Académique Documents
Professionnel Documents
Culture Documents
Abstract: The dissipated strain energy method (DSEM), a new method for determining the preconsolidation pressure,
is presented in this paper. Compared with the energy method, the DSEM uses dissipated strain energy and the slope of
the unloading–reloading cycle (in the strain energy – effective consolidation stress space) for the plot to minimize the
sample disturbance effects and eliminate the effect of elastic deformation. Dissipated strain energy, in terms of micro-
mechanics, is directly related to the irreversible process of consolidation and can be supported by theories dealing with
consolidation and compaction. The use of the unloading–reloading slope to simulate the elastic reloading for the
recompression stage can minimize sample disturbance effects. Examples presented indicate that the proposed new
method is less operator dependent than most of the existing methods.
Key words: dissipated strain energy, preconsolidation pressure, consolidation, energy method, dissipated strain energy
method.
Résumé : On présente dans cet article une nouvelle méthode pour déterminer la pression de consolidation, soit la
méthode d’énergie de déformation dissipée (DSEM). Comparée à la méthode d’énergie, la nouvelle méthode utilise
l’énergie de déformation dissipée, et la pente du cycle de déchargement-rechargement (dans l’espace énergie de
déformation-contrainte effective de consolidation) pour le graphique devant minimiser les effets de remaniement de
l’échantillon et éliminer l’effet dû à la déformation élastique. L’énergie de déformation dissipée en termes de micromé-
canique est directement reliée au processus irréversible de consolidation et peut s’appuyer sur des théories traitant de la
consolidation et du compactage. L’utilisation de la pente déchargement-rechargement pour simuler le rechargement élas-
tique pour le stade de recompression peut minimiser les effets de remaniement de l’échantillon. Les exemples présentés
indiquent que la nouvelle méthode proposée est moins dépendante de l’opérateur que la plupart des méthodes existan-
tes.
Mots clés : énergie de déformation dissipée, pression de préconsolidation, consolidation, méthode d’énergie, méthode
d’énergie de déformation dissipée.
[Traduit par la Rédaction] Wang and Frost 768
Can. Geotech. J. 41: 760–768 (2004) doi: 10.1139/T04-013 © 2004 NRC Canada
Wang and Frost 761
Fig. 1. Illustration of Casagrande’s (1936) observations. The Fig. 2. Schmertmann’s (1955) reconstitution of the field consoli-
relationship between section I (recompression curve) and the vir- dation curve.
gin compression curve is closely similar to that between the
unloading–reloading curve and the virgin compression curve.
which generally obscures the distinctness of the deformation Fig. 3. Sample disturbance effects on preconsolidation pressure
rate between pre-σc and post-σc sections. This phenomenon (adapted from Holtz et al. 1986).
also exists in the energy method proposed by Becker et al.
(1987), where total strain energy is used. Because the me-
chanical implication of σc lies in the fact that beyond σc the
irrecoverable deformation rate (against load increment) of
soil increases drastically, a separation of the irrecoverable
deformation from the total deformation will enhance the
contrast between the deformation characteristics of the pre-
and post-σc sections.
Figure 3 is adapted from Holtz et al. (1986), where the ef-
fects of different levels of disturbance on the assessment of
σc are illustrated. Figure 3 shows that the slope of the
recompression curve increases when the soil samples are
more severely disturbed (i.e., piston samples versus block
samples), which results in an underestimation of the precon-
solidation pressure. It can also be deduced from Fig. 3 that
the laboratory consolidation curve will be closer to the vir- Fig. 4. Uncertainty in defining the recompression line in the
gin consolidation curve if the soil is less severely disturbed work – effective stress plot.
or destructured. As most conventional tests do not study the
sample preparation and disturbance effects, for a given test
result, it is hard to evaluate how severe it is distorted by
sample disturbance.
Even if sample disturbance is minimized, it is still diffi-
cult to construct the straight lines in the cases of the log(1 +
e) − log p and work–p methods. Figure 4 illustrates this
effect using the work–p method to present data from a con-
solidation test performed on a silty clay (tube sample from
the field by Law Engineering and Environmental Service,
Inc., Atlanta, Ga.), showing the difficulties in selecting the
recompression line.
