(r—n) A(m) = ---- [3.11] A(O) Substituting, I- I' '1- 2 2 R - (R-i'i') asin 2 L R .. A(m) = 2 R 2 L I°R 4 2 _.2 __ R. ~ (R m) -2-R-m- [ 3.12 ] 2 w-R which simplifies to: __ 2 [ I __2 _| _2] [3.13] .A(m) = -- asi 1 - m - m- 1 - m w APPENDIX C Rewards and Benefits from Implementation of Findings in General Motors Plants 152 APPENDIX C - 1 Rewards and Benefits from Implementation of Findings in General Motors Plants The following pages contain letters, reports, and excepts from meeting minutes regarding the implementation of weld process control that followed from this research. The program was implemented in three phases at the BOC Lansing Fabrication Plant of General Motors. The three phases were: PHASE I - Get Documents in Order - Get Inspection in Order - Get Repair Operations in Order PHASE II - Fix Processes PHASE III Keep them Fixed Although these phases were identified specifically for this implementation program, they were intentionally labeled generically, so as to encourage their use in process areas other than welding. 153 APPENDIX C " 2 /‘ The Lansing Fab FeetureLine Sept. /Oct. 1990 I Buy American — :-‘ r a" JVerification , J 5.: 5"" W . takes on an M, i increasingly important role I L Weld Article inside '11) our ‘ 5mm 5mm personal staring in :54 Llfidiflk-fim, ‘ uur Hams and tfioug/iu ‘1 art witfi you I 154 APPENDIX C - 3 Weld Verification Update Information providai by Dave Kelly In Sept of 1989 the weld verification team started to verify approximately 1503 weld guns at Plant 1.2 and 3. This was done with the cooperation of our production. trades. engineers and management working togethcrto meet GM corporate standard that are required. Weld verification and welder mainte- nance are on gomg jobs. From the time that a welder is verified. it needs to be monitored to be sure that all ofthe weld parameters are maintained. The welder maintenance per- sonnel go through the welder and check the airpressure sethateach weld gun hasomlb of force. They take current readings so that they can set proper Starting weld schedules. All other aspects of the welding process are checked outto be sure that they are operating properly. (water flow, gun operation. weld Up alignment, etc) After all of the thing above are accom- plished each vm'iable needs to be monitored to make sure they are maintained at the proper setting. Process monitoring is currently visu- ally monitonng all the welders in Lansing through check sheets. The check sheets carry all the weld schedules andair pressure. The monitorde go to awelder and visually check the water flow indicators. they will check the weld controllers to make sure that that proper weld and stepper schedules me being maintained. We check with QC to see if all welds are acceptable and check air pressure gages for proper PSI. If during any of the checks. the monitor finds that any variable has been changed they document the findings on the maintenance log. “Ric log is than given to the floor supervisor to report to area maintenance so that they can correct the situation or find out the reason for the change. Each week the maintenance log is reviewed. All documentation from the monitor is kept on file for one year. On the cover Katherine Miller. C/H Rail Offline #3 welder operator talks with Skip Collins and Sheri Dubber. Process Monitoring. Otherkey figures that deal with the cm line maintenance are Mike Miller. setupoman. Lynn Marier. jobsetter. Dale Napier and Jim Sonday. die tryout and Jim Stewart. supervisor. 155 APPENDIX C - 4 MARCH 1991 Special pir‘asis: THE PUR ‘ IT OF ‘ WELD 0UA\ITY \ s V.» How Cadillac Won the Malcolm Baldnge Quality Award 57 PUBLISHED BY THE AMERICAN WELDING SOCIETY TO ADVANCE THE SCIENCE, TECHNOLOGY AND APPLICATION OF WELDING WELDING JOURNAL @ APPENDIX C - MalcoIm 3 \latior Idli V I — MM 0 Cu- WIW vbm Camera) Motors no president and general may at the Came Motor Car Dmvon Is greeted by secretary I Harem Salome \UIW Cruelty heard; terms or Lima Ruben 4 “0th al the ’Wi PRACTICAL WILD!“ Weld Quality Helps Cadillac Win the Malcolm Baldrige Award sir-00m malice spot welding played a key W role in the recent Maker" Devon-Pg 1 Forthe l990award more than 167(1)) companies requested appu- cations out only 97 convicted the rigor satin/NC»; mam {orator the wear, Iomul ous procme established by the US Department or Commerce Or the 97 onlvstxmadeittothemals Fourwinners Dive-on. BM Corp -see Table million resistance spot welds are made every day In the Detrort—Hamtramdr R9 1900 or them are applied roboticallv The baunce, tor the most part, are applied us- Ing hard automation. Philip Borowslu. weld verification co— ordinator. said the mean average oI good 97 8“. am; the Wall WELDNG IOURNAI. I 57 F1 1- In much?! on at 0:15 4! Cache ) 091rar+lamrramct mmpunr ”r... Elsewhere In the plant the exhaust sys- he Tyne .sh needed in weldng roots to quarter panels Theltiermetalusedthereissicon bronze In ada'non to the Detroe-l-lamtramclt plant the Baldnge eumners visited Ca- Phfip wendent olman- utacturng at the Detron- Hamtramclr plant. said The ex w crested in or: use ot statistical process control in rests- tance Thespot welrhg.’ Theexarnners wanted to make sue that or. statistical diedrs were stable and .‘added Al Titley advanced mm- SIIMARCH 1991 157 APPENDIX C - ”"666”,...— , utactmng engineer. Carmllac Motor Car Division, Troy. Much "The product had to be stmcturally and on tar- get The process is what they were most interested In The examiners were also Impressed the number of people trorn different tis- ciphes. both blue and white colar work- ether. who were Involved in the Ins I08 control ol weld quaky lnodeot .tehe eermw easily control experts who were not Iamtiar withw The body steels ot the wtasomobles sembled at the Detrort-Hamtramck As- steel The steels are generally SAE 1003 and 1010 mid steel Of course, It ismore drllicult to weld these types of galv anized steel than It Is to weld bare steel of the oat: ethev are worthitn termsolrehbfiy OltheI90reststancespotweliiigro- 6 s In use at the Detrort+lamtramck ro- bots. Their repeataqu Is rated at :0. 02 he welding line there are two Robogates IFig 2) which Initially weld the Irames E ch R e con~ taIns ten robotically controlled resistance W "S In termsoI' weI ' equipmenta not omrnent investment at the Detrort- robot- unrts are also helptng to Improve weld quarry on the Cadillac auto bodies With Its ability to localIze high ener maintenance cost of this Installation Caril- esthe (act that the trans or mgr-frequency DC gun permits It to pro— wrt rthout gram weld schedrlestn namperes gong through al the time-consuming steps requiredo to do the same using conven- tional controls 0t: statisth are based on destructiv Ive test ing.' said Boro wslti 'That Is the most reliable way to check whethery ouvha ve good or In In terms ofW resist ahc destructive test plays a critically Important olein quaity contr olat Cadillac It is the results from these tests that tprovtde the data for statistical process control vvithn test Is also a good means of checking for broken or stuck welds. one of six catego- nes oi "discrepant welds.’ Four other cwategories eoil-location welds omitted welds and bum-through; can be detected vtsualy An undersIzed weld. the sixth category ot discrepant soot welding operations are stil within their Ilmfls whether they are trendmg L9 or down Each year the DetrOIH-lamrrama plant I!) ll)? man-hours In chisel checks welaIng or In modm-ng existing QQJIO‘ merit tor new model runs great deal or money tor welding at Cdt’lllldC rs spenton the mechanics or the sembtvline tortooltngandmiunngand clamp-n8 A Iot ol plant Input goes Into con- stanttv updating tool design standards ' saTId One or the very key things Is a tormal weld tool venhcation procemre At the construction phase tooling goes througha tormal checkout toseeII Itdoes Indeed represent design Intent That pro- cess wi‘l catch design errors and construc- tion errors betore they ever reach the t Cadilacmadentothelnablorthe 1989 Bamge award To apply tor the award, corrvanies must suomu ‘3 page application whnd'i is actually a denied examnation at what thec convanies are al about The seven sections n the apphcation lichde leader- stun, mtorrnanoriva platen; man resouces quality assuance resdts The overseer or the award Is the Na- tional lristitue or Standardsand Technol- ogy Catt ,AnMd Madrilstrator oI the 1990 award was the American 50- oetv tor QaabtvCont rol. Wwautee The 1990boardolexanliersconsistedolrlie Detron-Hdiitrmidi Ar next monnis-AWS Ited to dé people, For tow ntorrrianon. cal (m 4434353 0 158 APPENDIX C - 7 MMWWMAw-dm Year “‘68 W inner Motorola Inc smitten s Co L‘adutac Victor Car Div-snort Cover at Motor sorC IBM Lora Rochester DII Is-ori Vl. allace federal Express Corp Manulacturmg Vlanutactunng small Business Vlanulactunns Vlanulartunng Manuiactung Manutactunng Small BusIness Ice f; J-nrherea-dowmesuarCadar. “ solancesporwets. 