CIRCLY 4:
Overview of improvement to damage calculation at layer interfaces
[First released in Version 4.0i (17 July 2001)].
Note:
If you are using an alternative CIRCLY User Interface (such as the RTA
Flexible Pavement Design (FPD) system) or you are using the CIRCLY engine
directly, you WILL NOT encounter the interface accuracy problems described
here.
Background
Until this version was released, previous releases of CIRCLY Version 4 (and all releases of Version 3) did not use the exact z-value when calculating critical strains and damage factors at a layer interface. Instead 0.1 mm was added or subtracted to the z-value. At the time this method of calculation was first introduced it was believed that this would give results that were within engineering tolerances of the values at the interface.
It has recently come to our attention that in some circumstances (usually only associated with the top layer), this method can give rise to numerical errors.
CIRCLY 4.0i now uses the exact z-values corresponding to layer interfaces, ensuring that results are more accurate.
As a consequence, critical strains and damage factors calculated by CIRCLY 4.0i (or later) will in many cases be different from values calculated by earlier releases of Version 4 (and all releases of Version 3).
Our testing to date shows that significantly different values will be obtained for relatively thin asphalt surface layers.
The gross numerical instability (see Example 1 below) that was displayed for a very thick surface layer could lead to the Automatic Thickness Design capability giving spurious thicknesses.
Examples
These examples have been chosen to give existing CIRCLY users an indication of the order of the differences given by the new version.
Example 1: Asphalt on Unbound Granular
The pavement structure is (variable thickness of) Asphalt (Modulus=3000 MPa) on 400 mm Unbound Granular (max. Modulus = 250 MPa) on Subgrade (Modulus=50 MPa). Traffic = 1e6 ESAs.
|
Asphalt Thickness |
Asphalt Damage Factor |
|
|
Version 4.0i |
earlier versions of |
|
| 20 | 0.207 | 0.159 |
| 25 | 0.423 | 0.357 |
| 30 | 0.763 | 0.743 |
| 35 | 1.14 | 1.13 |
| 40 | 1.36 | 1.34 |
| 50 | 1.35 | 1.32 |
| 60 | 1.07 | 1.05 |
| 80 | 0.580 | 0.555 |
| 100 | 0.274 | 0.262 |
| 150 | 0.0370 | 0.0377 |
| 200 | 0.00703 | .00706 |
| 500 | 9.98 e-6 | 9.95 e-6 |
| 700 | 5.23 e-6 | 4.93 e-6 |
| 1000 | 1.94 e-8 | 2.58 e+6* |
* Note: In previous versions, the change in sign of the exponent demonstrates the gross numerical instability that was displayed for a very thick surface layer (e.g. thickness > 750 mm).
Example 2: Cement-Treated on Subgrade
The pavement structure is (variable thickness of) Cement-Treated (Modulus=3500 MPa) on Subgrade (Modulus=100 MPa). Traffic = 1e6 ESAs.
|
Cement-Treated Thickness |
Cement-Treated Damage Factor |
|
|
Version 4.0i |
earlier versions of |
|
| 100 | 3.85 e7 | 3.59 e7 |
| 150 | 1.03 e5 | 1.05 e5 |
| 200 | 1.01 e3 | 0.98 e3 |
| 250 | 2.55 e1 | 2.51 e1 |
| 300 | 9.87 e-1 | 9.74 e-1 |
| 350 | 5.33 e-2 | 5.27 e-2 |
| 400 | 3.78 e-3 | 3.75 e-3 |
| 450 | 3.38 e-4 | 3.36 e-4 |
| 500 | 3.68 e-5 | 3.65 e-5 |
These differences in the damage factor have very little impact on design thicknesses, for example:
|
Traffic |
Cement-Treated Thickness |
|
|
Version 4.0i |
earlier versions of |
|
| 103 | 173.0 | 173.4 |
| 106 | 299.7 | 299.5 |
Conclusion
This change to the method of calculating damage at layer interfaces generally does not have a significant impact on resulting design thicknesses.
But with earlier versions a spurious thickness could be obtained using the Automatic Thickness Design capability if trial thicknesses showed gross numerical instability (see Example 1 above).
revised October 03, 2005
