Acta Geotechnica

AVISA: Anisotropic Visco ISA model and its performance at cyclic loading
--Manuscript Draft--

Manuscript Number: AGEO-D-19-00109

Full Title: AVISA: Anisotropic Visco ISA model and its performance at cyclic loading

Article Type: Original Research Paper

Keywords: soft soils; clay; silt; cyclic loading; viscosity; anisotropy; rate-dependency; time-
dependency; excessive pore water pressure; partially flow

Abstract: In this work a constitutive model with some interesting capabilities, such as the
simulation of the strain rate and time dependency, the incorporation of small strain and
of fabric effects without restriction on the overconsolidation ratio is proposed. The small
strain effects of clays has been simulated through an extended ISA plasticity
formulation (first version of ISA plasticity has been proposed for sand by Fuentes &
Triantafyllidis). The rate dependency is captured by a third strain mechanism (in
addition to the total and the hypoplastic strain) that is dependent on the stress and the
void ratio. The loading surface has been well defined in order to provide a clean and
complete formulation for the overconsolidation ratio. The degree of nonlinearity
presented in the reference model [2] turned to be unsuitable as it does not provide
information about the preloading state of the material. For this purpose it has been
reformulated to account on the influence of OCR. Some investigations regarding the
inherent anisotropy has shown that the effective stress path in p - q space prior to the
appearance of the cyclic mobility is signicantly inclined towards the left upper corner of
the p - q plane for Kaolin Clay and Lower Rhine Clay and that the inherent anisotropy
of clays can be described through the transversal isotropy. Consequently the
hypoelastic stiffness has been transformed to a transversal hypoelastic one.
Furthermore, the last cycles at cyclic mobility show a prevented dilatancy at loading
and a contractant behaviour after an elastic regime is overcomed during unloading. At
stress reversal both clays first exhibited an increase of the mean pressure in contrast
to sand whereas the stress path on reversal starts with the highest contractancy. This
can present a reason why most clays do not reach liquefaction p = 0 kPa due to cyclic
loading. The performance of the model has been inspected by simulating the database
with about 50 cyclic undrained triaxial (CUT) tests on the low plasticity Kaolin Clay
published by Wichtmann & Triantafyllidis including variation of the deviatoric stress
amplitude, initial stress ratio, displacement rate, overconsolidation ratio and cutting
direction and 4 CUT tests performed on the high plasticity Lower Rhine Clay, whereby
the influence of the displacement rate as well as the deviatoric amplitude has been
studied. The simulations showed to be in a very good agreement with the experimental
observations.

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
 Click here to download Manuscript AVISA_TT.pdf

1
2 AVISA: Anisotropic Visco ISA model and its performance at cyclic loading
3
4 Merita Tafilia,∗, Theodoros Triantafyllidisa,b
5
a Institute
6 of Soil Mechanics and Rock Mechanics (IBF), Karlsruhe Institute of Technology, Karlsruhe 76131, Germany
b Professor and Director of IBF
7
8
9
10
11
12 Abstract
13 In this work a constitutive model with some interesting capabilities, such as the simulation of the strain rate and time
14 dependency, the incorporation of small strain and of fabric effects without restriction on the overconsolidation ratio is
15
proposed. The small strain effects of clays has been simulated through an extended ISA plasticity formulation (first
16
version of ISA plasticity has been proposed for sand by Fuentes & Triantafyllidis). The rate dependency is captured
17
18 by a third strain mechanism (in addition to the total and the hypoplastic strain) that is dependent on the stress and
19 the void ratio. The loading surface has been well defined in order to provide a clean and complete formulation for the
20 overconsolidation ratio. The degree of nonlinearity presented in the reference model [2] turned to be unsuitable as it does
21 not provide information about the preloading state of the material. For this purpose it has been reformulated to account
22 on the influence of OCR. Some investigations regarding the inherent anisotropy has shown that the effective stress path
23 in p − q space prior to the appearance of the cyclic mobility is significantly inclined towards the left upper corner of the
24 p − q plane for Kaolin Clay and Lower Rhine Clay and that the inherent anisotropy of clays can be described through
25 the transversal isotropy. Consequently the hypoelastic stiffness has been transformed to a transversal hypoelastic one.
26 Furthermore, the last cycles at cyclic mobility show a prevented dilatancy at loading and a contractant behaviour after
27 an elastic regime is overcomed for unloading. At stress reversal both clays first exhibited an increase of the mean pressure
28 in contrast to sand whereas the stress path on reversal starts with the highest contractancy. This can present a reason
29 why most clays do not reach liquefaction p = 0 kPa due to cyclic loading. The performance of the model has been
30 inspected by simulating the database with about 50 cyclic undrained triaxial (CUT) tests on the low plasticity Kaolin
31 Clay published by Wichtmann & Triantafyllidis including variation of the deviatoric stress amplitude, initial stress ratio,
32 displacement rate, overconsolidation ratio and cutting direction and 4 CUT tests performed on the high plasticity Lower
33 Rhine Clay, whereby the influence of the displacement rate as well as the deviatoric amplitude has been studied. The
34
simulations showed to be in a very good agreement with the experimental observations.
35
36 Keywords: soft soils, clay, silt, cyclic loading, viscosity, anisotropy, rate-dependency, time-dependency, excessive pore
37 water pressure
38
39
40 1. Introduction On the other side, most natural clays show anisotropic
41 behaviour due to their mode of deposition and the elon-
42 Structures subjected to cyclic loading experience ir- gated shape of the particles [6, 7, 8]. This anisotropy,
43 reversible displacements resulting from the accumulated known as inherent anisotropy, along with the time depen-
44 strains within the underlying soil. A rapid cyclic load- 20 dent phenomena of soft soils have practical consequences
455 ing of a fully saturated soil leads under undrained condi- for i.e. the passive lateral thrust on piles, which has been
46 tions to a significant reduction of the effective stress. For the target of many investigations in the last decades. By
47 an accurate prediction of the settlement and the bearing
48 creeping slopes or construction of an embankment on such
capacity of an installed structure as for example a vibro- formations, passive time-dependent lateral pressure may
49
replacement with stone columns the cyclic behaviour of the 25 occur to the pile shaft. This may lead to deformations
50
10 underground consisting of clay as well as its time depen- of the pile foundation and of the superstructure. Be-
51
52 dent behaviour is of high importance. The majority of an- sides these effects, the behaviour of clays depends not
53 alytical settlement calculation methods pertain to primary only on the material state, but also on the strain ampli-
54 settlement only, thus the same value of the settlement im- tude. Researchers [9, 3, 10, 11, 12, 13, 14] have found that
55 provement factor tends to be applied to both primary and 30 clays behave elastically only under very small strain am-
56
15 creep settlements [4, 5]. plitudes |ε1 | < 0.01 %. Under medium strain amplitudes
57 0.01 % < |ε1 | < 1 %, the so-called small strain effects,
58 namely the stiffness increase due to reversal loading and
∗ Correspondingauthor
59 the reduction of the plastic strain rate take place. Finally,
Email address: merita.tafili@kit.edu (Merita Tafili)
60
61 Preprint submitted to Journal of LATEX Templates March 5, 2019
62
63
64
65
 35
1 when the soil is sheared under very large deformations, it without changing the model from non-viscous to viscous
2 tends asymptotically to the critical state, which may re- clays while keeping the same set of material parameters.
3 sult to a failure. Thus, in order to simulate each of these
4 effects the usage of an appropriate model is recommended
2. ISA-plasticity for the small strain stiffness
5 e.g. [15, 16, 17, 18, 19, 20]. Yet, each of these effects is
6
40 important for the investigations of the soft soil behaviour 95 In order to acurately describe the behaviour of gran-
7 in urban excavations with shoring walls subjected to a low ular soils under cyclic loading, in particular for a bet-
8 number of cycles. This cyclic loading may be attrubited ter prediction of the small strain stiffness and the shear
9 to the excavation shoring and backfilling processes or in modulus degradation curve, Niemunis and Herle [22] pro-
10 case of offshore foundations with a large number of cyclic posed the concept of the intergranular strain (IS). In 2015
11
45 loading due to wave and wind actions or for the long term100 Fuentes and Triantafyllidis [1] reformulated the intergran-
12 design of stone columns as soil replacement subjected to a ular strain and named it the ISA-model, hereafter denoted
13 large number of cycles carrying traffic loading. In this pa- as ISA-plasticity, introducing an elastic locus in the inter-
14 per a single model capturing all these effects is proposed.
15 granular strain space. Tab. 2 provides a summary of the
The paper examines the anisotropic material response of models developed unter the ISA-plasticity framework. In
16
50 clays resulting in transverse isotropy. Only one additional105 the following lines the ISA-plasticity formulation for sand
17
18 parameter is required to describe this type of inherent elas- [1, 23, 2, 24, 25] is summarized and its extensions to cap-
19 tic anisotropy [21]. Among all strain amplitudes the strain ture the specific behaviour of soft soils are described.
20 rate dependency of clays is experimentally investigated. The main feature of the ISA-plasticity formulation for sand
21 These effects are incorporated into the proposed model by is the incorporation of the elastic locus as a strain-type
22
55 an extension and modification of the (hypo)elastic stiff-110 yield surface. It describes a hypersphere with diameter R
23 ness tensor and the consideration of an additional strain and depends on the intergranular strain tensor h and the
24 mechanism, whose intensity depends on the material vis- kinematic hardening tensor c:
25 cosity as depicted in [2]. The model is proposed under
26 the platform of the ISA-plasticity, which has shown good FH = kh − ck − R/2 = 0 (1)
27
60 simulation results for cyclic loading and captures well the The evolution of the intergranular strain tensor is described
28 small strain effects for sands. These advantages of the in accordance with elastoplastic formulations introducing
29 ISA-plasticity has been extended and adjusted to the me-115 the consistency parameter λ̇H and the flow direction N:
30 chanical behaviour of clays.
31 The structure of this article is as follows: At the begin- ḣ = ε̇ − λ̇H N, (2)
32
65 ning, the formulation of the proposed model is explained.
33 whereby λ̇H fulfills the consistency condition ḞH = 0 and
Then, some details about the numerical implementation
34 is derived to:
and the integration scheme are given. The performance of
35 hN : ε̇i hN : ε̇i
the model is finaly illustrated in the case of undrained tri- λ̇H = = . (3)
36
axial cyclic simulations of two different clays: the medium N : N + N : c̄ 1 + HH
37
70
38 plasticity Kaolin Clay and the high plasticity Lower Rhine For the plastic condition FH = 0 an associated flow rule
39 Clay. The experiments for the Kaolin Clay are borrowed N = ∂FH /∂h and a kinematic hardening c of the yield sur-
40 from [3], whereas the experiments for the Lower Rhine120 face are introduced as a transition state until the bounding
41 Clay are performed by the authors and Wichtmann as will surface (BS) is reached:
42 be reported in a subsequent paper. The simulations are
43
75 carefully analyzed regarding the strain rate effects, dif- FHb = khk − d · R = 0. (4)
44 ferent overconsolidation ratios, various initial stress ratios
The BS has been formerly proposed in [1] according to
45 and mean preassures and different sample cutting direc-
the bounding surface plasticity characteristics [26] repre-
46 tions. The samples where subjected to various deviatoric
senting it similarly to the yield surface as a hypersphere
47 amplitudes.
48 125 in the IS-space, but with twice the diameter of the yield
80 The proposed model is able to describe the material be-
49 locus FHb = khk − R = 0. Following Eq. (4) we propose
haviour of viscous and non viscous clays under cyclic as
50 a changed size of the bounding surface through the new
well as monotonic loading, capturing also the present in-
51 parameter d. Hence, for sand this parameter can be set
herent anisotropy of clays. The model covers a wide range
52 to d = 1 yielding the formulation of [1]. For clays, the ex-
of strain amplitudes without restriction on the overcon-
53 130 perimental experience gained in this work has shown that
85 solidation ratio (OCR). Normal consolidated as well as
54 d tends to d = 2. The indroduction of a new size of the
slightly and heavily over consolidated clays behaviour can
55 bounding surface has required the adjustment of the fol-
be simulated. Simulations of tests with different initial
56 lowing equations compared to [1].
57 stress ratios show the good performance of the model in
The kinematic hardening tensor represents the center of
58 capturing the induced anisotropy of Kaolin Clay. From
135 the yield surface and reads:
59
90 the point of constitutive modeling, this is very interesting
60 because the proposed model is able to simulate each effect ċ = λ̇H c̄, (5)
61 2
62
63
64
65
 1 In order to fulfill the bounding constraint khk ≤ R the [28] χ = χ0 + εa (χmax − χ0 )). Laboratory tests dealing
2 hardening mechanism should evolve towards the image165 with cyclic behaviour of clays with different number of
3 tensor at the BS (projection in direction of N) denoted cycles [11] has shown that the plastic accumulation rate
4 by cb . Thereby, the hardening function is proposed to reduces for increasing number of consecutive cycles before
5
140 read: reaching the critical state. This effect can be described
6 → through a new state variable εa [28], which distinguishes
7 c̄ = β (cb − c)/(d · R), cb = R (d − 1/2) ε̇ (6)170 whether the soil is performing a few or severeal consecutive
8 cycles. Its evolution rate is controlled through a material
whereby β = f (β0 , σ) is here proposed as a variable that
9 parameter Ca and reads:
controls the hardening rate as will be defined in the sub-
10
11 sequent section. For isotropic conditions it takes the value ε̇a = Ca /R (1 − yh − εa ) kε̇k. (10)
12 of the material parameter β(q = 0) = β0 .
13 Thus, for a few cycles εa ≈ 0 ⇒ χ = χ0 and the IS expo-
14 nent proposed by [1] is recovered. εa → 1 is reached when
15 175 several consecutive cycles have been experienced and the
16 exponent then increases to χ → χmax . χ0 and χmax are
17 material parameters that can be calibrated for a few and
18 several number of cycles, respectively.
19 The increase of the elastic stiffness at unloading is achieved
20 180 through the scalar function m = mR +(1−mR ) yh , whereby
21 mR is a material parameter introduced first in [22, 9].
22 If the stress loops are performed at the critical state, even-
23 though the strain rate may be small |∆ε| < R the material
24 yield surface
response is fully mobilized. To capture this effect we pro-
25 bounding surface 185 pose a relation for the scalar function β, which controls
26 the hardening rate of the IS (compare Eq. (6)):
27
28 3
Figure 1: Yield surface and bounding surface of the extended ISA β = β0 + (2 − β0 ) (krk/krc k) , (11)
29 plasticity for clays. Example when the intergranular strain is fully
30 mobilized h = hb and c = cb with the stress ratio r and its image at the critical state
31 rc .
32 The formulation of the general model might at first sight
33
145 2.1. Mechanical model 190 appear high, but under some specific assumptions it re-
34 According to the ISA-plasticity, for material states in- duces to some model families that possibly are simpler to
35 side the IS yield surface k∆εk < R the response of the understand, as outlined in Tab. 1. For example if the yield
36 model is visco-elastic thus λ̇H = 0 and the intergranular surface radius goes towards R → 0 than the ISA-plasticity
37 strain evolves equaly to the applied strain. The (visco- is disregarded and we obtain the constitutive model for-
38 elastic) constitutive equation for these states reads:
150 195 mulation under mobilized states.
39
vis