159 APPENDIX C - 8 WELD TOOL VERIFICATION U N S GOOD COPPER (BIDDLE) MACHINE LABELED HANG STATION/GUN/HIT/SCR CHARTS AIR GAGES ALL IN GOOD SHAPE GUNS ALL IN ACCEPTABLE CONDITION GUNS ALIGNED, WELD LOCATIONS CORRECT COOLING WATER ACCEPTABLE (INFRARED) ITEM NUMBERS OF CAPS, SHANKS, ADAPTERS, SHUNTS WTIS - RECORD ALL WELD SCHEDULES/SCR'S BY GUN NUMBER GUN QUALIFICATION STEPS TEST CURRENT BALANCE BETWEEN SIMULTANEOUS WELDS. (GOAL: ALL WELDS WITHIN 10% OF EACH OTHER) TEST FOR MAXIMUM OUTPUT AT 81% HEAT FOR (5) CYCLES ON TOP OF A PREVIOUS WELD. IF MAX CURRENT COMPARES WELL WITH DESIGNED MAX. CURRENT (OR RED BOOK CURRENT RECOMMENDATION), THEN SET-UP WELD SCHEDULE ACCORDING TO RED BOOK RECOMMENDATIONS AT 20% LESS CURRENT. IF CURRENT OUTPUT IS LESS THAN DESIGNED (OR RED BOOK) THEN USE PULSATION (2-3 PULSES). SET WELD FORCE: 600# FOR LIGHT GAGE METAL; 900# IF GMT GREATER THAN .055". WHEN FINISHED - GET FINAL PART OK FROM WELD INSPECTION. NUGGET SIZES RECORDED AND SIGNED STATEMENT THAT ALL WELDS MEET SPECIFICATIONS. CLEAN UP WTIS AND POST PREVENTIVE MAINTENANCE SET-UP PREVENTIVE MAINTENANCE SCHEDULE AND GET SHEETS MADE UP. SET—UP PROCESS MONITORING WITH PRODUCTION APPROVAL. Subject: From: To: 160 APPENDIX C - 9 GM wold Information Group January 23, 1989 (GM-HIS) January Mtg H. L. Brinkmann Sad Attached Distribution Tho January mootino took placo at Hodar, Inc in Farmington Hills, M! on Thursday, January 18, 1989. Attondanco indicatod on tho attachod sign-in shoot. Following are the dates and locations of future nootingo: Fobruary 15, 1990 - 8830 AH Kirkhof-Flox-Cablo Troy Hilton Contact: Jack Goodrich (313)689-4666 March 15, 1990 - 6:30 AM 7 - Naod a volunteer April 19, 1990 - 8:30 A" ? - Mood a voluntoor Highlights of tho January flouting as follows: LANSING (DOC) FAB WELD VERIFICATION Hike Karaqoulia of 80C Lansing explained their wold tool verification, dun qualification, and preventive oaintonance plane. The capability of tho proceed to reduce discrepancies was dooonatratad with a ”before and after“ story for an actual part. A niftv part of Hika‘a presentation was a Dara-ater chart of "Rocoooondod Valuoa for Phaaa It Hold Tool Verification”. This chart nay provido the "operational“ paraaotarc aontionod earlier in thapa ninutoa. A neat feature is to show the roduction in required current for different tip coobinationci for axaaplo two ball nos. cape taka approxiaataly .3 X tho current shown on the chart; on. ball noaa and one flat cap taka approxioataly .73 x the current chown; otc.. Thorn wad ao ouch good stuff in Hika'a presentation that it was aqraod to attach his presentation oat-rial - including tha aforooontionod chart - to thaoa oinutop. 161 APPENDIX C - 10 25541 582 WELD QUALITY C/H SHOCK rowan - RH (from 09/01 - 09/30) III [III III I I I I .1 - I I I -d-d--i—- I-’ III- -II I--- III ‘1'! I" I--- III. IIIIII‘I II- III'III- ’Iil 'Ill'-|' II. III- III -1-_.——‘--r-r- III U--- I.’ I -L-L-a—J-J-J—-L-L-L-J-J-A--A—-L-u-t-a_a-.4--l.-|.-L-a-.l I l I I ‘P‘?’ -r I I I III II‘IUIIO III Il'lll-‘ III III... V'V'W’W“r'r'f'7‘1'fi’-I l . I I I I I I I D I I 0 I L—‘—J-J-—I—-h-Labn‘—J-J—J— I I I i I I I I 22323232323232323232323232 2 GROSS AVG = 0.42 DAY SHIFT, 3=AFTERNOON SHIFT 2: pm<¢_oawu.ozII . g I I I l I I 4 l I J I I l I 1 I I 4 I E ; IIILIIIIrIIIvI ' 3 I - -y- I I I I I I -L-L- I I I:;t ....... 2 I I I I I l I I I I E o p- .[-a. -4- I l l I I I -L B-L-L-J-J-J-- I I I I I III- III I‘-I 32 2=DAYSHFI 3=AFH¥WKXNHSHWT 5 0 2:2. .2. mu.oz mmmOOmm>m¥ - 330.on 1.: I. an . < .. 168 APPENDIX C - 17 November I. 1989 To: Distribution From: Jim Rypkema Subject: Measured Improvements Weld-Fab Shocktower Lines 09-01-89 to 09-30-89 / 10-01-89 to 10-31-89 As a result of weld verification process and ongoing scheduled maintenance for both North & South Shocktower Lines we have achieved the following results in the last 60 days: QUALITY: Based on documentation from destruct tests for failed welds on a daily shift baSlS: September 1989 October 1989 Total Failed Total Failed welds-All Samples Welds-All Samples R/H Shocktower 81 29 L/H Shocktower 68 48 Analzsis: R/H Side saw a reduction in failed welds of -52. L/H Side saw a reduction in failed welds of ~20. This represents a 64% improvement in the FVH Side. and a 29% improvement in the L/H Side. UPTIHE: The actual increase in parts run as a result of this maintenance and weld verification is as follows: September 1989 October 1989 +/- Parts Per Hour Parts Per Hour % R/H Side 207.7 221.8 +14.1 +6.7% L/H Side 204.9 222.7 +17.8 +8.7% The actual increase in capacity was 4.512 more R/H parts. and 5.696 L/H parts. By producing more parts per hour in same allotted time. we are decreasing the cost per part and increasing the net profit for Weld-FaDII As weld verification continues. and scheduled maintenance continues to accomplish their work. this major gain in performance will continue. Congratulations to both maintenance (skilled trades) and ooerat' s for a job well done! Operations Supervisor Cate SUDIOCI To 169 APPENDIX C - 18 mm OrganIzamn I.’ v H g I ChevaeI-Ponuec Census GIOUD fl January 22. 