Académique Documents
Professionnel Documents
Culture Documents
Abstract: This paper considers the consolidation of a layer of clay in which partially penetrating prefabricated vertical
drains (PVDs) are used in conjunction with a combined surcharge and vacuum preloading. Analytical solutions for partially
penetrating PVDs are derived by considering vacuum pressure (suction), time-dependent embankment surcharge, well resist-
ance, and smear zone. Three-dimensional seepage with a virtual vertical drain is assumed to reflect real seepage into the
soil beneath the tip of a PVD. Analytical solutions were then used to examine the length of the vertical drain and vacuum
pressure on soft clay to determine the consolidation time and degree of consolidation, associated settlement, and distribution
of suction along the drain. The proposed solutions are then employed to analyse a case history. Finally, an appropriate PVD
length in relation to clay thickness and drain spacing is recommended for various loading patterns.
Key words: consolidation, partially penetrating, pore-water pressure, time dependent, vacuum, vertical drains.
Résumé : Cet article discute de la consolidation d’une couche d’argile dans laquelle des drains verticaux préfabriqués
(DVP) à pénétration partielle sont utilisés, avec une surcharge combinée et un pré-chargement à vide. Des solutions analyti-
ques pour les DVP à pénétration partielle sont dérivées en considérant la pression à vide (succion), la surcharge du remblai
dépendante du temps, la résistance du puits et la zone de souillure. Le modèle suppose que l’infiltration en trois dimensions
avec un drain vertical virtuel reflète l’infiltration réelle dans le sol sous la pointe du DVP. Les solutions analytiques ont en-
suite été utilisées pour évaluer la longueur du drain vertical et la pression à vide sur l’argile molle, ainsi que pour déterminer
le temps et le degré de consolidation, le tassement associé, et la distribution de la succion le long du drain. Les solutions
proposées sont ensuite utilisées pour analyser un exemple de cas. Finalement, une longueur appropriée pour un DVP est re-
commandée en relation avec l’épaisseur d’argile et l’espacement entre les drains et ce, pour différents patrons de charge-
ment.
Mots‐clés : consolidation, pénétration partielle, pression interstitielle, dépendant du temps, vide, drains verticaux.
[Traduit par la Rédaction]
Can. Geotech. J. 48: 970–983 (2011) doi:10.1139/T11-011 Published by NRC Research Press
Geng et al. 971
mated the influence of well resistance on consolidation com- Fig. 1. Analysis scheme of consolidation of clay with partially pe-
pared to Zeng and Xie (1989). Subsequently, Tang and Onit- netrated PVDs combined with vacuum pressure and surcharge pre-
suka (1998) proposed an analytical solution using the loading. kh, horizontal coefficient of permeability of the soil; ks,
variables separation method, which can consider both vertical horizontal coefficient of permeability of remoulded soil; kv, vertical
and radial drainage in the PVD zone, whereas consolidation be- coefficient of permeability of the soil; kw, coefficient of permeability
low this zone was still assumed to be purely one-dimensional. of the vertical drain; p, vacuum pressure; re, radius of influence
Most of the studies to date adopt 1-D theory to simplify zone, kh is the horizontal coefficient of permeability of the soil; rs,
consolidation beneath the PVD. However, the seepage ve- radius of the smear zone; rw, radius of the vertical drain; u, pore
locity directly beneath the drain tip should be faster than pressure.
the soil further away, and ideally, such drainage should be
considered as a true three-dimensional rather than a 1-D
problem.
To the best of the authors’ knowledge, no previous re-
search includes the effect of vacuum pressure with partially
penetrating PVDs. Although most commercial finite element
method software can be used to solve this problem, it is often
cumbersome if a parametric study is to be incorporated.