40 σ̇ = m E : ε̇ − ε̇ . (7)
41 3. Constitutive model under mobilized states
42 Once the yield surface is reached, the kinematic hardening
43 of the centre of FH activates c̄ with the IS consistency The fully mobilized state is reached when the inter-
44 parameter λ̇H , which defines also the plastic evolution of granular strain h lies at the IS bounding surface FHb =
45 the IS state variable h. The area between the yield surface 0 ⇒ yh = 1 ⇒ m = 1. Hence the constitutive equation for
46
155 and the bounding surface is termed as a trasition area200 clays at fully mobilized states reads:
47 between the visco-elastic Eq. (7) and the fully mobilized
σ̇ = E : ε̇ − ε̇hp − ε̇vis .


48 state yh = 1: (12)
49
σ̇ = m E : ε̇ − yh ε̇hp − ε̇vis .


50 (8) This relation is visco-hypoplastic as described in [21]. The
51 evolution equation of the hypoplastic strain rate ε̇hp reads:
The bounding condition h = hb i.e. the fully mobilized
52 ε̇hp = Y mkε̇k (13)
state should be asymptotically reached ensured with the
53
160 scalar function 0 ≤ yh ≤ 1:
54 and for the viscous strain rate ε̇vis :
55 → khb − hk
yh = ρχ hN : ε̇ i,
1/Iv
ρ=1− , (9)

56 d·R 1
ε̇vis = Iv λ m (14)
57 OCR
58 whereby hb = d/2·R N is the image tensor of the IS at the
59 bounding surface. The exponent χ accounts indirectly for with the compression index λ and the viscosity index Iv .
60 the number of constitutive cycles the soil has experienced205 The same direction is used for both plastic strains [24, 2,
61 3
62
63
64
65
 1 Table 1: Simplification of the general constitutive model to some particular forms
2 Name Assumptions Simplified equation Remarks
3
4
5 Elastic R → ∞, Iv = 0 σ̇ = mR E : ε̇ time-independent, no accumulation
6 Visco-elastic R→∞ σ̇ = mR E : (ε̇ − ε̇vis ) time-dependent, no accumulation
7 Hypoplastic R → 0, Iv = 0 σ̇ = E : (ε̇ − ε̇p ) time-independent, monotonic loading
8
Visco-Hypoplastic R→0 σ̇ = E : (ε̇ − ε̇p − ε̇vis ) time-dependent, monotonic loading
9
10 ISA-plasticity none σ̇ = mE : (ε̇ − yh ε̇p − ε̇vis ) time-dependent, monotonic + cyclic loading
11
12
13 Table 2: Summary of the models developed unter the ISA-plasticity framework
14 Name Geomaterial Description Literature
15
16 ISA - Sand (2014) Sand Uses projections in the stress space (similar to [27]) [1]
17
18 ISA - HP Wolffersdorff (2015) Sand Without projections in the stress space [28]
19 ISA - HP Clay (2015) Clay Rate independent [23]
20 ISA - visco HP Clay (2016) Clay Inclusion of rate and time dependency [2]
21
22 ISA - aniso visco HP Clay (2017) Clay inherent anisotropy introduced this work
23
24
25 25, 21]: δil δjk ) The barotropic bulk modulus K of clays linearly in-
26  → creses with increasing mean stress p as shown in [2, 21]:
27 1 r
m = − (krc k − krk) + . (15) p (1 + e)
28 2 krc k K= (17)
29 λ (1 − Y0,max )
30 The equations governing the reference model [2] are sum-
marized in Appendix A. In this section the differences whereby Y0,max = (λ − κ)/(λ + κ) is the degree of non-
31
between the reference model and the AVISA model are235 linearity deducted for normal consolidated states upon un-
32
in particular presented. The new approach for the de- loading [2, 21]. The shear modulus G can be deduced from
33
210