1990 Held Tool Verification and Process Monitoring at Lansing Fab C-P-C Quality Directors - AsI & Fab C-P-C Weld Coordinators - AsI & Fab I would like to share with you and your plant soIe excellent work done on the above subject by Lansing Fab of B-O-C. Hike Karagoulis. welding Engineer. presented this inforIation at the GM Weld Informa- tion Group (w.I.G.) Ieeting on 1/18/90. The 9 page report is attached. This work is significant because it shows what can be done and typical quality gains that can be achieved by identifying and controlling the Iain process variables in welding. This report is based on the Lansing Fab‘s analysis of 1.100 weld guns. It may very well reflect those key process variables at your plant. The second and third graphs of discrepancies illustrate the typical quality inroveIent achieved by controlling the 4 Iain variables based upon deforIation testing. Hike c0IIented that the only areas which didn’t have coIparable inroveIents were those where short- cuts were Iade in the process. This process doesn’t require a large expenditure for equipIent. but it does require a coIIitIent of people. At Lansing Fab it took some reassignIent of people and an overtiIe coIIitIent to get the job done. The payoff seeIs to be in long terI benefits: quality improvement. less scrap/repairs and less productive down tine. It looks like a worthwhile investIent. Richard J. Lipinski C-P-C Operations Quality cc: Hike Karagoulis R. Heithaus S. wechsler Heaoauv'e's General Motors Comomvon JOOOIvsn OyIIeAvenue wenen MIc-.gar 48690 9020 APPENDIX D A Technical Summary of the Resistance Spot Welding Process 170 APPENDIX D - 1 A Technical summary of the Resistance Spot Welding Process Production resistance spot welding involves simultaneously: 1) a physical process, 2) an electrical process 3) a metallurgical process 4) a cultural manufacturing environment. D.1) The Physical Process In the spot weld process, two or more pieces of sheet metal are clamped together by copper electrodes as in Figure 0.1. The working contact diameter of the electrodes is often 0.25 inches (6.35 mm) for thin—sheet automotive steel (0.024-0.120 inches [0.6-3.0 mm] thick). Clamping force is generally 400-1300 pounds (174-565 kg). Electrical current, of the order of 2x105 amps/in2 is passed between the electrodes. The current causes intense heat build-up inside the clamped volume of metal. The heat produced results from Joulean (resistance) heating. In a short time (.1 - 1.0 seconds), melting occurs at the weld interface and the current may be stopped. After the current is stopped, solidification occurs very rapidly because the sheets are :relatively thin, the electrode force is high enough to provide good thermal coupling, and‘the copper electrodes are water cooled. If the surface temperature of the weld is 171 APPENDIX D - 2 WATER-COOLED COPPER ALLOY ELECTRODE \ ( q BASE METAL BASE METAL S 4.2—J L’V-vw WELD NUGGET ’\ j WATER-COOLED COPPER ALLOY ELECTRODE Figure D.l Schematic of resistance spot welding (Ref 62) . 172 APPENDIX D - 3 maintained below a critical "sticking" temperature, then electrode sticking will not occur and the welded metal may be easily unclamped. The electrodes are then re-useable for thousands of welds before being replaced or re-dressed. Surface temperature is mainly controlled by cooling water, force, current, and weld time. The main wear mechanisms for the electrodes are: mushrooming (a gradual increase in the tip area due to either warm or cold plastic flow of the copper), alloying, and contamination by oil or paint residue. D.2) The Electrical Process The governing equation for joulean heat release in a spot weld is Joule's Law: heat [watts] current [amps] resistance [ohms] ”+43: "II“ The resistance of a spot weld is within the range of 50 - 200 micro ohms. As far as electrical power loads are concerned, a spot weld is considered a low resistance load. According to the joulean equation above, this condition necessitates high current for adequate resistance heating. A typical setup for 1mm sheet steel might be 10,000 amps and 173 APPENDIX D - 4 1 volt at the weld. It is necessary in the design and maintenance of the welding circuit to keep the open-circuit resistance low, so as not to cause heat build-up in the hardware. Large cables with silver plated connections are recommended