Accepted Manuscript

Importance of Controlling the Degree of Saturation in Soil Compaction linked

to Soil Structure Design

Fumio Tatsuoka, Antonio G. Correia

PII: S2214-3912(18)30070-9
DOI: https://doi.org/10.1016/j.trgeo.2018.06.004
Reference: TRGEO 175

To appear in: Transportation Geotechnics

Received Date: 7 April 2018

Revised Date: 28 May 2018
Accepted Date: 13 June 2018

Please cite this article as: F. Tatsuoka, A.G. Correia, Importance of Controlling the Degree of Saturation in Soil
Compaction linked to Soil Structure Design, Transportation Geotechnics (2018), doi: https://doi.org/10.1016/
j.trgeo.2018.06.004

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 Importance of Controlling the Degree of Saturation in Soil
Compaction linked to Soil Structure Design
Fumio Tatsuoka1 and Antonio G. Correia2
1
Tokyo University of Science, Chiba, Japan; tatsuoka@rs.noda.tus.ac.jp,
2
University of Minho, Guimarães, Portugal; agc@civil.uminho.pt

Abstract
In the typical conventional fill compaction, the dry density, ρd, and the water content, w, are
controlled in relation to reference values, (ρd)max and wopt, determined by standard laboratory
compaction tests using a representative sample at a certain compaction energy level, CEL. Although
CEL and soil type affect significantly (ρd)max and wopt, they change inevitably, sometimes largely, in
a given earthwork project while field CEL may not match the value used in the laboratory
compaction tests. Compaction control based on only an index of strength/ stiffness of compacted soil
has such a drawback that, for a fixed value of ρd, the index may increase significantly as the degree
of saturation, Sr, of compacted soil becomes lower than the optimum degree of saturation (Sr)opt
defined as Sr when (ρd)max is obtained for a given CEL. In comparison, the value of (Sr)opt and the
unified compaction curve in terms of ρd/(ρd)max vs. Sr - (Sr)opt relation of compacted soil are rather
insensitive to variations in CEL and soil type. Besides, CBR (unsoaked and soaked), the unconfined
compression strength, the elastic shear modulus and the collapse upon submerging of compacted soil
and the cyclic undrained shear strength and the coefficient of hydraulic conductivity of saturated soil
are all controlled by ρd and “Sr at the end of compaction”. As a standard method, it is proposed to
control the values of w and ρd in such that Sr becomes (Sr)opt while ρd becomes large enough to ensue
soil properties required in design fully taking advantage of available CEL. It is argued that the
compaction control keeping Sr larger than a certain value (or the air void ratio va lower than a certain
value) without controlling ρd is relevant as a simplified method only if Sr is controlled to become
(Sr)opt while always keeping CEL high enough to ensure the soil properties required in design.

Keywords: compaction, degree of compaction, degree of saturation, strength, stiffness, water content

Introduction

Early 1920s the earthwork construction was purely empirical despite the recognition of the
importance of proper compaction of fine soils to improve its use as a construction material. Proctor
(1933) introduced a rational approach to soil compaction and applied it for the first time during the
construction of the Bouquet Canyon Reservoir in Los Angeles, USA. This approach established a
laboratory compaction test to control embankment construction by layers. The laboratory test aims to
determine the proper amount of mixing water to use when compacting the soil in the field to achieve
the densest state. In the typical conventional soil compaction procedure by end product specification
following Proctor (1933), the dry density, ρd, and the water content, w, of compacted soil are
controlled in relation to the maximum dry density, (ρd)max, and the optimum water content, wopt,
obtained by laboratory compaction tests performed on a representative sample at a certain specified
compaction energy level (CEL) (Fig. 1).

Fig. 1. Typical conventional soil compaction control

Fig. 2. Shifting of compaction curve by changes in CEL and soil type and the optimum degree of
saturation

In the meantime, needs for higher ρd to ensure smaller residual deformation for less maintenance
cost and higher stability against severer natural disasters (i.e., floods, heavy rainfalls and
earthquakes) of soil structure have become stronger (e.g., Tatsuoka, 2011, 2015). In parallel,
available compaction energy level (CEL) in the field has become higher. Therefore, higher ρd has
become achievable more efficiently if the lift is small enough (say, thinner than 30 cm) and the
number of pass is large enough for ρd to reach an asymptotic value. However, as illustrated in Fig. 2,
(ρd)max increases and wopt decreases with an increase in CEL and as the soil becomes more
compactable, typically with an increase in the coefficient of uniformity (Uc) without an excessive
increase in fines content (FC) or with an increase in the particle diameter for the same Uc (e.g.,
Tatsuoka, 2015). However, even if fixed CEL and soil type are specified in a given earthwork project,
the actual CEL and soil type inevitably vary, sometimes largely, while field CEL may not match the
2
value used in the laboratory compaction tests. Therefore, the in-situ compaction curve moves from
time to time and from place to place and it is usually very difficult to capture exactly transient in-situ
values of (ρd)max and wopt. Moreover, laboratory compaction tests to obtain compaction properties and
field direct measurements of ρd and w are both time-consuming therefore cannot be performed
frequently.
As an alternative method, it has become popular to control soil compaction based on the stiffness
of compacted soil frequently evaluated from the response acceleration of a compaction machine in
operation or by ‘portancemeter’, as well as by dynamic loading plate tests, including the falling
weight deflectometer tests (Gomes Correia, 2008; Gomes Correia and Magnan, 2012). However, as
shown later, the stiffness of compacted soil does not always increase monotonously with an increase
in ρd during compaction at a constant w. Besides, for a fixed value of ρd, the stiffness increases
significantly as the degree of saturation, Sr, of compacted soil becomes lower than the optimum
degree of saturation (Sr)opt (defined as Sr when (ρd)max is obtained for a given CEL of a given soil type,
Fig. 2). Therefore, it is not possible to evaluate the ρd value only by measuring the stiffness of
compacted soil. The strength/stiffness under saturated conditions is usually used in design of soil
structures that may get wet or saturated by heavy/prolong rains, floods, impounding etc. during the
long lifecycle span. At the Sr= (Sr)opt state, the saturated strength/stiffness of a given soil type
becomes nearly its maximum for a given CEL (as shown later). On the other hand, the
strength/stiffness decreases by wetting differently depending on the value of Sr relative to (Sr)opt. So,
it is not possible to identify the Sr= (Sr)opt state only by measuring the strength/stiffness of compacted
soil under unknown unsaturated conditions. In addition, a simple method to evaluate field values of
the coefficient of hydraulic conductivity, k, is not available. As shown later, a higher stiffness value
of compacted soil at a given CEL may mean a higher k value (i.e., a less favorable value). Then, a
higher stiffness is not the consistent compaction target in this respect. Indeed, the soil properties (i.e.,
strength, stiffness and hydraulic conductivity) of a given soil type are controlled by not only ρd but
also “Sr at the end of compaction”, therefore, both of them should be controlled in relation to the
design requirements of soil structure in soil compaction.
In this paper, the soil compaction characteristics and several basic soil physical properties of
compacted soil are analyzed to propose a field compaction control method relevant to not only
highways, railways and residential embankments but also those impeding the water flow (e.g., river
dykes and earth-fill dams). The followings are shown. The value of (Sr)opt and the unified compaction
curve in terms of “the ρd/(ρd)max vs. Sr - (Sr)opt relation” are rather insensitive to variations in CEL and
soil type. Besides, CBR (unsoaked and soaked), the unconfined compression strength, the elastic
shear modulus and the collapse upon wetting of compacted soil and the cyclic undrained shear
strength and the coefficient of hydraulic conductivity of saturated soil can be expressed by empirical
equations having ρd and “Sr at the end of compaction” as variables while not including CEL. Based
on these findings, a unified framework for fill compaction control is proposed. That is, the values of
w and ρd of compacted soil are controlled in such that the Sr value becomes (Sr)opt while the ρd value
becomes large enough fully taking advantage of available CEL to ensure the soil properties required
in design. Prior to compaction, the value of w of fill material is adjusted so that the above-mentioned
control of ρd and Sr can be achieved efficiently. It is argued that the compaction control keeping Sr
larger than a certain value (or the air void ratio va lower than a certain value) without controlling ρd is
relevant as a practical simplified method only if Sr is controlled to become (Sr)opt while always
keeping CEL constant and high enough to ensure the soil properties required in design. The issue
how to evaluate field compacted states and stress-strain and hydraulic properties of fill materials
including too large particles that are difficult to use in compaction tests and stress-strain/hydraulic
tests in the laboratory is beyond the scope of this paper.
3
Optimum Degree of Saturation and Unified Compaction Curve
A comprehensive series of full-scale compaction tests using a wide variety of compaction
roller used in field earthworks was performed in a large concrete pit (24 m-long, 3.5 m-wide and 1
m-deep) during a period from 1965 to 1990 (Nemoto and Sasaki, 1994). Overlying a 60 cm-thick
fixed base soil layer, a 30 cm-thick surface layer of a single soil type, sandy loam (Fig. 3a), was
prepared repeatedly at different water contents for respective compaction rollers. After spreading the
soil followed by preliminary compaction by eight passes of a light compaction machine with a
weight of 9.6 kN, the respective compaction rollers passed over the test soil layer. The values of ρd
and w in the upper 10 cm-thick soil layer were measured four times by the sand-replacement method
after the number of passes, N, became 0, 2, 4, 8 and 16. The values of ρd and w in the lower 10
cm-thick soil layer were measured only after N became 16.

100
土粒子比重 2.83
Percent passing by weight

Gs
80 液性限界
L. L. 29.9 %
Finer than 2 mm
塑性限界
P. L. 17.3 %
60
塑性指数
PI 12.6
40 流動指数
FI 7.3
Material used in full-
scale compaction
20
tests & laboratory
compaction tests
0
0.001 0.01 0.1 1 2 5 10 20 50
a) Particle size (mm)
2.2
5 Gs=2.798 B: 室内締固め試験
va (%) B: Laboratory compaction test
= 10 (機械化研究所
(1Ec ; 1Ec & Gs=2.83)
& Gs= 2.83)
3
(d)max= 1.996 g/cm
Unsoaked CBR
wopt= 12 %;

by Eq. 2 ＋1Ec; × 4.5Ec

density, , d (g/cm )
3

2.0 室内締固め試験
Lab. compaction test
(1)式：　CBR(%)=80 ((TUS
Sr= 100 % 2012年理科大; s=2.798)
2012, Gs=G2.798)
N=16
Gs= 2.83
▲, ○, ●, □, ■
Dry乾燥密度

40 A 8 振動タイヤローラ (重量15.6 ton）

Full-scale compaction:
B
the number of passing
による締固め曲線　
4
N=2, 4, 8 & 16:
（Nは通過回数）　
1.8 16L 90 A 16L: measurement in the lower
4.5Ec B Sr (%) 80 　　　2, 4, 8, 16:　上1/2層での測定値
20 half (10 cm-thick) at N= 16
2 = 70 　　　16L: N=16後の撒き出し層30cm
　　　　　　の下1/2層での測定値
A-A: compaction curve
1.0Ec (Sr)opt=81.3% （A-A)for
N=16に対する推定締固め曲線
N= 16
10
5
1.6
0 5 10 15 20
b) 含水比, w (%)
Water content,

4
 2.2 40
Full-scale tests Eq. (2): CBR=80
Eq.(1): 20
N= 60 1.00
2; 4 10
2.0 8; 16
16L 5
Dry density, d (g/cm3)

0.95

d/(d )max
TUS: Gs=2.798
1.8
4.5Ec ★ 0.90
Laboratory compaction tests
(Murata et al. 2011)
1.6 1.0Ec ▼ 0.1Ec; ◆ 0.2Ec; ★0.5Ec;☆☆1Ec 0.85 (Sr)opt= 81.3 % for all data sets
Laboratory compaction tests (1Ec) (d )max determined for each data set
(Sr)opt=
Gs=2.83; wopt= 12 %, (d)max= 1.996 g/cm3 81.3 %
1.4 0.80
0 20 40 60 80 100 -80 -60 -40 -20 0 20
c) d) Degree of saturation, Sr (%) Sr -(Sr)opt (%)

Fig. 3. Analysis of full-scale and laboratory compaction test data reported by Nemoto and Sasaki
(1994): a) grading curves of sandy loam; b) compaction characteristics with unsoaked CBR contours
on ρd - w plane (TUS 2012 denotes the data obtained at Tokyo University of Science in 2012); c)
compaction curves and unsoaked CBR contours on ρd - Sr plane; and d) ρd/(ρd)max vs. “Sr - (Sr)opt“
relations.

