EC7 Fundamental Issues and Its Implications On Users PDF

Keynote lecture Eurocode 7 fundamental issues and some implications for users
Eurocode 7 fundamental issues and some implications for users

B. Simpson
Arup, UK, [email protected]
Simpson, B (2012) Eurocode 7 fundamental issues

and some implications for users. Keynote Lecture, Proc
Nordic Geotechnical Meeting 2012. DGF Bulletin 27.
ABSTRACT
As Eurocode 7 becomes more widely used, questions raised by designers have highlighted issues
that require further debate and clarification. Most of these have existed, in one form or another,
for many years, but the advent of a new code, providing a common language, has brought them
into sharper focus. Some of these issues will be considered in this paper: the selection of
characteristic and design values of soil parameters, design in situations dominated by water
pressures, the relevance of the EQU limit state and the use of numerical analysis for ULS design.
The selection of parameter values for calculations frequently leads to debates among
geotechnical designers. Eurocode 7 attempts, in an rather qualitative way, to point towards a
target reliability for characteristic values, while providing a framework in which the precious
expertise of individual engineers can be fully exploited. Problems of water pressures and the EQU
limit state have a lot in common: how to make provisions for safety in situations where forces
largely balance one another and material strength plays a small, but often vital, part. Numerical
models are now widely used to study serviceability, but their use in checking ultimate limit states
has been questioned; how are partial safety factors to be applied, at what point in staged
calculations, and can they be used with advanced non,linear models of soil behaviour?
Each of these issues is discussed and some practical solutions suggested.
Keywords: Codes of practice & standards; Design; Strength and testing of materials;
Groundwater; Numerical modelling.
The paper refers to previous publications

in which more detail, and in some cases a
more rigorous account may be found:
Schuppener et al (2009), Simpson and
Hocombe (2010) and Simpson et al (2011).
References to specific paragraphs in EC7
will be shown thus: {...}.
1 INTRODUCTION
This paper considers topics that are
currently under debate in relation to the
application of Eurocode 7 Part 1 (EN1997 1
2004, referred to here as EC7). The issues
raised are fundamental to geotechnical
engineering, not artefacts of the new code,
though they may have been brought to a head
by attempts to systematise geotechnical
procedures. The paper will provide a review
of opinions on the issues discussed, and,
where possible the author will give his own
opinion with justification.
2 SAFETY FORMAT OF EC7

The safety format in EC7 uses a limit
state approach. Limit states are states of a
construction beyond which the behaviour is
considered unacceptable.
The aim of
analyses is therefore to show that these
states will not be exceeded.
Table 1. Factors proposed by CEN for the three Design Approaches.
DA1
Actions
Soil
Spread
footings
Driven
piles
Permanent
Variable
tan '
Effective cohesion
Undrained
strength
Unconfined
strength
Weight density
unfav
fav
unfav
Bearing
Sliding
Base
Shaft
(compression)
Total/combined (compression)
Shaft in tension
Comb 1
1,35
Comb 2
1,5
1,3
1,25
1,25
1,4
DA2
DA3
1,35
1,35
1,5
1,5/1,3*
1,25
1,25
1,4
Piles
1,3
1,4
1,25
1,4
1,3
1,3
1,4
1,1
1,1
1,1
1,3
1,6
1,1
1,15
1,1
Note: Values of all other factors are 1.0. Further resistance factors are provided for other types of piles, anchors etc.
* 1.5 for structural loads; 1.3 for loads derived from the ground.
applied to resistances rather than to material

strengths. There are some situations in
which factoring loads at source leads to
unreasonable situations, especially in the
design of retaining structures. For these,
EC7 allows the factors to be applied to the
effects of the loads, and this is used where
appropriate in Combination 1 of DA1.
In DA2, partial factors are applied to
loads and to ground resistances. In a variant
of DA2, DA2*, the equilibrium calculation
is
carried
out
using
unfactored
(representative) loads, and the factors are
applied to derived load effects. It has been
found that DA2 and DA2* are unsuitable for
slope stability problems and for use of
numerical methods, so most countries which
have adopted DA2 use DA3 for slope
stability and for numerical methods.
In DA3, factors are applied to material
strength and to loads simultaneously, in
contrast to the two combination approach of
DA1 in which they are applied to the two
separately and the results compared. A few
countries propose to use DA3 for all types of
designs, with factors quite different from
those shown in Table 1 in most cases.
EC7 requires that both ultimate limit states

(ULS) and serviceability limit states (SLS) be
considered. Most of its text refers to ULS,
for which the main approach is based on use
of partial factors. Opinions in Europe differ
about where and how these should be
applied, and this is left to national choice; the
values to be adopted for partial factors may
also be varied nationally. Three alternative
Design Approaches have been developed,
combining partial factors in different ways;
the factor values proposed in the European
document are shown in Table 1, modified as
noted below for DA1 Combination 1.
In Design Approach 1 (DA1), two
combinations of partial factors are
specified, and the design must be shown to
accommodate
both
combinations.
Essentially, they are used in the same way as
load combinations, but the concept is
extended to include material strengths and
resistances. Partial factors are generally
applied to either loads (before combination)
or ground strengths (before calculation of
resistances), though with some exceptions. In
countries that use DA1, the factors on ground
materials and strengths are generally set to
1.0 in Combination 1, as shown in Table 1.
For design of piles and anchors, factors are
2
3 CHARACTERISTIC
VALUES
AND
to determining characteristic properties of

materials in structural and geotechnical
design.
DESIGN
3.1 Background in design practice

Many codes and text books tell the
engineer how to analyse results of specific,
individual tests and derive parameters that
define the ground, particularly its strength
and deformation characteristics, so that these
can be used in calculations; some of this
information is repeated in EC7 Part2
(EN1997 2). In practice, the geotechnical
process is more complicated than this,
however. Available information is often
sparse in quantity, variable in quality and
reliability, some of it from precise
measurement at the relevant location in the
ground on the construction site, some
inferred from a general understanding of the
geology, some taken from text books, papers
or lecture notes, where it may have been
derived by back analysis of another event in a
similar, but not identical situation some
distance away, and so on. These sources may
complement one another, but they may also
be found to be inconsistent and contradictory.
Furthermore, even when relevant parameters
can be measured directly, perhaps by an in
situ test, their values may be changed by the
construction process itself, or by some future
event such as loading or excavation.
In structural design, it is commonly the
case that drafters of codes of practice have
more knowledge about the parameters of
strength and loads relevant to a particular
design, and their variability, than does the
designer. For example, code drafters may be
more knowledgeable about wind loading,
floor loading, variations in dimension of cast
in situ concrete, or seismic loading than is the
designer, and the same applies to the
variability of steel and concrete. However, in
geotechnical design, the designer knows the
location of the site, something of its geology
and ground water conditions and the results,
or paucity of results, of the ground
investigation, together with their likely
reliability.
This
information
varies
considerably from one design to another and
could not possibly be known by the code
drafter.
Because of this, there are
considerable differences between approaches
3.2 Definitions in EN 1990

