|
Per Ullidtz, The Technical University of Denmark
N.F.Coetzee, Dynatest Consulting, Inc.
Offered for presentation at the 1995 TRB session
on Structural Modelling Applications in Pavement Analysis and Design
| Abstract / Introduction / Deflection Uses | Backcalculation | Problems Encountered in Backcalculation |
| Critical Issues in Backcalculation | Summary and Conclusions | References |
Summary and Conclusions |
| I |
The use of backcalculated moduli is essential to the application of mechanistic principles to pavement evaluation.
Backcalculation techniques and software have advanced greatly over the past decade.
In spite of this, many routine problems are still encountered which are dealt with in a variety of more or less satisfactory ways.
However, the critical issues remain since they are the fundamental assumptions of the theoretical models typically used. Possibly the most "correct" approach in the future will involve use of stochastic, non-linear, dynamic finite element analysis applied at the particulate level, which would require the routine availability of massive computing power.
Application of neural networks to backcalculation shows some promise.
|
| II |
Backcalculation programs cannot be verified through theoretical means.
Even the most sophisticated theoretical models like dynamic, three-dimensional visco-elasto-plastic finite element programs are based on simplifications, such as not considering the particulate nature of most pavement materials.
No theoretical model constitutes the "truth", they are all simplified models of reality.
|
| III |
Most pavement materials do not have a modulus, the example of a handful of sand makes that obvious.
The apparent modulus depends on the stress condition, which is influenced by moisture or bitumen content, temperature, loading time, etc.
Trying to verify backcalculation procedures by comparing the moduli derived from the procedure to moduli determined by other means is, therefore, extremely difficult for most pavement materials.
|
IV |
The most promising method for verification of backcalculation procedures appears to be through comparison of stresses and strains predicted by the procedure to values measured in actual pavements.
The measurement of stresses and strains in pavements is very difficult, because the presence of a measuring instrument changes the stress or strain condition.
Nevertheless this appears to be the only solution, and instrumentation of pavements is presently taking place in many parts of the world.
|
| V |
It is important to verify the models used for calculating pavement response, if pavement engineering is to move away from being a craft to being a science.
It should be kept in mind, however, that the pavement response is seldom the final answer, but only an intermediate result used to predict the pavement performance.
It is equally important that the relationships between response and performance be verified, through the use of accelerated full scale testing (preferably on instrumented pavements) and through long term pavement performance studies. |
|
|
