Actes du colloque

Proceedings of the 18

International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013, volume 6, 2016

step highlights the benefit of loading tests on isolated inclusions

in producing an excellent structural design.

g) The results of these experiments reveal that reinforced soil

also undergoes lateral deformations around the periphery, with

these deformations to be incorporated into the inclusion design

(addressing the need to reinforce or not inclusions positioned

along the bank). A ratio of 0.25 was measured between the

maximum horizontal displacement and the settlement at the

center of the reinforced zone; this ratio is comparable to that

applicable beneath embankment slopes on compressible soils.

h) Moreover, these experiments have underscored the

importance of a robust geotechnical characterization of project

sites. The advantages of a static penetrometer have received

recognition, and the execution of oedometric tests is shown to be

crucial. Pressure meter testing offers a strong correlation with

the experiment on deep foundations (limit values of lateral

friction and/or load at the pile tip, shape of transfer curves and

plot of loading curves for isolated inclusions).

2.2

Instrumentation of actual structures

The results of full-scale experiments were complemented by

instrumenting actual building sites in order to collect additional

data on the behavior of inclusions under varied conditions. Over

ten structures could be instrumented, among which let's cite: a

wind turbine foundation, an apron for a facility handling weakly

radioactive waste, a reinforced concrete frame below an

embankment, a wastewater treatment tank, and an industrial slab

for assessing the impact of point loads (rack bottoms or cart

caster wheels).

Let's point out the difficulties inherent in these jobsites, the

most important of which is to preserve the sensors and their

connections throughout the successive phases of the works.

7.2.3

Physical models

7.2.3.1

Detailed description

Physical models were developed in the calibration chamber to

study: load transfer around an inclusion head, the influence of

distribution layer thickness, and for a given layer thickness the

differences between a slab and an embankment under

comparable mattress conditions.

The most valuable physical models were produced in the

centrifuge, where all similarity conditions were respected. The

capacity of the IFSTTAR centrifuge in Nantes reaches 100 g,

and it was decided to proceed with a 1/28-scale model to study a

group of 9 inclusions and then a 1/12-scale model for tests with a

mobile tray that enabled simulating soil settlement on inclusion

groups. In all, 35 centrifuge tests were performed for a detailed

parametric study focusing on: the type of structure being

supported (embankment or slab), inclusion spacing, distribution

layer height, and type of material found in this layer (natural

gravel or treated silt).

7.2.3.2

Lessons drawn

The models placed in a calibration chamber display a notable

difference between embankments and slabs with thinner

distribution layers, though this difference fades as the thickness

increases. These models also show that mattress granularity is a

key factor. A lower level of reversibility was also exhibited for

an embankment compared to a slab, which highlights the critical

role played by the slab (through its elastic reversible behavior),

as opposed to the embankment, in which the shear

accompanying load transfer is irreversible.

The series of mobile tray tests enabled validating the finding

that the Prandtl model for a spread footing could also be used to

evaluate the maximum stress on an inclusion head underneath a

slab. Moreover, it was established that the magnitude of strains

justified adopting the angle of friction at the critical state rather

than the peak angle of friction.

These results served to guide the choice of verification rules

explained in the set of Recommendations, as well as the rules

selected to verify consistency conditions for the simplified

design models.

7.2.4

Numerical models

Numerical models provide a vital complement to experiments

conducted on full-scale structures or reduced-scale models.

During ASIRI, it was understood that numerical 3D finite

element and finite difference models needed to serve as a

reference. Yet one crucial point pertained to the choice of

constitutive models and calibration for the parameters drawn

from detailed characterizations performed on the various

materials (distribution mattress, compressible soil) that prove

suitable for introduction into these models.

In the case of the tested structures, such considerations helped

verify the model's capacity to accurately reproduce test structure

behavior. Nonetheless, some models wound up requiring

extensive computation time (several weeks).

This procedure also allowed verifying edge effects by means

of comparing complete 3D models with actual 3D models or

axisymmetric 2D models of an elementary cell.

A study of model representativeness conditions with respect

to simulated behavior under the inclusion tip (model extension

and minimum number of elements) was also carried out. Its

findings suggested the need to opt for a compromise between

precision and amount of computation time.

The models evaluated in this manner could be applied to

structural situations outside the strict scope of the experimental

campaign. Such was the case, for example, with slabs subjected

to strip loads or point loads (rack bottoms), as well as with

spread footings positioned on a small number of inclusions

subjected to ordinary loadings (though not handled

experimentally, this case still needed to be studied in order to

yield Recommendation results, since these structures are

commonly encountered in industrial warehouse or logistics

projects).

The ASIRI Project also developed a number of discrete

element models. It is worth indicating that this latter type of

model showed a better capacity than continuous models to

describe the distribution mattress behavior observed in physical

models (i.e. sliding of particles at the edge of inclusion heads).

However, their implementation remains cumbersome and must

be reserved for special calibration or validation studies.

7.3

ASIRI Project publications

Research conducted as part of the ASIRI Project has given rise

to numerous internal reports presented subsequently at over 20

national and international conferences. Moreover, they have

provided the basic material for a number of doctoral theses.

A widely referenced book entitled "Recommendations for the

design, layout, execution and control of improved foundation

soils by means of rigid inclusions" was published by

Presses des

Ponts

in 2012. This publication contains 383 pages and eight

chapters, i.e.: 1) Description and development up to launch of

the National Project; 2) Mechanisms and operations; 3)

Computation models; 4) Design; 5) Justifications; 6) Soil

surveying; 7) Execution conditions; and 8) Controls and

instrumentation.

Volume 6 - Page 88