EN 1990 contains the concept that
material properties, or resistances, are first
entered into calculations as characteristic
values Xk, to which prescribed partial factors
are applied to obtain design values Xd. This
relationship is provided in Equation 6.3 of
EN 1990:
Xd = Xk/m
(1)
where is a conversion factor relating

values measured in tests to actual values in
the real construction, and m is a partial
factor for the material. EN1990 notes that
may be incorporated into m (giving M) or
into the characteristic value.
Thus the characteristic values are used in
the derivation of design values, which
incorporate all the safety elements required
by the Eurocodes. To make this process
with prescribed partial factors useful, it is
necessary that characteristic values are
defined as clearly as possible.
EN 1990 {4.2} says that the characteristic
value of a material parameter will generally
be a 5% fractile value (ie of test results),
unless otherwise stated in the other
Eurocodes relevant to particular materials.
3.3 Definition in EC7
The prime definition of characteristic
value in EC7 is: The characteristic value of
a geotechnical parameter shall be selected as
a cautious estimate of the value affecting the
occurrence of the limit state {2.4.5.2(2)}.
EC7 requires that data from laboratory and
field tests should be complemented by
well established experience {2.4.5.2(1)}.
These paragraphs make it clear that the
characteristic values required by EC7 are to
be estimated, requiring a degree of human
judgement, and they are to be cautious, not
simply best estimates, most probable or
statistically mean values. They are to be
cautious estimates of the value affecting
the occurrence of the limit state, that is, the
3
value actually operative in the ground, not
simply the values measured in tests, and they
are to take into account well established
experience as well as test results made for the
particular project.
Referring to Equation (1) above, taken
from EN1990, the requirement in EC7 for
the value affecting the occurrence of the
limit state is equivalent to incorporating the
conversion factor , in this case relating soil
test results to real ground behaviour, into the
characteristic value, as allowed by EN 1990.
Paragraph {2.4.3(4)} notes that assessment of
ground properties should take account of
the effect of construction activities on the
properties of the ground.
The experience to be considered in
estimating the characteristic value is noted in
{2.4.5.2(4)}:
geological
and
other
background information, such as data from
previous projects. This paragraph also lists
the following items as relevant to the
required estimate:
the variability of the measured property
values and other relevant information,
e.g. from existing knowledge;
the extent of the field and laboratory
investigation;
the type and number of samples;
the extent of the zone of ground
governing the behaviour of the
geotechnical structure at the limit state
being considered;
the ability of the geotechnical structure
to transfer loads from weak to strong
zones in the ground.
variation seen in tests on small specimens.

In these cases, the characteristic value
should be a cautious estimate of the mean
value for the zone of ground governing the
behaviour of a geotechnical structure at a
limit state {2.4.5.2(7)}. This is illustrated in
Figure 1, which shows a building to be
designed at the top of a slope formed in
estuarine beds, consisting largely of sands
but with some weaker clay inclusions.
Considering the overall slope stability, any
failure would have to pass through a
majority of the sands, and could average out
the effects of the weaker clay zones; so in
this case the relevant characteristic
parameter, possibly , would be a cautious
estimate of the mean for the slip surface.
However, in considering the individual pad
foundations for the building, it could be
possible that a pad would be located almost
exclusively on clay, so the characteristic
value for the foundation design would be
based on the strength of the weaker clay.
Another possibility is that the designer
chooses to have ground beneath each pad
probed to check for clay, and this is to be
dug out if it is found. In that case, the
characteristic value could be a cautious
estimate of the strength of parameters of the
stronger sand. It can be seen, therefore, that
the characteristic value depends on the
failure mode, the extent of the zone of
ground affected, and the way it has been
investigated.
3.4 Use of statistics
Although the definition of characteristic
value in EC7 is not basically statistical,
statistical methods could be useful in its
assessment, and help to define the term
cautious. Paragraph 2.4.5.2(11) says If
statistical
methods
are
used,
the
characteristic value should be derived such
that the calculated probability of a worse
value governing the occurrence of the limit
state under consideration is not greater than
5%.. Again, attention is drawn to the real,
overall behaviour in the ground
governing the occurrence of the limit state
so the preceding paragraphs about mean
values in a zone of influence still apply.
Figure 1. Building on estuarine beds.
Although the characteristic value is

defined to be cautious, not a statistical
mean value, it is noted that the ground has
the ability to average out some of the
4
combined by statistics. The designer must
be convinced that a cautious estimate of
the value affecting the occurrence of the
limit state is being adopted.
The requirement to consider information

from all available sources also still applies.
Paragraph 2.4.5.2(10) says that if statistical
methods are employed they should allow the
use of a priori knowledge of comparable
ground properties, and differentiate between
local and regional sampling.
This requires quite advanced use of
statistics. EC7 certainly does not encourage
the replacement of well established
experience and well researched information
by simple statistical analysis of the
immediately available test results.
Nevertheless, statistical analysis of test
results may provide one source of useful
information, to be considered alongside other
available sources. Various authors have
considered statistical approaches to a
cautious estimate of the mean value for the
zone of ground governing the behaviour of a
geotechnical structure at a limit state.
Schneider (1997) suggested that a value
taken to be 0.5 standard deviations from the
mean of directly relevant test results could be
used. It can be seen in Figure 2 that this is
quite different from a 5% fractile of the test
results, being much closer to the mean of the
results. In the context of North American
practice, Dahlberg and Ronold (1993) and
Becker (1996) for more general use proposed
the use of a conservatively assessed mean
(CAM) as the characteristic value, such that
for a normal distribution 75% of the
measured values would be expected to
exceed this value. This requires an offset of
0.69 standard deviations from the mean, for a
normal distribution, as shown in Figure 2.
Foye et al (2006) take up the same idea
proposing to use a CAM with 80%
exceedance, equivalent to 0.84 standard
deviations below the mean of the test results
for a normal distribution. More recently,
Tietje et al (2011) have discussed how
characteristic values can be derived for slope
stability problems, taking account of the
coefficient of variation of test results and
their spatial correlation.
In the authors view, these statistical
approaches are useful aids, but they must
never be allowed to replace or overrule the
use all information from all relevant sources,
even when the sources are not easily
Figure 2. Derivations of characteristic values
Bored
pile
Figure 3. Undrained shear strengths from

borehole samples on site.
3.5 A London example

Figure 3 shows the results of a series of
undrained shear strength measurements in
London Clay. The measurements were
made using unconsolidated undrained
triaxial tests. A statistical mean line has
been drawn through the data and it is clear
that undrained strength increases with depth.
A characteristic line is required, and this
should depend on how the characteristic
values will be used what is the limit mode
being considered? For example, if the
undrained strength is needed for calculation
5
of the shaft resistance of a pile, a value such
as the cautious (average) value shown on
the figure could be used. However, for a
mechanism that might take place in a small
zone of soil, such as at the base of a pile, a
more cautious value the cautious (local)
value should be adopted.
Figure 5. Comparison of results.
On the basis of these inconsistent data

sets, what value should be used as the
characteristic undrained strength?
The
values measured in the triaxial tests should
not be ignored, but the SPT results and the
data from adjacent sites should also affect
the decision.
The characteristic value
proposed for these data is shown on Figure
6. This is less than the initial assessments in
Figure 3, which were based on the triaxial
results only, and is closer to a lower bound
of this particular set of triaxial results.
Figure 4. SPT results from boreholes on site.

From these boreholes, results from
standard penetration tests were also available,
as shown in Figure 4. In London Clay, there
is usually a constant factor between standard
penetration and undrained shear strength
results; the factor is about 4.5 to 5.
However, if the mean line from the SPT
results is transferred onto the undrained
strength plot, as in Figure 5, it appears that
the normal correlation does not work. In
fact, the measured undrained strengths are
remarkably high: they are consistent with
very low water contents, which were
measured, but this might simply mean that
the samples had dried out on the way to the
laboratory, though there was no reason to
suspect this. Figure 5 also shows lines
representing mean values through data from
other nearby sites, both for undrained shear
strength and SPT results. The usual close
correlation applies to these, and it is clear that
the undrained strengths for the new site are
remarkably high.
Figure 6. Chosen characteristic values.
Engineers often need to follow this sort of
process when trying to interpret real data. It
may be that statistical methods could trace a
similar logical sequence. However, this
would require quite advanced methods and
any statistical approach which failed to take
account of the diverse array of data, typically
available, would be harmful to the design
process.
Generally the strength to be used in

Eurocode 7 is the maximum available to
prevent collapse, not a value mobilised in a
working state.
3.7 Design values of
Whereas the selection of characteristic
values is common to all the Design
Approaches of EC7, only approaches DA1
and DA3 require derivation of design values
of material properties. This section is
relevant to them.
Eurocode 7 allows two alternative means
of deriving design values of material
properties:
a) by application of a partial factor M, as
discussed above.
b) by direct assessment, in which case
the values of the partial factors
recommended ... should be used as a
guide to the required level of safety.
It is often asked where the use of a
critical state angle of friction fits into this
scheme. The suggestion is made by some
that as the angle of friction cannot fall any
lower this value could be used as a directly
assessed design value, requiring no further
M. This would mean that no margin of
safety would be applied to the strength of
soil believed to be in an initially loose state.
It is true that part of the uncertainty
about is the state of the soil, which
elevates its value above the critical state
value. However, in the authors experience,
ground investigation is often inadequate to
give the designers complete confidence in
the actual nature of the soil, such as its
grading, apart from its state of compaction.
On this basis, the design value of for
loose soil should be less than its anticipated
critical state value.
The critical state value is directly relevant
to the strengths of interfaces between the
ground and concrete cast against the ground
{6.5.3(10), 9.5.1(6)}. In the authors view it
is arguable that the reduction factor
applied to this could be less than the value
used generally in the body of the soil,
perhaps 1.15 instead of 1.25, provided its
design value is still less than the design
value for the body of the soil.
3.6 Which value of as characteristic?