The data points denoted as ▲,○, ●, □ and ■ in Fig. 3b denote the average values of ρd and w from a
test series using a heavy compaction machine (KVR15) with a total weight of 15.9 tonf comprising a
single front vibratory steel drum carrying a load of 63 kN (with a diameter of 1.2 m, a width of 1.99
m, a vibration force of 51.68 kN and a line load of 57.6 kN/m) and four rear pneumatic tires carrying
a load of 93.2 kN (with a diameter of 1.083 m and a width of 0.293 m each). The values of CBR
were measured three times after respective N values. The broken curves denote the contours of CBR
depicted based on an empirical equation derived from the measurements (i.e., Eq. 2 explained later).
A solid curve B-B denotes the laboratory compaction curve (Standard Proctor, 1Ec) on the material,
which is equivalent to approximately N= 3 for the upper 10 cm-thick soil layer in the full-scale
compaction test. A broken curve A-A denotes the compaction curve when N= 16 for the upper 10
cm-thick soil layer in the full-scale compaction test. The data points × and + denote the results of
laboratory compaction tests (1Ec and 4.5Ec) performed separately at Tokyo University of Science. In
Fig. 3c, the data presented in Fig. 3b have been re-plotted changing the abscissa from w to Sr while
adding other laboratory compaction test data of the same soil type (Murata et al., 2011). In Fig. 3c,
the compaction curves from the laboratory and full-scale compaction tests at different CELs exhibit
nearly the same shape and nearly the same (Sr)opt value (81.3 % on average). Then, the ρd/(ρd)max vs.
“Sr - (Sr)opt“ relations obtained under the different compaction conditions have become nearly the
same (Fig. 3d).
100
Miboro dam 100 Core material sieved
御母衣ダム、
passing by weight

Core material used in the field compaction test for laboratory tests
80 (Dmax= 150 mm) コア材料室内試験用材料
Miboro dam
Percent passing by weight

80
Gs = 2.65
重量通過百分率

60 D50 = 0.854 mm
60
Uc = 49.8
FC = 14.4
40 40
Percentage

20
20

0
0
0.01 0.1 1 10 100 1E-3 0.01 0.1 1 4.8 10
a) Partocle diameter, D (mm) b) Grain
粒径, size
D (mm)
(mm)

5
 2.3 1.02
Core material (Miboro dam) Core material (Miboro dam)
Z.A.V.L. Laboratory tests
1.00
2.2 Gs = 2.65 Under No. 4 (Dmax= 4.8 mm)
(Sr)opt = 84 % (from lab. tests)
0.6Ec
1.0Ec 0.98
2.1
Dry density , d (g/cm )

16
3

1.6Ec
12 2.4Ec 0.96
2.0 16 4.0Ec

d/(d)max
Field tests*
N=8
Under No. 4 portion 0.94
1.9 N= 8
N= 12 Range of lab.
0.92
N= 16
1.8 & field test data
12 Full grading (Dmax= 150 mm)
N= 8 0.90
1.7 N= 8 N= 12
N= 16 0.88
* 20 ton-sheepsfoot roller
1.6 Liftsheep’sfoot
before compaction: 20 cm roller
* Field tests using a 20 ton compaction
Fit thickness of soilN:
Lorentzian
Spread number
layer= 20ofcm,
passing
N= number of passes 0.86
1.5 -60 -40 -20 0 20
4 6 8 10 12 14 16 18
Sr-(Sr)opt (%)
c) d) Water content , w (%)

Fig. 4. Compaction curves by laboratory and field compaction tests at different CELs of core
material for Miboro dam site (Mikuni, 1962; Asao, 1964): a) core material used in the field
compaction tests; b) sieved core material used in laboratory compaction tests; c) ρd – w relations; and
d) ρd/(ρd)max vs Sr – (Sr)opt relations.

The same trend as seen from Fig. 3d was observed with many other soil types compacted at
different CELs (Tatsuoka, 2015). Figs. 4a and b show the grading curves of the core material used
for a rockfill dam with a crest height of 131 m completed in 1961 (Miboro dam, Japan) and its sieved
material for laboratory tests. Fig. 4c shows compaction curves from the field and laboratory
compaction tests at different CELs. The field compaction curves of the material having the same
grading as the sieved core material were obtained from the compacted states of field full-grading
material. They are consistent with those obtained by the laboratory compaction tests on the sieved
core material. The (Sr)opt value is similar among all these compaction curves. It may be seen from Fig.
4d that, except for two data points on “the wet side” from the field tests, all the compacted states
shown in Fig. 4c collapse into a rather unique ρd/(ρd)max vs. Sr - (Sr)opt relation. Importantly, the
relations from the field and laboratory compaction tests shown in Fig. 4d are similar to each other, as
in the case shown in Fig. 3d. These results indicate imply that the value of (Sr)opt and the ρd/(ρd)max vs.
Sr - (Sr)opt relation obtained by laboratory compaction tests at a certain CEL using a representative
sample could be applied to the field.
d: proportinally adjusted
2.4
Ohio State Highway Testing and
1.0
Research Laboraotry
2.2 Typical moisture - dry density curves Ohio State DOT: Typical moisture - dry density curves
(set C), groups A - Z from results of tests (set C), groups A - Z from results of tests on 10,000 soil samples
(Standard Proctor) on 10,000 soil samples 0.8
Dry density, d (g/cm )
3

2.0

0.6
w/(d)max

1.8 Sr=100 %
(Gs=2.749:
common value)
1.6 0.4 Linear fitting:
w/(d)max=C1wopt+C2=0.0122wopt+0.3638
1.4 w: density of water
○: (d)max state 0.2
(Sr)opt=1/C1=82.0 %
1.2 Gs=1/C2=2.749
(Sr)opt= 82 % (average)
0.0
0 10 20 30 40 0 5 10 15 20 25 30 35 40
a) Water content, w (%) b) wopt (%)

6
 d: proportinally adjusted d: proportinally adjusted
2.4 1.05
(Sr)opt= 82.0 % (average value) (d)max state Ohio State Highway Testing and Research Laboratory
Typical moisture - dry density curves (set C)
2.2 from results of tests (Standard Proctor) on
10,000 soil samples
1.00
Dry density, d (g/cm )

2.0
3

d/(d)max
1.8 0.95

1.6
0.90
1.4 Gs= 2.749 (common value)
Gs=2.749 (Sr)opt= the value for each soil group
(common value)
1.2 0.85
20 30 40 50 60 70 80 90 100 -70 -60 -50 -40 -30 -20 -10 0 10 20
c) d) Degree of Saturation, Sr (%)
Sr - [respective (Sr)opt] (%)

Fig. 5. Data of 1,000 compaction tests at 1.0Ec by Ohio State (Joslin, 1959): a) 26 average ρd – w
relations; b) estimation of average values of Gs and (Sr)opt; c) ρd vs Sr relations; and d) ρd/(ρd)max vs Sr
– (Sr)opt relations.

Moreover, Joslin (1959) reported 26 average compaction curves derived from laboratory
compaction curves (1Ec) of 10,000 soil types (Fig. 5a). From the reported values of (ρd)max and wopt
for each group, plots shown in Fig. 5b were obtained. The average values of the particle specific
gravity, Gs, and (Sr)opt for the whole data were obtained by linear fitting shown in this figure. This
average Gs value was used in the following data analysis. Fig. 5c shows the ρd vs. Sr relations
obtained from Fig. 5a. Fig. 5d shows the ρd/(ρd)max vs. Sr - (Sr)opt relations obtained from the relations
shown in Fig. 5c by using the (Sr)opt value of the respective groups. In Fig. 5d, all of the 26 relations
are very similar to each other without showing a specific trend of variation according to soil type.
Fig. 6 shows a similar data set of many sandy and gravelly soils obtained at Kobe University. In Fig.
6d, the ρd/(ρd)max vs. Sr - (Sr)opt relations presented in Figs. 4d, 5d and 6c and others presented in Fig.
13 of Tatsuoka (2015) are overlapped. These relations shown in Fig. 3d and 6d are not quite unique.
Therefore, it is difficult to obtain the precise relation for a given soil type from them without doing
relevant compaction tests. Yet, they are quite similar to each other when compared with their
relations on the ρd – w plane and considering that these data were obtained for a quite large variety of
soil type.

2.2
2.2

2.1
2.1
100
2.0
2.0
Percent passing by weight

) 3)

80
ρd (g/cm

1.9
1.9
3
d (g/cm

60 1.8
1.8
Dry density,

40 1.7
1.7

1.6
1.6
20
1.5
1.5
0
0.001 0.01 0.1 1 10 100 1.4
1.4
00 55 10
10 15
15 20
20 25
25 30
30
a) Particle size (mm) b) Water content,
w (%) w (%)

7
 F2 F2
1.05 rd/rdmax
1.05 rd/rdmax
rd/rdmax rd/rdmax
砂礫材料
Sands and gravels 砂礫材料
rd/rdmax rd/rdmax
室内及び原位置締め固め試験
rd/rdmax ----- rd/rdmax
オハイオ州道路局データ
1.00 1.001.00
rd/rdmax (10,000種類の土質; Joslin, 1959
rd/rdmax
rd/rdmax 日本のデータ
rd/rdmax
rd/rdmax 砂質ローム
rd/rdmax
(Dc)t= d/(d)max

(Dc)t= d/(d)max
室内, 1Ec（機械化研究所）
0.95 rd/rdmax rd/rdmax

(Dc)t= d/(d)max
0.950.95 室内, 1 & 4.5Ec（東京理科大学TU
rd/rdmax rd/rdmax
実大締め固め試験（機械化研究所
rd/rdmax rd/rdmax
rd/rdmax 稲城砂 (室内, 0.5 - 9Ec: TUS)
rd/rdmax
0.90 rd/rdmax
0.90 土器川礫 (室内, 0.5 - 9Ec: TUS)
0.90 rd/rdmax
JR粒度調整砕石 (室内, 0.5 - 9Ec;
(ρ 3 3 rd/rdmax 3 rd/rdmax
d)max (g/cm ) (g/cm )
最大乾燥密度 最大乾燥密度 (g/cm ) 御母衣ダムコア室内試験用試料 (
rd/rdmax rd/rdmax
御母衣ダムコア (原位置)
■ 1.9 - 2.12 rd/rdmax ■ 1.9 - 2.12 rd/rdmax
御母衣ダム周辺の8種類土質材料
0.85 ▲ 1.7 - 1.9 0.850.85 ● & ■:▲ 1.7 1.9 --------- : Fig. 5d; ■, ▲, 〇:▽Fig.
Fig.- 4d; 6c
rd/rdmax 　 rd/rdmax
藤沼ダムコア試料
〇 1.5 - 1.7 rd/rdmax 〇 1.5 -sandy
The others: 1.7 and gravelly soils tested at TUS 藤沼ダムランダム材
and
◇ rd/rdmax

rd/rdmax rd/rdmax
others presented in Fig. 13 of Tatsuoka (2015)
0.80 rd/rdmax
0.800.80 rd/rdmax
-70 -60 -50 -40 -30 -20 -10 0 10 20 rd/rdmax
-70
-70 -60
-60 -50
-50 -40
-40 -30
-30 -20
-20 -10
-10 00 10
10 20
20 30 rd/rdmax
Sr - (Sr)opt (%) rd/rdmax S - (S ) (%) rd/rdmax
c) d) rd/rdmax
Sr - (S
r ) r (%)
r opt opt
rd/rdmax
rd/rdmax rd/rdmax
Fig. 6. Sandy and gravelly soils tested at Kobe University (Shibuya and Horita, 2013; Tatsuoka et rd/rdmax rd/rdmax
al., 2013): a) grading curves; b) ρd – w relations for 1.0Ec (the symbols in Figs. 6a and 6b represent rd/rdmax
rd/rdmax
rd/rdmax
rd/rdmax
the same soil types respectively); c) ρd/(ρd)max vs S – (S ) relations; and d) overlapping of the data rd/rdmax
r rd/rdmaxr opt
rd/rdmax
rd/rdmax
rd/rdmax rd/rdmax
presented in Figs. 4d, 5d and 6c and others presented in Fig. 13 of Tatsuoka (2015). rd/rdmax rd/rdmax
rd/rdmax rd/rdmax
rd/rdmax rd/rdmax
rd/rdmax rd/rdmax
Fig. 7 shows the (Sr)opt vs. (ρd)max relations obtained from laboratory compaction tests of a wide
variety of soil type. With the whole data in a very large range of (ρd)max, the scatter of (Sr)opt is not
small and any specific effect of soil type on this scatter cannot be found. So it is not possible to
predict the (Sr)opt value only from a given value (ρd)max without doing any compaction test. Yet, it is
important to note that there is no specific effect of (ρd)max on the (Sr)opt value and the (Sr)opt value is
rather constant in each data set. This fact indicates that, in each earthwork project in which the
variations of soil type and CEL are limited, we can assume that the (Sr)opt value is constant. Two
broken curves denote the Sr (%) - ρd relations for constant air void ratios, va, equal to 5 % and 10 %:

 d  
 d  1 

va  1   (100  Sr )  1    w  (100%) (1)
  w  Gs    w  Gs
 


Along each constant va value, the Sr value decreases at a rate that increases with an increase
in (ρd)max. This implies that the value of va where (ρd)max is obtained for a fixed CEL, (va)opt, is not
constant but tends to decrease with an increase in (ρd)max.

8
 100
F-core 5 MC F-random
90 61
80 TG CG
2
9 IS 3
70 5%
7
60
4
(Sr)opt (%)

50 Data for different CELs (average (Sr)opt)

CG (crushed gravel), TG(Tokigawa); va=10%
40 MC(Miboro dam core); IS(Inagi sand8) (Tatsuoka, (Gs=2.65)
X 1Ec (sandy and gravelly soils) 2011)
30
+ Natural backfill materials (1Ec) (Kobe University)
20 　☆/★ Fujinuma dam, field compaction tests, core/randam (Fukushima Pref.)
10 Nos. 1 -7 & 9; Data for different CELs (Montana data, Mokwa, 2005)
◇ Ohio State DOT typical density curves (Gs= 2.749) (Joslin, 1959)
0
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4
3
(d)max (g/cm )
Fig. 7. (Sr)opt - (ρd)max relations from laboratory compaction tests on a wide variety of soil
(Tatsuoka, 2015).

All the compaction test data available to the authors, including those shown above, indicate
that the value of (Sr)opt and the ρd/(ρd)max vs. Sr - (Sr)opt relation are rather insensitive to changes in
CEL and soil type. Therefore, we can reasonably assume that they are essentially constant in a given
earthwork project in which fixed CEL and soil type are specified and their inevitable variations are
rather limited. In that case, by following the procedure depicted in Fig. 8, irrespective of unknown
variations in CEL and soil type, the true degree of compaction (Dc)t at a transient current compaction
state at given moment and location is readily obtained by substituting a measured in-situ value of Sr
into a given (Dc)t= ρd/(ρd)max vs. Sr - (Sr)opt relation when the (Sr)opt value and this unified relation
have been obtained by relevant laboratory compaction tests performed in advance. Besides, if field
values of both Sr and ρd are available, the value of (ρd)max can be obtained as ρd/(Dc)t. Then, the
compaction efficiency at different CELs by different compaction roller specifications and different
lifts for a given soil type can be evaluated by comparing these values of (ρd)max under respective
compaction conditions. The implications of Fig. 8 in soil compaction controlled are discussed more
in details later in this paper.

