A REAL-TIME MONITORING AND CONTROL SYSTEM

FOR RESISTANCE SPOT WELDING

K. Matsuyama, R. Obert#, J-H. Chun*

ABSTRACT

A new monitoring and control algorithm has been developed based on an integral form of an energy balance model to realize a low-cost real time monitoring and control system for resistance spot welding. The system captures welding voltage, welding current, and total plate thickness to calculate the mean temperature of a weld during welding. It predicts both weld diameter for the non-destructive evaluation of weld quality and splash occurrence for improvement of the working environment. After training with two stack welds of equal plates, the system can handle two stack welds of unequal thickness, multi stack welds, and other thickness welds without modification of the program parameters.

KEYWORDS

Resistance spot welding, Prediction of weld diameter, Prediction of splash occurrence, Integral form, Energy balance model, Quality monitoring, Improvement of working environment

INTRODUCTION

Industry belief has it that the occurrence of splash, or expulsion, yields good information on weld melting. Its evaluation has been used as quality assurance in resistance spot welding. Splash, however, causes some deterioration of the working environment and the quality of welds. Making a weld for comparison purposes is also wasteful of expensive energy. Furthermore, maintenance costs are higher than necessary because the metal powders caused by the splash degrade the moving parts of production robotics.

A new procedure addresses the problems inherent in splash during resistance spot welding. It solves deterioration issues based on the idea that splash or expulsion occurs when a weld part is overheated, even shortly, during welding.

The authors analyze the facts of splash with numerical simulations and experiments (Ref. 1). The article demonstrates that splash occurs when the corona bond zone at the faying interface suddenly melts. The article also suggests that continuous monitoring or prediction of weld part temperature, i.e., of the dynamic behavior of the temperature rising pattern, is important in predicting splash caused by overheating.

Temperature patterns and history can be continuously predicted if a monitoring procedure is associated with a prediction system based on a numerical simulation program as an identification routine of the weld part temperature (Ref. 2). The system described in the referenced paper

Mechanical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA 02139 #MS Student, Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA 02139

addresses welding phenomena, current distributions, temperature distributions, and variations over time for each tested weld. In addition to the physical property data of the workpiece, the system only requires the monitoring of welding voltage and current.

The procedure, however, has a very high processing cost when using a standard CPU. The length of the calculation time is 100 to 1000 times that of the actual weld time. The system is a powerful tool for predicting the weld diameter and temperature history of welds if a high speed CPU, i.e., a high speed Digital signal processing, DSP, device with a large amount of memory, is installed on a welding controller. Therefore, the system is comparatively quite expensive and so used only in special cases (Ref. 3).

Calculation times are dictated by the procedure time required to solve the heat conduction differential equation. A finite differential method requiring many mesh points to get accurate temperature patterns is commonly used. The same is true when the finite element method or finite boundary method is employed.

A new integral form of the heat equation is induced for achieving a real time treatment. The new concept calculates the energy balance in a weld part to simplify the relationships among heat input to the weld, heat loss from the weld part, and the temperature rise in the weld.

This paper describes fundamentals of nugget formation process to understand the basic concept, an energy balance model, new governing equations derived from the new concept, configuration of monitoring and control system, and experimental results of the adaptation. The effort has developed a new low cost monitoring/control system to simultaneously end splash occurrences and estimate weld diameters.

NUGGET FORMATION PROCESS IN RESISTANCE SPOT WELDING

Figure 1 illustrates a typical nugget formation process for resistance spot welding (Ref. 4). Similar nugget growth patterns occur regardless of the type of material used as the workpiece (Ref. 2). Figure 1(a-1) to (a-3) illustrates cross sections of welds at different weld times. de is the contact diameter at the electrode-workpiece interface and do the diameter at the faying interface. Nugget diameter and nugget penetration are d2 and pn, respectively. Figure 1(b) shows the nugget growth pattern with contact diameter data at the electrode-workpiece interfaces and at the faying interface. The observation timings with arrows can also be found in Figure 1(b). Broken lines in Figure 1(b) show the molten diameter and molten thickness of the nugget. After the solid and broken lines separate, solidification has begun, although the current conduction continues due to increasing contact diameters during welding.

The contact diameters are inconstant during welding. This variation in diameter has a large influence on the nugget formation process (Ref. 5). Measurement of the contact diameters shows that the contact diameter at the beginning stage of the weld cycle before forming a nugget is smaller than at the end. Furthermore, the contact diameter of tips used for many welds is usually larger than those used with the new electrodes. This change causes the weldable current range to increase. A similar effect causes a shift in the weld lobe for zinc-coated sheets. The initial contact diameters are also larger than those for uncoated sheets in spot welding zinc-coated steels.

Figure 2: Relationship between Nugget and Contact Diameters at Various Setting Current Levels

Changes in the contact diameter have such a large influence because the heat generation density in resistance spot welding is a function of the contact diameter raised to the fourth power. For this reason, the variable contact diameter concept needs to be considered.

Figure 2 illustrates the growth patterns of nugget and contact diameters at various welding current levels. Figure 2(a) shows a pattern in lower current condition than the critical splash current

condition. Figure 2(b), just under the critical splash current condition, and Figure 2(c) in a splashing current condition over the critical current. All figures show at the same electrode force condition.

The difference between the contact diameter and nugget diameter is greatest in a low current condition. This suggests that the mean temperature of a weld, defined as the weld zone bounded by the contact diameter and both interfaces between workpieces and electrode tips, is lowest among these three cases. The mean temperature in the splashing condition is the highest. This suggests that information about the mean temperature of a weld is useful to predict the welding state, including splash occurrence.

MATHEMATICAL CONSIDERATIONS BASED ON AN ENERGY BALANCE MODEL

The mean temperature of a weld can be estimated with both a finite difference method and two sets of monitoring data of the welding current and resistance between tips (Ref. 2). The latter procedure, however, requires a huge investment in numerical calculation. An expensive digital signal processor (DSP) is required to process the data. The problem can be resolved by employing a new algorithm based on an integral form of heat conduction equation.

Basic Equation written in Integral Form

The mean weld temperature is calculated in a target volume defined as a square zone enclosed by the four broken lines, shown in Figure 3, where de is the contact diameter at electrode-workpiece interfaces. The following heat Q is contained in the target volume at a time t after starting current flow:

where is the heat amount contained in the target volume, t is the time, v is the voltage between plate surfaces (measured voltage between tips [drop in tips], induced voltage by inductance), i is the welding current, K. is the heat conductivity in electrode tip, K is the heat conductivity in the