for stable, low resistance circuits operating under high current. The waveform used for resistance heating is generally not critical, so AC power is most often used for economic reasons. Low voltage, high amperage power is delivered to the weld through a step-down transformer. Typical turns-ratios for welding transformers are in the range of 20:1 through 200:1. A constant-current circuit is becoming increasingly desired in industry, rather than constant-voltage or constant-wattage circuits, so that the energy lost due to circuit variation such as cable wear will not alter the current or resulting voltage drop across the weld. The root mean square (RMS) of the current waveform is the most measurable heat input parameter. RMS current is adequately controlled during welding by precisely timing a point during each half-cycle of power when current is allowed to begin flowing. Thus, in resistance welding there is a brief period of off-time at the beginning of each half-cycle. Therefore although the line power is nominally sinusoidal, the weld does not receive a sinusoidal waveform (see Figure D.2). The off-time is proportional to the 174 APPENDIX D - 5 primary power 440v I (1 fire pulses I Iow heoi transformer primary fire pulses Inc reused transformer heat serti ng primary secondary ADVANCE - FIRE ANGLE - RETARDED <— DECREASE - FIRE ANGLE - INCREASE ——> INCREASE - HEAT - DECREASE Figure D.2 Waveforms in resistance welding (Ref 63). 175 APPENDIX D - 6 timing of the firing circuit. The off-time throttles back the system to a desired RMS current. Maximum RMS current occurs with no off-time; this condition of full power is referred to as 99% current or 99% heat (depending on which current programming convention is used). Pulsing signals control RMS current in modern welding equipment by switching on silicon controlled rectifiers (SCR's). The SCR's operate in pairs on the primary (input) side of the transformer. Upon receiving a pulse the SCR is "gated" (turned on) for whatever time remains in the half-cycle. Current automatically stops when the power alternates to the opposite polarity. A second SCR is then similarly gated to pass current in the next (opposite) half-cycle (see Figure 0.3). The SCR's receive their timing pulses from a microprocessor-based weld control called a weld timer. The primary function of the weld timer is to fire weld schedules to the SCRs. A weld schedule is essentially a set of pulses which will cause the SCRs to fire for a programmed length of time at a programmed level of RMS current. Weld time is programmed in cycles of the standard 60 Hz waveform (.0166 seconds), or the standard 50 Hz waveform (.0200 seconds). Current may be programmed in either percent heat, percent current, or in secondary amps. Therefore, typical weld schedules might read: 176 APPENDIX D - 7 mwmme parallel SCR's T @ hOO' L control ‘ \ fél ~\fire pulses Figure 0.3 Typical single phase SCR wiring diagram for resistance welding (Ref 63). 177 APPENDIX D - 8 WELD 1S CYCLES at 50% HEAT 0r. WELD 15 CYCLES at 71% CURRENT 0r: WELD 1S CYCLES at 9000 AMPS % HEAT and % CURRENT are related by the equation: 2 %H= 9.: [D.2] % CURRENT and CURRENT are related by the equation: I %I=__ [D.3] I(max) where Imax is the current at full power. Over a number of welds the required current for welding increases because of tip mushrooming. This swelling of the electrode face actually causes the weld diameter to decrease, since current density decreases. Therefore, in order to maintain a certain weld diameter, the control must be programmed to compensate by stepping up the current after a predetermined number of welds. By increasing the current, current density remains nearly constant throughout the life of the electrodes. Electrode wear is by far the main source of variation in the entire spot weld process. However, once good weld schedules and stepper schedules are developed, there is no longer any loss of process efficiency due to electrode wear since no discrepant welds are produced. The electrodes are simply replaced at the required interval and the stepper counter reset to zero welds. A negative aspect of using a 178 APPENDIX D - 9 current stepper is that power consumption increases as the current is raised. Even so, the energy required to produce a spot weld is relatively small, roughly only .0005 kw-hr, according to the calculation: E = I-V-t [ D.4 ] E = 10 1 0.00005 (kilo-amp volt hr) [ D.5 ] E = 0.0005 kw-hr [ D.6 ] The next major source of process variation is in the resistance of the welding cables and the inductive losses of the secondary