The question has been asked: Which
value of is the characteristic value? It is
sometimes necessary to chose from one of
the following, depending on circumstances:
peak, critical state or residual shear
strength
ultimate strength or a mobilised value
strength of intact material or strength on
joints
strength at first loading or after repeated
loading
stiffness of intact rock or of the jointed
material
stiffness on first loading, or on unload
reload
In all cases, the answer of Eurocode 7 is:
The one that is relevant to the prevention of
the limit state under consideration. EC7
does not differ in this respect from normal
practice. For some particular situations, the
code is able to specify which of these values
is relevant. For example, where concrete is
to be cast against ground, which might
therefore be disturbed, the critical state value
for the angle of shearing resistance is
required {6.5.3(10), 9.5.1(6)} .
This answer to the question is not the
same as: The one which would become
relevant if the limit state was not prevented.
For example, in most plastic clays, if a slip
occurred, the angle of shearing resistance
would eventually fall to the residual value.
Nevertheless, it is not necessary to design for
residual strength in clays which have not
previously slipped. Similarly, it may be
unnecessary to design for critical state values,
though brittleness and ductility must be
considered, as noted in {2.4.1(13)} and
{2.4.3(4)}.
For soil in a denser state, the design value
in the body of the soil will normally be
obtained from the characteristic value by
application of a partial factor . Often, this
will mean that the design will be less than
a cautious estimate of the critical state value,
ie d = k/ < crit,k. In the authors
opinion, this is a very useful additional safety
check (ie d < crit,k), and it would be good
if it were added, at least as an application
rule, to EC7.
overview will be provided here in an attempt

to help the reader understand the
conclusions and remaining debate. Simpson
et al discussed five simple problems
intended to highlight particular issues,
together with three more practical designs.
This work is limited to considering
conditions of hydrostatic water pressures or
steady state seepage, in which water
pressures are specified in calculations,
independent of the loading and stress strain
behaviour of the ground.
Situations
involving the time dependent response of
the ground are not discussed.
3.8 Further development by SC7

CEN/TC250/SC7 has set up Evolution
Group 11 on characterisation, chaired by
Lovisa Moritz of Sweden. One of the tasks
of this group is to consider the derivation of
characteristic values.
4.2 Requirements of EC7

EC7 recognises five types of ultimate
limit states:
EQU: loss of equilibrium in which the
strengths of materials is insignificant
STR: failure of structural elements
GEO: failure in the ground
UPL: failure due to uplift by water pressure
(buoyancy)
HYD: hydraulic heave
The particular problems of HYD will be
discussed in 4.9 below.
STR and GEO may occur together and
they are checked using load factors of 1.35
and 1.5 on unfavourable permanent and
variable loads, respectively, with 1.0 and 0.0
on the equivalent loads when acting in a
favourable manner. The load factors used
for UPL are generally lower: 1.1 for
unfavourable actions, generally water
pressure, and 0.9 for favourable, generally
weight of a potentially buoyant structure.
The way in which the STR/GEO factors are
to be applied when water pressure is a
leading action is a particular point of debate.
EC7 says: When dealing with ground
water pressures for limit states with severe
consequences (generally ultimate limit
states), design values shall represent the
most unfavourable values that could occur
during the design lifetime of the structure
{2.4.6.1(6)P}. Noting that design values
are values that already incorporate safety,
requiring no further partial factors, this
paragraph indicates direct assessment of
ULS design values on the basis of their
physical limits in extreme, but credible
4 DESIGNS DOMINATED BY WATER

PRESSURE
4.1 Introduction
In surveys of views on EC7, greater clarity
of requirements for safety provisions in
relation to water pressures is a frequent
request. As in the case of derivation of
characteristic values, the author believes this
request reflects a problem that pre dates EC7.
CEN/TC250/SC7 has set up Evolution Group
9 on water pressures, chaired by Professor
Norbert Vogt of Germany.
Water pressures raise two particular
problems:
a) They sometimes constitute large forces
that are critical to design and have well
defined maximum values, with little real
uncertainty.
b) Besides constituting forces, water
pressures reduce the strength of frictional
soils. Thus they have a double effect in
soil mechanics, and this is also true of
any partial factors applied to them.
The issues raised by designs dominated by
water pressure were considered by Simpson ,
Vogt and van Seters (2011) in a paper
requested
by
the
EC7
committee
CEN/TC250/SC7. Reference to that paper is
recommended for full details of the work,
including mathematics. A less mathematical
8
If these secondary actions or action
effects are large, failure could occur but the
fault may be seen to rest with the owners or
maintainers of the structure, or the vandals;
alternatively, the designer should have
foreseen them and was wrong to omit them
from the primary actions for which the
structure was designed.
However, if the secondary actions or
action effects are small, the owner would
reasonably expect the structure to be
sufficiently robust to withstand them. In this
context, large and small effects have to
be judged in relation to the magnitude of the
primary actions.
It follows that even where there is no real
possibility of unfavourable variation of the
primary actions, it may be necessary to
include some variation of them in design in
order to accommodate the possible
secondary actions that are not otherwise
included. The uncertainty of the way the
actions produce effects within a structure
also has to be accommodated.
The
variations could be applied either to the
actions themselves, in deriving design
values, or to the action effects.
situations. For serviceability limit states, the

design values are to be less severe,
corresponding to normal circumstances.
Expanding on this two paragraphs later,
EC7 provides an application rule: Design
values of ground water pressures may be
derived either by applying partial factors to
characteristic water pressures or by applying
a safety margin to the characteristic water
level {2.4.6.1(6)}. It therefore appears that
the designer can choose between direct
assessment, factoring water pressures or
adjusting water levels to derive ULS design
water pressures.
An earlier paragraph on actions is also
important: Actions in which ground and
free water forces predominate shall be
identified for special consideration with
regard to deformations, fissuring, variable
permeability and erosion {2.4.2(9)P}. This
has attached a significant note, outlining the
single source principle: Unfavourable (or
destabilising) and favourable (or stabilising)
permanent actions may in some situations be
considered as coming from a single source. If
they are considered so, a single partial factor
may be applied to the sum of these actions or
to the sum of their effects.
4.4 Explicitly accommodate the worst water

pressures that could reasonably occur
As noted in 4.2 above, EC7 requires for
ULS design that the design water pressures
design values shall represent the most
unfavourable values that could occur during
the design lifetime of the structure.
4.3 Robustness allow for secondary

actions and action effects
Even in cases where the magnitudes of the
primary actions are fixed with no possibility
of unfavourable variations, designs should be
sufficiently robust to accommodate unknown
and unpredictable secondary actions.
Furthermore, even where the magnitudes of
actions are fixed, the values of resulting
action effects within a structure may have
some uncertainty; that is, there is uncertainty
in the loading model.
In the cases considered here, the primary
unfavourable actions are derived from water
pressure, which in some cases may have very
clear limits. Secondary actions could include,
for example, sedimentation around a
structure in water, excavation of the ground
above a structure relying on the weight of
ground, minor vehicle or ship impacts,
considered too small to include in
calculations, or vandalism of various kinds.
Figure 7. Gravity wall retaining free water.
Figure 7 shows a wall supporting water