Therefore, this paper provides analytical solutions that con-
sider a partially penetrating drain combined with a vacuum
and surcharge preloading. Moreover, a virtual vertical drain
is considered to represent real three-dimensional seepage in
the soil beneath the PVD tip. The normalized settlement –
time factor curves, which can be used by practicing engi-
neers, are also provided.
rw r rs
where uw1(r, z, t) is the excess pore-water pressure within the
where ks is the horizontal coefficient of permeability of re- real vertical drain, us1 is the excess pore-water pressure in the
moulded soil, gw is the unit weight of water, us(r, z, t) is the smear zone improved by PVD, and kw is the coefficient of
excess pore-water pressure at any point in the smear zone, kv permeability of the vertical drain.
The average excess pore-water pressure at a given depth is Fig. 2. Comparison of model prediction of different methods: (a) de-
then given by gree of consolidation without smear zone and well resistance; (b) de-
Z rs Z re gree of consolidation with smear zone and well resistance. de,
1
½5 u1 ¼ 2prus dr þ 2prun dr diameter of influence zone; ds, smear diameter; dw, drain diameter;
pðre2 rw2 Þ rw rs U , degree of consolidation.
@us @un The continuity and boundary conditions at the interface be-
½10b ks ¼ kh neath the clay layer and the top of the bottom drainage layer
@r @r
(z = Lw) can be expressed by
where p is the vacuum pressure.
Continuity of pore-water pressure at r = rs ½10g z ¼ Lw : uw1 ¼ uw2
½10c us ¼ un
½10h z ¼ Lw : u1 ¼ u2
Continuity of pore-water pressure at r = rw implies
@uw1 @uw2
½10d us ¼ uw1 ; u2 ¼ uw2 ½10i z ¼ Lw : kw ¼ kv
@z @z
At the top of the clay layer z = 0 ensures
@u1 @u2
½10e uw1 ¼ p and u1 ¼ p ½10j z ¼ Lw : ¼
@z @z
Fig. 3. Finite element discretization for partially penetrating drain in unit cell: (a) nodes and integration points for a single eight-node biqua-
dratic displacement, bilinear pore pressure element; (b) mesh discretization and vacuum pressure distribution for short-drain analysis. qu, sur-
charge preloading pressure.
The initial conditions are ½11 uw1 ðZ; SÞ ¼ X1 ea1 Z þ X2 ea1 Z þ X3 ea2 Z þ X4 ea2 Z
b
½10k t ¼ 0; u1 ¼ u2 ¼ u0 ðzÞ ¼ q0 b
þ QðSÞ
where q0 is the initial value of preloading.
uwi ðZ; SÞ and b
The solutions for b ui ðZ; SÞ (i = 1, 2) obtained ½12 uw2 ðZ; SÞ ¼ Y1 eb1 Z þ Y2 eb1 Z þ Y3 eb2 Z þ Y4 eb2 Z
b
by Laplace transformation are (where complete derivations b
þ QðSÞ
are given in Appendix A)
Fig. 4. Comparison of model prediction between authors’ method and finite element (FE) analysis: (a) surcharge only (case a); (b) vacuum
only (case b); (c) vacuum and surcharge (case c).
b 1 @ 2b
uw1 ðZ; SÞ Fig. 5. Site plan of Tianjin East Pier project.
½13 u1 ðZ; SÞ ¼ b
uw1 ðZ; SÞ
B2 @2Z
b 1 @ 2b
uw2 ðZ; SÞ
½14 u2 ðZ; SÞ ¼ b
uw2 ðZ; SÞ
B2 @2Z
In the above, the parameters a1, a2, b1, b2, and B2 are de-
termined explicitly using algebraic functions (details in Ap-
pendix A). Z = z/H, S is the Laplace transform of the
dimensionless time factor Th ¼ Ch t=de2 (where Ch is horizon-
tal coefficients of consolidation and de is the diameter of the
b
influence zone); QðSÞ is the Laplace transform of the time-
dependent surcharge preloading q(t); and b uwi ðZ; SÞ and
b
ui ðZ; SÞ (i = 1, 2) are the Laplace transforms of uwi(Z, Th)
and ui ðZ; Th Þ (i = 1, 2), respectively.
