The straightening process is the main cause of residual stresses in the manufacture of rails. It is a non-trivial process with cyclic plastic loads, solid–solid contact and complex geometry, which computational simulation is often complex and time-consuming. In this work, a new methodology was developed by means of a quasi-static modeling instead explicit dynamic. This methodology was proved to be effective and fast. Sixteen cases were simulated, and a C-shaped pattern for longitudinal stress, as seen through the literature, was obtained in the most of them, even with large variations between the main parameters: the yield strength, the tangent modulus and the initial curvature of the rail. The longitudinal normal residual stresses were higher than the transversals ones, as expected. The results obtained by simulations were the basis for the use of a Gaussian process regressor to predict the residual stresses from any initial parameters. This tool confirmed that the parameters that more affect the final state of residual stresses are, in this order, yield strength, tangent modulus and curvature. This is relevant information, since the hardest data to obtain in practice are the initial curvature of rail. Both simulation methodology and the statistical Gaussian process tool could be useful to perform life fatigue analysis in rails, since this needs the initial state of residual stresses to be more reliable.

矫直过程是钢轨制造中残余应力的主要原因。这是一个具有周期塑性载荷，固-固接触和复杂几何形状的非平凡过程，其计算模拟通常是复杂且耗时的。在这项工作中，通过准静态建模而不是显式动态开发了一种新方法。这种方法被证明是有效且快速的。模拟了16种情况，即使在主要参数（屈服强度，切线模量和初始曲率）之间存在较大差异的情况下，大多数情况下也获得了通过文献看到的C形纵向应力图案。铁路。如预期的那样，纵向法向残余应力高于横向法向残余应力。通过模拟获得的结果是使用高斯过程回归器从任何初始参数预测残余应力的基础。该工具证实，依次影响屈服强度，切线模量和曲率的参数更多地影响残余应力的最终状态。这是相关信息，因为实际上最难获得的数据是钢轨的初始曲率。仿真方法和统计高斯过程工具都可用于进行钢轨的疲劳寿命分析，因为这需要残余应力的初始状态才能更可靠。切线模量和曲率。这是相关信息，因为实际上最难获得的数据是钢轨的初始曲率。仿真方法和统计高斯过程工具都可用于进行钢轨的疲劳寿命分析，因为这需要残余应力的初始状态才能更可靠。切线模量和曲率。这是相关信息，因为实际上最难获得的数据是钢轨的初始曲率。仿真方法和统计高斯过程工具都可用于进行钢轨的疲劳寿命分析，因为这需要残余应力的初始状态才能更可靠。