Due to the complexity of soil properties, the three prob- as the method of Butterfield (1979) in the mapped space:
lems described previously cannot be completely avoided. log(1 + e) − log p, mapped into E–p space. Unlike Casa-
Rational methods should minimize these effects, however. In grande’s method, which focuses more on the local properties
this regard, the following observed phenomena should be (the largest curvature) around the preconsolidation pressure,
mentioned: the Becker et al. energy method has adopted the average
(1) The disturbing effects due to stress release, etc. become slope of the recompression curve before the preconsolidation
smaller when the reloading stress exceeds the stress pressure.
where stress release starts. This suggests the use of un- The energy method is relatively new, more inclusive, and
loading–reloading slopes at a stress level higher than the conceptually very promising. Considering preconsolidation
in situ stress. as the maximum yielding stress to which a soil has ever
(2) Any quantities to be used should be physically or me- been subjected, work by Roscoe et al. (1958) and Tavenas et
chanically related to the process in the analysis. In the al. (1979) is supportive of this method. Apart from the more
case of the consolidation process, quantities causing the accurate determination of σc due to the advantages of the lin-
consolidation process should be more relevant. For ex- ear coordinates and the flexibility of the concept of strain
ample, the dissipated strain energy that causes the parti- energy in dealing with consolidation induced by other
cles to slide and rotate could be better quantitatively mechanical and nonmechanical effects such as evaporation
related to preconsolidation pressure. through energy equivalency, the theoretical equivalency of
the E–p linear relation to the e − log p and log(1 + e) − log p
Energy method linear relations is even more important in that no violations
are introduced against previous experimental observations.
The energy method (Becker et al. 1987) also uses Casa- The proof of the equivalency of the E–p linear relation to
grande’s (1936) first conclusion and the conclusion of Mesri the e − log p and log(1 + e) − log p linear relations is simple;
and Choi (1985) on the unique relation between end of pri- however, it is briefly described here for completeness and to
mary consolidation void ratio and effective stress. As for the provide an introduction to the linear relations between total
term “similar” in Casagrande’s second conclusion, Becker et strain energy, elastic strain energy, and dissipated strain en-
al. (1987) assume a linear relation between total strain en- ergy versus effective consolidation stress.
ergy E and the effective stress p (E–p) for the recompression There are basically two types of strain definitions, engi-
part directly from the laboratory recompression curve with- neering strain and true strain. When engineering strain is
out considering the unloading–reloading portion of the tests. adopted, the E–p linear relation is equivalent to the e − log p
In actuality, the Becker et al. energy method is the same linear relation:
improved by unloading at a reasonably estimated precon- Fig. 5. Idealized consolidation tests in strain energy (E) – p
solidation pressure, or the consolidation rate could be moni- space. OD, dissipated strain energy; OE, elastic strain energy;
tored so that unloading can take place at the stress where the OT, total strain energy.
consolidation rate has some significant variations. In this
regard, a reasonable approach to this problem is suggested
as follows: (i) perform unloading–reloading cycles at two
stresses; and (ii) obtain a correlation between effective stress
where unloading starts and the average slope of the
unloading–reloading cycle and then iterate to obtain σc . This
approach is derived from the following correlation equation
proposed by Schmertmann (1955):
C r1 er 2 + 1
[6] log = 2.5 log
Cr2 er1 + 1
Fig. 6. Idealized consolidation tests in dissipated strain energy of the work done during the process is dissipated in
(Ed) – p space. overcoming the friction between particles.
In plasticity theory, an irreversible process is usually de-
scribed by dissipated strain energy; most of the hardening
rules describing the conditions for evolution of irreversible
processes also adopt dissipated strain energy as an independ-
ent variable. Therefore, it is natural and easier to be adapted
to other theories by using dissipated strain energy to de-
scribe the consolidation process. In addition, as shown by
Janbu (1967), the yielding surface is actually contours of
dissipated strain energy. The computation of the dissipated
strain energy in a consolidation test also presents a method
for calculating the evolution of the yielding envelop.
Experimental justifications
Fig. 7. Actual consolidation tests in strain energy – p space. OD, Although a large database is very important in verifying a
dissipated strain energy; OE, elastic strain energy; OT, total method or a technique, the method’s theoretical rationality
strain energy. based on the fundamental phenomena of the process in the
analysis is also important. The authors realize the difficulties
for a widespread justification due to the lack of high-quality
data where preconsolidation pressure is known. Therefore,
instead of presenting as many experimental examples as pos-
sible, the authors present one example for each of the two
critical questions explored: the better linear relation of the
dissipated strain – effective consolidation stress, and the en-
hancement in the accuracy for assessing the preconsolidation
pressure. The authors encourage readers to analyze their
own database and establish their own judgment. Justification
by many people is better than justification by a few people;
in this respect, the authors appreciate the ideas of Becker et
al. (1989).