Considering the boundary conditions (eqs. [10a]–[10f]),
continuity at the interface between the underlying soil and
the sand blanket (eqs. [10g]–[10j]) together with the initial
Fig. 6. Soil properties before improvement. CD, chart datum ; UU Fig. 7. Settlement due to vacuum preloading combined with sur-
test, unconsolidated undrained test; w, water content; wL, liquid charge preloading: (a) surcharge preloading versus time; (b) vacuum
limit; wp, plastic limit. preloading versus time; (c) surface settlement versus time. DOC, de-
gree of consolidation.
Z aþI1
1 b
½17 ui ðZ; Th Þ ¼ ui ðZ; SÞeST dS ði ¼ 1; 2Þ
2pI aI1
pffiffiffiffiffiffiffi
where a is a real number and I ¼ 1. The settlement of
the soil is now given by
ZH
½18 sðtÞ ¼ 3 dz
0
Comparisons with previous studies Table 1. Vertical drain parameters and soil properties.
Figure 2 illustrates the comparative results of the overall Spacing, S 1.0 m (square)
average degree of consolidation due to the surcharge preload- Length of vertical drain 18 m
ing alone, obtained using different methods (with zero vac- Dimension of drain 100 mm × 4 mm
uum pressure). Discharge capacity, qw 100 mm3/year (per drain)
Hart et al. (1958) adopted Barron’s (1948) solution and Dimension of mandrel 120 mm × 50 mm
Terzaghi’s (1943) 1-D consolidation solution to calculate kh/kw 1.05 × 10–6
consolidation due to radial flow and vertical flow, respec- kh/kv 1.0
tively. The consolidation solution from a combination of ra- kh/ks 4.0
dial and vertical flow was obtained by Carrillo (1942)’s Vertical coefficients of 1.589 × 10–7 m2/s
method, where it was assumed that the drainage distance of consolidation, cv
the layer without vertical drains was equal to the thickness Horizontal coefficients of 2.514 × 10–7 m2/s
of the whole layer. The well resistance and smear effects consolidation, ch
were not considered either. The overall degree of consolida-
tion given by Hart et al. (1958) for zero vacuum pressure
(surcharge only) is smaller than that predicted by the solution
of Tang and Onitsuka (1998) and the present solution pro-
3. Vacuum (40 kPa) plus surcharge preloading (40 kPa).
posed by the authors (Fig. 2a). Zeng and Xie (1989) overes-
The comparison of normalized settlement between the ana-
timates the well resistance on consolidation, and hence
lytical model and numerical predictions for the above three
deviates considerably from the present solution for Th < 0.5
cases is shown in Fig. 4. For a given applied load, the nor-
(Figs. 2a and 2b). In the absence of vacuum pressure, the de-
malized settlement was calculated based on the ratio of settle-
gree of consolidation obtained by Tang and Onitsuka (1998)
ment divided by the ultimate settlement of the clay layer.
is higher than the authors’ solution that applies the concept
Overall, a good agreement between the analytical and numer-
of virtual vertical drain to capture three-dimensional seepage
ical techniques was found. It can be seen that the ultimate
under the drain tips. It is also shown in Fig. 2 that even the
normalized settlements obtained for cases (b) and (c) are al-
partially penetrating drains with vacuum pressure accelerate
ways less than unity, because of the vacuum pressure loss at
consolidation progress significantly faster than the partially
the bottom pervious boundary.
penetrating drain with surcharge only.
Fig. 10. Normalized settlement–time factor curves under vacuum pressure only.
Fig. 11. Normalized settlement–time factor curves under the surcharge combined with vacuum pressure.