loop. Wear in the cables or electrical connections of the secondary loop causes a direct loss in weld current, which in turn will reduce weld size from the initial setup. This resistance may be measured with a sensitive ohmmeter (in micro-ohms) as a preventive maintenance practice, or monitored through the weld control during normal production welding. Variation in inductive loss is due to a change in secondary loop area or the amount of magnetic material in the loop. Induction and resistance losses may together impede current flow to the point that the system output is inadequate for welding. In such a case, the required current may be attainable by either: an increase in transformer size, a decrease in loop area, a decrease in 179 APPENDIX D - 10 loop magnetism, or a decrease in loop resistance. A remedy specifically for excessive inductive losses is the use of direct current rather than alternating current. D.3) The Metallurgical Process There is a wealth of interesting metallurgical phenomena associated with resistance spot welding. According to basic principles in materials science, welding may be defined as a metallurgical joining of metals in intimate contact. It is common knowledge among materials scientists that atomistically clean metallic surfaces bond to one another readily upon contact in order to reduce surface energy. Using this as a first principle the in analysis of welding, every metallic welding process may be rationalized fundamentally as a two step process, where: 1) the surfaces are cleaned, and 2) brought into intimate contact. In the case of spot welding these two steps are fulfilled by resistance heating to the point of melting. Some of the events known to occur are (in relative chronological order): 1) Cold plastic flow of surface asperities during initial clamping. 2) Surface cleaning by electrical heating and pressure extrusion. 3) Warm plastic flow of contacting surface asperities. 180 APPENDIX D - 11 4) Phase transformations in the bulk volume upon rapid heating. 5) Dissolution and stirring of remaining surface impurities upon fusion (melting) at the weld interface. 6) Intense heat flow towards copper electrodes. 7) Phase transformations upon rapid cooling of the melt. D.4) The Impact of the Cultural Manufacturing Environment Operating a controlled process gives predictable results. Given that one of the goals of manufacturing is return on investment, then process control is important because it contributes to controlled, predictable profits. Predictability is crucial whenever competitiveness and market share are at stake. Let us first establish this discussion within a framework of basic definitions. PIOCOSI: A process is a series of manufacturing operations which result in a definable, useful product. Product: The product may either be a saleable end product or an intermediate product used in a downstream process. All useful products contribute eventually to a final marketable product. It is the sale to a customer of the final product which generates business resources and profits. Customer: A customer is any downstream user of the product. 181 APPENDIX D - 12 Customer Requirements: The customer's specifications and expectations make up the "customer requirements". Rey Variable: An input variable which affects the output to a significant degree. Minor Variable: An input variable which does not affect the output to a significant degree. Remote Variable: A variable which is outside the span of control of the employees at the process. Process Control: A devised method of keeping output characteristics of a process within desired limits, achieved by maintaining key inputs within a set of limits. In a controlled process the outputs are nearly predicted by the key inputs. However, a margin of uncertainty still exists due to the impact of minor variables which are not controlled, and due to statistical variation. The level of understanding necessary for process control must include both technical and human variables. A manufacturing process cannot be singularly controlled from either the technical or the human point of view. If so, the technical "laboratory" approach may neglect true plant conditions and the human "plant" approach may be seriously lacking in technical integrity. Because true processes are often socio-technical, both factors must be understood in order to be controlled. A controlled laboratory environment may provide the technical facts about materials processing interactions, but the laboratory must at some point include 182 APPENDIX D - 13 the plant environment in order to comprehend