34 scription of fabric effects, as well as the influence of the K and the Poisson ratio ν to read:
35 overconsolidation ratio (OCR), formulated with a param- 3 (1 − 2ν)
eter less than the required number of the reference model G=K . (18)
36 2 (1 + ν)
37 parameters, and the new relation for the degree of nonlin-
38
215 earity Y are described in detail. Most natural clays show anisotropic behaviour because
39 Furthermore it should be noted that experimental studies of their mode of deposition and the elongated shape of
40 [29, 30, 31] has shown that the viscous effects of (highly)240 the particles [21]. This anisotropy, known as inherent
41 overconsolidated clays vanish exponentially. For cyclic anisotropy or fabric, resulting from the deposition pro-
42 loading of a normalconsolidated sample this means that cess tends to induce a horizontal bedding plane in the soil
43
220 the viscous effects decrease exponentially with each cycle. layer and can be regarded as a transverse isotropy [6, 21].
44 In the constitutive Eq. (7) of the AVISA model the vis- For undrained monotonic loading the transverse isotropy
45
cous strain is active also for elastic conditions. Yet, Eq.245 influences the slope of the effective stress path. In case of
46 undrained cyclic loading the described elastic anisotropy
(14) suggests that its magnitude decreases with increasing
47 besides the effective stress slope affects the direction of
48 OCR exponentially. The exponent 1/Iv takes high values
225 for 0 ≤ Iv ≤ 0.5 and ensures the vanishing viscous strain the strain accumulation, whether it occurs in extension
49 or compression. Thus, a performance based design of a
50 for elastic conditions e.g. overconsolidated clays.
250 geotechnical structure is not possible without describing
51 the fabric well.
52 3.1. Inclusion of fabric effects for clays
Furthermore, experiments show that after a stress rever-
53 A hypoelastic stiffness has been introduced in the AVISA sal first an increase of the mean pressure occurs in most
54 model: clays in contrast to sand where the reversal starts with the
55
→→ K 255 highest contractancy. This can also present a reason why
56 Eiso = 3K 1 1 +2G Idev − √ (1 r + r 1) (16)
57 3Mc most clays do not reach liquefaction p = 0 kPa. Thus, also
58 →→
in order to capture the dilatancy/contractancy behaviour
59
230 with the fourth rank deviatoric unit tensor Idev = I− 1 1 of soils well, first the accurate description of elasticity and
60 and the unit tensor for symmetric tensors Iijkl = 1/2(δik δjl + of the inherent anisotropy is required.
61 4
62
63
64
65
260
1 For a complete transverse isotropy five material param- 3.3. Degree of nonlinearity Y
2 eters are required. However, according to the pioneer The degree of nonlinearity is proposed according to
3 work of Graham & Houlsby [6] for the special case of some hypoplastic requirements:
4 anisotropic clays a simplified transversal isotropic elastic-
5 ity is deduced, whereby only three parameters are needed: • isotropic states (q = 0): Y takes the minimum value
6
265 the Young’s modulus E = Ev , the Poisson ratio ν = νh 305 Y = Y0 , whereby the influence of OCR is also con-
7 and the anisotropic factor α obeying the relation: sidered: Y0 = Y0,max (p/pei )2 [2].
8
9 Gh /Gv = (Eh /Ev )1/2 = νh /νvh = α (19) • states lying at the loading surface Fb = 0 or at the
10 critical state surface Fc = 0: experimental observa-
11 The index h stands for the horizontal direction and v vor tions has shown that at these states the hypoplastic
12 vertical direction. Two of these material constants are310 degree of nonlinearity takes the value Y = 1.
13 already present in the proposed model with E = f (κ)
270
14 and ν. Thus, only one additional material parameter α • the degree of nonlinearity should provide information
15 needs to be introduced into the model. Now the question about the state of the material: the loading surface
16 arises how can a stiffness of a model be transformed to a provides the boundary where Y = 1 and inside the
17 transversal one? After some mathematical manipulations loading surface Y decreases towards Y = Y0 which
18 it becomes obvious that each hypoelastic stiffness can be315 is reached at the isotropic axis. This requirement is
19
275 transformed to a transversal one by scaling it according to firstly termed in this work.
20 the following schema [32]:
21 A suitable relation to fulfill these requirements is the fol-
22 Eabcd = Qabij : Eijkl : Qklcd (20) lowing equation:
23 
OCR − 1

24 Qabcd = µac µbd (21) Y = Y0 + (1 − Y0 ) exp − . (24)
krk
25
26 √ √ For isotropic states ||r|| → 0 the relation (24) tends to:
27 µ= α1 + (1 − α)ms ⊗ ms ,
28 lim Y = Y0 (25)
ms is the unit vector along the sedimentation axis e.g. if r→0
29
280 the sedimentation axis is horizontal then ms = {1, 0, 0}.
30 320 thus, the first requirement is fulfilled. States lying at the
31 Hence, only 1 additional parameter termed as anisotropic
loading surface or at the critical surface are normalconsol-
coefficient α compared to the reference model [2] is re-
32 idated OCR = 1, hence Y → 1. The incorporation of the
33 quired.
loading surface as a boundary for the degree of nonlinearity
34 is obvious from Eq. (24). Both the degree of nonlinear-
35 3.2. Overconsolidation ratio OCR
325 ity proposed in Eq. (24) and the reference model’s degree
36
285 Whereas in [2] an interpolation function is used for of nonlinearity Yrefmod [2] are illustrated in Fig. 2. De-
37 OCR introducing an additional material parameter, a more spite the OCR-independence of the Yrefmod , it possesses
38 precise and comprehensive OCR formulation is proposed also numerical issues approaching the isotropic axis as de-
39 herein. Following Hvorslev [33] we introduce the equiva- picted in Fig. 2 a) with dashed green lines. The orange
40 lent isotropic pressure pei = exp((ei0 − e)/λ). After in-330 limit Y = Y0 holds for both definitions equally.
41
290 corporating a loading surface for the stress Fb = 0 where
42
OCR= 1 holds and along the same lines of thoughts as de-
43 4. Inspection of the model’s performance compared
scribed in [21] the OCR = f (p, q, e, θ) is defined for general
44 with laboratory tests
45 3D states by:
46 pei

ei0 − ex

The model’s performance is in this section evaluated
47 OCR = + , pxei = exp (22) by simulating experimental results for two different clays:
pei λ
48 335 Kaolin Clay from [3] and Lower Rhine Clay (experiments
49 conducted by the authors and Wichtmann and not pub-
||r||
 