Table 2. Vertical drain penetration ratio (r) for different conditions to achieve a 90% dimensionless settlement.
r
de = 1.0 m de = 1.5 m
H (m) SP (80 kPa) VP (–80 kPa) VP and SP SP (80 kPa) VP (–80 kPa) VP and SP
10 0.8 Cannot achieve 0.9 and SP > 0.8 Cannot achieve 0.9 SP > 50%
90% normal- 50% total ap- 90% normal- total applied
ized settle- plied pres- ized settle- pressure
ment sure ment
20 0.85 0.9 0.9 SP > 40% 0.8 Cannot achieve 0.9 SP > 40%
total applied 90% normal- rotal applied
pressure ized settle- pressure
ment
Note: SP, surcharge preloading; VP, vacuum preloading; VP and SP, vacuum combined surcharge loading.
results are shown in Figs. 7c and 8. A time-dependent vac- shown in Fig. 8. The nonlinear distribution of pore-water
uum was applied based on the field measurements (Fig. 7b). pressure at greater depths (8–15 m) may be attributed to
The soil and drain parameters used in the analysis adopted ground-water pumping in the sand layer and the accuracy of
from Choa et al. (1990) are tabulated in Table 1. The settle- the field instrumentation.
ment predictions based on eq. [18] agreed with the field
measurements (Fig. 7c). When a PVD is longer, but does Determining the optimum partially
not totally penetrate the entire clay layer (i.e., r = 0.9), it
penetrated length of drain
should cause more settlement. However, if it were to ap-
proach the layer of fine sand, there would be vacuum loss In this study, the parameters assumed for the current model
from the bottom pervious boundary, which would create a are dw = 0.1 m, de = 1.0–1.5 m, ds = 0.3 m, kh/kv = 1.0, kh/
zero vacuum head at the bottom end of the PVD. In this kw = 10–6, kh/ks = 4, and H = 10–20 m, which are typical of
case, the analysis assumed a triangular distribution of suction the case study, Tianjin Port. When the PVD tip is at or close
down the drain (Indraratna et al. 2004, 2005); hence, the re- to the bottom pervious layer, there is a possibility of vacuum
sulting settlement would be smaller. The location of the pore- loss that affects the ultimate settlement. In this aspect, the
water measurement mentioned by Choa et al. (1990) is at the conventional degree of consolidation cannot be used as a per-
boundary of the drain influence zone (i.e., r = 0.5 m). The formance indicator as it always converges to unity. Therefore,
predicted pore-water pressure matches the field data, as the settlement normalized with the ultimate settlement deter-
Fig. 12. Effects on partially penetrating drains on time-dimensional settlement: (a) penetration ratio = 0.5; (b) penetration ratio = 0.9. n, re/rw;
s, rs/rw.
mined for the total applied load is employed to evaluate the 1. A normalized settlement of 0.9.
efficiency of the partially penetrated PVD. It is noted that 2. Th,req < 1.25 Th,i, where Th,req is the desired consolidation
the ultimate normalized settlement can be less than unity if time factor due to partially installed vertical drains and Th,i
there is vacuum loss. With the surcharge preloading only is the time factor required to achieve a normalized set-
(qu = 80 kPa), Fig. 9 illustrates the effect of partially in- tlement of 0.9 where PVDs fully penetrate the entire
stalled vertical drains on the time factor (Th), as represented clay layer that is followed by an impervious boundary.
by the time-normalized settlement curve. Two desirable crite- In the case of surcharge only, for a 10 m thick clay layer,
ria were employed for all three loading cases: (a) surcharge the PVD length could be reduced by about 20% of the entire
preloading only; (b) vacuum preloading only; and (c) vacuum soft clay thickness for both de = 1.0 and 1.5 m, without sig-
(40 kPa) plus surcharge preloading, as follows: nificantly increasing the consolidation time. For the 20 m
thick clay layer, the PVD length could be reduced by about the drain tips. It is important to note that the partially pen-
15% of the soft clay thickness. Figure 10 illustrates the con- etrating drains with vacuum pressure accelerate consolida-
solidation response due to vacuum preloading only. With tion progress significantly faster than the case of surcharge
vacuum preloading applied to a clay thickness of 10 m, it preloading only, especially when the bottom layer is impervious.
