Browse Projects > Detailed View

Universal Stress-Strain Data Base for Standard Classes of Backfill Soil and Levels of Compaction


Finite element analysis (FEA) is now a commonplace method to analyze and design buried culverts and soil bridges using soil-structure programs such as ABAQUS, CANDE, FLAC, PLAXIS and SAP to name but a few. Each FEA program offers a suite of soil-model formulations that can be used to characterize the behavior of the all-important backfill soil ranging from simple linear elasticity to fully nonlinear formulations such as incremental plasticity and variable modulus developments. Depending on the soil-model formulation, the analyst/designer must specify anywhere from two to twenty soil-model parameters in order to characterize a particular soil model. Herein lies the conflict because backfill soil requirements are usually specified in terms of textural composition (e.g., Unified Classification System) and degree of relative compaction (e.g., Standard Proctor Density). However, this information by itself is not sufficient to define the parameters of any soil-model formulation. Rather, a reliable stress-strain data base is required to determine the appropriate soil model parameters.

Unfortunately, there is no universally-accepted or AASHTO-approved stress-strain data base for characterizing various soil models to represent backfill specified by textural classification and percent compaction. Rather, each analyst/designer typically selects soil model parameters based on his or her individual experience and habits. Consequently, two different analyst/designers often come to opposite conclusions with regard to the safety of a particular culvert installation simply because they have chosen different soil-model parameters.


The objective is to produce a universal stress-strain data base for commonly specified backfill soils in culvert installations including CA (course aggregates), SW (gravelly sand), SM (silty sand), SC (silty-clayey sand), ML (sandy silt), and CL (silty clay), with initial compaction levels of 80%, 85%, 90%, 95% and 100% of Standard Proctor density. Standard triaxial stress-strain data (shear softening) shall be obtained for each soil-density combination for at least three confining pressures. In addition, stiffening stress-strain data shall be obtained from uniaxial strain tests or hydrostatic tests, preferably both. A sufficient number of repeated tests shall be performed and/or collected to establish statistically meaningful upper-bound and lower-bound stress-strain curves for each individual condition. Intermediate design curves shall be established that will serve as templates for analyst/designers to determine reasonable parameter values associated with their soil model formulation

This research is not intended to evaluate the efficacy of various soil model formulations or to determine soil model parameters for the various formulations. Rather, the purpose is to provide a reference stress-strain data base for backfill soils that can be used by designer/analysts to determine realistic soil model parameters applicable to the soil model formulations available in their FEA program. This applies to all soil model formulations including linear elastic, Mohr/Coulomb plasticity models, Cam-Clay and Cap-like models, Duncan/Selig variable modulus models and a host of other models published in the literature. This research objective is only applicable to reworked soils used for backfill soil that has been disturbed and mechanically compacted. It is not applicable to undisturbed insitu soils that may have experienced overconsolidation pressure and physical and/or chemical bonding over eons of time.


Research results will provide State DOTs and owners of culvert installations with a method of verifying the veracity of the soil model chosen by designer/analyst and hence have better knowledge of the installation’s safety. This is simply achieved by requesting the designer/analyst to show stress-strain plots of the soil model predictions in comparison with the universal stress-strain data base. Designers and analysts will have a method to determine the best soil model parameters that are consistent with the soil type and compaction level specified for the culvert installation.

Related Research:

To address this problem, James M. Duncan and subsequently Ernest T. Selig conducted extensive laboratory tests on backfill soil samples to determine the model parameters for their hyperbolic Young’s modulus and Bulk modulus functions for several soil classes and compaction levels. Their laboratory stress-strain data provides an excellent starting point for developing a universal stress-strain data base that includes softening stress-strain curves from standard triaxial shear tests as well as stiffening stress-strain curves from uniaxial-strain and hydrostatic testing procedures. Other published stress-strain data bases are also available in the literature associated with other soil model formulations such as Mohr/Coulomb and Cam-Clay plasticity models as well as

other variable modulus models. However, more testing and data gathering is required in order to provide statistically meaningful upper bound, lower bound and recommended stress-strain curves for all relevant soil classifications and compaction levels.

  1. Literature search - Undertake comprehensive search of relevant literature to collect and collate all experimental stress-strain data for backfill soils. Catalog existing data and identify all classes of backfill soil where more laboratory testing is statistically needed.
  2. Data-base organizational plan - Develop an overall plan for organizing, storing, transmitting and presenting the combined new and old stress-strain data base. The plan should include procedures for handling future growth of the data base.
  3. Laboratory testing - Undertake the proposed laboratory testing program using soil samples from diverse geographical locations. Establish laboratory testing protocols to be used by multiple soil-testing laboratories in order to get statistically expected variations.
  4. Final deliverable.Produce the stress-strain data base that is accessible by the engineering community and capable of producing plots for upper bound, lower bound, and recommended stress-strain curves for any selected class of backfill soil and level of compaction. Specific stress-strain curves to be plot-able include; (1) Standard triaxial compression - axial stress and radial strain versus axial strain for at least 3 confining pressures, (2) Uniaxial strain - axial stress and radial stress versus axial strain, and (3) Hydrostatic – confining stress versus volumetric strain.

This work is long overdue and serves to better integrate LRFD principles into Section 12 of the LRFD Specifications of the culvert world. It also will facilitate development of LRFR load rating systems for culverts. Moreover, application of this work extends to the entire geotechnical community where ever re-worked compacted soil is used.

Sponsoring Committee:AFS40, Subsurface Soil-Structure Interaction
Research Period:12 - 24 months
Research Priority:High
RNS Developer:Dr. Michael G. Katona, Dr. Ian Moore, Dr. Timothy J. McGrath, PE
Date Posted:01/16/2020
Date Modified:02/21/2020
Index Terms:Deformation curve, Backfill soils, Soil compaction, Databases, Soil tests,
Cosponsoring Committees: 
Bridges and other structures
Hydraulics and Hydrology

Please click here if you wish to share information or are aware of any research underway that addresses issues in this research needs statement. The information may be helpful to the sponsoring committee in keeping the statement up-to-date.