INTERNATIONAL SOCIETY FOR

SOIL MECHANICS AND

GEOTECHNICAL ENGINEERING

This paper was downloaded from the Online Library of

the International Society for Soil Mechanics and
Geotechnical Engineering (ISSMGE). The library is
available here:

https://www.issmge.org/publications/online-library

This is an open-access database that archives thousands

of papers published under the Auspices of the ISSMGE and
maintained by the Innovation and Development
Committee of ISSMGE.
1-D consolidation model with stress dependent recompression index and coefficient
of consolidation
Modèle de consolidation unidimensionnelle incluant des coefficients de re-compression et de
consolidation dépendant des contraintes.

Bosco Poon, Kim Chan

GHD Pty. Ltd., Australia, [email protected], [email protected]

ABSTRACT: According to Terzaghi’s consolidation theory, the coefficient of consolidation, cv is a composite parameter that depends
on both the coefficient of permeability, k and coefficient of volume compressibility of the soil, mv. Conventionally, k is related to void
ratio, e by a e – log k relationship and mv is calculated based on a bi-linear e –log ’v compressibility model with a recompression
index, cr and compression index, cc at the over-consolidated and normally consolidated stress ranges, respectively. Based on the
observations of an instrumented embankment site, this paper has shown that because of the inherent simplification of the soil
compressibility using a constant cr, the conventional model overestimates cv and underestimates settlement at the lightly over-
consolidated stress range near the pre-consolidation pressure. This paper further proposes a 1-D consolidation model with a stress
dependent cr in order to model appropriately the soil behavior at this stress range.

RÉSUMÉ : Suivant la théorie de la consolidation selon Terzaghi, le coefficient de consolidation cv dépends de la perméabilité k et du
coefficient de compressibilité volumique mv. Usuellement, k est relié à l’indice des vides e par une relation e – log k et mv est calculé en
utilisant un modèle bilinéaire e –log ’v incluant un coefficient de re-compression cr dans le domaine sur-consolidé, puis un coefficient
de consolidation cc dans le domaine normalement consolidé. S’appuyant sur les observations d’un chantier instrumenté, cet article
démontre que la simplification de la courbe de consolidation par le modèle bilinéaire tend à surestimer la valeur de cr et sous-estimer les
tassements dans le domaine légèrement sur-consolidé, proche de la contrainte de pré-consolidation. Cet article propose un modèle de
consolidation unidimensionnelle incluant un cr dépendant des contraintes qui représente plus correctement le comportement du sol dans le
domaine proche de la contrainte de pré-consolidation.
KEYWORDS: compressibility; consolidation; permeability.
consistent with the general understanding about the variation of
cv with pressure. For the over-consolidated stress range, the soil
1 INTRODUCTION has a lower compressibility with less reduction of mv with
decreasing e whilst the rate of change of k with e remains the
According to Terzaghi’s theory of 1-D consoldiation of soils, same as for the normally consolidated stress range. This results
the coefficient of consolidation, cv is thoertically related to the in an increasing cv with ˊ v, followed by a sudden drop of cv as
coefficient of volume compressibility, mv and the coefficient of the soil state changes from over-consoldiation (OC) to normal
permeability, k by consoldiation (NC). This calculated cv appears to be
cv = k/(mvw) (1) contradictory with the expected trend outlined in Figure 1 (Ladd
and DeGroot, 2003), which show that cv (OC) decreases to a
where w is the unit weight of the water and mv can be defined fairly constant cv (NC) value as the stress level increases up to ’p.
in tems of the compressibility index cc, or the recompression Therefore, the soil model may overestimate the consolidation
index cr, depending on the vertical effective stress ’v relative rate when the soil is in the slightly over-consolidated stress
to the preconsoldiaiton pressure, ’p as follows: range. The authors considered that this inconsistency is due to
0.434𝐶𝑟 the use of the simplified bi-linear recompression and
𝑚𝑣 =
𝜎′𝑣 (1 + 𝑒0 )
(’v ≤ ’p) (2a) compression e –log ’v curves in the model.
1 𝜕𝑒
𝑚𝑣 =
1 + 𝑒0 𝜕𝜎𝑣′
0.434𝐶𝑐
𝑚𝑣 = (’v > ’p) (2b)
𝜎′𝑣 (1 + 𝑒0 )