pressure. A drain is provided, with the
intention that the depth of water be limited
to 3m. However, if the drain should become
blocked and the water depth increases to 4m,
9
Figure 8. Submerged anchor block.
illustrated in Figure 9. Methods 3 and 4

both follow the single source principle,
noted in 4.2 above, but Methods 1 and 2 do
not.
the bending moment in the wall is increased

by a factor of 2.5. Clearly, a design that took
the expected water depth, 3m, and applied a
partial factor of 1.35 to the water pressure or
bending moment would be inadequate if the
4m depth occurred.
This example illustrates why it is essential
that designs explicitly accommodate the
worst water pressures that could reasonably
occur.
4.5 The single source principle
Figure 8 shows an anchor block, for which
the total weight W is a permanent stabilising
(favourable) force and the anchor force F is a
variable destabilising (unfavourable) force.
The characteristic total density of the block is
c and that of the water w. The water forces
are taken to be permanent.
The strength of the ground or structure are
not at issue, so the only ultimate limit state to
be considered for the anchor block is uplift,
UPL. For this, EC7 provides two factors for
permanent actions, abbreviated here as G;dst
(generally > 1) for the destabilising force and
G;stb (generally < 1) for the stabilising force;
the factor for the variable destabilising force
is Q;dst (> 1).
It is clear that the characteristic weight of
the block, Wk, will be multiplied by G;stb to
derive the design value for UPL, and the
characteristic anchor force, Fk, will be
multiplied by Q;dst. Four possible methods of
applying partial factors to the water pressures
could be considered, as listed in Figure 8 and
(a)
(b)
(c)
(d)
Figure 9. Factored water pressures on anchor

block. (a) Characteristic (and Method 4), (b)
Method 1, (c) Method 2, (d) Method 3.
In Figure 10, the allowable characteristic

anchor force, Fk, is plotted against the
Density ratio c/w; Fk is normalised by
dividing by Wk. For the purpose of this
figure, the values of partial factors have
been taken from the UK National Annex:
G;dst = 1.1, G;stb = 0.9, Q;dst = 1.5.
Figure 10a shows that for Method 1 the
allowable anchor force depends on the water
depth (normalised by dividing by the height
of the block). This occurs because different
factors are applied to the destabilising and
stabilising water forces. This is considered
to be physically unreasonable, except,
perhaps, in very rare circumstances for
10
Figure 10 Submerged anchor block allowable anchor force in relation to density of block.
(a) Method 1, (b) Methods 2 to 4, (c) Methods 2 to 4 assuming the anchor force is permanent.
indicating that for this problem it may not be

necessary to apply factors to water pressure,
either directly or indirectly. The resultant of
water actions, which is destabilising, is not
increased, so the overall factor of safety is
lower than obtained with Method 3.
It is concluded that methods which follow
the single source principle (Methods 3 and
4) are to be preferred. Method 4, in which
water pressure is not factored, appears to be
adequate for this example.
which the pressures above and below the

block are independent because they are not
from a single source.
As the water
becomes deeper, the allowable anchor force
reduces for the same block, and for d/H=5 no
force can be taken unless the density of the
block is more than twice that of water.
The results for Methods 2 to 4, shown in
Figure 10b, are independent of the water
depth. For Method 2, the allowable Fk tends
towards the unfactored value for low density
ratios. Figure 10c is similar, except that it is
assumed that the anchor force is permanent,
rather than variable (ie G;dst has been applied
to F in place of Q;dst). In this case, Method 2
provides very little safety for low density
ratios. A further important objection to
Method 2 is that it applies a reduction factor
(G;stb<1) to the buoyancy effect of the water,
which is clearly a destabilising effect.
Methods 3 and 4 both follow the single
source principle, and so avoid the need to
distinguish
between
stabilising
and
destabilising actions of water pressures.
Method 3 provides apparently reasonable
results, though in effect the density of water
is factored, which could lead to difficulties in
more complex situations where the strength
of soil is affected by water pressures. This
difficulty might be avoided if all actions of
connected water are combined to find a
resultant destabilizing uplift force, which is
then factored by G,dst. This method clearly
shows where safety on water pressures is
applied, by considering the block weight and
water uplift separately.
Method 4, with no factors on the water
forces, also provides reasonable results,
4.6 Partial factors on the density of water?

Referring to the example shown in Figure
11, Simpson et al (2011) discussed whether
the density of water should be factored in
order to provide a safety margin. Although
they agreed that that should be avoided, they
noted that some safety formats would factor
the water pressures derived from the
unfactored density, and others would factor
the forces derived from unfactored
pressures.
Figure 11. Gravity construction retaining

water.
EC7 has no middle third rule for

eccentrically loaded spread foundations,
though it says that special care should be
exercised if the resultant force does not lie
11
Where G and G,inf are partial factors on the
unfavourable and favourable actions. For all
cases:
within the middle two thirds, as illustrated in

Figure 11. Simpson et al (2011) suggested
that this limit should be applied for situations
dominated by water pressure.
Td Rd = Rk/P,t
4.7 Use of an offset in water level?

Figure 12 shows a deep basement
extending below the water table.
No
drainage is provided beneath the base slab, so
hydrostatic water pressures are expected.
Some unplanned variation in the water level
is possible, for example due to leakage from
a water main. The total weight of the
structure, which could include superstructure
built on the basement, is W and its area in
plan is A. If needed, tension piles are to be
provided to prevent uplift.
where P,t is a resistance factor on pile

capacity in tension.
In situations where U greatly exceeds W,
the precise sequence of calculation in which
the factors are applied and the value of the
partial factors may vary according to
national practice, but the outcome is much
the same. The case of W greatly exceeding
U, which would require compression piles if
the slab is suspended, is not considered here.
The problem is more debatable when the
characteristic (unfactored) values of W and
U are close, especially in formats that use
G;inf = 1.0, which is common. If Wk=Uk and
G>1 is applied to water pressure, tension
piles are needed, but if water pressure is not
factored or adjusted in some other way no
piles are needed, even if a factor is applied
to the resultant (UkWk), which in this case
equals zero.
To illustrate this problem, suppose n piles
are to be provided each with a characteristic
resistance in tension Rk. For the purpose of
plotting results of calculations, it is
convenient to define Ww=wAD; this is not
the buoyancy force, which is Uk=wAh.
When Uk=Wk, h/D = Wk/Ww. In Figure 13 the
number of piles required, n, represented by
nRk/Ww, is plotted against h/D for a typical
case in which Wk/Ww=0.25. The values of
partial factors used here are adopted for
illustration only, and may not represent any
particular national practice. Some countries
prefer to view tension piles as providing a
favourable action, which would also lead to
adoption of different factors. In Figure
13(a), the critical area of the graph is shown
as an enlarged detail as Figure 13(b).
In the unfactored case, piles only become
necessary when h/D > Wk/Ww = 0.25 in this
example. If factors are applied to the
unfactored resultant force in the piles,
together with pile resistance factors, a line
such as line (b) is obtained, for which R=1.7
was used for the piles in this illustration.
Figure 12. Deep basement subject to uplift.
The uplift force beneath the basement is

given by U = wAh, where w is the weight
density of water.
If the characteristic uplift force U
approaches or exceeds the characteristic
weight W, the tension force T in the piles has
to be derived. For ULS UPL we find:
UkG,dst = WkG,stb + Td
(2)
where G,dst and G,stb are partial factors on the

permanent disturbing and stabilising actions.
This means that
Td = UkG,dst WkG,stb
(3)
It is also possible to consider the problem

as ULS STR/GEO. Then we get:
Td = UkG WkG,inf
(5)
(4)
12
22
0.4
0.4
(a)
nRk/Ww
k/Ww
(b)
n.Rk/Ww
n.Rk/WwnR
(b)
(c)
(c)
11
(d)
(d)
(e)
(d)
(f)
(c)
(a)
(b)
0.2
0.2
(a)
(e)
(e)
(f)
(f)
00
00
0.2
0.2
0.4
0.4
0.6
0.6
h/D
h/D
0.8
0.8
11
0
0
0.1
0.1
(a)
0.2
0.2
h/D
h/D
0.3
0.3
0.4
0.4
(b)
Figure 13. Number of piles required (normalised). (a) unfactored, (b) pile resistance factored, (c) G =
1.35 on water pressure, (d) water table adjusted, (e) UPL, (f) G;fav = 0.8 on weight.
The gradient of this line depends on the

