Lessons Learned Regarding Seismic Deformation Analyses of Embankment Dams from Re-Evaluation of the Upper and Lower San Fernando Dams Performance Case Histories: On-Demand

Speakers: Raymond Seed, P.E., Professor Emeritus, University of California, Berkeley; and Khaled Chowdhury, PhD, Senior Geotechnical Engineer, USACE and UC Berkeley

PDHs: This webinar is eligible for 2 PDH credits. 

Analyses of expected seismic performance of earthen dams subject to potential liquefaction hazard are routinely performed for high hazard dams. The State of Practice has increasingly evolved from simplified methods to the use of fully-coupled seismic pore pressure generation and nonlinear seismic deformation analyses (NDA) using either finite difference or finite element analysis frameworks. To evaluate the accuracy and reliability of these NDA methods, the Upper and Lower San Fernando Dams field performance case histories from the 1971 San Fernando earthquake were re-analyzed in these current studies by means of fully nonlinear seismic deformation analyses, using (1) a suite of four analytical models, (2) a suite of three cyclic pore pressure generation relationships, and (3) a suite of three post-liquefaction residual strength relationships. The results of these analyses have shown that a successful implementation of non-linear deformation analyses would involve (1) how cyclic pore pressure generation and liquefaction triggering is accounted for in different constitutive modeling approaches, (2) treatment of critical state and dilatant behaviors in soils during shaking, (3) evaluation and implementation of post-liquefaction residual strength (Sr), (4) modeling of potential strain softening of the non-liquefiable (e.g. clayey) soil layers, (5) continuation of the analyses through post-shaking conditions, (6) dealing with numerical difficulties associated with very large embankment displacements within the continuum analysis framework, and (7) suitable engineering assessment and interpretation of the analytical results. Failure to suitably accomplish any of these can result in potentially misleading and/or potentially unconservative findings. When these are suitably accomplished, seismic deformation analyses were found to be capable of producing excellent engineering insights and a good basis for engineering decision-making and/or mitigation design.

Key Takeaways:

•     Seismic nonlinear deformation analyses using numerical modeling can provide useful and reliable engineering results.
•     Accomplishing reliable results requires attention to a number of characterizations, modeling, and analysis details.
•     Modeling of non-liquefiable layers should also incorporate strain softening effects in earthquake and post-earthquake analyses.
•     Selection of appropriate combinations of liquefaction triggering, post-liquefaction strengths, and constitutive models is important to achieve reliable engineering results.
•     The common continuum analysis framework (e.g. FLAC) also has intrinsic limitations, and suitable engineering interpretation is also needed.

• 1. Significance of the Upper and Lower San Fernando Dam seismic performance case   histories during the 1971 San Fernando Earthquake
  • (a) Contributions to the national seismic dam safety programs still ongoing today
  • (b) An excellent “test” of analytical tools: two similar dams, built with similar materials and methods, co-located, but with two very different performance outcomes (one dam with a liquefaction-induced upstream flow failure, and one dam with only limited to moderate deformations)
• 2. Analytical methods employed for back-analyses
  • (a) Four different analytical models
  • (b) Three different liquefaction triggering relationships
  • (c) Three different post-liquefaction residual strength (Sr) relationships
  • (d) Consistent set of analytical procedures and protocols for all analyses
• 3. Re-evaluation of input motions for the 1971 San Fernando earthquake based on modern understandings and state of practice, including near-field effects
• 4. Back-analyses of the Lower San Fernando Dam
  • (a) Six different combinations of analytical models and relationships 
  • (b) Observed and analytical mechanisms of deformation and failure, and degree of matches between analytical results and observed field performance
  • (c) Four successful back-analyses; excellent matches with observed field performance
  • (d) Key lessons learned
• 5. Back-analyses of the Upper San Fernando Dam
  • (a) Nine different combinations of analytical models and relationships
  • (b) Observed and analytical mechanisms of deformation and failure, and degree of matches between analytical results and observed field performance
  • (c) Six successful back-analyses; excellent matches with observed field performance
  • (d) Key lessons learned
• 6. Overall lessons learned 
  • (a) Accuracy and reliability of fully nonlinear seismic deformation analyses
  • (b) Key analytical and engineering issues and details that must be addressed for successful outcomes
  • (c) Engineering interpretation and use of analysis results