where e0 is the initial void ratio. For the derivation of k in

equation (1), Tavenas et al (1983) indicated that it can be
related to the void ratio, e by:

e = e0 + ck log (k / k0) (3)

where ck is the permeability index and k0 is the initial

permeability value. The data presented in Tavenas et al (1983), Figure 1. cv calculated based on bi-linear e –log ’v compressibility
indicate that the above linear e – log k relationship holds curves and void ratio dependent permeability function
irrespective of stress history. Substituting Equations (2) and (3) This paper proposes a new soil model that consists of a
into (1), and following the derivation similar to that given in curvilinear reompression curve in the lead up to a linear virginal
Walker et al (2012), a theoretical cv versus ’v relationship can compression line in the e –log ’v space. It is demonstrated that
be obtained as shown in Figure 1. For the normally the proposed model is able to capture appropriately the cv
consolidated stress range, both k and mv decrease rapidly with response at the lightly consolidated stress range.
decreasing void ratio, hence cv is fairly constant and this is

- 3031 -
Proceedings of the 19th International Conference on Soil Mechanics and Geotechnical Engineering, Seoul 2017

2 CONSTITUTIVE RELATIONSHIP index properties, compression ratio (CR= cc/(1+e0)) and

recompression ratio (RR= cr/(1+e0)) with reduced level (RL) obtained
A variable recompression index cr with ’v is proposed as: from CRS tested samples. Figure 3 shows the normalized
compression curves from representative CRS tests within the lower
(1−𝑚)
𝑐𝑟 (𝜎𝑣′ ) = 𝑐𝑐 [ + 𝑚] (4) soft clay (undrained shear strength Su of about 15kPa) between RL-
𝑒 (𝑂𝐶𝑅−1)𝑛
4m and RL-10 m. The curves exhibit non-linear response in the
where cc is compression index and is considered as constant with normally consolidated range with largest deformation just passing ’p
stress; OCR is the over-consoldiation ratio, i.e. OCR = ’v / ’p; and then reduces with increasing ’v, which is the consequence of
m is the ratio of initial cr value to cc, i.e. cr(’vo) / cc; and n progressive soil restructuration. Using the conventional Casagrande’s
controls the rate of change of cr with ’v. As to be discussed in graphical approach, the assessed RR, CR and pc’ are 0.05, 0.65 and
Section 3, the parameter n can be obtained from oedometer test 80kPa, respectively. Note that for the purpose of trial embankment
results. A higher n value (typically n = 4 or above) will give a prediction over the Ballina clay as outlined in Section 4, the stress
more accute change in the compressibility curve from OC to range is less than 200 kPa after the application of 3 m high
NC states than a lower n value. The mv as defined by Eq 2 can embankment load, hence justifying the use of a constant CR for this
be rewritten as stress range just passing’p.
1 𝜕𝑒 𝜕𝑙𝑜𝑔𝜎𝑣
,
1 0.434 𝜕𝑒 Using the proposed model, a varying cr with ’v (Eq. 4) is
𝑚𝑣 = , , = , , (5) employed while maintaining a constant cc for the virgin
1+𝑒0 𝜕𝑙𝑜𝑔𝜎𝑣 𝜕𝜎𝑣 1+𝑒0 𝜎𝑣 𝜕𝑙𝑜𝑔𝜎𝑣
consolidation curve. The fitted parameters of Eq. 4 for the
𝜕𝑒 expermental data are n = 0.65 (similar to RR/CR ratio in the bi-
But 𝑐𝑟 (𝜎𝑣, ) = (for ’v ≤ ’p) (6)
𝜕𝑙𝑜𝑔𝜎𝑣, linear model) and m = 5 (by fitting the curvature approaching
Substituting equations (4) and (6) into (5) gives ’p). The proposed model is considered to give a better
representation of the soil compressibility at the lightly over-
𝑚𝑣 =
0.434𝑐𝑐
[
(1−𝑚)
+ 𝑚] (7) consolidated stress range. The conventional bi-linear e –log ’v
,
𝜎𝑣 (1+𝑒0 ) 𝑒 (𝑂𝐶𝑅−1)𝑛 model is only an approximation at the OC – NC transition, and
may potentially underestimate settlement as depicted in Figure 3.
It can be seen from Eq. 7 that for OCR = 1, mv reduces to volume As discussed in Pineda et al., 2016, The CRS tests were conducted
compressibility for virgin consolidation defined by Eq. 2b. at a displacement rate of 0.004 mm/min. The ’p indicated from these
In the proposed model, cv is computed from the mv defined in tests are higher than when tested at a slower displacement rate, but no
Eq. 7 and k defined in Eq. 3 using the relationship defined by important changes in the shape of the compressibility curves. Pineda
Eq. 1. This soil model hs been incorporated in an in-house suggested to apply a reduction factor of 0.84 on ’p to correct for the
program LMCON, which solves the partial differential equation strain rate effect.
for 1-D consolidation via finite diference method. The resultant
function for cv can be appraised as follows. (a) Index pr oper ties (% ) (b) Compression Ratio (CR) (c) Recompression Ratio (RR)
Let cv(OC) and cv(NC) be the cv values at the over-consolidated state Cc/(1+e0) Cr /(1+e0)
0 20 40 60 80 100120140 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.02 0.04 0.06 0.08
and normally consolidated state, respectively. Also, as cv(NC) is fairly 1 1
constant with ’v ≥ ’p, the stress dependence of cv in the 0 0
recompression and compression zones can be expressed as -1 -1
-2 -2
-3 -3
𝑐𝑣(𝑂𝐶) 𝑘𝑂𝐶 𝑚𝑣(𝑁𝐶 𝛾
𝑎𝑡 𝜎′𝑝 ) 𝑤 -4 -4
= ×
RL (mAHD)

