Record Details
Multivariable model for optimizing the percussive welding of aluminum
ScholarsArchive at Oregon State University
| Field | Value |
|---|---|
| Title | Multivariable model for optimizing the percussive welding of aluminum |
| Names |
Comstock, James Martin
(creator) Stone, Solon A. (advisor) |
| Date Issued | 1964-08-07 (iso8601) |
| Note | Graduation date: 1965 |
| Abstract | A multivariable model describing the system response (weld tensile strength of 6061-T6 aluminum) of a percussive stud welding system as a function of five natural (external) system variables was developed. The five natural system variables were 1) static weight of the drop-shaft (W), 2) initial stud base height (H₀), 3) capacitance (C), 4) equivalent d-c circuit resistance (R), and 5) initial capacitor voltage (V₀). In order to develop the multivariable model, a Single Factor experiment was used to bring the system response to the neighborhood of a maximum. The maximum observed tensile strength was 2231 ± 18 pounds. This value was seven percent higher than the nominal tensile strength rating for the 6061-T6 aluminum stud material and seven percent less than the "cross-grain" tensile strength of the 6061-T6 work-plate material, To describe the effects of the independent variables (W, H₀, C, R, and V₀) and their interactions upon the dependent variable (tensile strength), a complete Factorial experiment was performed. An Analysis of Variance showed that (in order of importance) V₀, V₀², RV₀, W, R, CRV₀, H, WV₀, H₀CRV₀, WH₀ CV₀ and WC contributed significantly to the observed variability of the system response. In order to describe the response surface by a working multivariable model, a weighted Multiple Regression Analysis of the Factorial data was performed, The result of the Regression Analysis was a quadratic model relating the weld tensile strength of 6061-T6 aluminum (y[subscript t][superscript (pred)]) to the five standardized natural variables of the welding system. The quadratic model was y[subscript t][superscript (pred)] = -1483.1 + (446.7x₁ - 57.6x²₁) + (501.3x₂ - 71.2x₂²) + (200.2x₃ - 7.2x²₃) - (508.5x₄ + 2.9x²₄) + (2096x₅ - 456.5x²₅) + (- 5.5x₁x₂ + 5.6x₁x₃ + 9.7x₁x₄ - 64.3x₁x₅ - 7.6x₂x₃ - 38,1x₂x₄ - 20.9x₂x₅ - 3.6x₃x₄ - 60.1x₃x₅ + 242.2x₄x₅) Examinations of metallurgical photomicrographs of weld specimens suggested that the multivariable model would have to be restricted to specific regions of the response surface since voids, intergranular cracking, and transgranular cracking was observed in the weld zones, Comparisons of observed and predicted tensile strengths, using the quadratic model, suggested that future multivariable models might very well have to include the higher order interactions to achieve a better fit to the observed response surface. |
| Genre | Thesis/Dissertation |
| Topic | Aluminum -- Welding |
| Identifier | http://hdl.handle.net/1957/48468 |
