Prediction and Optimization of Sandblasting Process for Mar M247 Nickel-based Superalloy Based on Regression Analysis
The work aims to establish a high-precision model of the multivariate sandblasting process for Mar M247 Ni-based superalloy used as gas turbine blade and used it to design the process combination that can realize surface quality of high roughness and low fraction of retained grit.The sandblasting specimens were prepared by an automated precision controlled sandblasting system made in China and then the surface quality was examined by roughness detector and scanning electron microscopy and the relationship between the sandblasting processing and the surface quality was established based on the mathematical statistics theory.With the sandblasting angle increasing from 30° to 90°,both the roughness and fraction of retained grit increased and the increasing trend was more pronounced at low angle range(30°-60°)as compared to that at high angle range(60°-90°).With the pressure of 0.3 MPa and the sandblasting gun moving velocity of 5 mm/s as example,as the sandblasting angle increased from 30° to 90°,the fraction of retained grit increased from 12.84%to 24.31%,the average size of retained grit increased from(32.10±0.47)μm to(60.77±0.50)μm,and the roughness increased from(5.88±0.48)μm to(7.80±0.41)μm.With sandblasting gun moving velocity increasing from 1 mm/s to 20 mm/s,the fraction of retained grit showed a trend of firstly increasing and then decreasing.In contrast,the roughness decreased firstly at low gun moving velocity,then increased with further increase in gun moving velocity,and again decreased at high gun moving velocity.With the pressure of 0.3 MPa and the sandblasting angle of 60° as an example,as the sandblasting gun moving velocity increased from 1 mm/s to 20 mm/s,the fraction of retained grit firstly increased from 20.70%to 23.54%,and then decreased to 21.94%,the average size of retained grit increased from(51.75±0.26)μm to(58.85±0.45)μm and then decreased to(54.85±36)μm.In the meantime,the roughness firstly increased from(7.55±0.24)μm to(9.11±0.52)μm and then decreased to(8.72±0.32)μm.For the prediction precision of roughness,the prediction error of linear regression model was 5.97%,while that of polynomial regression model was 5.15%.For the prediction precision of the fraction of retained grit,the prediction error of linear regression model was 5.76%,while that of polynomial regression model was 3.08%.Based on existing data,the polynomial and linear regression models can be used for prediction.By comparing the linear regression model with the polynomial regression model,a higher prediction precision is obtained for the polynomial regression model.By analyzing the data predicted by polynomial regression model,the ideal processing parameter window is the sandblasting pressure range of 0.2-0.3 MPa,sandblasting angel range of 45°-60° and sandblasting gun moving velocity range of 15-20 mm/s.Under the selected conditions,the high roughness and low fraction of retained grit are resulted,which assists in forming a high bonding strength of interface between coating and substrate.
万方数据
维普