50 lished yet). The low plasticity Kaolin Clay used for the ex-
e+ = A1/nf ei , A = 1 − . (23)
51 ||rb0 || periments has a liquid limit of wL = 47.2 %, a plastic limit
52 of IP = 12.2 % and a grain density of ρs = 2.590 g/cm .
3
53
295 Hence, OCR is a function of the void ratio e and the stress
54 invariants p, q and the Lode angle θ accounting for mul-340 The highly plastic Lower Rhine Clay stems from the ex-
tidimensional loading paths. Thus, one parameter less is avation side of an opencast lignite mine and has a liquid
55
56 required compared to the reference model [2] as the con- limit of wL = 56.1 %, a plastic limit of IP = 34.1 % and a
3
57 troversial parameter nocr introduced in [2] is omitted in grain density of ρs = 2.675 g/cm [3]. The permeability of
58
300 Eq. (22). Kaolin and Lower Rhine Clay has been determined by the
59 345 authors in a permeability test in a cylinder with variable
60 pressure head during the test resulting to k = 1.3 · 10−9
61 5
62
63
64
65
 1 Table 3: Material parameters
2
3
Description Approx. range Kaolin Lower Rhine Clay
4
5 Transversal (hypo)elasticity
6 λ compression index 10−6 − 1 0.13 0.26
7 κ swelling index 10−7 − 0.6 0.05 0.04
8 νh Poisson ratio 0 − 0.5 0.3 0.2
9
α anisotropic coefficient 0−5 1.8 0.7
10
11 Intergranular strain (IS):
12 mR stiffness factor 1−7 3 5
13 R IS yield surface radius 10−6 − 10−2 2 × 10−4 2 × 10−4
14
d IS bounding surface radius 1 − 10 3 4
15
β0 IS hardening parameter 0−2 0.076 0.076
16
17 χ0 min IS exponent 1 − 10 7 7
18 χmax max IS exponent 20 − 50 40 35
19 Ca accumulation rate factor 0 − 0.2 0.001 0.005
20 Critical and loading surface:
21
Mc CS slope 0.5 − 2 1.0 0.95
22
23 ei0 max void ratio at pref = 1 kPa 0.5 − 5 1.76 2.47
24
25 fb0 loading surface factor 1−3 1.5 1.45
26 Viscosity:
27
Iv viscosity index 0−2 0.015 0.025
28
29
30
31 m/s and k = 3.6 · 10−11 m/s, respectively. Both materials The simulated effective stress paths (red) and the exper-
32 have been tested in disturbed form, i.e. using reconstituted imental results (blue) are given in Fig. 3 for tests C02,
33 samples. The material parameters used for the simulations C04 − C08. It is evident that the effective stress path
34
350 are listed in Tab. 3. 375 before cyclic mobility is significantly inclined towards the
35 left upper corner of the p − q plane [3]. This observation
36 4.1. Simulations with Kaolin Clay is reproduced very well with the AVISA model through
37 In several test series performed by Wichtmann & Tri- the anisotropic coefficient of α = 1.8 resulting in Eh =
38 antafyllidis [3] the stress amplitude q0 , the displacement 3.24 · Ev . The reference model could not reproduce the de-
39 rate ṡ, the initial mean pressure p0 , the OCR, the initial380 scribed slope of the effective stress path [2]. The inherent
40 anisotropy effects also the axial strain path, whereby the
355 stress ratio η0 = q0 /p0 and the cutting direction of the
41 sample first accumulates in extension. Due to the p − q in-
samples have been varied as described in [3, 10]. Each of
42 clination, first the compression triaxial critical state line is
43 these tests has been simulated with the AVISA model and
showed a good agreement with the experimental results reached, thus the accumulation of the axial strain changes
44
(see the following sections). Only one material parameter385 to compression as demonstrated in Fig. 4. The AVISA
45
set listed in Tab. 3 has been used for all simulations. The model captures well all of these features of the material
46
360

notation of the tests with CXX e.g. C02 is the same as in behaviour, whereas the reference model accumulated only
47
48 [3], so that missing information can be depicted from [3]. in extension [2]. For a performance based design of an e.g.
49 flooded opencast mine unter earthquake loading the accu-
50 4.1.1. Variation of deviatoric stress amplitude qampl 390 rate description of these effects is indispensable.
51 The tests C01 − C08 were performed with different de- The tests C05 and C06 have been performed with the same
52
365 viatoric stress amplitudes qampl = {30, 40, 45, 50, 60, 70} deviatoric amplitude qampl = 50 kPa but with different
53 kPa. The initial conditions OCR0 = 1, p0 = 200 kPa, initial void ratios e0 = 1.145 and e0 = 1.252, respec-
54 η0 = 0, ṡ = 0.1 mm/min were the same for all sam- tively. As reported in [3] lower initial void ratios due to
55 ples. In the simulations the same number of cycles as395 higher overconsolidation ratios lead to a higher number
56 documented in [3] has been applied. For the initializa- of cycles to reach cyclic mobility or the failure criterion
57
370 tion of the OCR the simulations used the initial void ratio of |ε1 | = 10 % used in the experiments. The simulations
58 are of course completely in agreement with the described
e0 = ei0 − λ ln(p0 /(1 kPa) OCR).
59 behaviour, presented also in Fig. 5 a) in terms of accu-
60
61 6
62
63
64
65
 a) to sand whereas the reversal starts with the highest con-
1
2 1500 tractancy. This can present a reason why most clays do
Deviatoric stress q [kPa]

3 not reach liquefaction p = 0 kPa. Therefore, the ques-

4 1000 430 tion arises if there is an angle for the cutting direction of
5 500
the sample at which the clay reaches a total degradation
6 of the mean stress similarly to sand. In Sec. 4.1.7 we
7 0 will notice a reduction of the relaxation of the mean stress
8 compared to vertically cutted samples for the same initial
9 -500
435 and boundary conditions.
10
-1000
11 4.1.2. Variation of displacement rate ṡ
12 0 400 1200 Six tests with variation of displacement rate between
13 ṡ = {0.01, 0.05, 0.5, 0.01, 0.05, 0.5} mm/min has been
Mean stress p [kPa]
14 documented in [3] and simulated with the AVISA model.
15 b)
440 Isotropically normalconsolidated samples with an initial
16
1.0 mean stress of p0 = 200 kPa were tested with a deviatoric
17
18 amplitude of qampl = 45 kPa (C09 − C12) and qampl = 50
19 0.8 kPa (C13 − C15).
20 Fig. 6 shows the results in the effective stress plane, whereas
0.6
21 445 the corresponding deviatoric stress - axial strain relations
22 0.4 are for example for tests C13 and C14 illustrated in Fig. 7.
23 In order to distinguish between the effects of stress ampli-
24 tude and displacement rate either tests C09 − C12 or C13
25 0 − C15 are compared. Due to the low plasticity of Kaolin
26 450 an effect of the displacement rate can be seen rather com-
27
Figure 2: Definition of the degree of nonlinearity: a) contour plot of paring the lower rates ṡ = 0.01 and 0.02 mm/min with a
28 Y in p − q space, b) Y in dependence on the stress ratio η. (green = number of cycles to failure Nf = 81 and 58 (C09 − C12)
29 Y from the reference model [2], blue = Y proposed in Eq. (24)) or Nf = 24 and 40 (C13 − C15). The increasing number
30 of cycles to failure with decreasing frequency has been ob-
31 455 served by other authors too [12, 13, 11, 14]. In agreement
32
400 mulated pore water pressure against the number of cycles
with this tendency is an acceleration of the accumulated
33 N and Fig. 5 b) whereas the accumulated axial strain
pore water pressure and of the accumulated axial strain for
34 against N is illustrated. This behaviour is observed also
lower displacement rates as depicted in Fig. 8. The sim-
35 by other authors [11, 12, 34]. Moreover, Fig. 3 d) shows
ulations show a good agreement with all effects shown by
36 a better agreement for the first cycle between experiment
37 460 the experiments. The frequency dependence of the model
405 and simulation. Due to the experimental data scatter of
38 is established through the viscous strain εvis with the evo-
the initial void ratio the first cycle in some simulations do
39 lution relation given in Eq. (14). The inherent anisotropy
not coincide exactly with the experiment.
40 resulting in a compressive accumulation of the axial strain
The tests pair C03 and C04 were performed with the same
41 and an inclination of the effective stress path to the upper
amplitude of qampl = 45 kPa and showed for the higher ini-
42 465 left of the p − q diagram is very well captured by the simu-
410 tial void ratio of e0 = 1.224 corresponding to the test C03
43 lations with the AVISA model. Responsible for its simula-
an increasing number of cycles up to failure in opposition
44 tion is the transversal (hypo)elasticity with the anisotropic
to the above described effect for tests C05 and C06. The
45 coefficient of α = 1.8 (isotropic would be α = 1 and would
46 authors addressed this occurrence to the natural scatter of
coincide with the reference model, for which no simulations
47 experimental data [3]. The AVISA simulations showed a
470 with different ṡ are shown in [2]).
48
415 good agreement with the test C04 as illustrated in Fig. 3
49 b). 4.1.3. Variation of initial mean pressure p0
50 The last cycles at cyclic mobility show a prevented di-
Six isotropically normalconsolidated samples with vari-
51 latancy at loading and a contractant behaviour after an
ation of the initial mean pressure p0 = {50, 75 100, 125,
52 elastic regime is overcomed for unloading. This behaviour,
150, 250} kPa have been tested with deviatoric ampli-
53
420 resulting also in an axial strain ”hystheresis” in Fig. 4 is
475 tudes resulting in a constant amplitude-pressure ratio of
54 not accounted by the model as for its description the pre-
ζ = qampl /p0 = 0.2 and a displacement rate of ṡ = 0.1
55 cise description of the energetical correct hyperelasticity of
56 mm/min.
the Kaolin is necessary. The authors are currently working
57 A comparison between test C21 with p0 = 250 kPa pre-
in this direction. Yet it is of high importance to note that
58 sented in Fig. 9 f) with Nf = 995 and C5 with p0 = 200
425 at the stress reversal clays first exhibit under undrained
59 480 kPa presented in Fig. 3 c) with Nf = 68 shows the sig-
conditions an increase of the mean pressure in contrast
60 nificant influence of the initial mean stress. A decrease of
61 7
62
63
64
65
 1
2
a) C02: qampl = 40 kPa b) C04: qampl = 45 kPa c) C05: qampl = 50 kPa
Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

