Developing and Improving Asset Deterioration Models
An important function of an asset management system is the ability to predict asset deterioration rates so changes in condition over time can be modeled for use in planning and programming activities. In the absence of data, models can be developed based on expert judgment, but as historical performance trends are established based on actual data, the expert models should be replaced by or calibrated against the real data.
The AASHTO Transportation Asset Management Guide: A Focus on Implementation (2011) introduced the following thought process to help an agency evaluate their deterioration models and determine whether improved data is needed to enhance future forecasts:
- If there is disagreement with the timing for recommending a treatment, what is the difference? Does a difference of one to two years make a substantial difference to the program? This type of difference is typically the result of the program optimization models.
- If there is disagreement with the treatment, are the differences substantial, such as deck repairs versus bridge replacement? These differences are often the result of treatment rules but may indicate that deterioration rates are not correct. The deterioration model parameters may need to change (e.g., change traffic considerations or geographic location) or there may have been some work performed that the model is not aware of.
- If the differences are irreconcilable, the agency may decide to investigate the model setup and analysis further or may conduct research to see how other agencies have resolved similar issues.