Fatigue cracking is a primary asphalt pavement distress and various models have been developed to predict the fatigue life of asphalt mixtures using laboratory tests. One such model is the simplified viscoelastic continuum damage (S-VECD) model, which has been implemented in the pavement performance prediction program, FlexPAVE^TM. The S-VECD model test protocols (AASHTO TP 133 and AASHTO T 400) and data processing tool (FlexMAT^TM) are widely used around the world. Over the past three decades, this model has been continuously improved and refined. However, questions remain on the model’s ability to predict the material response when stress or strain are used as the model input. Also, different fitting procedures for the model calibration were found to affect the model’s prediction accuracy. In this study, analysis was conducted using data for four typical North Carolina mixes based on single replicate tests and multiple replicate tests to compare the model’s prediction accuracy based on stress versus strain as the model input and by considering fitting errors in the different calculations. The results show that using strain as the model input, which automatically incorporates portions of permanent strain, yields more accurate predictions compared to using stress as the input, regardless of the fitting algorithm. Additionally, in the analysis of individual test data, which is not affected by replicate specimen variability, the model’s predictions match the measured data well, as long as the fitting errors of the damage characteristic curve are controlled. When the data from replicate tests are analyzed together, although specimen variability compromises the S-VECD model’s prediction accuracy, failure can still be reasonably determined when strain is used as the model input.

疲劳开裂是沥青路面的主要破坏因素，已经开发出各种模型来使用实验室测试来预测沥青混合料的疲劳寿命。一种这样的模型是简化的粘弹性连续损伤 (S-VECD) 模型，该模型已在路面性能预测程序 FlexPAVE ^TM中实施。S-VECD 模型测试协议（AASHTO TP 133 和 AASHTO T 400）和数据处理工具（FlexMAT ^TM) 在世界范围内广泛使用。三十多年来，这一模式不断得到完善和完善。但是，当使用应力或应变作为模型输入时，模型预测材料响应的能力仍然存在问题。此外，还发现模型校准的不同拟合程序会影响模型的预测准确性。在本研究中，使用基于单次重复测试和多次重复测试的四种典型北卡罗来纳州混合物的数据进行分析，以比较模型基于应力与应变作为模型输入的预测准确性，并考虑不同计算中的拟合误差。结果表明，使用应变作为模型输入，自动包含部分永久应变，与使用压力作为输入相比，无论拟合算法如何，都会产生更准确的预测。此外，在不受重复试样变异性影响的单个测试数据的分析中，模型的预测与测量数据很好地匹配，只要损伤特征曲线的拟合误差得到控制。当对重复试验的数据进行分析时，虽然试样变异性会影响 S-VECD 模型的预测精度，但当使用应变作为模型输入时，仍然可以合理地确定失效。只要控制损伤特性曲线的拟合误差即可。当对重复试验的数据进行分析时，虽然试样变异性会影响 S-VECD 模型的预测精度，但当使用应变作为模型输入时，仍然可以合理地确定失效。只要控制损伤特性曲线的拟合误差即可。当对重复试验的数据进行分析时，虽然试样变异性会影响 S-VECD 模型的预测精度，但当使用应变作为模型输入时，仍然可以合理地确定失效。