(Dc)t=ρd/(ρd)max
(Sr)opt
3) Sr measured 2) Inferred field
in the field compaction curve

1.0
4) Field value of (Dc)t=
ρd/(ρd)max for CEL & soil
type at each location
Sr= 100 %

→ (Dc)t = 100 % if Sr= (Sr)opt

1) Compaction curve by
laboratory test (e.g., 1Ec)

Degree of saturation at compacted state, Sr

Fig. 8. Evaluation of (Dc)t= ρd/(ρd)max from a given Sr value.
Compaction to Sr= (Sr)opt by higher CEL
→ Higher ρd (i.e., higher strength & stiffness)
9
Physical Properties as a Function of Sr at the End of Compaction

Micro-structure of compacted soil

The soil properties of compacted soil are strongly influenced by the micro-structure produced by
compaction in addition to the compacted dry density (or the degree of compaction). Fig. 9
schematicaly illustrates the effects of “Sr at the end of compaction” on the compacted micro-structure.
The observations supporting Fig. 9 are shown in Fig. 10. In Fig. 9, when compacted to point C (Sr <
(Sr)opt), a coherent micro-structure is formed with fine particles sticking to coarse particles by high
matric suction due to a low Sr. This micro-structure is relatively stable with fine particles reinforcing
the contact points between coarse particles while forming relatively large voids, resulting in
relatively high strength and stiffness with a relatively high value of saturated hydraulic conductivity
(as shown below). Although the terminology ‘flocculated’ has been used to describe this
micro-structure of clay, this is not used in this paper dealing with a wide range of soil type including
sandy and gravelly soils. On the other hand, when compacted to point D (Sr > (Sr)opt), a dispersive
micro-structure is formed with fine particles filling voids of the fabrics of coarse particles in a
dispersive manner by low matric suction due to a high Sr. This micro-structure is relatively unstable
resulting in relatively low strength and stiffness because the contribution of fine particles to the
stability of the fabrics of coarse particles is relatively low, while the voids become relatively small
resulting in a relatively low value of saturated hydraulic conductivity (as shown below).

D: dispersive
(Sr)opt High- micro-structure
saturated
Dry density, ρd

Compaction Sr= 100%

curve by a
certain CEL
C D C: coherent
micro-structure
Low-saturated

Water content, w

Fig. 9. Schematic diagram showing the effects of “Sr at compacted state” on the micro-structure of
compacted soil.

Figs.10a, b and c (Matsumura and Tatsuoka, 2018) show the horizontal slice images by the X-ray
CT scanning of the specimens of K3 soil compacted to (Dc)1Ec≈ 100 % (i.e., ρd≈ [ρd]1Ec= 1.059
g/cm3) at w= 0.8, 1.0 and 1.2 times (wopt)1Ec= 40.5 %, so Sr was lower than, equal to, and higher than
(Sr)opt= 74.3%. K3 soil is one of the silty sands of which cyclic undrained strength was evaluated as
explained later. The specimens (10 mm in diameter and 60 mm high) were scanned immediately
after compaction. The resolution of images obtained was 7.7 m/pixel. In these images (shown in 16
bit grey-scale), the brighter white areas are materials with a higher density. The black areas
correspond to pore air or water. Among the three images, differences are noticeable in the zones
consisting of fine particles surrounding coarse particles. When compacted at Sr lower than (Sr)opt (Fig.
10a), the fine particles form aggregations due to strong suction with some adhering to zones around
inter-particle contact points between coarse particles creating relatively large voids. Some
aggregations contain coarse particles while the others are fully separated from coarse particles. As a
result, to achieve a given dry density, a higher compaction energy becomes necessary. This
10
compacted state is represented by point C in Fig. 9. When compacted at Sr higher than (Sr)opt (Fig.
10c), due to weak or negligible effects of suction, the fine particles are spread in the voids of the
fabrics of coarse particles, not forming such aggregations with pores in between as when compacted
at Sr < (Sr)opt. This compacted state is represented by point D in Fig. 9. It may be seen from Fig.
10b that the soil compacted at Sr = (Sr)opt) shows the intermediate soil fabrics between the two cases
described above. More details are reported in Matsumura and Tatsuoka (2018).

1mm 1mm 1mm

Coarse Coarse Pore Coarse

particle Pore particle particle Pore
a) b) c)
Fig. 10. Horizontal slice images of compacted K3 specimens (ρd= [(ρd)max]1Ec= 1.059g/cm3)
(Matsumura and Tatsuoka, 2018): a) w= 0.8･(wopt)1Ec or Sr = 60.2 %; b) w= (wopt)1Ec= 40.5 % or Sr
= (Sr)opt= 74.3%; c) w=1.2･(wopt)1Ec or Sr = 91.3%

In the following, it is shown that many of the major physical properties of compacted soil (i.e.,
shear strength, stiffness, strength/stiffness drop and collapse upon wetting, saturated hydraulic
conductivity etc.) are controlled by the Sr value at the end of compaction relative to (Sr)opt in addition
to the compacted value of ρd.

100 100
N= 8 実線群：　(1)式（Nに独立）
Solidcurves:
Solid curves:Eq.
Eq.121
Eq. N= 8 Solid curves: Eq. 2
実線群：　(1)式（Nに独立） 1
90 N= 8 実線群：　(1)式（Nに独立） w (%) = 7
55
w (%)=6-8 55 80 Sr (%) = 25 65
80 Sr (%) = 25 65
8-10 Sr(%) = 20-30 35
Sr(%) = 20-30 35 45
10-12 45 30-40
30-40 9 75
12-14 40-50
(%)

60
CBR (%)

60 40-50 75
60
CBR(%)

14-16 50-60
50-60
16-18 60-70 82.5
CBR

60-70
18-20 82.5 70-80
40
40 70-80
20-22 80-85
30 80-85 11
22-24 85-90 87.5
85-90 87.5 20 90-100
20 90-100 95
13 95
15
Sr= 100 %
0 0
01.2 1.5 1.8 2.1 1.2 1.4 1.6 1.8 2.0 2.2
1.2 1.4 1.6 1.8 2.0 2.2 3
Dry乾燥密度 3
density, , d (g/cm ) 3 Dry乾燥密度
density,, d (g/cm )
Dry乾燥密度
density,, d (g/cm )
a) b)
Fig. 11. Unsoaked CBR vs. ρd relations when N=8 from the full-scale compaction tests (Fig. 3) for:
a) different w values; and b) different Sr values.

CBR
Although CBR may not be well representative of the stress-strain behavior at small strains, which
11
is often taken into account in the recent mechanistic design of soil structure, it is still a good index
for the behavior at large strains including the peak strength of compacted soil. A large amount of
unsoaked CBR data obtained from the full-scale compaction tests described in Fig. 3 is analyzed
below. Usually, the strength and stiffness of compacted soil is expressed by empirical equations
using ρd and w as the variables. So, in Fig. 11a, the unsoaked CBR values for different w values after
the number of pass N= 8 obtained from all the full-scale compaction tests are plotted against ρd.
Similar figures are obtained for other N values (Tatsuoka, 2015). The solid curves denote an
empirical equation (i.e., Eq. 2 explained later). With an increase in ρd during compaction at a fixed w,
CBR increases until ρd becomes a certain value then starts decreasing. This peculiar trend can be
explained as follows. The unsoaked CBR is a function of at least: 1) ρd; 2) the matric suction during
the measurement of CBR; and 3) the micro-structure produced by compaction. By factors 2 and 3,
the unsoaked CBR (i.e., the strength and stiffness) of compacted soil decreases with a decrease in the
matric suction associated with an increase in Sr. So, despite an increase in ρd during compaction at a
fixed w, the CBR value decreases once the negative effects on CBR of “factors 2 and 3 by an
increase in Sr” becomes more dominant than the positive effects of “factor 1 by an increase in ρd”.
120 120
Sr(%)=30-40
CBR (%)= Sr(%)=70-80
100 100
N=0 1.2(ρd(–
CBR=1.2 d0.4) )9.5
- 0.49.5
N=1
80 N=2 80 N=0 CBRCBR=0.4
(%)= 0.4(ρ – 0.4)
( d- 0.4 )9.5
9.5
d
CBR (%)

N=4 N=2
CBR (%)

N=8 N=4
60 60
N=16 N=8
N=16
40 40

20 a)
20
b)
0 0
1.2 1.4 1.6 1.8 2.0 2.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2
3
Dry乾燥密度
density, , d (g/cm ) Dry乾燥密度
3
density,, d (g/cm )

Fig. 12. Unsoaked CBR vs. ρd relations when Sr=: a) 30 - 40 %; and b) 70 – 80 %.

1.5

Data point
CBR(Sr) (%)
(%)

1.0
関数, ffCBR

Fitted relation
Function

Boltzmann 関数
Fitted Boltzmann function
0.5 y = A2 + (A1-A2)/[1 + exp{(x-x0)/dx}]

A1= 1.26215 ±0.05888

A2= 0.04176 ±0.0824
x0= 64.79547 ±2.1094
dx= 9.96664 ±2.16158
0.0
0 20 40 60 80 100
締固め時飽和度
Degree of saturation , Sr (%)
when compacted, Sr (%)

Fig. 13: A function expressing the effects of Sr on unsoaked CBR

The data shown in Fig. 11a have been re-plotted changing the parameter from w to Sr in Fig. 11b.
For any value of Sr, unsoaked CBR consistently increases with an increase in ρd. Besides, all the
CBR vs. ρd curves for different Sr values exhibit nearly the same shape exhibiting a large decrease in
CBR with an increase in Sr. So, in Fig. 12, the unsoaked CBR values when Sr= 30 - 40 % and 70 -
80 % from the full scale compaction tests are plotted against ρd. Similar results were obtained for
other Sr values (Tatsuoka, 2015). Importantly, these unsoaked CBR vs. ρd relation for different Sr
values are independent of compaction roller type and the number of pass N (i.e., independent of
CEL). Accordingly, these test data were fitted by the following empirical equation comprising two
12
uncoupled functions of Sr and ρd:

CBR  fCBR ( Sr )  ( d /  w  b)c (2)

where b= 0.4 and c= 0.9 (positive material constants); and fCBR is a function of “Sr at the end of
compaction”. In Fig.13, the values of fCBR for different ranges of Sr obtained from the relations
shown in Fig. 12 and others are plotted against Sr. It may be seen from Figs. 11a and 11b and Figs.
12a and 12b that Eq. 2 expresses reasonably the whole data. Importantly, the value of unsoaked CBR
for given values of ρd and Sr can be estimated by Eq. 2 without knowing the value of CEL, which is
usually very difficult to estimate in the field. Besides, by Eq. 2, the value of ρd can be obtained from
given values of CBR and Sr, therefore, from given of CBR and w.
Eq. 2 is the empirical equation for a specific soil type. A possible more general equation applicable
to different soil types may have the following form:

CBR  gCBR (soil type)  fCBR. N (Sr  (Sr )opt ) {( Dc )1Ec  B}C (3)

where gCBR is a function of soil type; fCBR.N is a function of the difference Sr – (Sr)opt; (Dc)1Ec is the
degree of compaction for Standard Proctor; and B and C are constants, where C is the same as c in
Eq. 2. More study is necessary in this respect.
100

Eq. (2)(1)式（Nに独立）
(constant w) w (%) = 7
80

(Sr)opt= 83.3 %
最適含水比状態： 9
(Sr)opt=81.3%
CBR (%)

60

40
11
A
20
13
15
Sr=
= 100
100%%
0
1.2 1.4 1.6 1.8 2.0 2.2
乾燥密度, d (g/cm3)

Fig. 14. Unsoaked CBR vs. ρd relations for several constant w and Sr values by Eq. 2

Fig. 14 shows the relationships between the unsoaked CBR and ρd for several w values presented
in Fig. 11a together with the relation for Sr= (Sr)opt and 100 %, all obtained by Eq. 2. With an increase
in ρd during compaction at a fixed w, the unsoaked CBR value increases until it becomes the
maximum at a certain ρd value where Sr is still lower than (Sr)opt. The Sr= (Sr)opt state, where ρd=
“(ρd)max for the transient value of CEL”, is reached at point A. Beyond point A, unsoaked CBR
decreases fast due to an increase in Sr in spite of an increase in ρd. It may be feasible to find the Sr=
(Sr)opt state by detecting the start of decreasing in CBR that follows its transient maximum. However,
this procedure requires very careful continuous observation of CBR. These trends indicate that the
Sr= (Sr)opt state (i.e., the (ρd)max state for a given transient CEL) cannot be detected only by measuring
the values of unsoaked CBR or the stiffness of compacted soil. In other words, the maximum
compaction efficiency is not achieved by seeking for the maximum unsoaked CBR or stiffness state
of compacted soil.
The contours of unsoaked CBR obtained from Eq. 2 are depicted in Figs. 3b and 15a. On the other
hand, soaked CBR is a relevant index of the strength and stiffness after wetting, which is usually
13
used in the design of soil structures that would get wet or fully saturated by heavy/prolonged rains,
floods, impounding etc. during their lifecycle span. In Fig. 15a, the contours of soaked CBR are also
depicted that were obtained by incorporating the effects of soaking into the function fCBR in Eq. 2.
The soaking effects were evaluated by a series of laboratory CBR tests performed on unsoaked and
soaked samples produced by compacting the same soil type as used in the full-scale compaction tests
to various ρd - Sr states (Tatsuoka, 2015). As seen from Fig. 15a, the shapes of the unsoaked and
soaked CBR contours depicted on the ρd and w plane are very complicated in such that each contour
crosses a vertical constant w line two times. This trend reflects that, in Fig. 14, at a constant w,
unsoaked CBR does not increase monotonously with an increase in ρd, so two values of ρd are
obtained for each unsoaked CBR. In Fig. 15a, curve A is the compaction curve fitted to the data
points when N= 16 for the upper 10 cm soil layer in the full-scale compaction test (Fig. 3b), while
curve B is the compaction curve by laboratory compaction tests (1.0Ec). Fig. 15b shows the
relationships between the unsoaked and soaked values of CBR and “w during compaction” along
compaction curves A and B. In these two cases at different CELs, the following common trends can
be seen. When compacted to Sr < (Sr)opt, the unsoaked CBR is relatively high, whereas it drops
largely upon soaking. Collapse deformation upon wetting is also significant if ρd is low (as shown
later). When Sr= (Sr)opt, CBR drops only slightly upon soaking, while the soaked CBR is close to its
maximum at a slightly lower w along each compaction curve. Besides, collapse deformation upon
wetting is insignificant, in particular when well compacted (as shown later). When Sr > (Sr)opt, the
drop of CBR upon soaking is negligible, whereas CBR decreases largely with an increase in w along
each compaction curve. Besides, the CBR value decreases with an increase in CEL at a constant w.
These results indicate that the Sr= (Sr)opt state is the relevant compaction target to obtain a
near-maximum strength/stiffness after wetting under given compaction conditions irrespective of the
value of CEL (which is usually unknown). Nowak and Gilbert (2015) also showed a similar principle
in soil compaction control as above.
2.2 80
Soaked CBR (%) A: Full-scale compaction test Full-scale test
= 80
Unsoaked or soaked CBR (%)

(upper half at N= 16) Top 10 cm, N=16

wopt=9.7%
(Sr)opt=81.3% Curve A-A:
B: Laboratory for curve A-A
60 unsoaked
compaction tests
Dry density,d (g/cm3)

40
(1.0Ec)
2.0
Unsoaked CBR (%) =80 Sr= 100 % Laboratory compaction
test (1Ec)
Gs= 2.83
20 40 Curve B-B:soaked

40 A wopt=12%
Curve A-A:
1.8 10 soaked for curve B-B
90 20
20
5 80
Curve B-B:
10 Sr (%)= 70 soaked
5 0
1.6 0 5 10 15 20
0 5 10 15 20
a) b) Water content, w (%)
Water cotent, w (%)
Fig. 15. a) Contours of unsoaked and soaked CBR on the ρd - w plane; and b) CBR – w relations
along two compaction curves, A and B, shown in Fig. 15a.

