Reanalysis of Braced Diaphragm Wall Excavation in Abu Dhabi

Sometimes Analysis Results are close for Deep Excavations Case History in Abu Dhabi

Portion of an excavation in Abu Dhabi required excavation of a 20m basement with a 1.0m thick diaphragm wall supported by two levels of steel struts. Soils at the site consist of fill over a mix of sandstone, claystone, gypsum, and mudstones. The design water level was at -2m but measured water levels were about 0.5 to 2m lower during construction.

The excavation was initially analyzed with FREW and Plaxis 8.2 in 2009 and loads and displacements were measured. Results were published by Dr. Walid Jaber & Amr Al Degwee by NCSS International in 2015 “Behavior of Instrumented Anchored and Strutted Wall in Soft Rock”, 2015 DFI Conference.

While some information was missing from the paper, this article compares the initial SOE analysis results against four runs with limit equilibrium, non-linear, and finite element models with DeepEX.

The initial models appeared to have slightly different construction sequences vs. the final design which partially explains some of the differences in the initially reported FREW values vs. the current DeepEX analysis.

Results from all analyses are summarized in Table 1:

Table 1: Summary of Analysis Results

In this effort, Limit-Equilibrium with FHWA & Caltrans produced very close bending moments with the initial FE analysis.

The FREW analysis appears to considerably underpredict lower support reactions vs. all DeepEX methods and unreasonably overestimates all bending moments by more than 30%. All DeepEX analyses produced relatively similar reactions.

Measured displacements were much smaller vs. all predictions, indicating that the soil strength design assumptions were on the safe design.

Toe wall displacements predicted in the FE models were not observed in measured inclinometer displacements. Such displacement analysis results are known to be the result of modelling limitations by the Mohr-Coulomb elasticity model that was employed.

The current analysis illustrated that water modelling and pressures were dominant drivers in obtained moments and strut loads since the relatively large cohesion of site soils exert little driving soil force.