can be seen that even when the vertical drain penetrates 90% A case history taken from Tianjin Port, China, was dis-
of the entire clay thickness, the desired 90% normalized set- cussed and analysed using the analytical solutions proposed
tlement could not be attained when de = 1.5 m. When the in this study, capturing the time-dependent vacuum pressure
drain spacing is 1.0 m and penetration ratio is 0.9, a normal- variation and the surcharge preloading history based on the
ized settlement of 90% can be achieved, but a significant pe- field data. The analytical predictions agreed well with the
riod of consolidation is required (Th,req = 1.75 Th,i). For the measured settlement and excess pore pressure. If the PVDs
case of a 20 m clay thickness, a 90% penetration of PVD fully penetrate towards the bottom of the soft clay layer to
would give a normalized settlement of 90% for both de = 1.5 reach the underlying pervious sand layer, then the resulting
and 1.0 m, but with a longer consolidation time of Th,req = 2 Th,i. settlement at any given time would become smaller due to
This is attributed to the loss of vacuum pressure through the loss of vacuum (suction).
the pervious layer. Generally, the PVD penetration length can be reduced by
Figure 11 shows the normalized settlement curves deter- up to 20% of the total clay thickness without significantly af-
mined by the combined surcharge and vacuum preloading. fecting the time for settlement to achieve a normalized settle-
Regardless of PVD spacing, for the 20 m clay thickness, a ment of 90%. For surcharge preloading combined with
penetration ratio of 0.9 can provide a normalized settlement vacuum pressure, the PVD length can only be reduced by
of 0.9 within an increased consolidation time (Th,req) less about 10% of the entire soft clay thickness for vacuum to to-
than 20% of Th,i. For the 10 m thick clay layer with 1.0 m tal applied pressure ratio of 0.4. It can be seen that vacuum
drain spacing, the applied surcharge pressure should be more preloading alone may not be effective when there is a perme-
than 40% of the total applied pressure (vacuum plus sur- able layer at the bottom of the clay.
charge), to obtain a normalized settlement of 0.9 within
1.25Th,i. However, for the 10 m thick clay layer with 1.5 m Acknowledgement
drain spacing, the surcharge pressure has to be more than The authors wish to thank the Australia Research Council
50% of the total applied pressure to achieve 90% normalized (Australia), Road and Traffic Authority (Sydney), Coffey
settlement within 1.25Th,i. This analysis demonstrates that to Geotechnics, Queensland Department of Main Roads, and
overcome the vacuum loss caused by the bottom pervious Douglas Partners for their continuous support.
layer, the combination of vacuum with surcharge pressure is
most beneficial when the PVD length is less than 90% of the
total clay thickness. Generally, the vertical drain penetration References
ratio can be reduced by up to 20% of the total clay thickness Barron, R.A. 1948. Consolidation of fine-grained soils by drain wells.
without significantly affecting the settlement time for sur- Transactions of the American Society of Civil Engineers, 113:
charge loading alone. In contrast, for surcharge load com- 718–742.
bined with vacuum pressure, the vertical drain length can be Biot, M.A. 1941. General theory of three-dimensional consolidation.
reduced by 10% of the total soft clay thickness with an ap- Journal of Applied Physics, 12(2): 155–164. doi:10.1063/1.
propriate vacuum and surcharge combination (see Table 2). 1712886.