Shear strength of compacted soil

Tatsuoka (2011 and 2015) showed that the shear strength and stiffness of saturated specimens of a
number of well-graded sandy and gravelly soils evaluated by drained triaxial compression (TC) tests
are also a function of ρd and “Sr at the end of compaction”, like the soaked CBR values. Jenkins
(1996) prepared specimens of three types of well-graded cohesive soils by combining sand and
14
gravel particles with a silty clay to have different amount of fines content (25%, 35% and 45%) and
compacting them at different water contents at CEL equal to 1.0Ec. He showed that, irrespective of
soil type and compacted states, the values of unsoaked CBR are uniquely related to the undrained
shear strength evaluated by UU TC tests with a confining pressure equal to 50 kPa. Combined with
the fact that the unsoaked CBR values can be expressed by Eq. 2, this result suggests that the
undrained shear strength of compacted soil can be expressed by a function similar to Eq. 2. Abe et al.
(2014) reported unconfined compression strength qu of moist samples compacted to different states
denoted by symbol ■ in Fig. 16a (n.b., points a- d are discussed later). The material was silty sand
(Gs= 2.7; 28% clay; 26.5% silt; 44.6% sand; and 0.9% gravel; PI= 28.5; and wP= 30.5%) retrieved
from highway embankment.
1.5 50
Silty sand (embankment

(kN/m2) qu (kPa)
10 5 常磐自動車道上り線92.4KP qu= gqu(ρd)・fqu(Sr)
for Joban Highway) D(D )1Ec=%100 % D
(Dc=c)100 %
1Ec= 100 %
15 c 盛土・シルト質砂 c=c 100
3 gqu(ρd): see Fig. c)
d (d=1.439 g/cm3) (d=1.439 g/cm ) fqu(Sr)= [64.0 –(S0.97751･S
va (%) Sr= 100 %; va= 0 % 40 qu=A*{64.0-0.97751* )r+0.00414*r(Sr)2}
1.4
3 3)

qustrength,
= 20 (Gs=2.7)
+ 0.00414･(Sr)2]/12
d (g/cm

b d
)

95 (1.367)
density,dρ(g/cm

1Ec
30
qu when Sr=

Constant Sr
compression
1.3 90 (1.295) (Sr)opt= 81
(Sr)opt= 81 % 95 (1.367)
一軸圧縮強度,
乾燥密度,

Constant w 20 c
A=1.9
22.8
85 (1.223)
Dry

1.2 a
wopt= 26.3 % 90 (1.295)
10 1.0
12.0
b
Unconfined

Sr (%) 0.48
85 (1.223) 5.76
= 60 70 80 90
a 2.3
0.19
1.1 0
15 20 25 30 35 40 45 50 50 60 70 80 90 100
含水比,
Water w (%)
content, w (%)
a) b) 締固め時飽和度,
Degree of, Sr (%)Sr (%)
50
qu (kN/m2)

40 d
quSwhen
r= 62.4S%での関係
for constant Sr (kPa)

r= 62.4 %
qurでの一軸圧縮強度,

30 Sr= (Sr)opt= 81%での

全データのフィッティング曲線：
“q when S = (S ) =
u r r opt
y=f (d)= -715.76+1969.0*x
d 81 %” – ρd relation
20 -1802.3246*x^2+549.65324*x^3
fitted to all the data
Dc=(D100 %
c)1Ec= 100 %
(ρd= 1.439 g/cm 3)
3
(d=1.439 g/cm )
一定のS

10
85 (1.223) 95 (1.367)
a
90 (1.295)
0
1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45 1.50
3 3
乾燥密度, d (g/cm
Dry density, ) )
ρd (g/cm
c)
Fig. 16. Unconfined compression strength of compacted soil from highway embankment (Abe et
al., 2014): a) compacted states relative to compaction curve (1Ec); b) qu vs. “Sr at the end of
compaction” relations for different dry densities; c) qu vs. dry density relations for Sr= 62.4 % and
(Sr)opt= 81 %..

The analysis of the data showed that qu can be expressed by Eq. 4, comprising two uncoupled
functions of ρd (or (Dc)1Ec) and Sr, like Eq. 2.

qu  gqu ( d )  fqu (Sr ) (4)

where fqu(Sr) is the function expressing the effect of “Sr at the end of compaction” on qu, which was
obtained by fitting to the measured relationships between qu and Sr for different constant ρd values
(Fig. 16b). gqu(ρd) is the function expressing the effect of ρd on qu, obtained by fitting to the
15
relationship between the value of qu when Sr= (Sr)opt= 81 % (which is equal to gqu(ρd)･fqu(Sr= 81%))
and ρd (Fig. 16c). The values of “qu when Sr= 81 %” shown in Fig. 16c were obtained as {measured
qu/fqu(Sr)}･fqu(Sr＝81%) for all the data shown in Fig. 16b. In Fig. 16c, the “qu when Sr= 62.4 %” – ρd
relation obtained based on Eq. 4 is also shown. By Eq. 4, without knowing CEL, the qu value of this
soil type at a given compacted state can be obtained.
Fig. 17b shows the drained triaxial compression (TC) strength of moist specimens compacted as
shown in Fig. 17a and the drained and undrained TC strengths of saturated compacted specimens
plotted against (Dc)1Ec of the soil used in the full-scale compaction tests (Fig. 3). The number of data
is not sufficient due to a limited amount of the test material available. Despite the above, the
following trends may be seen. The drained shear strength largely drops by wetting. The effects of Sr
at the end of compaction on the moist drained TC strength are noticeable while the effects become
smaller by wetting. When saturated, the effects of (Dc)1Ec on undrained strength is much larger than
the drained strength. Further study will be required to find a relation, such as Eq. 4, for the TC
strength, then the relationship between the unsoaked CBR and the undrained and drained shear
strength after wetting of cohesive soil.

砂質ローム 600
2.2
100% Sandy loam
三軸圧縮試験供試体 Sandy loam
砂質ローム
qmax (kPa)

（排水試験） states of
Compacted
60% 白抜き：不飽和
TC tests (σc’= 50 kPa)
三軸圧縮試験供試体
Sr = TC specimens （排水試験）
塗りつぶし：飽和
白抜き：不飽和
qmax (kPa)

40% (see Fig. b for TC test

（非排水試験）
塗りつぶし：飽和 Drained strength
density, d (g/cm3)

2.0
十字マーク：飽和
conditions) 400
（非排水試験） (unsaturated as
strength,

十字マーク：飽和 compacted)
圧縮強度,
Dry 乾燥密度,

Compressive

1.8 200 Drained strength

4.5Ec (saturated)

(Sr)opt = 81.3 % Undrained strength

1.0Ec
(saturated)
1.6 0
4 8 12 16 80 90 100 110
3
a) 含水比, w (%)
Water content, b) Degree 乾燥密度, d (g/cm
of compaction, (D ) ) (%) c 1Ec

Fig. 17. TC tests (σc’= 50 kPa) on sandy loam used in full-scale compaction tests (Fig. 3)
(Tatsuoka, 2015): a) compacted states; and b) TC strength plotted against (Dc)1Ec.

Cyclic undrained strength

Matsumura and Tatsuoka (2018) showed that the cyclic undrained triaxial strength, SR20, of five
sandy soil types (Fig. 18a) can be expressed by Eq. 5 comprising two uncouple functions, similar to
Eq. 3:

SR20  g SR. N [( Dc )1Ec , soil type]  f SR [Sr  (Sr )opt , soil type] (5)

The specimens were prepared by compacting to different states referring to the respective
compaction curves (1Ec) (Fig. 18b) and making fully saturated. SR20 is defined as the cyclic shear
stress ratio SR= σd/(2σ’c) necessary to cause a double amplitude axial strain DA= 5 % at a number of
loading cycle Nc= 20, where σd is the single amplitude of symmetric cyclic deviator stress; and σ’ c is
the isotropic effective confining pressure equal to 50 kPa. Each SR20 value was obtained from
respective SR – Nc relations, as shown in Fig. 19a (for K3 soil), which is typical of the test results.
gSR.N is the function expressing the effects of the degree of compaction (1Ec), (Dc)1Ec, and soil type

16
on SR20 when Sr= (Sr)opt (Fig. 19b). The gSR.N - (Dc)1Ec relation is noticeably different among the five
soil types.
fSR is the function representing the effects on SR20 of Sr and soil type expressed by the
relationship between the ratio SR20/gSR.N and Sr - (Sr)opt (Figs. 19c and d). With two soil types having
the fines content, FC ≤ 17.4 % (K1 and K2), fSR is rather independent of Sr - (Sr)opt. On the other hand,
with three soil types having FC ≥ 31.0 % (I, K3 and K4), with an increase in Sr, fSR decreases slightly
when Sr < (Sr)opt while it decreases significantly as Sr becomes higher than (Sr)opt. This trend is
consistent with the function fCBR(Sr) of unsoaked CBR (Fig. 13) and due likely to such different
micro-structures produced by compaction to different Sr values relative to (Sr)opt as shown in Figs. 9
and 10. Combined with the fact that the value of (Dc)1Ec attained by the same compaction effort
becomes the maximum when Sr= (Sr)opt, the trends of fSR described above indicate that the maximum
value of SR20 for the same compaction effort is obtained by compaction to “Sr at the end of
compaction”= (Sr)opt, particularly by avoiding compaction to Sr> (Sr)opt when FC is high.

D= 0.075 mm Standard Proctor (1Ec)

Zero air void
Percentage passing in weight

Sr= 100 %
Dry density, ρd (g/cm3)

(Gs= 2.509
~ 2.578)

Particle size, D (mm)

a) b) Water content, w (%)

Fig. 18. a) Grading curves; and b) compaction curves (1Ec) of silty sands of which cyclic
undrained triaxial strengths were evaluated (Matsumura and Tatsuoka, 2018).

0.6 gSR.N= SR20 when Sr= (Sr)opt

w (%) ρd(g/cm3)
Cyclic undrained triaxial strength

= a･[(Dc)1Ec/100 – b/(ρd)max]c
43.5 1.054
gSR.N= SR20 when Sr= (Sr)opt

0.5 41.3 1.044

a, b & c: material constants
r)SA/(2σ’c)

40.5 0.999
SR

43.4 0.980
(σa – σratio,

33.2 1.053
0.4 34.1 1.019
32.7 0.945
SR= stress

28.5 1.003
0.3 46.6 1.061
46.4 1.010
Cyclic

48.0 0.965
0.2

0.1
1 10 100
Number of loading cycles Ncto
toDA=
cause DA=5% Degree of compaction, (Dc)1Ec (%)
Number of loading cycles 5%
a) b)

17
 Fines content ≤ 17.4 %: Fines content ≥ 31.0 %:
△:I

●: K4
fSR[Sr – (Sr)opt]

fSR[Sr – (Sr)opt]
〇: K3
K1 K2

For this plot,

fSR= SR20/[a･{(Dc)1Ec/100 – b/(ρd)max}c]

S – (S ) (%) r r opt S – (S ) (%)

c) r r opt
d)
Fig. 19. Cyclic undrained triaxial strength of saturated sandy soils (Matsumura and Tatsuoka,
2018): a) SR – Nc relations for K3 soil compacted to different states; b) function gSR.N expressing the
effects of (Dc)1Ec; and c) and d) function fSR expressing the effects of “Sr at the end of compaction”.

Collapse upon submerging

Okamoto et al. (2014) performed a comprehensive series of 1-D compression tests to evaluate
vertical strain εv upon submerging at different over-burden stresses of various soil types (Fig. 20a).
The specimens (4 cm-high and 6 cm in diameter) were compacted to different states relative to the
respective compaction curves (1Ec) (Fig. 20b). The analysis of the data showed that Eq. 6
comprising four uncoupled functions reasonably expresses the whole data of ε v (%):

 v  f .1 (soil type)  f .2 (Sr  (Sr )opt )  f 3[( Dc )1Ec ]  f .4 ( v )

v v v v
(6)

where the function fεv.1 expresses effects of soil type, which increases with an increase in FC and
equal to 0.21 for Heiso and Masado soils; the function fεv.2 expresses the effects of Sr, which is equal
to “(Sr - (Sr)opt) + 5 %” when “Sr - (Sr)opt” ≤ – 5 %; and 0 when “Sr - (Sr)opt” ≥ – 5%; the function fεv.3
expresses the effects of (Dc)1Ec, which is equal to [95 - (Dc)1Ec (%)]/15 when (Dc)1Ec ≤ 95 %; and 0
when (Dc)1Ec ≥ 95%; and fεv.4 expresses the effects of σv, which is equal to [σv (kPa)/150]m where m=
0.3.
Figs. 21a and b show the contours of εv when σv= 150 kPa of Heiso and masado soils obtained
from Eq. 6. εv increases significantly as (Dc)1Ec decrease from 95 % and “Sr at the end of compaction”
decreases from “(Sr)opt – 5 %”. On the other hand, εv becomes negligible either when (Dc)1Ec becomes
higher than 95 %; or when Sr becomes higher than (Sr)opt – 5 %. So, by compaction to Sr= (Sr)opt and
(Dc)1Ec higher than 95 %, the condition of not only εv= 0 but also high strength and stiffness can be
ensured.