To investigate whether a better linear relation between the
dissipated strain energy and the effective stress could be ob-
tained using the proposed method, a typical consolidation
OD (i.e., dissipated strain energy line), and (vii) the p′ coor- test (see Fig. 8 for the e − log p plot from this test) carried
dinate of the intersection of line Rp and OD is the precon- out at the Mississauga, Ontario, laboratory of Golder Asso-
solidation pressure. ciates and provided by D.E. Becker was analyzed using dif-
ferent methods. The coefficients of determination from the
Characteristics of the DSEM different methods are presented in Table 1. The coefficient
The characteristics of the DSEM can be summarized as of determination is a criterion for the quality of linear corre-
follows: lation. A perfect linear correlation has a coefficient of deter-
(1) In strain energy – effective stress space, consolidation mination equal to 1. Natural strain was used in calculating
tests start at the point that corresponds to the dissipated the total strain energy, the elastic strain energy, and the dissi-
strain energy at the preconsolidation stress. This point is pated strain energy. In addition, total work, elastic work, and
the base point that serves as reference for other strain plastic work are calculated for the correlation analysis. The
energy calculations. The total strain energy method does analytical results indicate, in practice, the total work, elastic
not have such a characteristic. work, and plastic work can be directly used for the correla-
(2) The total strain energy method by Becker et al. (1987) tion analysis and the plots to obtain σc. This makes the
uses only the recompression curve, whereas the DSEM DSEM very convenient. The example also indicates that, al-
uses both the recompression curve and the unloading– though the energy method is theoretically equivalent to the
reloading curves from the same test and thus will de- e − log p and log(1 + e) − log p methods, in practice the meth-
crease the systematic errors. ods are not equivalent. The coefficients of linear correlation
(3) By knowing the value OR, σc can be obtained numeri- show the relative merits of these methods, but this is difficult
cally, which makes computer implementation simpler: to determine by looking at the graphs. In this example, it
σc = OR / tan(DOP′). is shown that both second loading and unloading follow a
(4) For each step, the operator has a more objective proce- better linear relation and there is no apparent enhancement
dure to follow. Therefore the method is less operator de- in the linear relation for the strain energy plots (dissipated or
pendent. total strain energy).
(5) The DESM has a strong theoretical basis, in that the Since no other information is available for judging the ra-
consolidation process is an irreversible process and most tionality of the preconsolidation pressures obtained, it is dif-
Table 1. Preconsolidation pressures (Pc) and coefficients of determination from the different methods.
Coefficient of determination
Method Pc Initial loading Second loading First unloading Second unloading
e − log p 380 0.800 0.820 0.750 0.963
Log (1 + e) − log p 367 0.799 0.818 0.750 0.965
Energy method (EM) 480 0.973 (0.973) 0.995 (0.995) 0.915 (0.929) 0.997 (0.997)
DSEM 435 0.973 (0.973) 0.995 (0.995) 0.915 (0.929) 0.997 (0.997)
Note: The values in parentheses are for the work–p (total work, elastic work, plastic work) correlation and indicate
that this correlation can be used without changing much of the linear relation, but the method is greatly simplified.
ficult to evaluate the relative merits of the different methods Table 2. Preconsolidation pressures deter-
through this example. To investigate how well the different mined by the different methods.
methods assess the preconsolidation pressure, the results
from a man-made sample consolidated in the laboratory at Preconsolidation
Georgia Tech to a maximum stress of 400 kPa have been an- Method pressure (kPa)
alyzed using the different methods. The results are presented Casagrande 469
in Table 2 and indicate that the DSEM gives the closest esti- Burmister 560
mate of the preconsolidation pressure. Becker 494
Although it cannot be concluded that the newly proposed Butterfield 467
method will always work better than other methods based on Intersecting tangent 492
these two examples, the authors invite readers to analyze DSEM 408
their own tests to establish their confidence in using the pro-
posed method. It is also the authors’ pleasure to analyze any
data that readers are willing to provide. can be analytically established through micromechanics. The
To give better illustrations of the calculations of the vari- newly developed method is less operator dependent and can
ous strain energies, a detailed example is presented in Ap- account for sample disturbance effects to a certain degree.
pendix A. The examples investigated herein show that the new method
has potential and should be investigated further.
Conclusion
Acknowledgment
A new method, called the dissipated strain energy method
or DSEM, has been developed and presented in this paper. Dr. Becker’s kind presentation of some of the test data for
Dissipated strain energy is mechanically the cause of consol- this study from an engineering project is sincerely appreci-
idation; its linear relation with effective consolidation stress ated.
using Eid = Eit − Eie and Eie = C n pi /(1 + e0). These values Fig. A1. Strain energy versus consolidation stress for the exam-
should be corrected by adding a term equal to OR in Fig. 5. ple test.
RO can be obtained by regression analysis of the last three
points on the OD plot, i.e., (2, 0.0107), (4, 0.0388), and (8,
0.0976) (see Table A1, the Load column and the ADSE col-
umn). In this case, OR = 0.0187 ksf.
The accumulative total strain energy corrected (ATSEC) =
ATSE + OR, which is equivalent to coordinate transform
from R origin to O origin (see Fig. 5), and the accumulative
dissipated strain energy corrected (ADSEC) = ADSE + OR,
the meaning of which is as given previously.
Figure A1 plots the strain energies in the energy – consol-
idation stress space. The p coordinate that corresponds to the
Rp–OD intersecting point, 1.3 ksf, is the preconsolidation
stress.