values of the factors, but when h/D = Wk/Ww
= 0.25 no tension piles will be provided and
there is no reserve of safety for deviation
from the characteristic values of water
pressure and weight. This is considered to
represent
an
unacceptable
situation.
However, for a high water table (h/D
approaching 1), this case is taken as a
reasonable guide to the number of piles
needed.
If the water pressure beneath the base is
multiplied by a partial factor G=1.35, a line
such as line (c) in Figure 13 is obtained; in
plotting this line a lower value of pile
resistance factor R=1.3 has been adopted, in
acknowledgement of the increased value of
G. In this case, a reserve of safety is
provided when h/D = Wk/Ww, requiring some
tension piles. However, the number of piles
might be regarded as excessive for the case
of a high water table, h/D approaching 1,
where the water pressure beneath the base
becomes physically unreasonable.
An alternative approach could be to avoid
factoring water pressure but to require an
increase in the water head h. For example,
line (d) in Figure 13 shows the results when
the free height above the water table (Dh) is
reduced by 10%. This has an advantage in
the case where h is large (eg h/D=1) that it
does not enhance the water pressures
unreasonably, requiring too many piles. The
amount by which the water head should be
raised is difficult to specify for general
application in a code of practice, however. If

this approach is preferred, it may be
necessary to rely more heavily on the
expertise of the designer to decide what
margin is appropriate. This is consistent
with the approach of EC7 {2.4.6.1(6)P}
using direct assessment of design values:
When dealing with ground water pressures
for limit states with severe consequences
(generally ultimate limit states), design
values shall represent the most unfavourable
values that could occur during the design
lifetime of the structure.
In relation to EC7, the discussion above
relates to the STR/GEO requirements
normally used for finding the number and
required resistances of piles. EC7 has
another requirement for uplift cases, UPL,
which is normally understood to require a
factor G;dst > 1 applied to uplifting water
pressure and a factor G;stb < 1 applied to
stabilising total weight. Line (e) in Figure
13 is plotted for typical values G;dst = 1.1,
G;stb = 0.9, with the resistance factor for the
piles R=1.7. This requirement can produce
sensible results provided that (a) it is agreed
that piles are to be designed using loading
derived from UPL and (b) an appropriate
system and values of factors is adopted in
applying these loads to pile design. As with
other schemes involving factors on water
pressure, it becomes unreasonable when the
water table approaches ground level (h/D=1)
and may demand more piles than are really
needed.
13
In this problem, it is necessary to change
the water pressure or the building weight
from their characteristic values in order to
increase safety when Uk is close to Wk. A
possible alternative, not considered by
Eurocode 7 but recommended for further
consideration, would be to apply a reduction
factor to the weight of the building, say 0.8,
while leaving the water pressure unfactored.
This is shown as line (f) in Figure 13, plotted
with R=1.7. This provides safety when h/D
= Wk/Ww, but it avoids factoring water
pressure and has a smaller effect than some
of the alternatives, such as UPL, when
h/D=1.
Of all the approaches presented here, the
author recommends adoption of either (d) an
increase in the water head h or (f) a reduction
factor applied to the weight of the building of
about 0.8.
included here because it has been a further

point of controversy. This equation relates
to states of hydraulic heave in the ground
caused by hydraulic gradient, as illustrated
in Figure 14.
Figure 14. HYD in EC7
4.8 The star approach DA2* , DA1* ?

In any scheme of load factoring, it can be
argued that equilibrium has to be broken at
some point in the chain from density to
design action effect. The present authors
view is that geotechnical calculations can
best be carried out by preserving equilibrium
with unfactored waster pressures up to the
point of deriving structural action effects.
This is consistent with the note in EC7
{2.4.2(9)}, discussed in 3.2 above, a single
partial factor may be applied to the sum of
these actions or to the sum of their effects.
This scheme could be applied to EC7s
Design Approach 1, where it is only relevant
to Combination 1.
It could be termed
approach DA1*, by analogy with DA2* in
which load factors are always applied at a
late stage in the calculation. An important
difference is that DA1*, as intended here, is
only used for problems dominated by water
pressures, in which the uncertainty of the
action effects of the water, together with
other secondary actions, are probably more
important than any adverse variation of the
water pressures themselves that have not
already been incorporated.
EC7 has an equation for checking

hydraulic heave which appears in total stress
and effective stress forms Equations 2.9a
and 2.9b. It also gives the value of partial
safety factors, but it is unclear about how
these factors should be applied. This has led
to much confusion, with debate about
whether 2.9a or 2.9b is correct.
EC7 {2.4.7.5(1)P} states: When
considering a limit state of failure due to
heave by seepage of water in the ground
(HYD, see 10.3), it shall be verified, for
every relevant soil column, that the design
value of the destabilising total pore water
pressure (udst;d ) at the bottom of the column,
or the design value of the seepage force
(Sdst;d) in the column is less than or equal to
the stabilising total vertical stress (stb;d) at
the bottom of the column, or the submerged
weight (Gstb;d) of the same column:
udst;d stb;d
(2.9a)
Sdst;d Gstb;d
(2.9b)
Annex A of EC7 provides values for
partial factors to be used for HYD, G;dst =
1.35 and G;stb = 0.9. But the code does not
state what quantities are to be factored. Orr
(2005) reported calculations for the situation
of potential hydraulic heave shown in Figure
15. He found that the calculated allowable
4.9 HYD EC7 Equation 2.9

Discussion of EC7 Equation 2.9 was not
considered by Simpson et al (2011), but it is
14
If the factors are applied to excess pore
pressure, or excess head, and to buoyant
weight, the allowable height of water H
following Orrs calculations is 6.84m, using
either form of the equation. If, instead, the
factor G;stb is applied to total density, the
allowable height of water H is only slightly
less at 6.1m, nowhere near the unreasonable
value of 2.78m obtained by Orr on his
interpretation of Equation 2.9a. Once it is
decided to what parameters the factors are
applied, the choice of Equations 2.9a or 2.9b
makes no difference.
Figure 15. Hydraulic problem considered by

Orr (2005).
4.10 Conclusions on water pressures

Prescription of safety in design for
problems dominated by water pressure is a
subject of ongoing debate.
For
completeness, the conclusions reached by
Simpson et al (2011) are listed here.
Agreement was reached on the following
points:
1. The effects of water pressures are very
important in geotechnical design. Their
actual values can have significant
uncertainties, and values outside the
range anticipated in design can cause
major failures.
2. Partial factor design applies factors to a
small number of leading, or primary
actions. In real design situations,
secondary actions of relatively small but
unpredictable nature and magnitude
should also be accommodated; that is, a
degree of robustness it required. Often,
these are accommodated by increasing
the partial factors applied to primary
actions or action effects.
3. Designers must explicitly accommodate
the worst water pressures that could
reasonably occur. Reliance on factors of
safety together with less extreme water
pressures or water levels may give a
false sense of security.
4. Application of partial factors to the
density of water should generally be
avoided.
5. One useful way to maintain a prescribed
degree of safety is to require an offset in
water pressure, raising or lowering the
water surface or piezometric level.
6. The single source concept should be
applied whenever possible.
height of water H could vary from 2.78m to

