Damage Models

Damage models predict damage accumulations based on pavement responses inputs. Typically this is achieved by providing estimate of the incremental damage caused by certain number of applications of a given critical pavement response. In addition, each damage model needs to specify how damages caused by various load levels are combined over time.

These damage models are typically developed based on laboratory test data. Damage can be broadly understood as physical changes, usually material degradation, in pavement layers caused by either traffic loading or environmental loading. Examples of damage observed in laboratory tests include stiffness reduction in beam fatigue tests (AASHTO T 321), or accumulation of permanent axial strain in repeated load tri-axial test using AMPT (AASHTO T 378).

Damage models need to be calibrated using data collected from well-controlled full scale pavement testings. Examples of these include but are not limited to various accelerated pavement testings. This calibration is necessary because there are significant differences in factors such as boundary condition, strain/stress state, and strain/stress path between laboratory tests and full scale pavement testing.

It is preferable that the data used to calibrate damage models come from small sections that are practically homogeneous in terms of material properties, layer thicknesses, trafficking history, and environment conditions. Any significant variations in these factors make it difficult to fully characterize the test section and account for the variabilities properly.

This part of the M-E method has both mechanistic (the damage model parameters determined using laboratory test data) and empirical (the calibration) parts.