the "real world". Key variables in a laboratory environment are not necessarily key variables in the plant. In addition, an important advantage of extending the laboratory into the plant environment is the possibility of further developing the process. The goal is to alter the set of key variables to a configuration that is socially practical and economically controllable. Some of the symptoms of a manufacturing process out of control are: product rework, sorting of incoming and outgoing stock, and excessive reliance upon product inspection to ensure quality. It is vital to competitive manufacturing that all critical processes are controlled, in order that costly practices may be minimized or eliminated. The purpose of describing the cultural manufacturing environment is to underscore that for a process to be controlled it must be sufficiently understood. The problem for many manufacturing processes is that there exist both technical and social variables. The feasibility of process control in any manufacturing process necessitates the following nine requirements: 1) To control a process, all key variables must be known, readily measurable, and controllable. The process should be developed to the point of having as few key variables as possible, since each key variable is to be monitored and controlled. 2) The process must be simply understood in terms of basic concepts by those who are given the job of monitoring and controlling it. 3) 4) 5) 6) 7) 8) 9) 183 APPENDIX D - 14 The process must be properly defined to include both human and technical key variables. The setup should be developed and refined to desensitize any remote variables. Examples of remote variables in spot welding might be material, metal fit, coatings, and line voltage. It is not feasible to completely control a process having key variables which are remote. An remote key variable can only be tolerated if its variation has been minimized. However, once a key variable has insignificant variation it is arguably no longer a key variable. A key variable under control might be re-classified as a minor variable, once a reliable trend has been thoroughly established. Preventive maintenance items are the best kind of key variables. This is because they may be adequately controlled during periods of scheduled downtime, using a planned inventory of spare parts. Variation in preventive maintenance may be minimized through procedures, training, supervision, and management support. The process manager must execute a plan to monitor and control each key variable at its optimum value. The healthiest mode of operation (both technically and socially) is to continually minimize variation at every step in the process. After periods of continuous improvement, it may be possible to demote more key variables to minor variable status. Having fewer key variables simplifies process monitoring requirements. APPENDIX E Elements of Process Control in Manufacturing 184 APPENDIX E - 1 Elements of Process Control in Manufacturing The major goal of this study was to understand the resistance spot welding process well enough to control it in the socio-technical environment of the automotive factory. It was necessary to consider in some detail all perceived sources of significant variation in order to distill reality from the sea of perceptions. Having done this, the results help justify the technical validity of the findings. All of the academic questions have not been answered on the subject, but the level of understanding has been sufficiently raised for good, economical process control. From a business standpoint the project has been a success in that the problem was properly defined and solved, the results implemented, and the benefits realized and measured. Therefore since the welding problems which triggered and sustained this study have been solved, it is not economically necessary at this time to pursue more in-depth academic knowledge of this process. Rather, it will probably take a few years' time to allow the implementation to catch up with the level of knowledge currently available. To list the generic elements that have made this process control plan a success, an effective plan must have: 1) Simplicity. Can the plan be understood by the users? technically accurate approach will fail if the key employees cannot understand it in their own terms. For example, in this plan the employees were taught to A 2) 3) 4) 185 APPENDIX E - 2 expect some visible weld expulsion, and to question whenever there wasn't any. Technical Integrity. Is the understanding accurate? It really