6.84m due to application of the same factors,
taken from EC7, at different points in the
calculation.
Taken at face value, Equations 2.9a and
2.9b are alternative representations of the
same physical requirement.
They are
expressed in terms of parameters that are
inter dependent, and design values are used.
In the absence of prescription in EC7, Orr
(2005), and others have interpreted these
equations to mean
G;dst udst;k G;stb stb;k
and
G;dst Sdst;k G;stb Gstb;k
where the subscript k indicates characteristic,
unfactored values for the parameters.
The two requirements expressed in this
way have significantly different effects, as
illustrated by Orr, because the factors are
applied to different quantities. If the factors
are applied to the same quantities, Equations
2.9a and 2.9b are alternative statements of the
same requirement.
As in examples presented above, the main
problem arises here if a partial factor is
applied to characteristic water pressure udst;k.
A factor can be applied to excess water
pressure or excess head, and this is
equivalent to factoring the seepage force
Sdst;k.
On the right hand side of the
requirement, the issue is whether to factor
total density or buoyant density of the
ground. Since the proposed factor, 0.9, is
close to unity, this is less important; in very
light weight soils factoring buoyant density
has no effect, which suggests that it is safer
to factor total density.
15
7. The star approach (DA2* or DA1*,
introduced here) has advantages when
dealing with problems dominated by
water pressures because it avoids the
application of partial factors to the
density of water or to water pressures.
8. In uplift problems, it is necessary to vary
either water pressures or the magnitudes
of favourable, stabilising weight, in order
to ensure safety in view of possible
secondary actions. In order to avoid
factoring water pressures, the possibility
of a reduced factor on favourable weight,
perhaps between 0.8 and 0.9 should be
considered.
9. To prevent toppling failure of structures
loaded laterally by water pressure, a
middle 2/3rds rule could be considered,
applied to unfactored actions.
10. Although there are obvious advantages in
making codes of practice as precise and
prescriptive as possible, the need for
engineering expertise and careful
evaluation of the full range of credible
scenarios cannot be replaced. This is
particularly true of situations in which
water pressure has a dominating role.
The following points were not agreed and
remain to be debated and researched further.
In some cases, appropriate conclusions may
depend on other features of the safety formats
adopted, for example the differing Design
Approaches of Eurocode 7.
11. Whether it is desirable to apply factors to
water pressures. Several approaches that
avoid this have been discussed, but in
some approaches factors are applied to
water pressures in some circumstances.
12. Whether it is reasonable to apply partial
factors to forces (action effects) directly
derived from water pressures. It is agreed
that this may raise problems, which were
discussed, but the authors could not agree
that it can always be avoided.
13. The use of the star approach, factoring
action effects, in cases where it is directly
equivalent to factoring water pressures,
either complying with the single source
principle or not compliant. The problem
particularly relates to situations in which
equilibrium is not maintained throughout
the geotechnical calculations of stability,
including sliding, bearing, toppling and
uplift.
Less concern is felt about
application of factors to action effects
internal to structures, such as bending

moments in walls and slabs or forces in
piles.
5 THE EQU PROBLEM

5.1 Outline of the issue
EN 1990 defines a mode of failure
termed the EQU limit state. It relates to
situations in which the effects of actions
coming from a single source effectively
balance, cancelling each other out in some
respects, and little or no strength of material
is apparently needed.
Limit state EQU is described in EN 1990
and EN 1997 1 by two slightly different
definitions:
EN 1997 1, 2.4.7.1(1)P gives: Loss of
equilibrium of the structure or the ground,
considered as a rigid body, in which the
strengths of structural materials and the
ground are insignificant in providing
resistance
EN 1990, 6.4.1(1)P gives: Loss of static
equilibrium of the structure or any part of
it considered as a rigid body, where:
minor variations in the value or the
spatial distribution of actions from a
single source are significant, and
the strengths of construction materials
or ground are generally not
governing.
Load factors required by EQU differ from
those of the main ultimate limit states
involving strength of materials, STR and
GEO, as shown in Table 2.
Table 2. Load factors in limit states STR, GEO
and EQU (simplified)
Permanent
unfavourable
favourable
Variable
unfavourable
favourable
STR/GEO
EQU
1.35
1.0
1.1
0.9
1.5
0
1.5
0
EQU is particularly relevant when the

actions come from a single source, so, in
accordance with EC7 {2.4.2(9)P} noted in
4.2 above, no imbalance would be created
by application of different factors. The
classic situation of a balanced cantilever
16
shown in Figure 16. If both the forces W
come from a single source, then regular
application of load factors would lead to
calculation of no bending moment in the
column and equal compressive forces in the
piles. However, this would be an unstable
equilibrium. The purpose of EQU is to
ensure stability with an adequate margin of
safety by applying different factors to the two
forces W, even though they come from a
single source.
Figure 16.
foundation.
Balanced
structure
on
5.2 Variety of views

Treatment of EQU in the geotechnical
context has been discussed by Schuppener et
al (2009), who noted two alternative
concepts:
Concept 1 proposes verifying only
EQU in those cases where loss of
static equilibrium is physically
possible for the structure or part of it,
considered as a rigid body. Similarly
Concept 1 proposes verifying only
STR/GEO in situations where the
strength of material or ground is
significant in providing resistance.
Concept 2 proposes verifying EQU in
all cases; it is interpreted as a load
case. Where minor strength of
material or ground is involved, the
combined EQU/STR/GEO verification
may be used, if allowed by the
national annex.
For the situation shown in Figure 16,
Concept 1 would require that the column
and piles must be designed as though the
two loads W are independent, with factors of
1.0 and 1.35, even though they come from
the same source, which implies that the
designer does not expect them to be unequal.
This gives forces in the piles of
Wr(1.1750.35a/b). However, Concept 1
states that this is not to be applied if no
material strength is involved.
Concept 2 requires that the piles be
designed for the smaller factors of EQU,
giving significantly smaller forces of
Wr(10.2a/b). Concept 2 suggests that EQU
is just another load case rather than a
different limit state, and that all designs
should be able to accept this load case.
piled
The problem this creates is that a likely

design response to any imbalance in EQU is
to provide some material resistance to assure
stability. For example, in Figure 16 if the
factoring system creates a differential
between the two forces, a possible response
is to provide bending resistance in the
column and the ability to take larger
compression, or possibly tension, in the piles.
This, however, is inconsistent with the
definition of EQU.
The problem also affects structural design
and has been recognised by Gulvanessian et
al (2002, section 7.5.1).
5.3 The authors view

The present author considers that
Concept 2 is to be preferred, for three
reasons.
(a) Concept 1 disregards the single source
principle, which was set up in EN 1990
specifically
to
avoid
creation
of
unnecessarily severe loading conditions
when loads come from a single source, that
is, they are clearly correlated.
(b)
Concept
1
implies
sharp
discontinuities in design requirements. For
example, in relation to Figure 16, if a/b =
17
4.99 both concepts check EQU and conclude
that there is no tension requirement for the
piles. For very slightly higher a/b = 5.01,
Concept 1 requires that the piles be designed
for a significant tension of 0.58Wr. In
contrast, for this very small change in a/b
Concept 2 requires very little tension
capacity, 0.002Wr.
(c) In some cases, Concept 1 requires that
the foundations be designed for more severe
loading when their strength is limited than
when their strength is considered unlimited,
implying that there is no strength
requirement. Schuppener et al considered the
retaining structure shown in Figure 17. In
Concept 1 the problem was judged to be one
of overturning if the ground was infinitely
strong, for which the load factors applied
were those of EQU. But if the ground was
not infinitely strong in bearing, the larger
load factors of GEO had to be considered. In
the authors view, it is unreasonable to
change the loading in this way as a function
of the strength of the ground. This issue does
not arise with Concept 2.
With the (1.1, 0.9) combination, tension