𝑐𝑣(𝑁𝐶) 𝑚𝑣(𝑂𝐶)𝛾𝑤 𝑘𝑝

RL (mAHD)
-5 -5
0.434𝐶𝑐 -6 -6
𝑘𝑂𝐶 𝜎′𝑝 (1+𝑒0) -7 -7
= × 0.434𝑐𝑐 (1−𝑚) -8 -8
𝑘𝑝 , [ +𝑚]
𝜎𝑣 (1+𝑒0 ) 𝑒(𝑂𝐶𝑅−1)𝑛 -9 -9
𝑘𝑂𝐶 1 1
= × (1−𝑚) × (8) -10 -10
𝑘𝑝 [ +𝑚] 𝑂𝐶𝑅 -11 -11
𝑒(𝑂𝐶𝑅−1)𝑛
-12 -12
-13 -13
where kOC is the permeability at over-consolidated state and kp -14
INCLO2_CRS (lab)
-14
Adopted RR profile
PI LL
is the permeability at pre-consolidation pressure. kOC/kp can be -15 Adopted CR profile INCLO2_CRS -15

calculated based on a relationship similar to Eq. 3 as Figure 2. Plots of (a) index properties; (b) CR vs. RL; (c) RR vs. RL
𝑒𝑂𝐶 −𝑒𝑝 𝑘𝑂𝐶
= log ( )
𝑐𝑘 𝑘𝑝
𝑒𝑂𝐶−𝑒𝑝
bi-linear e –log ’v model underestimates settlement at ’p
𝑘𝑂𝐶 1.1
Hence, = 10 𝑐𝑘
(9) Varied cr with 'v - Eq.4
𝑘𝑝 ’p
where eOC is the void ratio within the recompression range and 1
ep is the void ratio at preconsolidation pressure that can be
obtained from oedometer test. Combining Eq. 9 and Eq. 8 gives 0.9
𝑒𝑂𝐶−𝑒𝑝 Constant c r
1 1
𝑐𝑣(𝑂𝐶) = 𝑐𝑣(𝑁𝐶) 10 𝑐𝑘
× (1−𝑚) × (10) 0.8
[ +𝑚] 𝑂𝐶𝑅
𝑒(𝑂𝐶𝑅−1)𝑛
e/e0