The variation of the time-dependent average pore-water Carrillo, N. 1942. Simple two and three-dimensional cases in the
pressure (u) with depth is shown in Fig. 12. The pore pres- theory of consolidation of soils. Journal of Mathematics and
sure equals the applied vacuum pressure at the top boundary Physics, 21: 1–5.
of the clay layer, whereas the pore pressure is zero at the bot- Choa, V., Wong, K.S., and Low, B.K. 1990. New airport at Chek Lap
tom pervious boundary. As expected, the excess pore pres- Kok, geotechnical review and assessment. Consulting report to
Maunsell Pte Ltd., Singapore.
sure in PVD-improved zone rapidly alters with the applied
Chu, J., and Yan, S.W. 2005. Application of the vacuum preloading
vacuum pressure.
method in soil improvement projects. In Ground improvement –
case histories. Edited by B. Indraratna and J. Chu. Elsevier,
Conclusions London. Vol. 3, pp. 91–118.
An analytical solution for partially penetrating vertical drains Durbin, F. 1974. Numerical inversion of Laplace transform: An
efficient improvement to Dubner and Abita’s method. The
for vacuum preloading and surcharge preloading was presented.
Computer Journal, 17(9): 371–376.
The results are then compared with the finite element predic-
Hart, E.G., Kondner, R.L., and Boyer, W.C. 1958. Analysis for
tions. As the PVDs partially penetrate the clay, the area directly partially penetrating sand drains. Journal of the Soil Mechanics
beneath the PVD tips necessitates a three-dimensional consoli- and Foundations Division, ASCE: 84(4): 1–15.
dation analysis. A virtual three-dimensional seepage mecha- Indraratna, B., and Redana, I.W. 1998. Laboratory determination of
nism was assumed to represent the real flow condition of smear zone due to vertical drain installation. Journal of
the soil beneath the PVD tips. Subsequently, a comparison Geotechnical Engineering, 125(1): 96–99.
between the current analytical solutions and the numerical Indraratna, B., and Redana, I.W. 2000. Numerical modeling of vertical
model was made. It was found that the authors’ solution in drains with smear and well resistance installed in soft clay. Canadian
comparison with field data was more accurate due to the Geotechnical Journal, 37(1): 132–145. doi:10.1139/t99-115.
appropriate consideration of the vacuum distribution along Indraratna, B., and Rujikiatkamjorn, C. 2008. Effects of partially
the drain length and the realistic flow condition beneath penetrating prefabricated vertical drains and loading patterns on
vacuum consolidation. In Proceedings of GeoCongress 2008: Using the Laplace transform technique, the solution for
Geosustainability and Geohazard Mitigation, New Orleans, La., 9– eq. [A1] is given by
12 March 2008. GSP 178. Edited by K.R. Reddy, M.V. Khire, and 2
A.N. Alshawabkeh. American Society of Civil Engineers, Reston, @ 4b
uw1 SK3 h22 @ buw1 SK3 h22b
½A3 þ B 1 þ B2 þ B2 uw1
Va. pp. 596–603. @Z 4 n2 @Z 2 n2
Indraratna, B., Bamunawita, C., and Khabbaz, H. 2004. Numerical
modeling of vacuum preloading and field applications. Canadian b
B QðSÞ ¼0
2
Geotechnical Journal, 41(6): 1098–1110. doi:10.1139/t04-054. @ 2b
uw1
Indraratna, B., Rujikiatkamjorn, C., and Sathananthan, I. 2005. ¼ B2 ðb
u1 b
uw1 Þ
Analytical and numerical solutions for a single vertical drain @Z 2
including the effects of vacuum preloading. Canadian Geotechni-
2
cal Journal, 42(4): 994–1014. doi:10.1139/t05-029. @ 4b
uw2 SK3 h22 @ buw2 SK3 h2
Onoue, A. 1988. Consolidation of multilayered anisotropic soils by ½A4 þ B 3 þ B4 þ B4 2 2b uw2
@Z 4 n2 @Z 2 n
vertical drains with well resistance. Soils and Foundations, 28(3):
75–90. b
B QðSÞ ¼0
4
Runesson, K., Hansbo, S., and Wiberg, N.E. 1985. The efficiency of
@ 2b
uw2
partially penetrating vertical drains. Géotechnique, 35(4): 511– ¼ B4 ðb
u2 b
uw2 Þ
516. doi:10.1680/geot.1985.35.4.511. @Z 2
Tang, X.W., and Onitsuka, K. 1998. Consolidation of ground with
where
partially penetrated vertical drains. Geotechnical Engineering
Journal, 29(2): 209–231. kh kh kh
Terzaghi, K. 1943. Theoretical soil mechanics. John Wiley & Sons, K1 ¼ ; K2 ¼ ; K3 ¼
kv kw ks
New York.