3 80 80 80
4
5 40 40 40
6 0 0 0
7
8 −40 −40 −40
9
10 −80 −80 −80
11 0 40 80 120 160 200 0 40 80 120 160 200 0 40 80 120 160 200
12 Mean stress p [kPa] Mean stress p [kPa] Mean stress p [kPa]
13
14 d) C06: qampl = 50 kPa e) C07: qampl = 60 kPa f) C08: qampl = 70 kPa
Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

15
80 80 80
16
17 40 40 40
18
19 0 0 0
20
−40 −40 −40
21
22 −80 −80 −80
23
0 40 80 120 160 200 0 40 80 120 160 200 0 40 80 120 160 200
24
Mean stress p [kPa] Mean stress p [kPa] Mean stress p [kPa]
25
26
27 Figure 3: Effective stress paths in undrained cyclic tests with different stress amplitudes (OCR0 = 1, p0 = 200 kPa, η0 = 0, ṡ = 0.1 mm/min).
28 (simulation=red, experiment=blue)
29
30
31 a) C07: qampl = 60 kPa b) C08: qampl = 70 kPa
Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

32
33 80 80
34
40 40
35
36 0 0
37
38 −40 −40
39 −80 −80
40
41 −6 −2 2 6 10 −6 −2 2 6 10
42 Axial strain ε1 [%] Axial strain ε1 [%]
43
44
45 Figure 4: Deviatoric stress vs. axial strain in undrained cyclic tests with different stress amplitudes (OCR0 = 1, p0 = 200 kPa, η0 = 0, ṡ = 0.1
46 mm/min). (simulation=red, experiment=blue)
47
48
49 the mean pressure leads to an acceleration of the accumu- ditions were chosen to p0 = 100 kPa, qampl = 30 kPa and
50 lated pore water pressure and of the residual strain. This ṡ = 0.1 mm/min.
51 behaviour is also captured with the model simulations il- The expected dilative response in the first cycle of undrained
485
52 lustrated in Figs. 9 and 10. 495 cyclic loading of overconsolidated samples is recognisable
53 The mismatch in the first cycle of the effective stress paths in the negative accumulated pore water pressure at N = 1
54 can be attributed to the fluctuation of the experimental in Fig. 11 a). Afterwards the material response is contrac-
55 initial void ratio. tive ṗ < 0 ⇒ u̇acc > 0. The number of cycles to failure
56 Nf = {116, 531} grows significantly with the OCR0 =
57 4.1.4. Variation of initial overconsolidation ratio OCR0 500 {1.5, 2.5}. The rate of εacc
1 and uacc decreases with OCR0
58
490 The influence of the initial overconsolidation ratio was resulting in a stiffer response of the material for higher
59 studied in 2 tests with OCR0 = {1.5, 2}. The initial con- OCR0 . All these effects are properly described by the
60
61 8
62
63
64
65
 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16 Figure 5: Undrained cyclic tests with different stress amplitudes (OCR0 = 1, p0 = 200 kPa, η0 = 0, ṡ = 0.1 mm/min): a) normalized
accumulated pore water pressure uacc /p0 as a function of the number of cycles, b) accumulated axial strain as a function of the number of
17 cycles. (simulation=solid lines, experiment=dashed lines)
18
19
20 a) C09: ṡ = 0.01 mm/min b) C11: ṡ = 0.05 mm/min c) C12: ṡ = 0.5 mm/min
21
Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

22 80 80 80
23 40 40 40
24
25 0 0 0
26
27 −40 −40 −40
28 −80 −80 −80
29
30 0 40 80 120 160 200 0 40 80 120 160 200 0 40 80 120 160 200
31 Mean stress p [kPa] Mean stress p [kPa] Mean stress p [kPa]
32
33 d) C13: ṡ = 0.01 mm/min e) C14: ṡ = 0.05 mm/min f) C15: ṡ = 0.5 mm/min
Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

34 80 80 80
35
36 40 40 40
37
0 0 0
38
39 −40 −40 −40
40
41 −80 −80 −80
42 0 40 80 120 160 200 0 40 80 120 160 200 0 40 80 120 160 200
43 Mean stress p [kPa] Mean stress p [kPa] Mean stress p [kPa]
44
45
46 Figure 6: Effective stress paths in undrained cyclic tests with different displacement rates (OCR0 = 1, p0 = 200 kPa, η0 = 0, qampl = 45
47 kPa). (simulation=red, experiment=blue)
48
49
50 models simulations, solid lines in Fig. 11. The effective stress paths are presented in Fig. 12 a) −
51 c), whereby the slope and the relaxation of the mean pres-
52 4.1.5. Variation of initial stress ratio η0 sure are properly reproduced with the model. Both, the
53
505 Normal consolidated samples with initial mean pres-515 experiments and the simulations, showed that the failure
54 sure of p = 200 kPa has been anisotropically consolidated criterion is reached due to excessive axial strain accumula-
55 to different deviatoric stresses leading to various initial tion and not due to large strain amplitudes as observed for
56 stress ratios in compression η0 = {0.375, 0.25, 0.125} and isotropically consolidated samples (compare Fig. 4 with
57 in extension η0 = {−0.125, −0.25, −0.5}. Then cyclic Fig. 13). Furthermore, the accumulation of the pore wa-
58
510 undrained shearing has been performed with qampl = 30520 ter pressure almost stabilized after a certain number of
59 kPa and displacement rate of ṡ = 0.1 mm/min. cycles and consequently a complete relaxation of p is not
60
61 9
62
63
64
65
 1
2 a) C13: ṡ = 0.01 mm/min b) C14: ṡ = 0.5 mm/min
Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

3 80 80
4
5 40 40
6 0 0
7
8 −40 −40
9
10 −80 −80
11 −6 −2 2 6 10 −6 −2 2 6 10
12 Axial strain ε1 [%] Axial strain ε1 [%]
13
14
15 Figure 7: Deviatoric stress vs. axial strain in undrained cyclic tests with different displacement rates (OCR0 = 1, p0 = 200 kPa, η0 =
16 0, qampl = 45 kPa). (simulation=red, experiment=blue)
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34 Figure 8: Undrained cyclic tests with different displacement rates (OCR0 = 1, p0 = 200 kPa, η0 = 0, qampl = 45 kPa): a) normalized
35 accumulated pore water pressure uacc /p0 as a function of the number of cycles, b) accumulated axial strain as a function of the number of
36 cycles. (simulation=solid lines, experiment=dashed lines)
37
38
39 expected. amplitude. The existence of a lower limit of the number
40 The simulation in Fig. 12 c) shows a partially flow of the of cycles to failure for example Nf = 19 is indicated, as
41 sample when approaching the extension critical state line. for both tests C35 and C36 19 cycles are passed through
42
525 This behaviour is not evidenced by these experimentes,545 until the failure criterion of |ε1 | = 10 % is reached.
43 but it has been documented in some other works [35, 36]. The accumulated pore water pressure stabilizes after a cer-
44 tain number of cycles and it indicates the existence of a
45 4.1.6. Variation of stress amplitude qampl at anisotropic threshold value of uacc = 0.5 p0 for each test. The simu-
46 consolidated samples lations are in very good agreement with the experimental
47 The performance of the model is tested at anisotropic550 evidence presented by [3].
48
530 initial stress ratios also for different stress amplitudes qampl =
49 {20, 25, 30, 35, 40} kPa (notation in [3] C33, C34, C25, 4.1.7. Samples cut out in horizontal direction
50 C35 and C36, respectively). The stress-strain relation is The inherent anisotropy incorporated in the model re-
51 exemplary illustrated in Fig. 13 for C35, whereby the ax- quires the calibration of the material parameter α termed
52 ial strain amplitude is observed to remain constant during as anisotropic coefficient and the definition of the unit vec-
53
535 the cycles resulting in a constant but considerable rate555 tor along the sedimentation axis ms (not to be mixed up
54
of the irreversible axial strain. All these effects are cap- with the tensorial flow rule denoted with m). The cali-
55
56 tured by the simulation with the AVISA model. For other bration of α is in detail described in [21] and once more
57 deviatoric amplitudes a similar behaviour as depicted in summarized in Appendix B. The vector ms depends on
58 Fig. 14 b) was observed. Thereby the accumulated axial the sedimentation axis of the material and we define for a
59
540 strain is plotted versus the number of cycles. Obviously560 vertical cutting direction ms = {1, 0, 0} and consequently
60 the number of cycles to failure decreases with increasing for a horizontal cutting direction ms = {0, 0, 1}. In [3]
61 10
62
63
64
65
 1
a) C16: p0 = 50 kPa b) C17: p0 = 75 kPa c) C18: p0 = 100 kPa
2 40 40 40
Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

3
4 20 20 20
5
6 0 0 0
7
8 −20 −20 −20
9
10 −40 −40 −40
11 0 20 40 60 0 20 40 60 80 0 40 80 120
12 Mean stress p [kPa] Mean stress p [kPa] Mean stress p [kPa]
13
14 d) C19: p0 = 125 kPa e) C20: p0 = 150 kPa f) C21: p0 = 250 kPa
40 80 80
Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

Deviatoric stress q [kPa]