0.7
Eq. 10 can be used to obtain, as a preliminary, the varying cv(OC)
value with OCR if cv(NC) is known and assumed constant, which 0.6 CRS test at RL -4.39
is not necessarily the case. Note that LMCON calculates Cv CRS test at RL -4.99
from mv (Eq. 7) and k (Eq. 3) via Eq. 1, instead of using Eq. 10. 0.5
CRS test at RL -6.13
CRS test at RL -6.82
Proposed model Eq. 4
3 CRS TESTS ON BALLINA CLAY 0.4
10 100
Constant rate of strain (CRS) oedometer tests on high quality 'v (kPa)
piston-sampled Ballina clay (Pineda et al., 2016) were Figure 3. Normalised compressibility curves from CRS compared with
simulated using the proposed model. Figure 2 shows the plots of proposed model and convention bi-linear e –log ’v model

- 3032 -
Technical Committee 214 / Comité technique 214

To define the permeability function (Eq. 3) as part of the 4 TRIAL EMBANKMENT ON BALLINA CLAY.
inputs to LMCON, the e0 and k0 values are typically 3.0 and
1E-9 m/s, respectively, for the soft soil between RL-4 m and A 3 m high embankment was constructed over Ballina clay treated
RL-10 m, as shown in Figure 4. The ck is about 1 according to with prefabricated vertical drains (PVD). The embankment was
CRS data presented in Pineda et al., 2016. thoroughly instrumented with vibrating wire piezometers (VWP),
settlement plates, extensometers, horizontal profile gauge and
Initial Void Ratio Permeability variation with RL inclinometers to measure the behavior of the underlying 8.5 - 10m
e0 Permeability (m/s) thick very soft estuarine silt/ clay deposits during staged loading and
0.0 1.0 2.0 3.0 4.0 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 three years of subsequent consolidation. Figure 7 shows the
1 1 geotechnical section of the instrumented embankment. This paper
0 0
-1 -1 focuses on the extensometer and VWP installed in the middle of the
-2 -2 embankment.
-3 -3
1.2m
RL (mAHD)

-4 -4 1.2m
RL (mAHD)

-5 -5 4m 6m 6m 16m 6m 6m
-6 -6
-7 -7
-8 -8
-9 -9 ~RL+ 0.4m
-10 -10 1.3m Sandy Clay / Sandy Silt ~RL 0m
-11 -11 ~RL - 0.9m VWP6a at RL-1.6m
-12 -12 CRS Yield (INCLO2) Extensometer
-13 CRS - INCLO2 -13 IL CREEP Estuarine Magnet at RL-4.1m
8.5m VWP6a at RL-5.6m
-14 Pineda et al. 2016 -14 CRS Yield (MEX9)
-15 -15 BAT Extensometer 9.6m
CRS - MEX09 Adopted profile Magnet at RL-7.1m
-16 -16 VWP6a at RL-9.6m

Figure 4. Initial void ratio and initial permeability from CRS tests Sandy Clay / Extensometer Magnet
4.2m at RL-10.1m 2.7m
Clayey Sand
Figure 5 shows the plot of cv versus ’v normalized with in-
situ ’p, for the CRS tests within soft soil between RL-4 m and Sand (MD–D) 6.3m 4.2m
RL-10 m. It can be seen that cv reduces as ’v increases up to
’p and beyond which cv is fairly constant. The cv(NC) at the
normally consolidated stress range is about 0.4 – 0.7 m2/year, Pleistocene Clay
which is consistent with the empirical correlation proposed by
NAVFAC (1971) as shown in Figure 6 for a liquid limit (LL) of Figure 7. Geotechnical section of the instrumented trial embankment
up to about 120% (see Figure 2a for index properties). Also
shown in Figure 5 are the simulation of cv with increasing ’v. The soil stratigraphy and much of the soil deformation and
LMCON with varying cr gives agreeable cv responses at the OC drainage properties have been outlined in Section 2. In addition,
stress range as well as post-yielding. Conversely, the Figure 8a shows the ’p profile assessed from CRS and oedometer
conventional bi-linear e –log ’v compressiblity model gives an tests. While ’p of the CRS are higher than that of the oedometer tests,
inconsistent and opposite trend to the laborary results, in which good agreement is indicated after ’p of the CRS results are corrected
cv increases with ’v until ’p is reached, followed by an abrupt for strain rate effect by applying a reduction factor of 0.84. Also
drop in value at the NC stress range. shown in Figure 8a is the in-situ ’v with RL. The corresponding
OCR with RL is shown in Figure 8b.
Over consolidated Normally Preconsolidation stress (kPa) OCR
consolidated (a) (b)
0 20 40 60 80 100 120 140 1 10
100
1 1
0 0
-1 -1
-2 -2
10 -3 -3
cv (m2/year)