occurs for a/b>5. However, for the (1.35,
1.15) combination, the forces in the piles are
Wr(1.150.2a/b), so there is no tension until
a/b>5.75. This difference might be more
significant than was intended by the code
drafters. In the UK National Annex, the
(1.35, 1.15) combination has been accepted
for building design but not for bridges.
6 USE OF FEM FOR ULS
6.1 Introduction
Numerical methods, taken here to refer to
finite element and finite difference
computations, are often used with
unfactored parameters to analyse service
conditions, checking for serviceability limit
states.
Some results taken from these
analyses may be factored to provide design
values for ultimate limit state checks; this
applies mainly to structural forces and
bending moments. However, the use of
numerical methods for full ULS analyses, in
which the strength or resistance of the
ground is modified, is comparatively new
and still a subject of debate. To study this
and related issues, CEN/TC250/SC7 has set
up Evolution Group 4 on numerical
methods, chaired by Dr Andrew Lees of
Cyprus.
Use of numerical methods with EC7
raises the following issues:
1. Can numerical methods be used for
ULS analysis in accordance with all of
EC7s Design Approaches?
2. If the strength of ground is to be
reduced, how is this to be applied in
practical computations?
3. Does factoring ground strength lead to
the wrong failure mechanism?
4. For staged construction or excavation,
at what stage should factoring be
applied?
5. Can advanced soil models be used with
partial factor methods?
6. How are factors to be applied for
undrained
behaviour
and
time
dependent consolidation?
These issues are in current debate, with
no generally agreed answers. They will be
Figure 17. Potential overturning of retaining

structure.
5.4 An alternative formulation

EN 1990 permits national annexes to
accept an alternative set of factors for EQU
combined with STR/GEO, using 1.15 on
favourable permanent actions and 1.35 on
unfavourable permanent actions. Referring
to Figure 16, this would give the same
bending moment in the column as the basic
GEO combination of (1.1, 0.9), supporting
the case for Concept 2 discussed above.
Schuppener et al (2009) show that this
formulation can alleviate some of the
geotechnical problems of EQU.
18
example of this approach is the c,
reduction in Plaxis.
In method (a), the aim is to show that no
ULS occurs with the reduced strengths. The
various strength parameters involved, cu, c,
, other soil parameters and structural
strengths, can all be factored by different
amounts if required.
Normally the
occurrence of a ULS will not actually be
modelled. This method can show that the
design complies with code requirements, but
it may not make clear how much additional
reserve the design has, or how much further
economy might be available.
Method (b) models an actual ULS and the
aim is to show that the reduction factors that
apply to material strengths at the ULS are
greater than those required by the code. The
difference between the code requirements
and the actual reduction factors at failure
may give some indication of the reserve of
safety and the further economy available.
Even so, it will not lead directly to an
improved design; this has to be achieved by
further analysis of a revised geometry.
It is important that the appropriate partial
factors are applied to each strength
parameter involved, such as cu, c, , and
possibly structural strengths. This is easily
achieved with method (a); facility for
method (b) will be dependent on the
software.
As an example, Figure 18 shows the
displacements computed for an 8m deep
excavation with a diaphragm wall and single
prop. The design moment of resistance of
the wall is 1000kNm/m. Following method
(a), with a strength reduction factor of
1.25 the mobilised bending moment is
772kNm/m (Fig. 19). No ULS occurs, so
the design is verified. Figure 20 shows the
development of a mechanism as has
gradually been reduced to a value equivalent
to = 1.45, following method (b); for this
analysis the computed bending moment is
1729kNm/m, far above the moment of
resistance of the wall. The method (b)
analysis, with an excessive reduction in the
strength of the soil and an unacceptable
bending moment, yields no useful
information, failing to clarify whether the
outlined below and the author will offer

opinions.
Following this, a practical
example of computations to Design
Approach 1 will be presented.
6.2 Can numerical methods be used for all
Design Approaches?
Numerical methods can be used relatively
easily for ULS computations if this merely
requires using factored values for the input to
the program, or simply factoring the
structural action effects resulting from the
geotechnical program. Design Approach 2
(DA2) requires factors to be applied to
quantities that are internal to the geotechnical
analysis such as active and passive forces or
pressures, and bearing resistance for spread
foundations. So it is generally accepted that
full numerical analyses of ultimate limit
states cannot be undertaken for DA2. Most
countries that use DA2 require use of DA3
for numerical analysis.
Design Approach 1 was the only approach
in the ENV version of EC7 published in
1995. In its development, the possible use of
numerical methods was considered, so it can
be used relatively easily. The use of DA1 is
not very different from the combined use of
DA2 and DA3, except that DA1 requires
checking of two calculations, whereas
combined use of DA2 and DA3 could imply
acceptance of a design that passes according
to one DA but fails according to the other.
The legal implications of such as situation
might be debatable.
6.3 How should strength factors be applied?
Design Approaches 1 and 3 require the
strength of the ground to be reduced by
partial factors. The aim is to show that no
ULS occurs (strictly, none is exceeded) when
the ground is assumed to have its reduced
design strength.
Two alternative methods are available for
application of strength reduction factors:
a) Reduce the ground strengths used as
input to the numerical computation. This
could be done by the user or
automatically by the program.
b) Carry out the computation first with
unfactored strengths, then progressively
reduce them until a ULS is reached. An
19
Computing a failure mechanism by
increasing the loading to an unrealistic level
with unfactored strengths models an equally
wrong state. The requirement of EC7 is to
show that for the design conditions a ULS
will not occur; the mechanism by which it
would occur in even more extreme
circumstances is not important.
In design situations, there may be many
criteria to be met in checking against the
occurrence of ULSs, relating to the various
soil types and structural elements involved.
For example, in analysis of a retaining
structure, it is not useful to know that a
failure would occur for a certain reduction of
soil strength, if the computed forces or
bending moments in the same situation
would not be acceptable.
This could
potentially be overcome by modelling the
structural behaviour with elastic plastic
elements, using appropriate design strengths
and considering any brittleness. If this is not
done, an elastic model of the structure will
often be suitable for method (a), but method
(b), which takes the computation to failure,
might be more problematic.
Max wall displacement 48mm
Figure 18. ULS analysis of 8m deep excavation

= 1.25.
0
= 1.25
= 1.45
10m
500
1000
1500
kNm/m

bending moments.
6.5 Staged construction

Numerical methods are often used to
study the serviceability of designs for staged
excavations.
Figure 21 shows two
alternative strategies that have been used to
carry out ULS analyses with factored soil
strengths. In Strategy 1, all soil strengths
from the start and throughout the
computation. In Strategy 2, the computation
is initially carried out with unfactored
strengths, then, in separate branching
computations, soil strengths are factored at
critical stages, considered separately. These
alternatives have been discussed previously
by many Bauduin et al (2000), Simpson and
Yazdchi (2003) and, in the context of tunnel
design, Cheung et al (2009).
Referring to 6.3 above, method (a) can be
used with Strategy 1 or 2, but method (b)
can only be used with Strategy 2.
Current opinion generally favours
Strategy 2, as confirmed by recent
discussions in EC7 Evolution Group 4, for
two main reasons.
(a) It is feared that
applying factors in earlier stages might have

= 1.5.
design is acceptable or not, as in fact the

method (a) analysis showed it was.
6.4 The wrong failure mechanism?
A
common
objection
to
ULS
computations, particularly to method (b), is
that with factored strengths they give the
wrong failure mechanism. In fact, there is
no correct failure mechanism, because
failure is not the correct state to occur.
20
%%% Strategy 1 %%%
Compute using
factored strength
Factor material
strengths
and it is reasonable to take this as the

factor on drained strength, however derived.
But more interpretation might be needed if
the national annex gives two different
values. In any event, it will be important to
ask whether the strengths calculated by the
model are more or less reliable than those
used in normal practice. Some adjustment
to the factors should be made in the light of
this.
%%%%%%%%%%%%%% Strategy 2 %%%%%%%%%%%%%%%%

Compute using
characteristic
parameters
Compute using
factored
parameters
Initial state
Initial state?
Could be critical
for wall
bending moment
Excavate to 5m
wall cantilevering
Excavate to 5m
wall cantilevering
Install prop at 4m
depth
Install prop at
4m depth
Excavate to 10m
Excavate to 10m
Could be critical
for wall length,
bending moment
and prop force
No further factors
on strut forces or
BMs
Apply factors on
strut forces or
BMs
No further factors
on strut forces or
BMs
6.7 Undrained behaviour and consolidation