18
 100 2.0
○ Heiso dam　 ○ Heiso dam (Gs= 2.655)　
● Eiheiji　 ● Eiheiji (2.737)　
80 △ Masado (weathered granite)　　　 △ Masado (weathered granite)
Percent passing by weight

(2.690)　　　

Dry density, d (g/cm )

▲ Borrowed fill material 1.8

3
▲ Borrowed fill material (2.611)
□ Shiba embankment □ Shiba embankment (2.676)
60

1.6
40
Sr= 100 %

1.4
20

0 1.2
0.001 0.01 0.1 1 10 20 30 40
Particle diameter, D (mm) Water content, w (%)

Fig. 20. a) Grading curves: b) compaction curves (1Ec) of soils of which collapse properties were
evaluated by 1-D compression tests (Okamoto et al., 2014).

20% 40% 60% 80% Sr= 100 % Sr= 100 %

20% 40% 60% 80%
2.0 2.0
(Sr)opt= 86.34 % (Dc)1Ec= (Sr)opt= 87.2 %
(Dc)1Ec=
100% 1.9 0.2 % Compaction curve
0.2 % Compaction curve 100%
1 % 0.5 % (1Ec) 1 % 0.5 % (1Ec)
Dry densityd (g/cm )

1.8 0 % 95% 1.8 0 %

3

95%
Dry densityd (g/cm )

0%
3

0%
90% 1.7 90%
2% 2%
1.6 85% 1.6 5%
5% 85%

80% 10 % 80%
10 % 1.5
v=15 % v=15 %
1.4 1.4

Heiso dam 1.3 Masado (weathered granite)

1.2 1.2
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Water content, w (%) (wopt)1Ec (wopt) w (%)
a) 1Ec
b) Water content, w (%)

Fig. 21. Contours of 1D compression upon submerging when σ v= 150 kPa obtained by analysis of
the data reported by Okamoto et al. (2014): a) Heiso dam (Gs= 2.655; (Sr)opt= 86.34 %; (ρd)max=
1.860 g/cm3; and wopt= 13.9 %); and b) Masado (Gs= 2.69; (Sr)opt= 87.2 %; (ρd)max= 1.855g/cm3; and
wopt= 14.6 %).

Elastic shear modulus

The elastic shear modulus, G0, is more representative of the stress-strain behaviour at small
strains of the subgrade of road and railway under working conditions than CBR or shear strength
(e.g., Biarez et al. 1999; Gomes Correia, 2008). Heitor et al. (2013, 2015) reported the values of G0
measured by the bender element method under unconfined conditions of a silty sand compacted to
various states with different water contents along four compaction curves (Fig. 22a). Fig. 22b shows
the values of G0 measured at different Sr values plotted against ρd. These G0 (MPa) vs. ρd relations
have been grouped based on “Sr (%) at the end of compaction”, similar to Fig. 11b for unsoaked
CBR, and fitted by solid curves expressing Eq. 7, similar to Eq. 2 for unsoaked CBR:

G0  fG 0 (Sr )  ( d / w  b)c (7)

where b and c are the material constants and it was found that b= 0.4; and c= 9.5 for Eq. 2 are also
relevant to Eq. 7. In Fig. 22c, the values of fG0 obtained as “measured G0”/(ρd/ρw - b)c are plotted
19
against Sr. The whole data, except for two data points indicated by an arrow, were fitted by a solid
curve expressing the function fG0(Sr) of Eq. 7. It may be seen from Fig. 22b that Eq. 7 reasonably
expresses the whole data except for the two data points mentioned above. Heitor et al. (2013, 2015)
showed that these G0 values are controlled by matric suction and obtained a more comprehensive
empirical equation including suction as a variable. With Eq. 7, the effects of suction are implicitly
involved in the two functions of Sr and ρd. When based on Eq. 7, the G0 value can be obtained only
from the values of Sr and ρd without evaluating the value of suction and CEL. It is expected that an
empirical equation for field G0 values under working conditions, where the confining pressure is not
equal to zero, is similar to Eq. 7.
In Fig. 22a, the G0 contours by Eq. 7 are depicted on the ρd vs. w plane. The shape of these G0
contours is similar to the one for the unsoaked CBR contours (Figs. 3b and 15a). Fig. 22d shows the
G0 vs. ρd relations for different values of w and Sr obtained from Eq. 7. With an increase in ρd at a
fixed Sr, the G0 value consistently increases. This trend is very similar to those of the unsoaked CBR
(Fig. 11b). On the other hand, with an increase in ρd at a fixed w, the G0 value does not increase
monotonously, but it exhibits the maximum before Sr becomes (Sr)opt. This implies that, when Eq. 7
is representative of the G0 values of soil compacted in the field, the compaction target where Sr=
(Sr)opt is attained after G0 has started decreasing during compaction at constant w. This trend is the
same as unsoaked CBR of compacted soil (Fig. 14). Although it is a delicate procedure, the (Sr)opt
state may be detected from the start of decreasing in continuous measured G0 values.
Fig. 22e shows the G0 vs. w relations along two compaction curves for CEL= E2 and E4, similar
to the relations for unsoaked CBR shown in Fig. 15b. The peak G0 value is obtained when w is
noticeably lower than wopt (i.e., when Sr is noticeably lower than (Sr)opt) irrespective of CEL. These
trends indicate that the compaction target where Sr= (Sr)o is not attained by seeking for the maximum
G0 state in soil compaction control. Obviously, to identify the current compacted state relative to the
compaction target state on the ρd vs. w plane, the G0 value is not sufficient, but the Sr value, or other
information from which Sr can be obtained, is also necessary. It may also be seen from Fig. 22e that,
when w is larger than about 11 %, the G0 value decreases with an increase in CEL from 0.41Ec to
1.4Ec, which implies the phenomenon of over compaction. These implications with G0 in field soil
compaction control are similar to those with unsoaked CBR.
F1
2.1 Silty sand (SP-SC), Penrich, Australia 200
Gs= 2.7, 89 % sand; 7 % silt; & 4 % clay Sr (%)= 50 60 70
1Ec= standard Proctor 80
Initial shear modulus, G0 (MPa)

51.3
2.0 G0=fG0(Sr)(d/w - 0.4)9.5
of unconfined moist specimen

150 66.1
G0 (MPa); Sr (%)
Dry density, d (g/cm )

S=
3

1.9 200 r 100 76.8 85

% 68.4
100 42.4 77.1
1.8 100
63.8 84.4 87
E4 (1.4Ec) Sr (%)= 34.3
50.4 69.6 76.6 80.1
1.7 (S
E3 (0.89Ec) )
r o 50.0
=8 pt 50 86.1
50 0% 70.4
1.6 E1 (0.25Ec) 85.9 86.8
S= 86.6
E2 (0.41Ec) r 67 %
G0 (MPa)= 20 87.0 87.6 86.8 86.3
1.5 0
6 8 10 12 14 16 18 1.5 1.6 1.7 1.8 1.9 2.0 2.1
Compacted dry denisty, d (g/cm3)
a) Water content, w (%) b)

20
 F1
15 200
9.5
G0=fG0(Sr)(d/w - 0.4) [G0 (MPa); Sr (%)] Sr (%)= 50 60 70
2.0

Initial shear modulus, G0 (MPa)

(Sr)opt

of unconfined moist specimen

9.5
G0=fG (Sr)(d/w - 0.4)
0
150 = 80
1.5 G0 (MPa); Sr (%)

fG0(Sr) (MPa)
10
fG0(Sr) (MPa)

1.0 85

0.5 100 w (%)= 9 10 11

87
wopt (1Ec) 12
5 0.0
75 80 85 90 = 11.6 %
Sr (%) 13
50

14
0 0
30 40 50 60 70 80 90 100 1.6 1.8 2.0
3
Sr (%) Compacted dry denisty, d (g/cm )
c) d)
200
CEL= E4(1.4Ec) 9.5
G0=fG (Sr)(d/w - 0.4)
0
(1Ec= Standard Proctor)
Initial shear modulus, G0 (MPa)

G0 (MPa); Sr (%)
of unconfined moist specimen

150

(wopt)E4= 10.6%
(Sr)opt= 80%
100

(wopt)E2= 13.05%
CEL= E2(0.41Ec)
50

0
6 8 10 12 14 16 18

e) Water content, w (%)

Fig. 22. Elastic shear modulus, G0, of unconfined moist specimens of a silty sand reported by
Heitor et al. (2013, 2015): a) compacted states where G0 was measured along four compaction
curves; b) G0 vs. ρd relations for different Sr values; c) the function expressing the effects of Sr on G0;
d) G0 vs. ρd relations for different values of w and Sr; and e) G0 vs. w relations along compaction
curves for E2 and E4.

Coefficient of hydraulic conductivity of saturated soil

Fig. 23a shows compaction curves for five different CELs of the sieved core material (SCM, Fig.
4b) for Miboro dam. Fig. 23b shows the relationship between the coefficient of hydraulic
conductivity under saturated condition, k, in the logarithmic scale and “w during compaction” along
these five compaction curves. These k values were evaluated by falling head tests on unconsolidated
specimens (JIS A 1218, 2009). When compacted at CEL= 1Ec, the k value becomes the minimum at
point A, where w is slightly higher than “wopt for 1Ec”. Compaction wetter than the optimum is
usually recommended for fills impeding the hydraulic flow, such as fill dam cores, river dykes and
clay liners (e.g., Mitchell et al., 1965; Daniel and Benson, 1990; Benson and Trast, 1995). However,
wopt decreases with an increase in CEL. If the field CEL is 4Ec, compaction at “w around at point A”
results in only a small increase in ρd and a small decrease in k from respective values for 1Ec,
whereas a higher ρd value and a lower k value are obtained by compaction at “w at point B” (i.e.,
“wopt for 4Ec”). The minimum k value for 4Ec, which is noticeably lower than the minimum k value
for 1Ec at point A, is obtained by compaction by 4Ec at “w at point C”, which is noticeably lower
than “wopt for 1Ec”. Besides, as the in-situ CEL is usually unknown, it is difficult to predict the value
of k for a given w by referring to such a plot that is different for different CELs as shown in Fig. 23.

21
 2.0
Cut core material B (Sr)opt= 84%
(Dmax= 4.8 mm) C
for Miboro dam
1.9
Sr= 100%
A

Dry density, d (g/cm )

3
a) 4.0Ec
1.8

b) 2.4Ec 90%
d) 1.0Ec
c) 1.6Ec
1.7
80%
e) 0.6Ec 70%
60%
a) Sr= 50%
1.6
4 6 8 10 12 14 16 18
Saturated hydraulic conductivity, k (cm/sec)

1E-3
Water content, w (%)
e) 0.6Ec
1E-4
d) 1.0Ec

a) 4.0Ec
1E-5
b) 2.4Ec

1E-6 c) 1.6Ec
A
B
1E-7 C
Optimum degree of saturation,
(Sr)opt= 84 %
1E-8
4 6 8 10 12 14 16 18
b) Water content, w (%)

Fig. 23. ρd vs. w and log(saturated k) vs. w relations for different CELs of sieved core material
(SCM) for Miboro dam (Mikuni, 1962) analyzed by Tatsuoka (2015) (see Fig. 4b for the grading
curve).

Fig. 24a shows the logk vs. “Sr at the end of compaction” relation obtained by changing the
abscissa from w to Sr of the data plotted in Fig. 23. A much more systematic and simpler trend may
be seen. Benson and Trast (1995) showed a summarized relation for thirteen compacted clays similar
to those presented in Fig. 24a and pointed out that “Sr at the end of compaction” is the major
influencing factor for k. In Fig. 24a, a scatter in the k values for a fixed Sr is due mostly to a variation
in ρd. To evaluate the effects of ρd on k, the k values in different ranges of Sr are plotted against ρd
(Fig. 24b). As the slopes of the fitted linear relations for the different ranges of Sr are similar, the data
was fitted by Eq. 8:

log k  log f k ( Sr )  5.02(1.872  d /  w ) ( 8)

where k is in cm/sec; and fk is “the k value when ρd= [(ρd)max]1Ec= 1.872 g/cm3 ” that is a function of
“Sr at the end of compaction”. In Fig. 24c, the values of fk obtained by substituting the values of k
and ρd of each data point shown in Fig. 24a into Eq. 8 are plotted against Sr. The fk vs. Sr relation is
rather unique for these different CELs. This means that, without knowing CEL, the saturated k value
at a given compacted state can be estimated by substituting the measured values of ρd and Sr into
such an empirical equation as Eq. 8 that has been obtained by relevant laboratory and/or field tests
performed in advance. In Fig. 24c, fk is nearly constant as long as Sr< 60 % and starts decreasing as
22
Sr increases from about 60 %. The decreasing rate becomes very high as Sr becomes larger than about
70 %. This trend can be explained by the effects of “Sr at the end of compaction” on the
micro-structure of compacted soil illustrateted in Fig. 9. As suggested by Alonso et al. (2013), it is
likely that, when the global Sr value increases toward a certain limit (i.e., aroud 70 % in this case),
the clusters of fine particles absorb the increased volume of pore water maintaining the coherent
micro-structure. As the global Sr value exceeds this limit, the clusters of fine particles stops fully
absorbing the increased volume of pore water and, due to a decrease in the matric suction, the
clusters start collapsing dispersing the fine particle into the voids of the fabrics of coarse partiles.
Then, the saturated k value decreases significantly. Recently, Scarcella et al. (2018) showed that the k
values of two Italian soils (silty and sandy soils) measured in the course of oedometer testing can be
represented by equations in the form of Eq. 8.