15
16
17 20 40 40
18
19 0 0 0
20
21 −20 −40 −40
22
23 −40 −80 −80
0 40 80 120 0 40 80 120 160 0 50 100 150 200 250
24 Mean stress p [kPa] Mean stress p [kPa] Mean stress p [kPa]
25
26
27 Figure 9: Effective stress paths in undrained cyclic tests with variation of the initial mean pressure (OCR0 = 1, η0 = 0, ṡ = 0.1 mm/min).
28 (simulation=red, experiment=blue)
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
Figure 10: Undrained cyclic tests with variation of the initial mean pressure (OCR0 = 1, η0 = 0, ṡ = 0.1 mm/min): a) normalized
47 accumulated pore water pressure uacc /p0 as a function of the number of cycles, b) accumulated axial strain as a function of the number of
48 cycles. (simulation=solid lines, experiment=dashed lines)
49
50
51 despite of samples cut out in vertical direction, three tests dent in the p−q plane as illustrated in Fig. 15 and a decay
52 2 × η0 = 0 (C40 and C41) and 1 × η0 = 0.5 (C42) on sam- of the accumulation rate of the axial strain can be observed
53 ples cut out in horizontal direction have been performed in Fig. 16 b). These observations are of course mirrored in
54
565 as well. the accumulated pore water pressure too. Moreover, the
55 The experimental results and simulations with the AVISA575 inclination of the effective stress path is opposite to the
56 model are presented in Figs. 15 and 16. The lower initial inclination of samples cut out in vertical direction (com-
57 void ratio of the test C41 implies a lightly overconsolidated pare Fig. 3 with Fig. 15). The final effective stress loop of
58 initial state with OCR = exp[(1.76 − 1.07)/0.13]/200 = horizontal samples is practically mirrored along the p-axis.
59
570 1.48. As a result a stiffer response of the material is evi- Due to the higher elasticity, the rates of the accumulation
60
61 11
62
63
64
65
 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17 Figure 11: Undrained cyclic tests with variation of the initial overconsolidation ratio (p = 100, kPa, q
0 ampl = 30 kPa, η0 = 0, ṡ = 0.1
18 mm/min): a) normalized accumulated pore water pressure uacc /p0 as a function of the number of cycles, b) accumulated axial strain as a
19 function of the number of cycles. (simulation=solid lines, experiment=dashed lines)
20
21
22
23
24
25
26
27
28
29 "flow"
30
31
32
33
34 Figure 12: Effective stress paths in undrained cyclic tests with variation of the initial stress ratio (OCR0 = 1, p0 = 200, kPa ṡ = 0.1 mm/min).
35 (simulation=red, experiment=blue)
36
37
38 Consequently, for a given amplitude or another boundary
39 condition the horizontal samples can withstand a much
40 585 higher number of cycles to failure than the vertical ones.
41 All these observations are very well reproduced with the
42 AVISA model. Furthermore, the final relaxation of the
43 mean pressure at the failure criterion of |ε1 | = 10 % is
44 lower for horizontal samples. This occurrence gives a hint
45 590 to the answer of the question if a specific angle for the
46 cutting direction exists, at which a total relaxation near
47 to liquefaction p → 0 would be experienced.
48
49 4.2. Simulations with Lower Rhine Clay
50
51 Figure 13: Deviatoric stress vs. axial strain in undrained cyclic The experiments with the high plasticity Lower Rhine
52 tests with different deviatoric amplitudes on samples consolidated 595 Clay has been performed by the authors and Wichtmann in
53 anisotropically (OCR0 = 1, p0 = 200 kPa, η0 = 0.375). (simula- IBF, KIT and will be published once a complete database
54 tion=red, experiment=blue) is finished. In the following some results of undrained
55 cyclic triaxial tests with different deviatoric amplitudes
56 as well as with variation of the displacement rate are pre-
580 of pore water pressure as well as of the axial strain rate600 sented. By testing a specified stress amplitude q
57 ampl , the
58 are much lower for the horizontally cutted samples com- loading direction was changed once the defined deviatoric
59 pared to the vertical ones (compare Fig. 5 with Fig. 16). stress q was reached. The cyclic loading was stopped when
60
61 12
62
63
64
65
 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Figure 14: Undrained cyclic tests with different deviatoric amplitudes on samples consolidated anisotropically (OCR0 = 1, p0 = 200 kPa,
17 η0 = 0.375): a) normalized accumulated pore water pressure uacc /p0 as a function of the number of cycles, b) accumulated axial strain as a
18 function of the number of cycles. (simulation=solid lines, experiment=dashed lines)
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38 Figure 15: Effective stress paths in undrained cyclic tests with samples cut out in horizontal direction (OCR0 ≈ 1, p0 = 100, kPa, η0 = 0,
39 qampl = 45 kPa, ṡ = 0.1 mm/min). (simulation=red, experiment=blue)
40
41
42 a certain value of axial strain | ε1 |= 10 % was reached, as
Table 4: Programme of undrained cyclic triaxial tests on Lower
43 done also for the Kaolin tests [3]. The displacement rates Rhine Clay
44
605 were chosen low enough to guarantee a homogeneous field
45 Test p0 OCR0 q0 qampl ṡ i Nf
of pore pressure within the samples. The material pa- h mm
46 rameters used for all the simulations are depicted in Tab. [kPa] [−] [kPa] [kPa] [−]
47 min
3. Tab. 4 provides a short overwiev of the experiments
48 RBT-4 100 1 0 60 0.02 1.5
performed on Lower Rhine Clay with the specifications
49
610 of initial conditions in terms of: mean pressure p0 , over- RBT-5w 100 1 0 60 0.05 0.5
50
consolidation ratio OCR0 and deviatoric stress q0 . The
51 RBT-7 100 1 0 45 0.02 2.5
52 deviatoric amplitude qampl , the displacement rate ṡ and
the number of cycles to failure Nf which the experiments RBT-8 100 1 0 40 0.02 4
53
54 were subjected to are also documented in Tab. 4.
55
615 Fig. 17 shows the results for the test RBT-4 with a
56 deviatoric amplitude of qampl =60 kPa and Nf ≈ 1.5 un-620 % descreases to approx. 0.5, see Fig. 18. The decay of
57 til the failure criterion of |ε1 | = 10 % is reached. For the the number of cycles to failure with increasing frequency
58 same amplitude but with a 2.5 times faster displacement of the same amplitude has been observed by other authors
59 rate the number of cycles to reach an axial strain of 10 too and in accordance with this tendency is an accelera-
60
61 13
62
63
64
65
 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17 Figure 16: Undrained cyclic tests with different deviatoric amplitudes samples cut out in horizontal direction (OCR0 ≈ 1, p0 = 100, kPa,
18 η0 = 0, qampl = 45 kPa, ṡ = 0.1 mm/min): a) normalized accumulated pore water pressure uacc /p0 as a function of the number of cycles, b)
accumulated axial strain as a function of the number of cycles. (simulation=solid lines, experiment=dashed lines)
19
20
21 tion of the accumulated pore water pressure to observe. shown that the effective stress path before cyclic mobility
22
625 Thus, the accumulated axial strain increases for lower dis-660 is significantly inclined towards the left upper corner of
23
placement rates. The simulations show a good agreement the p − q plane as documented also in [3] for Kaolin Clay
24
with all effects shown by the experiments. The frequency and that the inherent anisotropy of clays can be described
25
26 dependence of the model is established through the vis- through the transversal isotropy as theoretically described
27 cous strain. Figs. 19 and 20 present the experimental and also in the pioneer work of Graham & Houlsby [6]. Due
630
28 simulated behaviour of Lower Rhine Clays for lower de-665 to the p − q inclination, first the compression triaxial crit-
29 viatoric amplitudes corresponding to qampl = 45 and 40 ical state line is reached, thus the accumulation of the
30 kPa. As expected the number of cycles leading to failure axial strain changes from extension to compression. For
31 increases with decreasing amplitude. Among other effects, a performance based design of an e.g. flooded opencast
32 also the inherent anisotropy resulting in a compressive ac- mine unter earthquake loading the accurate description
33
635 cumulation of the axial strain is very well captured by the670 of these effects is indispensable. This holds not only for
34 simulations with the AVISA model. the low plasticity Kaolin Clay as documented in [3] but
35 also for the high plasticity Lower Rhine Clay evaluated in
36 this work. Consequently the hypoelastic stiffness has been
37 5. Concluding Remarks
transformed to a transversal hypoelastic stiffness. The ex-
38 tensions and novelties compared to the reference model [2]
The present work proposes a constitutive model with675
39 are i.a. in detail presented in Appendix A.
some interesting capabilities, such as the simulation of
40 Furthermore, the last cycles at cyclic mobility show a pre-
640
41 the strain rate and time dependency, the incorporation
of small strain and of fabric effects, permitting the user vented dilatancy at loading and a contractant behaviour
42
to evaluate the influence of each effect on a simulation, after an elastic regime is overcomed during unloading. It
43
without changing the model and keeping the same set of680 is of high importance to note that at stress reversal both
44
45 material parameters. clays investigated in this paper first exhibit an increase of
46
645 The small strain effects of clays has been simulated through the mean pressure in contrast to sand’s behaviour whereas
47 an extended ISA plasticity formulation (first version of the stress reversal starts with the highest possible contrac-
48 ISA plasticity has been proposed for sand by Fuentes & tancy. This can present a reason why most clays do not
49 Triantafyllidis [1]). The rate dependency is captured by a685 reach liquefaction p = 0 kPa. Particularly the fact that the
50 third strain mechanism (in addition to the total and the final relaxation of the mean pressure at the failure crite-
51
650 hypoplastic strain) that is dependent on the stress and the rion of |ε1 | = 10% is lower for horizontally cutted samples
52 void ratio. The loading surface has been well defined in than vertically cutted ones gives a hint to the answer of
53 order to provide a clean and complete formulation for the the question if a specific angle for the cutting direction ex-
54 overconsolidation ratio. The degree of nonlinearity pre-690 ists, at which a total relaxation near to liquefaction p → 0
55 sented in the reference model [2] turned to be unsuitable would be experienced.
56
655 as it does not provide information about the preloading The performance of the model has been inspected by simu-
57 lating the database with about 50 cyclic undrained triaxial
state of the material. For this purpose it has been refor-
58 (CUT) tests on the low plasticity Kaolin Clay published
mulated depending on the OCR.
59 by Wichtmann & Triantafyllidis. The Kaolin database in-
60 Some investigations regarding the inherent anisotropy have695
61 14
62
63
64
65
 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17 Figure 17: Effective stress path and deviatoric stress - axial strain relations for the undrained cyclic test RBT-4 on Lower Rhine Clay with
18 qampl = 60 kPa and ṡ = 0.02 mm/min. (simulation=red, experiment=blue)
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37 Figure 18: Effective stress path and deviatoric stress - axial strain relations for the undrained cyclic test RBT-5w on Lower Rhine Clay with
38 qampl = 60 kPa and ṡ = 0.05 mm/min. (simulation=red, experiment=blue)
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57 Figure 19: Effective stress path and deviatoric stress - axial strain relations for the undrained cyclic test RBT-7 on Lower Rhine Clay with
58 qampl = 45 kPa and ṡ = 0.02 mm/min. (simulation=red, experiment=blue)
59
60
61 15
62
63
64
65
 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17 Figure 20: Effective stress path and deviatoric stress - axial strain relations for the undrained cyclic test RBT-8 on Lower Rhine Clay with
18 qampl = 40 kPa and ṡ = 0.02 mm/min. (simulation=red, experiment=blue)
19
20
21 cludes variation of the deviatoric stress amplitude, initial730 • transversal (hypo)elasticity: requires the deter-
22 stress ratio, displacement rate, overconsolidation ratio and mination of 4 material parameters, whereby the com-
23 cutting direction. Furthermore, 4 CUT tests performed on pression index λ can be calibrated upon an com-
24 the high plasticity Lower Rhine Clay have been simulated. pression path with a constant rate of strain and the
25
700 These experiments have been performed by the authors swelling index κ is defined as the slope of the cor-
26 and Wichtmann with the same device as [3], whereby the735 responding unloading path. Two remaining param-
27
influences of the displacement rate as well as the deviatoric eters can be defined in conjunction with each other,
28
amplitude have been studied. The simulations showed to as they both influence the slope of the effective stress
29
30 be in a very good agreement with the experimental obser- path in stress space and the shear modulus. Thus,
705
31 vations. for the Poisson ratio νh and the anisotropic coef-
32 The research has shown that the loops at the cyclic mobil-740 ficient α the effective stress slopes ∆q/∆p of two
33 ity represent an interaction between inherent anisotropy monotonic or cyclic undrained triaxial tests with dif-
34 and dilatancy. The later one is not considered in the ferent cutting directions need to be measured and
35 present work and will be a matter of future research. then the hypoelastic formulation can be used:
36     
37 ∆p Epp Epq ∆εv
710 Appendix A. Summary of constitutive relations = (B.1)
38 ∆q Eqp Eqq ∆εq
of the reference and the proposed
39 model
40 with ∆εv = 0 in undrained tests. Along these lines of
41 Table A.5 provides a summary of the relations consti-745 thoughts, the authors provided in [21] a Mathemat-
42 tuting the reference model compared to the constitutive ica code presenting a very simple and self-explanatory
43
715 equations of the in this paper proposed AVISA model (in method for the calibration of νh and α.
44 red the reformulations or extensions of the AVISA model).
45 • the intergranular strain: requires the calibration
46 of 7 material parameters whereby the stiffness fac-
47 750 tor mR , the radius of the yield surface R and the
48 Appendix B. Short overwiev of the calibration of hardening parameter β0 can be calibrated through a
49 AVISA material parameters secant shear modulus degradation curve (Resonant
50
720 The AVISA model requires the calibration of 15 pa- Column Test). The parameter R corresponds to the
51 rameters (one more than the reference model [2]) listed amplitude of the threshold strain which encloses the
52 in Tab. 3. However, the ”controversial” nOCR parame-755 elastic locus of the material and when results of the
53
ter termed as viscous exponent is eliminated due to the RC test are not available, a value of R = 10−4 is
54 recommended for the elastic behaviour of clays. The
explicit solution of the loading surface according to the
55 parameter
725
56 void ratio, see Sec. 3.2. Despite of the bounding surface
57 radius d all parameters have been used in former works β0 controls the strain amplitude required to reach
58 [1, 23, 28, 2, 24, 25, 21] and will be briefly summarized760 the mobilized states. A relation for β0 was provided
59 herein by separating them according to their role within in [1]. The minimum χ0 and maximum χmax ex-
60 the model: ponent control the shape of the secant shear modu-
61 16
62
63
64
65
 1 Table A.5: Comparison between the constitutive equations of the reference model [2] and the AVISA model.
2
3
Summary of constitutive relations
4
5
6 reference model [2] AVISA
7 Constitutive equation:
8 hp vis