Lw H ch t
Zeng, G.X., and Xie, K.H. 1989. New development of the vertical r ¼ ; h2 ¼ ; Th ¼ 2 ;
drain theories. In Proceedings of the 12th International Conference H dw de
on Soil Mechanics and Foundation Engineering (ICSMFE), Rio de z 2 H 2 8K1 h22
Janeiro, Brazil, 13–18 August 1989. Balkema, Rotterdam, the Z ¼ ; B 1 ¼ ch 2 ¼ ;
Netherlands. Vol. 2, pp. 1435–1438. H re Fa1 cv Fa1 n2
kv 2 2 H 2 8h22 ðn2 1ÞK2
B2 ¼ ðn 1Þch 2 ¼ ;
kw re Fa1 cv Fa1 n2
Appendix A
After mathematical integration, the partial differential equa- 2 H 2 8K3 h22
B3 ¼ ch ¼ ;
tions containing only uw and ui can be expressed as follows: re2 Fa2
cv Fa2
2 2 H 2 8h22 ðn2 1ÞK4
@ 4 uwi @ 3 uwi 2 kv 2 @ uwi B4 ¼ ðn2 1Þch ¼
½A1 cv 4 2 ch 2 1 þ ðn 1Þ re2 Fa2 cv Fa2 n2
@z @z @t re Fai kwi @z2
2 kh @uwi dq uw1 ðZ; SÞ, b
b uw2 ðZ; SÞ, b
u1 ðZ; SÞ, b b
u2 ðZ; SÞ, QðSÞ, and S are the
þ ðn2 1Þ 2 ¼0 ði ¼ 1; 2Þ
re Fai kvi @t dt Laplace transforms of uw1(Z, Th), uw2(Z, Th), u1 ðZ; Th Þ,
u2 ðZ; Th Þ, q(Th), and Th, respectively.
and
The solutions for eqs. [A2] and [A3] are
@ 2 uwi 2 kh
½A2 ¼ ðn2 1Þ 2 ðui uwi Þ ½A5 uw1 ðZ; SÞ ¼ X1 ea1 Z þ X2 ea1 Z þ X3 ea2 Z þ X4 ea2 Z
b
@z 2 re Fai kwi
b
þ QðSÞ
where
2
n kh 3 n ½A6 uw2 ðZ; SÞ ¼ Y1 eb1 Z þ Y2 eb1 Z þ Y3 eb2 Z þ Y4 eb2 Z
b
Fa1 ¼ ln þ lnm
m ks 4 n 1
2 b
þ QðSÞ
m2 kh m2
þ 2 1 1 2
n 1 ks 4n b 1 @ 2b
uw1 ðZ; SÞ
½A7 u1 ðZ; SÞ ¼ b
uw1 ðZ; SÞ
kh 1 1 B2 @2Z
þ 1 2
ks n 2 1 4n
b 1 @ 2b
uw2 ðZ; SÞ
2 ½A8 u2 ðZ; SÞ ¼ b
uw2 ðZ; SÞ
3 n 1 1 B2 @2Z
Fa2 ¼ lnn þ 1
4 n2 1 n2 1 4n2 where Xi and Yi (i = 1, 2, 3, 4) are the parameters that are
needed to be solved by using the boundary conditions, and
For the underlying soil with a virtual drain kw2 = k, where
n = re/rw, m = rs/rw, s = rs/rw, cv = kv/mvgw, and ch = kh/mvgw.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffiffi
u
u SK1 h2 r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
u SK1 h2 2 SK h 2
t n2 þ B1 þ B2 þ n2 þ B1 þ B2 4 n12 2 B2
2
a1 ¼
2