1E-3 1E-3
k (cm/sec)

Slope= -5.02

(cm/sec)
conductivity, kk (cm/sec)
1E-4 1E-4 (independent
勾配= -5.02の関係of Sr)
k (cm/sec)
Saturated coefficient of

Saturated coefficient of log(k)=3.5

log(k)= 3.5- –5.02(d//ρ
5.02(ρ ):
d w w):
1E-5 1E-5 Sr= (Sr)opt
(average = 84%の時の平均的関係
relation when Sr= 84 %)
conductivity,

飽和化後の透水係数,
御母衣ダムコア材を4.8mm
Compaction energy level Sr (%)
飽和透水係数,

以下に篩った室内試験用試料 (d)max
35 - 40
1E-6 □　　締固め方法
0.6Ec 1E-6
40 - 50
○ 1.0Ec (Standard Proctor)
3層×15回 50 - 60
△ 1.6Ec
hydraulic

3層×25回 (1Ec相当） 60 - 70
hydraulic

1E-7 ▽ 2.4Ec
3層×40回 1E-7 70 - 80
log(k)=2.8 - 5.02(d/w):
3層×60回
◇ 4.0Ec 矢印は最適飽和度(84%) 80 - 86
3層×100回 ↓ (Sr)opt= 84%
付近のデータを示す 86 - 90 Sr= 90 %の時の平均的関係
(average relation when Sr= 90 %)
90 - 91
1E-8 1E-8
20 30 40 50 60 70 80 90 100 1.6 1.7 1.8 1.9 2.0
a) 締固め時飽和度,
Degree of saturation at the end of Scompaction,
r
(%) Sr (%) b) 乾燥密度/水の密度, d3/)w
Dry density, ρd (g/cm
-3
御母衣ダムコア材 k
( Sr )  5.02(1.872 d /  w )
Miboro dam, sieved coref kmaterial
　　4.8mm以下に篩った
(for laboratory tests; Dmax= 4. 8 10 mm)
(log(cm/sec))

　　室内試験用試料 log[f k ( Sr )]  log(k )  5.02(1.872   d /  w )

[log(cm/sec)]

-4

(S r)opt
最適飽和度
-5 log[fk(Sr)}= -4.4 =(84
84%)%
Average
平均的関係
d (g/cm
3
) Relation
(Gauss関数で
)]
(Srr)]

フィッティング)
log[fkk(S

1.67 - 175
-6 1.75 - 1.80
log[f

1.80 - 1.85 Sr > 81%のデータの

Linear fitting for Sr>(Sr)opt
1.85 - 1.90 線形フィッティング:
4.86 –- 0.128x
y=y=4.86 0.128x
1.90 - 1.97
-7
30 40 50 60 70 80 90 100
c) Degree of saturation at the end Sof (%)
締固め時飽和度, compacted, Sr (%)
r

Fig. 24. Saturated coefficient of hydraulic conductivity of sieved core material (SCM) for Miboro
dam (Fig. 23): a) logk vs. “Sr at the end of compaction” relation; b) logk vs. ρd relations for different
ranges of Sr; and c) fk(Sr) (=logk when ρd= [(ρd)max]1Ec=1.872 g/cm3) vs. Sr relations.

23
 100

Percent passing by weight

80
(8)
60
(7)
(6)
(5)
40
(4) (3)
(2)
20 (1)

1E-3 0.01 0.1 1 10 100

a) Particle size, D (mm)

1 k  k0  10 Fk ( Sr ) {[( d )max ]1 Ec  d }/  w Range

大多数の of
most data
データの範囲
k,r)Fk(Sr)

0
Fk(S
function of Sr for

-1
飽和度の影響関数,

Average relation for sieved core

御母衣ダムコア材
material (SCM) for Miboro dam
（SCM材）の平均曲線
-2
Critical
全データ対する Sr
for all the data:
限界飽和度
-3
Influence

(Sr)cr=70%

-4
SCM材のS
Fitted linear >81%のデータ
relation r(slope= -0.128)
for SCM with Sr> 81 %
の線形フィッティング
（勾配: - 0.128)
-5
10 20 30 40 50 60 70 80 90 100
Degree of saturation at飽和度,
the endSrof(%)
compaction, Sr (%)
Symbol Material name (Sr)opt k0 α [(ρd)max]1Ec/ρw
(%) (cm/sec)
● SCM 84.0 10-4.40 5.02 1.872
☆ No.1 84.8 10-3.75 1.94
Miboro dam*

○ No.2 95.4 10-2.86 2.12

× No.3 87.1 10-4.52 5.02+ 1.94
＋ No.4 92.3 10-4.15 1.53
△ No.5 87.7 10-4.41 1.70
□ Type A** 91.0 10-4.85 3.38 1.47
▽ Type B** 85.0 10-5.8 5.675 1.75
+ assumed to be the same as SCM
* Mikuni (1962) & JGS (1979); **Daniel & Benson (1990)
b)
Fig. 25. a) Eight soils availabe at Moboro dam site (Mikuni, 1962; JGS, 1979); and b) effects of Sr
at the end of compaction on the coefficient of saturated hydraulic conductivity k of a wide variety of
soil types (Tatsuoka, 2015).

Tatsuoka (2015) generalized Eq. 8 to Eq. 9 based on analysis of the data of other materials having
a wide range of particle size shown in Fig. 25a in addition to those of SCM (Figs. 23 and 24),
together with soil type A (a mixture of sand, silt and clay; wL= 55%; PI= 27) and type B (a more
sandy soil than type A; wL= 34%; PI = 18) reported by Daniel and Benson (1990):

log k  log k0  Fk ( Sr )    {[( d )max ]1Ec  d }/  w (9)

24
where Fk is the function expressing the effects of Sr on k with Fk(Sr=0)= 0; and k0 is k when Sr= 0 and
ρd= [(ρd)max]1Ec. The k0 value and the parameter α are different among these different soil types
having largely different particle sizes. The analysis of the data reported by Benson and Trast (1995)
indicated that, with clayey soils used for liners, k is also a function of ρd and Sr, whereas the
parameter α tends to increase as Sr becomes smaller than (Sr)opt. Fig. 25b shows the Fk vs. Sr relations
where Fk is computed by Eq. 9. In this plot, the data of soils (6), (7) and (8) shown in Fig. 25a are not
included due to that the k value is available only in a limited range of Sr with these soils. Note again
that Eq. 9 does not include CEL as a variable. It may be seen from Fig. 25b that the Fk vs. Sr relations
of these different soil types all exhibit a similar trend in that, with an increase in Sr from 0 %, Fk is
kept nearly zero until Sr becomes about 70 %, then it starts decreasing at a similar large rate.
In summary, the coefficient of saturated hydraulic conductivity, k, of compacted soil is controlled
by ρd and “Sr at the end of compaction”, similarly as other properties as shown in the preceding
sections. Consequently, in field soil compaction control, it is necessary and efficient to monitor and
control the values of ρd and Sr of compacted soil so that the strength/stiffness and k value required in
design are realized. In so doing, the pre-compaction adjustment of w of the fill material is required
for efficient control of ρd and Sr of compacted soil. Based on the above, the following method can be
proposed as practical soil compaction control of fills impeding the hydraulic flow:
1) For clayey soils to be used for liners where the structural stability is not a design issue, to obtain
the minimum k at a given CEL, soil is compacted in such that Sr becomes higher than (Sr)opt by a
certain value, which is about 5 % with the data presented in Figs. 23b. If a high CEL is
available and the soil is compacted in such that Sr becomes (Sr)opt, the k value can become
smaller than the minimum value at a lower CEL while smaller than the allowable maximum
value.
2) For fills that need not only a low hydraulic conductivity but also a high strength/stiffness, such as
core zones of earth- or rock-fill dams and river dykes, soil is compacted in such that Sr becomes
(Sr)opt. This is because, when Sr= (Sr)opt, soaked CBR is close to the maximun for a given CEL
while k is not significantly higher than the minimun for a given CEL and usually can become
sufficiently low by compaction at a practically available high CEL. The compaction target Sr=
(Sr)opt is applicable also to fills not impeding the hydraulic flow, such as ordinary road/railway
embankments.

When it is necessary to quickly estimate in-situ values of saturated k at many places even though
they may be approximated values, it may be too slow to evaluate them by substituting measured
values of Sr and ρd into such an empirical equation as Eq. 8 or 9. Fig. 26a shows the values of
unsoaked CBR and saturated k obtained by substituting the values of Sr and ρd along respective
compaction curves for 1Ec and 4.5Ec into Eqs. 2 and 8 plotted against Sr – (Sr)opt. These equations,
obtained for similar but different silty sands, are analyzed below, as both of such two empirical
equations as these for the same soil type are not available to the authors. Therefore, the following
discussions assume that the two relations shown in Fig. 26a are equally valid to each soil type. Fig.
26b shows the relationships between the values of unsoaked CBR and log(k) for the same value of Sr
– (Sr)opt along the compaction curves for CEL= 1Ec and 4.5Ec obtained from the relations shown in
Fig. 26a. Then, we can adopt the following strategy to quickly estimate in-situ k values:
1. Obtain such relations as depicted in Fig. 26b by field trial compaction tests and related tests in the
laboratory.
2. Control the CEL value to be the same as the one in the field trial compaction test by specifying the

25
 lift and the type, the running speed and the number of pass of compaction roller and keep the soil
type as much as similar to the one used in the field trial compaction test.
3. Substitute measured unsoaked CBR values into such empirical relations as shown in Fig. 26b.
In Fig. 26b, line c-c denotes the compacted states where Sr = (Sr)opt. The point where line c-c crosses
each log(k) – unsoaked CBR relation at a given CEL should be located between two vertical lines
representing the upper allowable bound for unsoaked CBR specified to ensure sufficiently low k
values and the lower bound allowable bound for unsoaked CBR specified to ensure sufficiently high
strength and stiffness. This requirement is satisfied more easily by compaction at a higher CEL.
Along the relation in a range between these two bounds for a given CEL, k increases with an increase
in unsoaked CBR. This implies that it is not the target of compaction to seek for higher unsoaked
CBR values unconditionally, but the compaction should be controlled so that the unsoaked CBR
becomes the value when Sr = (Sr)opt (if so specified) while confirming the unsoaked CBR to be
between the upper and lower allowable bounds.
y = Intercept + B1*x^1 + B2*x^2 + B3*x
数式 ^3 + B4*x^4
プロット CBR4.5Ec/10
10 -2
log(k) (k in cm/sec) and unsoaked CBR (%)/10

重み 重み付けなし
切片 3.33626 ± 0.07273
Unsoaked CBR for 4.5Ec: (Sr)opt= 81.3 % B1 -0.40037 ± 0.01945
k: Sieved core material for Miboro dam
8 B2 -0.01158 ± 0.00221
B3 -7.92382E-5 ± 7.10515E-5
B4
-3 Unsoaked CBR: Full-scale compaction tests
7.93528E-8 ± 6.91457E-7
残差平方和 0.98255
6 R二乗(COD) 0.9908
補正Ｒ二乗 0.98905
log(k) (k in cm/sec)
b a
4 a -4 b a
2
Unsoaked CBR for 1Ec c 1Ec b
-5
0 d
c Sr=(Sr)opt
-2 Saturated k for 1Ec -6 d
b 4.5Ec
-4 a c
c -7
-6
d d
Saturated k for 4.5Ec: (Sr)opt= 84 %

-8 -8
-60 -40 -20 0 -10 0 10 20 30 40 50 60 70 80
Sr - (Sr)opt (%) Unsoaked CBR (%)
a) b)
Fig. 26. a) Unsoaked CBR by Eq. 2 and saturated k by Eq. 8 estimated for 1Ec and 4.5Ec plotted
against Sr - (Sr)opt; and b) relationships between unsoaked CBR and saturated k for 1Ec and 4.5Ec
obtained by combining the relations shown in Fig. a).

Control of ρd, w and Sr

In this section, a new end product specification method controlling ρd, w and Sr in a unified
manner is proposed modifying the conventional method.

Compaction target
To ensure the performance required to a given soil structure, the soil compaction should be linked
to the soil properties employed in design. In many design codes/manuals, however, this link is not
explicitly described. As discussed earlier, it is proposed to specify the target value of Sr basically to
be equal to (Sr)opt, or a value close to that depending on the design conditions of a given soil
structure, irrespective of CEL and soil type for the following reasons.
1) Unlike the optimum water content, the value of (Sr)opt, at which (ρd)max is obtained, is essentially
constant irrespective of inevitable variations in the soil type and CEL in a given earthwork.
2) For a given CEL, when “Sr at the end of compaction” = (Sr)opt, the soaked CBR is close to its
26
 peak value (Fig. 15b); the cyclic undrained triaxial strength SR20 of saturated silty sand is
essentially equal to the maximum (Fig. 19); and the collapse upon submerging of compacted soil
becomes negligible (Fig. 21).
3) Soil structures impeding the hydraulic flow need sufficiently low k values. For a given CEL, the
k value when “Sr at the end of compaction” = (Sr)opt is somehow higher than its minimum (Fig.
23). However, it becomes quite feasible by better compaction at Sr= (Sr)opt to decrease the k value
to a value lower than the minimum value at a lower CEL while lower than the maximum value
allowed in design.
In so doing, the target value of ρd should be determined in such that the strength/stiffness (and
hydraulic when necessary) properties required in the design of a given soil structure can be ensured.
In step 3) above, conversly, “Sr at the end of compaction” that is slightly higher than (Sr)opt may be
targeted so that the minimum k value is obtained while ensuring the strength/stiffness values required
in the design.