9 σ̇ = mE : ε̇ − yh ε̇ − ε̇ σ̇ = mEtrans : ε̇ − yh ε̇hp − ε̇vis
10 Intergranular strain relations:
11
12 FH = kh − ck − R/2
13 FHb = khk − 2 · R FHb = khk − d · R
14
ḣ = ε̇ − λ̇H N
15
16 ċ = λ̇H c̄
17 c̄ = β (cb − c)/(R) c̄ = β (cb − c)/(d · R)
18 → →
19 cb = R/2 ε̇ cb = R (d − 1/2) ε̇
20 khb − hk khb − hk
21 ρ=1− ρ=1−
2R d·R
22 hb = R N hb = d/2 · R N
23 →
24 yh = ρχ hN : ε̇ i
25 m = mR + (1 − mR ) yh
26
27 ε̇a = Ca /R (1 − yh − εa ) kε̇k
28 χ = χ0 + εa (χmax − χ0 ))
29 3
β = β0 β = β0 + (2 − β0 ) (krk/krc k)
30
31 Constitutive relations for mobilized states:
32 →→ K K
→→
33 E = Eiso = 3K 1 1 +2G Idev − √ (1 r + r 1) Eiso = 3K 1 1 +2G Idev − √
(1 r + r 1)
3Mc 3Mc
34 Eabcd = Qabij : Eijkl : Qklcd
35
36 Qabcd = µac µbd
√ √
37 µ = α1 + (1 − α)m ⊗ m
38 p (1 + e) 3 (1 − 2ν)
39 K= , G=K
λ (1 − Y0,max ) 2 (1 + ν)
40 hp
41 ε̇ = Y mkε̇k
42  1/Iv
1
43 ε̇vis = Iv λ
OCR
44  →
45 1 r
m = − (krc k − krk) +
46 2 krc k
nocr
krk ei0 − ex
   
47 pei pei pei
48 OCR = + 1− OCR = + , pxei = exp
p p krb k pei λ
49 
||r||

+ 1/nf
50 e =A ei , A = 1 −
51 ||rb0 ||
 nY
krk OCR − 1
  
52 Y = Y0 + (1 − Y0 ) Y = Y0 + (1 − Y0 ) exp −
53 krb k krk
54
55
56 lus degradation curve and should be calibrated with how fast the accumulation is produced during the
57
tests including a few and a considerably increasing cycles and can be calibrated through trial and error
58
765 number of cycles, respectively, for further details see procedure of an undrained triaxial test. The ratio d
59
60 also [28]. The accumulation rate factor Ca controls770 of the radius of the bounding surface and the yield
61 17
62
63
64
65
 1 surface d should be calibrated through a RC test [3] T. Wichtmann, T. Triantafyllidis, Monotonic and cyclic tests
in the transition between the visco-elastic and the on kaolin: a database for the development, calibration and ver-
2 ification of constitutive models for cohesive soils with focus to
3 visco-hypoplastic behaviour defining the end of this cyclic loading, Acta Geotechnica 13 (5) (2017) 11031128.
4 transition area as depicted in Fig. B.21. 810 [4] B. Sexton, A. Bryan, M. Karstunen, N. Sivasithamparam, Stone
5 column settlement performance in structured anisotropic clays:
6 Experiment (sketch) the influence of creep, Journal of Rock Mechanics and Geotech-
nical Engineering 8 (2016) 672–688.
7 Model
[5] J. Castro, M. Karstunen, N. Sivasithamparam, Influence of
8 815 stone column installation on settlement reduction, Computers
9 and Geotechnics 59 (2014) 87–97.
Gsec/G max [-]