Reduced Level (m)

-4 -4
Reduced level (m)

-5 -5
-6 -6
-7 -7
1 -8 -8
-9 -9
-10 -10
-11 -11
-12
0.1 -13 -12
0.1 1 -14INCLNO2_CREEP_Oedometer test -13
'v / 'p (insitu) -14
-15 INCLO_CREEP_Oedometer test
CRS_INCLO2_5.49 m CRS_INCLO2_6.63 m -16pc' from CRS without strain rate correction -15
OCR from CRS without strain rate correction
CRS_INCLO2_4.89 m CRS_INCLO2_7.27 m -17pc' from CRS with strain rate correction -16
Conv. model with constant Cr Porposed model with varied Cr -18In-situ vertical effective stress -17
OCR from CRS with strain rate correction
Figure 5. Variation of cv with ’v during CRS loading (Pineda et.al 2016) Figure 8. Pre-consolidation pressure and over consolidation ratio profile
100.0 Undisturbed samples: Prefabricated vertical drains were installed at the trial
cv in range of recompression
lies above this line embankment in a square pattern 1.2 m apart, to a depth of 15 m.
cv (m2/year)

10.0 The PVD installation has caused remoulding of the soft clay
cv in range of virgin
Completely compression
surrounding the drains. Indraratna et al. (2015) have conducted
remoulded research of smear zone characteristics of actual drain
samples: c v lies
1.0 below this line installation at the trial embankment site by studying the
0.4 extracted soil samples collected around the drains. The study
has indicated that the smear zone was about 5-7 times greater
0.1 than the equivalent dimensions of the mandrel, or about 11
0 20 40 60 80 100 120 140
times greater than the drain well radius. In addition, the ratio of
LiquidLimit (LL) in-situ permeability to disturbed permeability in the smear zone
Figure 6. NAVFAC (1971) correlation of cv vs. LL is about 1.7 -2.

- 3033 -
Proceedings of the 19th International Conference on Soil Mechanics and Geotechnical Engineering, Seoul 2017

The authors have conducted a Class A prediction and subsequently conventional and proposed models underestimate the soil
a back-analysis for the trial embankment at Ballina. Details of the compression. This is attributed to plastic deformation associate
works have been provided in Chan et al. (2017). This paper provides with soil yielding at shallow depth during fill placement (Chan
highlights of the back-analysis results, and focuses only on the et al. 2017), which is unable to be captured by the present
behavior of estuarine clay between RL-4 m and RL-10 m beneath the models for 1D consolidation. Nevertheless, the proposed model
centerline of the embankment. gives good prediction in terms of the slope of the compression-
Figure 9 shows the measurement of the excess pore pressure time curve post-plastic yielding. Moreover, the prediction given
at RL-5.6m (see Figure 7 for piezometer location), which has by the proposed model is closer to the measurements than that
been corrected for the increase in total head due to the of the conventional model
settlement of the piezometer under load. Also shown in the
figure are the predictions of excess pore pressure using the 0.2 4

Fill Thickness (m)

conventional bi-linear e –log ’v compressiblity model and the 0.1 2
proposed model with varied cr with ’v (<’p) that has been

Layer Compr ession (m)

0.0 0
incorporated in LMCON. Note that whilst both models are for
1D consolidation, the vertical stress that is applied on each -0.1 -2
discretized soil layer has been factored by the Boussinseq stress -0.2 soil compression due to -4
influence factor to account for the effect of 2D loading. plastic yielding
-0.3 -6
The conventional model gives a much faster pore pressure Extensometer measurement
dissipation at the onset of consolidation when ’v <’p, leading -0.4 Proposed model - varied Cr -8
to an overall under-prediction of excess pore pressure compared -0.5 Conv. Model - constant Cr -10
to the measurements. This under-prediction is due to the over- Fill thickness
-0.6 -12
estimation of cv at the OC stress range outline in Figure 5.