Lower and upper allowable bounds for Sr

As discussed in details in the preceding sections, it is necessary to control the degree of saturation
(Sr) of compacted soil for relevant soil compaction. To this end, it is effective to specify the target of
Sr, which is basically equal to (Sr)opt as discussed above, and the lower and upper allowable bounds
for all Sr values measured in a given earthwork project, as discussed below. In Fig. 27a, compaction
curves for 1Ec and 4Ec shown in Fig. 23a and several contours of k obtained from Eq. 8 for SCM for
Miboro dam are depicted. In a typical conventional soil compaction control, such an acceptable zone
as zone A is specified based on the results of laboratory compaction tests using a certain CEL
(typically 1Ec). In such a method, compaction to states around point c in zone A are allowed,
although these states may be reached due to insufficient CEL at too low values of ρd and Sr where the
k value could be too high, the strenght and stiffness after wetting could be too low and collapse upon
submerging could be too large. On the other hand, when Sr is higher than about 80 %, the contours of
k are rather in parallel to constant Sr curves. So, area B including unfavorable states around point c
can be effectively eliminated by specifying a lower allowable bound for Sr (e.g., 80 % in this case).
Even with soil structures that do not require the control of k, it is effective to specify a relevant lower
bound for Sr to avoid an excessive decrease in the strength/stiffness and large collapse upon wetting
and to ensure sufficient high strength/stiffness after wetting.
Fig. 27b shows compaction curves from laboratory compaction tests (1Ec) and by the number of
roller pass N=16 in full-scale compaction tests together with contours of unsoaked and soaked CBR
of sandy loam shown in Fig. 15a. The typical acceptable zone A illustrated in Fig. 27a is also shown.
In some codes/manuals, the area between the compaction curve and the zero air void curve is
unreasonably eliminated from zone A. It is relevant to eliminate such a zone as D from zone A by
introducing the upper allowable bound for Sr (e.g., 90 % in this case) to avoid too low strength and
stiffness of compacted soil while preventing over-compaction.
The degree of compaction (Dc)1Ec = ρd/[(ρd)max]1Ec where [(ρd)max]1Ec is the (ρd)max value obtained
by laboratory compaction tests using 1Ec is often used as the index for the strength and stiffness of
compacted soil. (Dc)1Ec is obtained as:

(Dc)1Ec= [ρd/(ρd)max]･{(ρd)max/[(ρd)max]1Ec}= (Dc)t･R (10)

27
where (Dc)t= [ρd/(ρd)max] is a function of Sr – (Sr)opt (Fig. 8); and the maximum dry density ratio R=
(ρd)max/[(ρd)max]1Ec is a function of CEL in the field, CELf. Eq. 10 implies that (Dc)1Ec is controlled by
Sr and CELf. A sufficiently high R value can be ensured by keeping CELf to a sufficiently high
constant value by specifying the lift, the compaction roller type, the running speed and the minimum
number of pass for a given specifies allowable soil type. Then, a sufficiently high value of (Dc)1Ec
can be obtained by ensuring a sufficiently high (Dc)t value by keeping the field Sr value close to the
(Sr)opt value by specifying the lower and upper allowable bounds for Sr, as indicated by two thick
solid curves in Fig. 27b (or specifying the lower and upper allowable bounds for air void ratio, va, as
discussed below). In other words, the compaction control without directly controlling ρd is relevant
as a simplified practical method only if Sr is controlled to become (Sr)opt and CELf is always kept
constant and high enough to ensure the soil properties required in design.

4 5 6
2.0 k (cm/sec)
3 (Sr)opt= 84% -4
1 10
Eq. 8 2 10
-4.5

2 -5
1.9 3 10
ρd3)(g/cm3)

Sr= 100% 4 -5.5

10
4.0Ec -6
5 10
-6.5
1 6 10
d (g/cm

1.8
A 90%
Dry Density,

1.0Ec
5
B 4
1.7 c 3
0.9 x [(d)max]1Ec

(wopt)1Ec (w opt))1Ec
(wopt 1Ec 80
80% ％
+3%
+4%
Sr= 50% 60% 70%
1.6
4 6 8 10 12 14 16 18
w (%)
a) Water content, w (%)

2.2
Soaked CBR (%) A: Full-scale compaction test
= 80 (upper half at N= 16)
(Sr)opt=81.3%
B: Laboratory
compaction tests
Dry density,d (g/cm3)

40 E D (1.0Ec)
2.0
Unsoaked CBR (%) =80 Sr= 100 %
A Gs= 2.83
20

40 A
1.8 10
0.9 x [(ρd)max]1Ec 90
20 (wopt)1Ec+3%
5 80
10 Sr (%)= 70
5
1.6
0 5 10 15 20
b) Water cotent, w (%)
Fig. 27. a) Compaction curves for 1Ec and 4Ec and contours of different k values of SCM for
Miboro dam; and b) compaction curves for 1Ec and N=16 and contours of different unsoaked and
soaked CBR of sandy loam used in full-scale compaction tests (modified from Fig. 15a).

With zone A shown in Figs. 27a and 27b, the upper allowable bound for water content is specified
at the optimum water content for 1Ec, (wopt)1Ec. This specification does not encourage the
compaction to such a zone as E, located between the lower and upper bounds for Sr, shown in Fig.
27b. In zone E, w is lower than (wopt)1Ec while ρd is similar to, or higher than, [(ρd)max]1Ec. It is quite
28
feasible to reach zone E by using CEL that is higher than the value used in the laboratory compaction
test (i.e., 1Ec in this case) while available under ordinary modern earthwork conditions. Later in this
paper, a soil compaction control method to encourage to go to zone E is proposed.

Air void ratio control

In the construction of long embankment for roads, railways and river/coastal dykes, the soil type
may change often and largely. In that case, soil compaction control based on laboratory compaction
tests on many encountered different soil types becomes extremely time-consuming. It is particularly
the case in recent fast earthworks using compaction rollers heavier than before (e.g., Mokwa and
Fridleifsson, 2005, 2007). For these reasons, in many construction codes/manuals, an upper
allowable bound (e.g., 15 %) is specified for the air void ratio va (Eq. 1). However, va is not uniquely
related to Sr unless when fully saturated (i.e., when va = 0 % and Sr= 100 %) and not well related to
matric suction as Sr. Besides, as seen from Fig. 7, the va value when Sr= (Sr)opt decreases with an
increase in (ρd)max. Therefore, va is not better than Sr to describe the compaction properties and the
mechanical and hydraulic properties of compacted soil. Yet, for a limited variation in ρd in respective
earthwork projects using nominally the same soil type and CEL, the shape of constant va curves on
the ρd –w plane is similar as the one of constant Sr curves (as seen in Fig. 16a). In that case, the
control of va is practically equivalent to the control of Sr. Even so, due to several confusions and
mixing-ups with respect to the implications and objectives, the va control is often not properly
understood and practiced as discussed below.
1) In many codes/manuals, only an upper allowable bound for va is specified without specifying the
lower allowable bound. Then, the compaction target may be misunderstood to be the state of va=
0 % (i.e., Sr= 100 %), which is obviously not a preferred compacted state. A lower allowable
bound for va (i.e., an upper allowable bound for Sr) should also be specified while indicating the
compaction target, such as the Sr= (Sr)opt state (i.e., the va= corresponding (va)opt state).
2) Related to the above, a decrease in va during compaction at a constant w may result in a decrease
in the strength/stiffness. For example, in Fig. 16a, the va value decreases along a compaction path
at constant w a→b→c (e.g., at w= 26.3 % = “wopt for 1Ec”). Path a→b→c comprises two
sub-components: (I) path a→d, along which the unconfined compression strength qu increases
consistently due to an increase in ρd at a constant Sr (i.e., at a constant va) (Fig. 16c); and (II) path
d→c, along which qu decreases consistently due to an increase in Sr (i.e., due to a decrease in va)
at a constant ρd (Fig. 16b). So, an increase in qu along path a→b→c is not due to a decrease in va,
but it is due to an increase in ρd, whereas the effect of ρd is somehow masked by a decrease due
to an increase in Sr (i.e., due to a decrease in va). Indeed, the original objective of va control is to
control the Sr value of compacted soil to approach the (Sr)opt value irrespective of CEL and soil
type. The same discussion as above based on unsoaked CBR values is made in Tatsuoka (2015).
3) In some codes/manuals, only an upper allowable bound for va, or both upper and lower bounds,
are specified without specifying the lower allowable bound for ρd (or Dc). However, to realize the
soil properties required in design, it is insufficient to control only the value of va (or Sr) unless the
field CEL is always kept to such a sufficiently high value that can ensure the soil properties
requited in design. The va control alone may allow the use of soft soil at too high water content.
In that case, the compacted ρd (or Dc) value may not become high enough even when Sr is
controlled to becomes (Sr)opt.

29
Unified specification of allowable boundaries for w, ρd and Sr
The four major causes for poor compaction are: 1) too low water content and 2) too high water
content of the fill material; 3) too low CEL resulting from the use of a too light compaction machine
with a too small number of pass and a too large lift; and 4) the use of soil that is too difficult to
compact. In the control of soil type, w and CEL to restrain these factors, it is necessary and inevitable
to allow their variations within relevant acceptable limits. Although this key factor is specified in
most of the codes/manuals, its background is rarely explained systematically. In the following, this
issue is discussed in a unified framework. Figs. 28a, b and c schematically show the contours as
control boundaries of three typical soil properties often considered in design: soaked CBR (as a
parameter expressing the saturated strength/stiffness); collapse settlement by wetting; and saturated
hydraulic conductivity, constructed on the ρd – w plane based on these properties depicted in Figs.
15a, 21 and 27, respectively. In Fig. 28, the target compaction state is located on the Sr= (Sr)opt line
while inside all of the acceptable zones for these soil properties. So, to restrain these four actions
while allowing their inevitable but limited variations, it is proposed to introduce the control
boundaries for w, ρd and Sr in relation to the target compaction state. The table inset in Fig. 28 lists
the objectives and effectiveness of the respective control boundaries introduced in soil compaction
control. Here, “WL and WU for w” denote the lower and upper allowable bounds for water content;
“DL for Dc” denotes the lower allowable bound for the degree of compaction, and “SL and SU for Sr”
denote the lower and upper allowable bounds for the degree of saturation. These allowable bounds
are explained in the following.

Sr= (Sr)opt Sr= (Sr)opt Sr= (Sr)opt

ρd
ρd
ρd

Compaction target Compaction target Compaction target

Compaction curve Compaction curve Compaction curve

(e.g., 1Ec) (e.g., 1Ec) (e.g., 1Ec)
③
②
①&④

Soaked CBR w
Settlement by submerging w Saturated k w
a) b) c)
Required properties (1) High (2) Small collapse, (3) Low (4) No
of compacted soil strength & & (2’) small saturated danger of
stiffness after strength reduction hydraulic over-
Control boundaries wetting upon wetting conductivity compaction
For fill material WL for w Important1) Essential Important
to be compacted WU for w Essential2) Essential
DL for Dc Essential Important Important Important
For compacted
fill SL for Sr Essential Essential
SU for Sr Important Essential

Notes 1) and 2): These control boundaries are either essential or important to ensure soil properties 1 - 3 and no
over-compaction.

Fig. 28. a) – c) Allowable boundaries and the corresponding acceptable zones for three typical soil
properties often considered in design and construction.

Proposal of a new soil compaction control

Fig. 29a shows the basic framework proposed for efficient and effective soil compaction control
(Tatsuoka, 2015; Tatsuoka and Correia, 2016). This method is organized to encourage the
30
compaction to a high ρd value by taking advantage of practically achievable high compaction effort
while keeping the Sr= (Sr)opt condition (i.e., at the water content lower than the value usually
recommended). Importantly, this proposal does not contradict the conventional procedures but it
attempts to unify them into a single consistent framework comprising steps 1 – 11 depicted in Fig.
29a and explained below:
Step 1: The basic compaction curve is obtained by compaction tests performed in the laboratory on a
soil sample representative of a given project at a certain CEL (usually 1Ec or 4.5Ec). It is
preferable that this CEL is as much as representative of the field value.
Step 2: Line of Sr= (Sr)opt is depicted based on the compaction curve obtained at step 1.
Step 3: The compaction target T is determined where Sr= (Sr)opt in the ordinary practice; and ρd=
(ρd)target that can ensure the soil properties required in design (as described in Fig. 28). With soil
structures impeding the water flow, the target value slightly larger (by 5 %, for example) than
(Sr)opt may be specified.
Step 4: The field target compaction curve, which passes target point T, is obtained by assuming that
the unified compaction curve, i.e., the ρd/(ρd)max vs. Sr - (Sr)opt relation (Fig. 8), and the value of
(Sr)opt do not change by possible variations of CEL and soil type in the field.
Step 5: The lower allowable bound for ρd, denoted as DL, is determined, where Dc= ρd/(ρd)target x
100 % is equal to a specified value, typically 95 %. Several data sets supporting this number are
reported in Tatsuoka (2015).
Step 6: Point B where w= “the value at target T, wtarget” is obtained along DL.
Step 7: The constant Sr curve that passes point B is defined as the lower allowable bound for Sr, SL.
Curve SL crosses the field target compaction curve at point C. If SL is otherwise specified, for
example, to satisfy requirements for a sufficiently low hydraulic conductivity as depicted in Fig.
27a and if this Sr value is higher than the value at point B, the Sr value for SL (i.e., the Sr value at
point C) is reset to be equal to that new Sr value.
Step 8: The lower allowable bound for w, WL, is specified to pass point C: i.e., wWL= “w at point C”.
Step 9: The upper allowable bound for w, WU, where w= wWU= wtarget + x, is specified to avoid too
low strength/stiffness and to prevent over-compaction. Trial field compaction tests may be
necessary to obtain the reliable value of x in a given project. Only the fill material having w
between wWL and wWU is allowed to be compacted.
Step 10: The upper allowable bound for Sr, SU, is specified to pass point D, where WU crosses the
target compaction curve. Alternatively, the upper allowable bound for Sr, SU, could be firstly
specified as (Sr)opt + 5 %, for example, then point D is obtained where SU crosses the target
compaction curve. Then, WU is obtained to pass point D.
Step 11: The acceptable zone for compacted soil comprising boundaries WL, SL, DL, WU and SU,
denoted by A, is specified. Basically, WL is not necessary to define the acceptable zone, because
compacted soils in zone E (where w < wWL) may satisfy the required soil properties. In actuality,
however, it is very difficult to reach zone E even by compaction at a very high CEL. Besides,
compacted fill may be required to be homogeneous to prevent differential deformation. In that
case, the soil compacted to zone E may be too stiff. So, it is wise to specify an acceptable zone A
that is truncated by WL.

