What the 80% and 90% Reliability Equations of IRC:37-2018 Imply: Back-Calculated Scatter in Fatigue and Rutting Life, Achieved Reliability of the Code's Worked Examples, and the Weight of Traffic and Strain Errors
- 1Former Officer, Chaitanya Godavari Grameena Bank (CGGB)
Abstract
IRC:37-2018, India's guideline for flexible pavement design, gives two versions of each performance equation, one for 80% and one for 90% reliability, but it does not state the statistical model behind them. I show that if pavement life at a given strain is lognormal with a common slope, the pair of equations identifies the scatter exactly: the standard deviation of log10 life is 1.121 for bituminous fatigue and 1.069 for subgrade rutting. One standard deviation therefore corresponds to a factor of about 13 in life, and the median fatigue life is 27 times the life given by the 90% equation. The 2018 fatigue pair implies almost the same scatter as the 2012 pair (1.121 against 1.120). I derive the reliability actually achieved when the computed strain is below the allowable strain. For the code's worked example II.3 (131 msa, 90% target), the computed strains of 146 and 243 microstrain give reliabilities of 0.907 for fatigue and 0.953 for rutting; for example II.4 they are 0.966 and 0.995. A 20% under-estimate of the critical strain lowers the achieved reliability only from 0.90 to 0.84–0.83, and traffic uncertainty (growth rate s.d. 1.5 points, damage-factor coefficient of variation 30%) moves it from 0.900 to 0.902, because the scatter in material response dominates every other source. All results follow from the published equations under the stated lognormal assumption, which the guideline neither confirms nor contradicts.
Related
- Association Between Admission Calculated Serum Osmolality and New-Onset Atrial Fibrillation in Patients with Non-ST-Segment Elevation Myocardial Infarction Undergoing Percutaneous Coronary Intervention
- Verifying calculated dose to the breast in stereotactic radiotherapy: A practical phantom‐based method
- Identifiability assessment and physics-informed residual machine learning for calculated Ti–Zr–Nb–Ta–X thermodynamics (X = Mo, Hf)
- Achieved systolic blood pressure and cardiovascular outcomes in advanced chronic kidney disease
- Tunable slow-light waveguide achieved by magneto-photonic crystals
BibTeX
@misc{aguru2026what,
title = {{What the 80% and 90% Reliability Equations of IRC:37-2018 Imply: Back-Calculated Scatter in Fatigue and Rutting Life, Achieved Reliability of the Code's Worked Examples, and the Weight of Traffic and Strain Errors}},
author = {Manoj Kumar Aguru},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23582980},
url = {https://zenodo.org/records/23582980},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - What the 80% and 90% Reliability Equations of IRC:37-2018 Imply: Back-Calculated Scatter in Fatigue and Rutting Life, Achieved Reliability of the Code's Worked Examples, and the Weight of Traffic and Strain Errors
AU - Manoj Kumar Aguru
PY - 2026
DA - 2026/10/20
PB - Zenodo
DO - 10.5281/zenodo.23582980
UR - https://zenodo.org/records/23582980
AB - IRC:37-2018, India's guideline for flexible pavement design, gives two versions of each performance equation, one for 80% and one for 90% reliability, but it does not state the statistical model behind them. I show that if pavement life at a given strain is lognormal with a common slope, the pair of equations identifies the scatter exactly: the standard deviation of log10 life is 1.121 for bituminous fatigue and 1.069 for subgrade rutting. One standard deviation therefore corresponds to a factor of about 13 in life, and the median fatigue life is 27 times the life given by the 90% equation. The 2018 fatigue pair implies almost the same scatter as the 2012 pair (1.121 against 1.120). I derive the reliability actually achieved when the computed strain is below the allowable strain. For the code's worked example II.3 (131 msa, 90% target), the computed strains of 146 and 243 microstrain give reliabilities of 0.907 for fatigue and 0.953 for rutting; for example II.4 they are 0.966 and 0.995. A 20% under-estimate of the critical strain lowers the achieved reliability only from 0.90 to 0.84–0.83, and traffic uncertainty (growth rate s.d. 1.5 points, damage-factor coefficient of variation 30%) moves it from 0.900 to 0.902, because the scatter in material response dominates every other source. All results follow from the published equations under the stated lognormal assumption, which the guideline neither confirms nor contradicts.
ER -