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffiffi
u
u SK1 h2 r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
u SK1 h2 2 SK h 2
t n2 þ B1 þ B2 n2 þ B1 þ B2 4 n12 2 B2
2
a2 ¼
2
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffiffi
u
u SK1 h2 r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
u SK1 h2 2 SK h 2
t n2 þ B3 þ B4 þ n2 þ B3 þ B4 4 n12 2 B4
2
b1 ¼
2
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffiffi
u
u SK1 h2 r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
u SK1 h2 2 SK h 2
t n2 þ B3 þ B4 n2 þ B3 þ B4 4 n12 2 B4
2
b2 ¼
2
Considering the boundary conditions (eqs. [10a]–[10f]), the continuity conditions at the interface between the underlying soil
and sand blanket (eqs. [10g]–[10j]) and the initial condition (eq. [10k]), the following matrix obtained from eqs. [A5] and [A6]
can be written as
½A9 x88 jT ¼ PT
where
2 3
1 1 1 1 0 0 0 0
6 a2 a 21 a 22 a 22 0 7
6 1 0 0 0 7
6 7
6 0 0 0 0 x35 x36 x37 x38 7
6 7
6 7
6 0 0 0 0 x45 x46 x47 x48 7
6 7
x¼6 7
6 x51 x52 x53 x54 x55 x56 x57 x58 7
6 7
6x x62 x63 x64 x65 x66 x67 x68 7
6 61 7
6 7
6 x71 x72 x73 x74 x75 x76 x77 x78 7
4 5
x81 x82 x83 x84 x85 x86 x87 x88
x35 ¼ b1 eb1 ; x36 ¼ b1 eb1 ; x37 ¼ b2 eb2 ; x38 ¼ b2 eb2 ; x45 ¼ b31 eb1 ; x46 ¼ b31 eb1 ; x47 ¼ b32 eb2 ;
x48 ¼ b32 eb2 ; x51 ¼ ea1 r ; x52 ¼ ea1 r ; x53 ¼ ea2 r ; x54 ¼ ea2 r ; x55 ¼ eb1 r ; x56 ¼ eb1 r ;
x57 ¼ eb2 r ; x58 ¼ eb2 r ; x61 ¼ a21 ea1 r =B2 ; x62 ¼ a21 ea1 r =B2 ; x63 ¼ a22 ea2 r =B2 ;
x64 ¼ a22 ea2 r =B2 ; x65 ¼ b21 eb1 r =B4 ; x66 ¼ b21 eb1 r =B4 ; x67 ¼ b22 eb2 r =B4 ; x68 ¼ b22 eb2 r =B4 ;
x71 ¼ k1 a1 ea1 r ; x72 ¼ k1 a1 ea1 r ; x74 ¼ k1 a2 ea2 r ; x75 ¼ k2 b1 eb1 r ; x76 ¼ k2 b1 eb1 r ;
x77 ¼ k2 b2 eb2 r ; x78 ¼ k2 b2 eb2 r ; x81 ¼ a1 ea1 r ð1 a21 =B2 Þ; x82 ¼ a1 ea1 r ð1 a21 =B2 Þ;
x83 ¼ a2 ea2 r ð1 a22 =B2 Þ; x84 ¼ a2 ea2 r ð1 a22 =B2 Þ; x85 ¼ b1 eb1 r ð1 b21 =B4 Þ;
x86 ¼ b1 eb1 r ð1 b21 =B4 Þ; x87 ¼ b2 eb2 r ð1 b22 =B4 Þ; x88 ¼ b2 eb2 r ð1 b22 =B4 Þ
j ¼ ½ X1 X2 X3 X4 Y1 Y2 Y3 Y4