100

1000
10
Conversely, the proposed model employed in LMCON gives a
Time (days)
more agreeable pore pressure response with the measurements.
Figure 11. Comparison of predictions and extensometer results between
Nodal Excess Pore Pressure Vs Time RL-4m and RL-7m
80
Excess Por e Pressure (kPa)

5 CONCLUSION
60
A new 1-D soil compressibility curve consisting of a stress
40 dependent recompression index cr at the over-consolidated stress
range has been developed. This compressibility model, when
20 used in conunction with a log-linear void-ratio – permeability
function, is able to capture implicitly the trend of reduction of cv
with ’v at the over-consolidated stress range, which has been
0 verified with high quality CRS oedometer test results. Conversely,
0.01 0.1 1 10
Time (year s) the conventional compressibility model with constant cr will give
Applied fill Load Proposed model with varied Cr an opposite trend of increasing cv with ’v at the over-
Field VWP6 at RL-5.6m Conv. Model with constant Cr consolidated stress range. Furthermore, the newly proposed soil
model will capture slightly more settlement than that of the
Figure 9. Comparison of predictions and measured excess pore pressure
conventional model at the transition from OC to NC soil states.
Figure 10 shows the measured soil compression between RL- The predictive capability of the proposed soil model has been
7 m and RL-10 m, which has been obtained from the measured demonstrated by numerical simulation of settlement and excess
settlements of the extensometer magnets installed at the pore water pressure monitored during the staged loading and 3
corresponding reduced levels. LMCON gives very good years of subsequent consolidation for a trial embankment
prediction of the soil compression under load. In contrast, the constructed over Ballina Clay treated with PVD.
conventional model underestimates the compression by about
10%. This is due to the inherent simplification of the
conventional model for the transition from OC to NC at ’p as 6 REFERENCES
delineated in Figure 3. Chan, K., Poon, B. & Perera. D. 2017. Numerical prediction and back
analysis of embankment using simple calculations – practitioner’s
0.2 4 approach. Computer and Geotechnics. In print.
Fill Thickness (m)

0.1
2 Indraratna, B., Perera, D., Rujikiatkamjorn, C. & Kelly, R. 2015. Soil
disturbance analysis due to vertical drain installation. Proc.
Layer Compression (m)

0
0.0 Institution of Civil Eng. Geotech. Engineering, 168 (3), 236-246.
-2 Ladd, C.C. and DeGroot, D.J. 2003. Recommended Practice for Soft
-0.1 -4 Ground Site Characterization. The Arthur Casagrande Lecture,
-6
Proceedings of the 12th Panamerican Conference on Soil
-0.2
Extensometer measurement
Mechanics and Geotechnical Engineering, Boston, MA, Vol. 1, pp.
Proposed model - varied Cr -8 3-57.
-0.3 Conv. Model - constant Cr NAVFAC 1971. Soil mechanics, foundations and earth structures.
-10
Fill thickness Design Manual DM-7, Naval Facilities Engineering Command.
-0.4 -12
Pineda, J.A, Suwal, L. P., Kelly R. B., Bates L., and Sloan S. W, 2016.
10

100

1000

Characterisation of Ballina clay, Géotechnique, 66: 556-577

Time (days) Tavenas, R., Jean, P., Leblond, P., and Leroueil, S. 1983. The
Figure 10. Comparison of predictions and extensometer results between permeability of natural soft clays, Part II: Permeability
RL-7m and RL-10m characteristics. Can. Geot. J., 20(4), 645-660.
Walker R., Indraratna B. and Rujikiatkamjorn C. (2012). Vertical drain
Figure 11 shows the measured soil compression between RL- consolidation with non-Darcian flow and void-ratio-dependent
4 m and RL-7 m obtained from extensometer results. Both the compressibility and permeability. Géotechnique 50 (1), 89-97.

- 3034 -