doesn't matter how practical conclusions are if they are technically unsound. A house built on sand will not stand (64). Cultural Appropriateness. Can people really perform as required? In this plan the key variables were all shown to be maintainable process parameters. In support of the journeyman labor establishment present in our plant, the extra maintenance required to support this plan was divided equitably between the various trades. For example, the electricians were given responsibility for cable and current maintenance, the pipefitters were given cooling water maintenance, and machine repairmen were given the gun cylinders etc. In addition, preventive maintenance groups were established in order to proactively keep equipment running properly. Consequently, the social needs of the key employees were met (eg. ownership, pride and teamwork), while the technical needs of the process were also well maintained. Economic Feasibility. What is the cost of not controlling the process? In this project there were clear economic incentives to support process control. Some incentives were: downtime, quality and rework, not to mention the public liability of poor welds on automobiles. See Appendix C for a presentation of economic benefit from this program. APPENDIX P Conclusion About Applied Research Strategies 186 APPENDIX F - 1 Conclusion About Applied Research Strategies In this study, it was crucial to invest the proper amount of time and effort working directly with those who stood to benefit from the work. When doing applied research, frequent personal contact is always recommended for at least two reasons: 1) It is normally advisable to get a good first-hand look at an open-form problem before attempting to solve it. This helps ensure that the problem has been properly defined. Whenever there are layers of people between the problem and the problem solver, there are significant opportunities for critical information to become overlooked, misinterpreted or misrepresented. Once the facts and assumptions about a problem are misunderstood the risk level for the researcher rises rapidly. Few things are more aggravating in the business of applied research than solving the wrong problem! 2) When conducting research in the plant environment, keep the information flowing both ways. Teamwork and teambuilding are essential in the midst of a vast manufacturing complex, since the researcher cannot possibly be personally present all of the time. VITA Michael John Karagoulis was born in Detroit, Michigan in 1957. Attending the University of Michigan, he graduated as Bachelor of Science in Materials and Metallurgical Engineering in 1979, and Master of Science in Metallurgical Engineering in 1980. Since 1981 he has worked at the Buick- Oldsmobile- Cadillac Group of General Motors Corporation, in a variety of materials-related engineering areas, notably eight years in the field of resistance welding. He is known in the welding community for his research and development in the resistance seam welding of polymer-coated terne plate for automotive fuel tanks, having been credited for pioneering this process, which has been widely used throughout General Motors since 1984. He has written two major papers on this subject and presented them at the American Welding Society. He has presented work on non-destructive testing of spot welds, using both ultrasonic and the resistivity methods. He is also credited within General Motors with a highly successful application of molybdenum alloy inserts for projection welding electrodes, which gave a 50-fold increase in electrode life over conventional copper electrodes. To date, his most personally rewarding contribution to the welding field is this present work on spot welding process control. The results summarized in this dissertation have contributed significantly to General Motors through the elimination of reworked welds and redundant waste. This work has shown that discrepant weld rates of 0.00% to 0.02% are economically achievable and controllable under production conditions. This quality level is more than 2 orders of magnitude better than previously considered possible. Consequently, engineers are now driven to design comparable vehicles using fewer welds and more reliable welding processes. By taking extra care to monitor and control the four key variables revealed by this study, General Motors is becoming more competitive in its core automotive business. To date, portions of this present study have been presented at the TMS fall meeting, the American Welding Society, and within General Motors.