**CODEN:** GTJOAD

Format |
Pages |
Price |
||

PDF Version | 4 | $25 | ADD TO CART |

Soil variability biases both the best-fit-by-eye and linear regression of *p-q* triaxial test data plots (where *p* represents the abscissa of the center of a Mohr circle and *q* is the radius) to substantially over-estimate the friction angle ø. This is because linear regression of *q* on *p* assumes that variability lies in the *y*-axis direction, whereas if lateral stress σ_{3}' is constant, variability is 45° to the *y*-axis. A correct regression is quickly performed by converting *p-q* data points to polar coor-dinates, adding 45° or another angle indicated by the mean stress paths to the coordinate angle, converting back to rectangular coordinates, and performing the linear regression. The correlation coefficient thus obtained is lowered because it correctly assesses data variability. Without correction the error in ø may be as much as 10°.

**Author Information:**

Handy, RL *Professor of civil engineering and director of the Geotechnical Research LaboratoryMember of ASTM, Iowa State University, Ames, Iowa*

**Stock #:** GTJ10789J

**ISSN:** 0149-6115

**DOI:** 10.1520/GTJ10789J

ASTM International is a member of CrossRef.

AuthorTitle Linearizing Triaxial Test Failure Envelopes

Symposium , 0000-00-00

Committee D18