Referring to Fig. 29b, suppose that, in a given earthwork project, the value of CEL is always
kept to a rather constant value, CELT, and the target compaction curve for CELT, which passes
compaction target T, has been obtained. Compaction at constant CEL is becoming more feasible in
recent well-controlled earthworks. Even in that case, however, the actual compacted state may not be
located on the target compaction curve due to inevitable variations in the soil type. In Fig. 29b, the
31
actual compacted state is supposed to be located at point X that is below the field target compaction
curve due to that the actual soil type is less compactable than the one for which the field target
compaction curve is determined. The true degree of compaction, (Dc)t, defined for the actual soil
type and CEL at state X is equal to ”ρd at point X”/”ρd at point Y (= (ρd)max for the actual soil type
and CELT)”. Then, even if point X is located below Point C, as far as “Sr at point X”> (Sr)SL, the (Dc)t
value at state X is higher than the (Dc)t value at state C (= “ρd at point C”/(ρd)target), so could be much
higher than the lower allowable bound for apparent Dc (typically 95 %). If the Sr value at state X is
equal to (Sr)opt, the (Dc)t value at state X is 100 % under the condition of CEL= CELT (i.e., ρd at state
X is equal to (ρd)max for the actual soil type and CELT). The compaction state when point X is located
above the field target compaction curve can also be evaluated by the value of Sr relative to (Sr)opt. It
is one of the advantages of the proposed method that a high compaction state can be ensured by
keeping Sr close to, or equal to, (Sr)opt to achieve a high (Dc)t value while keeping CEL to a high
constant value.
ρd

2) Sr= (Sr)opt

E 3) T: Compaction target
11) Acceptable zone
for compacted fill 10) SU
D
4) Field target compaction C
curve, passing target T 5) DL
A
(CELT > 1Ec in this case)

6) B
1) Compaction curve by
laboratory test (e.g., 1Ec) 7) SL
wtarget w
8) WL passing point C 9) WU
a)
Sr= (Sr)opt
ρd

1) T: Compaction target
2) Field target compaction
curve passing target T
(ρd) target
(CELT > 1Ec in this case)
State Y, where ρd = (ρd) max
C by CELT of the actual soil type
Compacted state X,
where Sr > (Sr)SL

3) DL
Compaction curve by
5) SL
laboratory test of the actual
soil type and CEL (= CELT)
target w w
b)
Fig. 29. a) Soil compaction control to ensure the soil properties required in design; and b) the true
degree of compaction (Dc)t for actual soil type and compaction energy level CELT.

Discussion on European practice

According to the European Standard EN 16907-1, the earthwork constructions are controlled by
earthworks specifications. These specifications describe the design requirements that are to be
satisfied and that should be monitored by an effective form of quality assurance procedure (EN
16907-5). There are three main types of specifications for earthworks: end product, method and
performance. The topic of this paper is the end product specification as referred before. In most of
32
the European countries, this type of specification has been developed defining the degree of
compaction necessary for a given material by reference to criteria linked with serviceability and
ultimate limit states. In general, the criteria normally define overall targets to be achieved, which are
commonly defined relative to minimum acceptable final dry density (DL), often combined with a
maximum acceptable final va (or a lower limit of Sr (SL)), plus lower and upper acceptable water
content values (WL, WU) in order to achieve acceptable mechanical and hydraulic properties and to
minimize post compaction changes (EN 16907-5). So, usually a minimum acceptable final va (or an
upper limit of final Sr (SU) is not specified. The followings are comments on these issues:
1) Tatsuoka (2015) and Tatsuoka and Correia (2016) argue that the degree of saturation (Sr) is
better than the air void ratio (va). That is, as shown in Fig. 7, the value of (Sr)opt is similar for a wide
range of soil type, whereas the value of (va)opt consistently decreases with an increase in (ρd)max. In
addition, the strength and deformation characteristics and the hydraulic conductivity of saturated
compacted soil is controlled by ”the Sr value at the end of compaction” in addition to the Dc value
and soil type.
2) Yet, this European method specifying an upper allowable limit of va is consistent with the
procedure specifying a lower allowable limit of Sr proposed by Tatsuoka (2015) and Tatsuoka and
Correia (2016) as depicted in Fig. 29.
3) In addition to the upper allowable limit of va, which is practically equivalent to the lower
allowable bound for Sr (SL), it is necessary to specify a lower allowable bound for va, which is
equivalent to an upper allowable limit of Sr (SU), as suggested by Tatsuoka (2015) and Tatsuoka and
Correia (2016) and depicted in Fig, 29.

Conclusions
The following conclusions can be derived from analysis of a large amount of laboratory and field
test data presented in this paper:
1) The optimum degree of saturation (Sr)opt is defined as the degree of saturation Sr where (ρd)max is
obtained for a given compaction energy level (CEL) and a given soil type. The value of (Sr)opt
and the unified compression curve in terms of the relationship between the true degree of
compaction (Dc)t= ρd/(ρd)max and Sr - (Sr)opt are rather insensitive to limited variations in CEL and
soil type.
2) The strength and stiffness before and after soaking, collapse upon wetting and cyclic undrained
strength and hydraulic conductivity after saturation can be expressed by functions having
variables of ρd and “Sr at the end of compaction”. These results suggest a strong correlation
between Sr and suction showing very important implications in future developments in the
non-saturated soil mechanics approach of related mechanical and hydraulic properties.
3) Even if inevitable variations in CEL and soil type are unknown, the value of (Dc)t under given
compaction conditions can be estimated only from the value of Sr when the value of (Sr)opt and
the unified compression curve are known.
4) To effectively control the mechanical and hydraulic properties of compacted soil, it is necessary
to control not only compacted ρd but also “Sr at the end of compaction”. It is proposed to
control “Sr at the end of compaction” to become (Sr)opt and ρd to become large enough to ensure
soil properties required in design, together with pre-compaction control of water content of fill
material. This proposed method encourages the use of higher compaction efforts with keeping the
Sr= (Sr)opt condition to increase ρd at a water content lower than a value usually recommended by
conventional methods.

33
5) The compaction control specifying the upper allowable bound for the air void ratio va, or the
lower allowable bound for Sr, without controlling ρd can be relevant only when Sr is controlled to
become (Sr)opt while the field CEL is always kept constant while high enough to ensure the soil
properties required in design.
6) The compacted state in terms of ρd and Sr and mechanical and hydraulic properties at that state
for a given soil type cannot be estimated only by the strength or stiffness at that compacted state.
The continuous stiffness measurement technology, which is available and of very high
engineering value, is needed to be coupled with other relevant information, such as Sr or CEL in
order that the soil compaction control becomes quite efficient and effective.

Acknowledgements

The information and suggestions provided by Prof. Tateyama, K. (Ritsumeikan University, Japan),
Dr Matsumoto, N. (Japan Dam Engineering Center), Prof. Shibuya, S. (Kobe University, Japan),
Prof. Kikuchi, Y. (Tokyo University of Science), Prof. Koseki, J. (University of Tokyo, Japan), Prof.
Ling, H.-I. (Colombia University, USA), Dr Yoshida, T. (Kajima Technical Research Institute,
Japan) and Dr. Hirakawa, D. (Chuo University) are highly appreciated.

Reference
Abe, T., Fujioka, K., Arai, T., Kobayashi, H. and Mishima, N. Changes in the strength characteristics
due to changes in the water content of compacted embankment fill, Proc. 49th Japan National
Conference on Geotechnical Engineering, Kita-Kyushu, July, 2014, pp.1013-1014.
Alonso, E. E., Pinyol, N, M, and Gens, A. Compacted soil behaviour: initial state, structure and
constitutive modeling, Géotechnique 63, No.6, 2013, 463-478.
Asao, I. About Miboro dam (paper related to No. 31 issue in 8 th ICOLD), Large Dam, Journal of
Japanese Commission on Large Dams, No.27, 1964, 33-44 (in Japanese).
Benson, C. H. and Trast, J. M. Hydraulic conductivity of thirteen compacted clays, Clays and Clay
Minerals, Vo. 43, No. 6, 1995, pp.669–681.
Biarez, J., Liu, H., Gomes Correia, A., Taibi, S., Stress-strain characteristics of soils interesting the
serviceability of geomaterials structures. Proc. II Int. Conf. on Pre-failure Deformation
Characteristics of Geomaterials, IS Torino 99, Jamiolkowski, et al. (eds). Balkema, Rotterdam,
1999, (1) 617-624.
Daniel. D.E. and Benson, C.H. Water content-density criteria for compacted soil liners, Journal of
Geotechnical Engineering, Vol. 116, No. 12, 1990, pp.1811-1830.
EN 16907-1: 2016. Earthworks. Part 1. Principles and general rules, European Standard (Norme
Europenne), European Committee for Standardization (CEN), 2016.
EN 16907-5: 2015. Earthworks. Part 5. Quality control, European Standard (Norme Europenne),
European Committee for Standardization (CEN), 2015.
Gomes Correia, A. (2008). Innovations in design and construction of granular pavements and
railways. Advances in Transportation Geotechnics: Proc. of the 1st International Conference on
Transportation Geotechnics (eds: Ellis, E., Yu, Hai-Sui, McDowell, G., Dawson, A.R. and
Thom, N.), CRC Press, London, UK, 2008, pp.1-13.
Gomes Correia, A. and Magnan, J-P. Trends and challenges in earthworks for transportation
infrastructures. Advances in Transportation Geotechnics 2 (eds: Miura, S., Ishikawa, T.,
Yoshida, N., Hisari, Y. and Abe, N.), CRC Press, London, UK, 2012, pp.1-12.
Heitor, A., Indraratna, B., Rujikiatkamjorn, C. Laboratory study of small-strain behavior of a
compacted silty sand. Can. Geotech. J. 50 (2), 2013, pp.179–188.
34
Heitor, A., Indraratna, B., Rujikiatkamjorn, C. and Golaszewski, R. Assessment of the
postcompaction fill characteristics at the Penrith Lakes development site, Chapter 13, Ground
Improvement Case Histories- Compaction, Grouting and Geosynthetics (Buddhima Indraratna et
al., ed.), Elsevier, 2015, pp.399-4280.
Japanese Geotechnical Society (previously Japanese Society for SMFE). Interpretation of the results
of ground investigation and soil testing and their application cases, Chapter 8 Permeability tests,
1979, pp.336-337 (in Japanese).
Jenkins, P. The strength of well graded cohesive fills, Ground Engineering, March, 1996, pp.38-41.
JIS A 1218. Test methods for permeability of saturated soils, Laboratory Testing of Geomaterials,
Japanese Geotechnical Society, Vol.1, 2009, pp.449-461 (in Japanese).
Joslin, J. G. Ohio’s typical moisture-density curves, ASTM STP239, Proc. of Symposium on
Application of Soil Testing in Highway Design and Construction, 1959, pp.111-118.
Mikuni, E. Study on the properties of fill dam core material and its compaction (No. 1), Monthly
Journal of JGS, Tsuchi-to-Kiso, 10(1), 1962, pp.4-12 (in Japanese).
Mitchell, J., Hooper, G. and Campanella, R. Permeability of compacted clay, Journal of Soil
Mechanics and Foundation Div., ASCE 91, 1965, pp.41-65.
Mokwa, R. L. and Fridleifsson, S. Soil air voids method for compaction control, Research report
(FHWA/MT-05-010/8117-23) prepared for The State of Montana, Department of
Transportation, 2005.
Mokwa, R. L. and Fridleifsson, S. Earthwork compaction evaluation using soil air voids, Canadian
Geotechnical Journal, 44: 2007, pp.151-159.
Matsumura, S. and Tatsuoka, F. Effect of compaction conditions and fines content on cyclic
undrained strength of saturated soils, Journal of Soil Dynamics and Earthquake Engineering,
112, 2018, pp.152-161.
Murata, T., Tateyama, K., Yamamoto, T., Yoshida, T. and Kawano, K. Experimental method of
predicting the compaction effect in fill work practice, Geotechnical Engineering Journal, JGS,
5(4): 2011, pp.589-601 (in Japanese).
Nemoto, T. and Sasaki, T. Soil compaction characteristics, Celebrating 30th Anniversary Volume,
Construction Method and Machinery Research Institute, 1994, pp.45-58 (in Japanese).
Nowak, P. and Gilbert, P. Earthworks: a guide, 2nd edition, ICE Publishing, 2015.
Okamoto, K., Lee, J.-H. and Shibuya, S. Influence of surcharge load and initial moisture condition
on settlement of fill materials at submergence, Proc. 49th Japan National Conference on
Geotechnical Engineering, Kita-Kyushu, July, 2015, pp.399-400.
Proctor, R. R. Fundamental principles of soil compaction, Engineering News-Record, New York,
Vol.111, No.9, August 31, 1933, pp.55-58.
Scarcella, E. G., Giust, I., Giusti, S., Lo Presti, D. C. and Squeglia, N. Strength and permeability
parameters of compacted, partially saturated silty and sandy soils, Rivista Italiana di Geotecnica,
Gennaio – Marzo, 2018, pp.54-65.
Shibuya, S. and Horita, T. Internal technical report on soil compaction, Geotechnical Engineering
Laboratory, Kobe University, 2013.
Tatsuoka, F. Laboratory stress-strain tests for the development of geotechnical theories and practice,
Bishop Lecture, Proc. 5th International Conference on Deformation Characteristics of
Geomaterials, Seoul, Korea, Sept., 2011, pp.3-50.
Tatsuoka, F., Fujishiro, K., Sudo, M., Kawabe, S., Kikuchi, Y., Shibuya, S., Lee, J.-H. and Horita, T.
Soil compaction No. 6, Geotechnical Engineering Note No. 6, Kisoko, December, 2013, pp.92 –
97 (in Japanese)

35
Tatsuoka, F. Compaction characteristics and physical properties of compacted soil controlled by the
degree of saturation, Keynote Lecture, Deformation characteristics of geomaterials, Proc. of 6th
International Conference on Deformation Characteristics of Geomaterials, Buenos Aires, 2015,
pp. 40-78.
Tatsuoka, F., Correia, A. G. Importance of Controlling the Degree of Saturation in Soil Compaction,
Advances in Transportation Geotechnics III (eds. Correia, A. G.), Procedia Engineering, 143,
Elsevier, 2016, pp.556-565.

36