Effect of Grinding Conditions of a TC4 Titanium Alloy on its Residual Surface Stresses

Abstract

The grindability of titanium alloys that are classified as hard-to-machine materials is studied in high-speed cylindrical grinding using a cubic boron nitride (CBN) wheel. The investigation is concerned with residual surface stresses, including the construction of its empirical model, orthogonal experiments with a CBN grinding wheel at a speed of 45-150 m/s, and prediction with the back propagation (BP) network. The results of residual surface stress measurements obtained in grinding experiments and simulation analysis for five sets of grinding conditions are compared, it can be seen that the empirical model is partially applicable to a Ti-6Al-4V titanium alloy (TC4) under examined grinding conditions. Generally, the calculation results with the empirical model exhibit a significant deviation from the data of actual measurements in some cases. The BP network possesses the function of complex nonlinear mapping and adaptive learning. So the BP network is adopted to predict the relation between residual surface stresses and three key grinding conditions accurately enough. The accuracy of the network is verified, which lays the foundation for its in practical application.Introduction. Titanium alloys are difficult to grind due to their poor thermal properties and high chemical activity. The primary challenges in grinding titanium alloys are: the high specific energy, the high temperature in grinding area as well as the heavy grinding wheel adhesion and wear. Grinding is a common finishing method for titanium material, in order to obtain a precise surface quality. The surface residual stress is one of the main evaluations indices of the surface quality, which has a high impact on the performance [1, 2] of the workpiece. Lin and Lee [3] studied the effect of tool flank wear on the surface residual stresses of the machined surface. Ee et al. [4] used the finite element method to investigate the impact of sequential cuts, cutting conditions, etc., on the residual stresses induced by orthogonal machining. Vosough et al. [5] investigated the impact of high-pressure water-jet on the surface residual stress and concluded that the high-pressure jet increases the level of residual compression stresses in both cutting and feed directions, and thus the high-pressure water jet-assisted machining of titanium alloys is beneficial. Kang and Ren [6] analyzed the causes of the grinding residual stress of Ti-alloy specimen and optimized the grinding residual stresses of titanium alloy by selecting the reasonable grinding conditions, CBN grinding wheel and using the high-quality grinding fluid with extreme-pressure (EP) agent. Hu and Yuan [7] analyzed the forming mechanism of residual stresses produced in the grinding process and concluded that higher constraints to the deformation in grinding process generate lower residual stresses, while the wheel material can diffuse into the surface layer of the workpiece to increase the compressive residual stresses.

Predictive Performance of Artificial Neural Network and Multiple Linear Regression Models in Predicting Adhesive Bonding Strength of Wood

“…For the present study, it is possible to conclude that the relationship between the bonding strength and process variables may be complex, and thus it may not be captured by MLR. As stated before, ANNs are capable of accurate mapping of complex or nonlinear relationship between the variables [28]. Therefore, findings of the present study are reasonable and compatible with those of earlier investigations.…”

“…ANNs are known as a massively parallel distributed processor that is able to simulate the behavior of the biological neural network [27]. They have a good capability of nonlinear mapping, which allows the network to capture the complex relationship in data structure accurately [28]. The function of ANNs resembles the human brain in two respects: (1) the network acquires the knowledge by a learning procedure, and (2) connection strengths are used to store the acquired knowledge [27].…”

Review on grinding-induced residual stresses in metallic materials

“…Using this method, a typical example was provided by Chen and Rowe [42] for En9 (medium carbon steel, AISI 1055), where the simulated maximum grinding temperature was 247°C and the predicted maximum tensile residual stress was 178 MPa. Additionally, Li and Jia [234] adopted a BP network model to establish the relationship between the grinding-induced residual stresses and the key grinding parameters; at the same time, this method was also utilized to predict the grinding forces and grinding burn of workpiece [234][235][236][237][238]. The accuracy of the BP network analysis results has been verified in grinding of titanium alloy Ti6Al-4V.…”

A mathematical model for the cylindrical grinding responses of 18CrNiMo7-6 alloy steel

“…Wen et al [31] analyzed calculation formula of ultrasonic grinding force and the temperature field of ultrasonic grinding transient moving heat source, and established a thermal-mechanical coupling residual stress prediction model of ultrasonic grinding using finite element method. Li et al [32] suggested an empirical model of workpiece surface residual stress based on the experimental results obtained from high speed cylindrical grinding of titanium alloy. Alauddin et al [33] proposed a grinding force model by conducting grinding tests on K1045 steel with an alumina wheel.…”