1
10 [6] J. Graham, G. Houlsby, Anisotropic elasticity of a natural clay,
11 Géotechnique 33 (2) (1983) 165–180.
visco-elastic visco- [7] A. Amorosi, S. Rampello, An experimental investigation
12 hypoplastic 820 into the mechanical behaviour of a structured stiff clay,
13 Géotechnique 57 (2) (2007) 153–166.
14 [8] J. Duncan, H. Seed, Anisotropy and stress reorientation in clay,
15 transition J Soil Mech Found Div ASCE 92 (SM5) (1966) 21–52.
16 [9] A. Niemunis, Extended hypoplastic models for soils, habilita-
0
825 tion, monografia 34, Ruhr-University Bochum (2003).
17 10-6 10-5 10-4 10-3 10-2 [10] T. Wichtmann, Soil behaviour under cyclic loading - experimen-
18 [-] tal observations, constitutive description and applications, Verf-
19 fentlichungen des Instituts fr Bodenmechanik und Felsmechanik
20 am KIT. Habilitation. Heft 181. (2016).
21 830 [11] T. Wichtmann, K. Andersen, M. Sjursen, T. Berre, Cyclic tests
on high-quality undisturbed block samples of soft marine norwe-
22 gian clay, Canadian Geotechnical Journal 50 (4) (2013) 400–412.
23 Figure B.21: Shear modulus degradation curve and the definition of [12] C. Gu, J. Wang, Y. Cai, Z. Yang, Y. Gao, Undrained cyclic
R [1] and d.
24 triaxial behavior of saturated soft clay under variable confining
25 835 pressure, Soil Dyn Earthq Eng 40 (2012) 118–128.
26 [13] M. Zergoun, Y. Vaid, Effective stress response of clay to
775 • the critical and loading surface introduce 1 and undrained cyclic loading, Can Geotech J 38 (1994) 714–727.
27 2 additional material parameters, respectively. The [14] N. Duong, M. Suzuki, N. Hai, Rate and acceleration effects on
28
critical state slope in triaxial compression Mc can residual strength of kaolin and kaolin-bentonite mixtures in ring
29 shearing, Soils and Foundations 58 (2018) 1153–1172.
be calibrated through a triaxial test with samples840
30 [15] A. Niemunis, C. Grandas, L. Prada, Anisotropic Visco-
31 sheared up to high strains ensuring that the critical hypoplasticity, Acta Geotechnica 4 (4) (2009) 293–314.
780
32 state (e = ec and η = ηc ) is reached. The maximum [16] A. Whittle, M. Kiavvadas, Formulation of MIT-E3 constitutive
void ratio ei0 corresponds to the maximum void ra- model for overconsolidated clays, Journal of the Geotechnical
33 Engineering Division ASCE 120 (1) (1994) 173–198.
34 tio at the reference pressure pref = 1 kPa at the845 [17] G. Grimstad, S. Abate Degago, N. S, K. M, Modeling creep and
35 virgin isotropic consolidated line OCR = 1. For its rate effects in structured anisotropic soft clays, Acta Geotech-
36 calibration the most suitable test is an isotropic com- nica 5 (2010) 69–81.
pression test with constant rate of strain. In absence [18] S. Raude, F. Laigle, R. Giot, R. Fernandes, A unified thermo-
37
785
850 plastic/viscoplastic constitutive model for geomaterials, Acta
38 of this test, an oedometer test can be used instead as Geotechnica 11 (2016) 849 – 869.
39 explained in [2, 21]. The loading surface factor fb0 [19] N. Sivasithamparam, M. Karstunen, P. Bonnier, Modelling
40 controls the peak strength of the overconsolidated creep behaviour of anisotropic soft soils, Computers and
41 soil. Thus, it can be calibrated using triaxial data of Geotechnics 69 (2015) 46 – 57.
42 855 [20] T. Schanz, P. Vermeer, P. Bonnier, The hardening soil model:
790 initially overconsolidated samples. formulation and verification., in: Beyond 2000 in Computa-
43 tional Geotechnics. Ten Years of PLAXIS, A.A. Balkema, Am-
44 • the viscosity requires the calibration of only one sterdam, the Netherlands, 1999, pp. 281–290.
45 parameter denoted as viscosity index Iv . It describes [21] M. Tafili, T. Triantafyllidis, A simple hypoplastic model with
46 the intensity of creep for normalconsolidated states860 loading surface accounting for viscous and fabric effects of clays,
47 Submitted for IJNAMG.
and can be determined through a creep stage or con- [22] A. Niemunis, I. Herle, Hypoplastic model for cohesionless soils
48
795 solidation process evaluating the secondary settle- with elastic strain range, Mech. Cohesive-frict. Mater. 2 (4)
49 ment curve. Other methods as the consideration of (1997) 279–299.
50 two different isotachs can be used as well [9, 21, 2,865 [23] W. Fuentes, M. Hadzibeti, T. Triantafyllidis, Constitutive
51 model for clays under the isa framework, Vol. 80 of Holistic
23]. Simulation of Geotechnical Installation Processes - Benchmarks
52
and Simulations, Springer International Publishing, Germany,
53 2015, pp. 115–130.
54 References 870 [24] M. Tafili, T. Triantafyllidis, Constitutive model for viscous
55 clays under the isa framework, Vol. 82 of Holistic Simula-
800 [1] W. Fuentes, T. Triantafyllidis, ISA model: A constitutive model tion of Geotechnical Installation Processes - Theoretical Results
56 for soils with yield surface in the intergranular strain space, Int
57 and Applications, Springer International Publishing, Germany,
J. Numer. Anal. Meth. Geomech. 39 (11) (2015) 1235–1254. 2017, pp. 324–340.
58 [2] W. Fuentes, M. Tafili, T. Triantafyllidis, An isa-plasticity-based875 [25] M. Tafili, T. Triantafyllidis, On constitutive modelling of
59 model for viscous and non-viscous clays, Acta Geotechnica anisotropic viscous and non-viscous soft soils, Vol. 1 of Numer-
60
805 13 (2) (2017) 367–386.
61 18
62
63
64
65
 1 ical Methods in Geotechnical Engineering IX, Taylor & Francis
Group, Porto, Portugal, 2018, pp. 139–148.
2 [26] Y. Dafalias, Bounding surface plasticity. I: Mathematical foun-
3
880 dation and hypoplasticity, J. Eng. Mech. 112 (9) (1986) 966–987.
4 [27] Y. Dafalias, M. Manzari, Simple plasticity sand model account-
5 ing for fabric change effects, J. Eng. Mech. 130 (6) (2004) 622–
6 634.
[28] M. Poblete, W. Fuentes, T. Triantafyllidis, On the constitutive
7
885 modeling of multidimensional cyclic loading, Acta Geotechnica
8 11 (6) (2016) 1263–1285.
9 [29] S. Krieg, Viskoses bodenverhalten von mudden, seeton und
10 klei, Veröffentlichungen des Instituts für Bodenmechanik und
11 Felsmechanik am KIT. PhD Thesis. Heft 150. (german) (2000).
890 [30] A. Niemunis, S. Krieg, Viscous behaviour of soils under oedo-
12 metric conditions, Can. Geot. &Journal 33 (1996) 159–168.
13 [31] J. Atkinson, Peak strength of overconsolidated clays,
14 Géotechnique 57 (2) (2007) 127–135.
15 [32] A. Niemunis, C. Grandas-Tavera, T. Wichtmann, Peak stress
16
895 obliquity in drained and undrained sands. smulations with neo-
hypoplasticity, Vol. 80 of Holistic Simulation of Geotechnical
17 Installation Processes - Benchmarks and Simulations, Springer
18 International Publishing, Germany, 2015, pp. 85–114.
19 [33] M. Hvorslev, Physical components of the shear strength of sat-
20
900 urated clays, ASCE Research Conference, Shear Strength of
21 Cohesive Soils, 1960, boulder Colorado.
[34] L. Guo, J. Wang, Y. Cai, H. Lui, Y. Gao, H. Sun, Undrained de-
22 formation behavior of saturated soft clay under long-term cyclic
23 loading, Soil Dyn Earthq Eng 50 (2013) 28–37.
24
905 [35] M. Hyodo, H. Tanimizu, N. Yasufuku, H. Murata, Undrained
25 cyclic and monotonic triaxial behaviour of saturated loose sand,
26 Soils and Foundations 34 (1) (1994) 19–32.
[36] Y. Vaid, J. Chern, Effect of static shear on resistence to lique-
27
faction, Soils and Foundations 23 (1) (1983) 47–60.
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61 19
62
63
64
65