Soils are essentially effective stress
materials responding to effective stresses, so
advanced models of soil behaviour are
almost always expressed in terms of
effective stress parameters.
When soil
deforms slowly enough that the deformation
does not cause change of water pressure, it
is said to be drained. When it deforms
quickly enough that no water can enter or
leave the soil elements it is said to be
undrained. The undrained shear strength
of soil, cu, can be measured, and this results
from its initial state and its effective stress
parameters (eg c, ).
Effective stress parameters (eg c, ) are
sometimes called drained parameters.
This is unhelpful terminology, since these
parameters govern behaviour and can be
used in models or drained, undrained and
partially drained states.
Reliable computation of undrained
strength from effective stress parameters is
very difficult, since it is affected by many
features of soil behaviour, including
anisotropy and dilation. Because undrained
strength can be measured directly, with
moderate reliability, it is often preferable to
input it directly as a parameter in numerical
analysis than to try to compute it. However,
this is not compatible with the use of
effective stress models.
EC7 generally requires a higher factor on
undrained strength (eg 1.4 on cu) than on
effective stress parameters (eg 1.25 on c,
tan). The drafters assumed that effective
stress parameters would be used only for
drained states and did not comment on the
use of approaches that computed undrained
strength for them. In numerical analysis,
there is therefore a temptation, for economy,
Figure 21. Strategies for analysis of staged

construction.
either an unreasonable or an optimistic effect

on later stages; the author supports this
reasoning. (b) In some cases the ULS
computations can be run as adjuncts to SLS
computations; however, this will not be the
case if design situations for the two limit
states require differences of geometry (such
as unplanned overdig), loading or water
pressures. A third possible reason is that
only Strategy 2 can be run with method (b),
such as the Plaxis c, reduction procedure.
6.6 Use of advanced soil models for ULS
Models of soil behaviour much more
realistic than linear elastic Mohr Coulomb
are increasingly used for SLS computations.
Although some of these, such as the Plaxis
hardening soil models, have simple strength
parameters (c, ) as input, many do not (eg
Cam clay models or the BRICK model,
Ellison et al 2012).
Practice to date has been to use only
elastic Mohr Coulomb models for the ULS
part of the computation, even when this was
an adjunct to an SLS run that used a more
advanced model.
EC7 provides factors specifically for c,
and cu, so if these are not input parameters of
the program some degree of interpretation
will be needed. In the case of cu, this is the
(relevant) undrained strength, so it would not
matter how this was derived. If the national
annex gives the same value for factors on c
21
to use an effective stress model for undrained
behaviour with a lower factor (eg 1.25 on c,
tan) than would be used if undrained
strength were input directly (eg 1.4 on cu).
The author considers this to be potentially
unsafe. The higher factor (eg 1.4) was
considered appropriate for characteristic
values of cu based on measurement, which is
generally more reliable than values computed
from effective stress parameters, so it is
unreasonable to adopt a lower value for the
latter.
The conclusion to be drawn from this is
that when undrained behaviour is being
modelled using effective stress parameters
(eg c, ) the partial factor applied to them
should give a reduction in undrained strength
at least equivalent to that required by EC7 for
cu. This requires some testing, but as a first
approximation the factor to be applied to c,
tan might be about the same as that
required for cu (eg 1.4).
If time dependent consolidation is to be
modelled in a ULS numerical analysis, it will
again be necessary to use an effective stress
model. The values of partial factors to be
applied have not yet been considered.
and 27 to 32m deep to underside of base

slab (Fig. 22). The assumed excavation
temporary works consisted of three levels of
temporary steel props.
Wall thickness
required by initial design was 1.2m. This
was subsequently increased to 1.6m for
consistency with other structures on the
Italian high speed network and to account
for a possible prolonged cessation of work at
final excavation stage (Hocombe et al 2007).
Figure 22. Florence station cross section with

design Mohr,Coulomb soil parameters and
stratigraphy.
Eurocode 7, Design Approach 1 (EC7,

DA1) was adopted by the design team.
Design of the structure to Combinations 1
and 2 (C1 and C2) was more onerous than
regulations applying at the time in Italy.
The current Italian national standard
specifies design of retaining structures and
their supports to C1 only while C2 is
checked for global failure due to collapse of
the soil, not of the structure. It was,
however, considered prudent for this project
that the structure be checked for both
combinations in accordance with EC7 DA1.
The decision to use EC7 DA1 did not,
however, increase the number of analyses to
be carried out since C2 was needed in any
event to evaluate the depth of embedded
walls for lateral stability.
7 CASE STUDY:
FLORENCE HIGH
SPEED RAIL STATION
7.1 Introduction
The analysis of a large station box in
Florence, Italy, designed to Eurocode is
described by Simpson and Hocombe (2010),
from which the following abbreviated version
is drawn. This case study presents salient
features of the design and the method
adopted. Partial factors were applied to soil
properties at all stages of excavation in the
ULS analyses. Results are presented here of
a subsequent comparative study into the
effects of applying partial material factors
only at specific excavation stages.
The proposed station lies on a high speed
rail line currently nearing completion
between Milan and Naples and is situated just
north of the historic centre of Florence. The
Client for the station is Rete Ferroviaria
Italiana (RFI) with construction scheduled for
2010. The structure is 454m long, 52m wide
7.2 Effect of factoring at discrete stages

In EC7 DA1, application of partial
factors to soil properties is required in ULS
C2, and these factors were applied at all
construction stages in the original design.
Subsequently, an investigation was carried
out into the effects of applying partial
material factors only at specific excavation
22
stages rather than at all stages. The results of
these C2 analyses (with factored soil but
unfactored wall moments and shears) are also
compared with the results of C1 analyses
(unfactored soil, factored wall moments and
shears). The investigation considered three
propped excavation stages based on a 1.2m
thick diaphragm wall, applying the
characteristic Mohr Coulomb soil parameters
presented in Figure 22 in the pseudo finite
element program Oasys FREW (Pappin et al.,
1986).
Bending moments derived using FREW
are presented in Figure 23. The ULS C1
results have been factored up by the specified
partial factor of 1.35 for comparison with
those from ULS C2.
Maximum positive wall moments (tension
on excavated face) are marginally higher
from C1 compared to those from C2 in which
partial factors are applied at all stages, the
solid lines in the figure. Negative wall
moments are, however, greater in C2 at the
middle prop level (+31m). This may be a
result of lower margin on lateral stability and
higher wall deflection during the deeper
stages of excavation in C2 compared to C1.
Applying partial factors on soil strengths
in C2 only at the respective excavation stages
gave similar wall moments in stages 1 and 2
to the C2 analysis with factors applied at all
stages. Application of partial factors only at
excavation stage 3, however, resulted in
larger negative wall moments at the lowest
temporary prop, with greater wall deflection
than the analysis with partial factors at all
stages. Inspection of predicted soil pressures
suggests that with partial factors applied
during stages 1 and 2 more soil arching onto
the higher props occurs, allowing greater
reduction of soil pressure below the active
limit than when partial factors are applied
only in stage 3.
Results of the comparison in design prop
forces using FREW are presented in Figure
24. The ULS C1 results have again been
factored up by the specified partial factor of
1.35 for comparison with ULS C2. The C2
forces with factored soil strength at all stages
are higher at the lower two levels than those
from C1.
Figure 23. ULS wall bending moments
Figure 24. ULS prop forces.
In the C2 analyses with soil strength

factors only at discrete stages the design
prop forces are similar at the upper two
levels of props to those from C2 analysis
with partial factors at all stages. In the
lowest level of props, however, the force is
significantly greater if partial factors are
applied only at this stage, possibly due to the
effects of soil arching mentioned above.
7.3 Conclusions from case study
For this particular study, the results
obtained when partial factors were applied at
discrete stages were more severe than when
they were applied throughout the
computation. More studies of this type are
needed to determine whether this is a
general rule. In the opinion of the author,
however, both types of analysis are valid
checks on the code requirements. In a more
complete ULS analysis, bending plasticity of
the wall could be included, and this would
probably give more similar bending
moments from the two C2 analyses. The
same may not be true for the strut loads,
however, and since struts may provide a
somewhat brittle response this remains an
area to be investigated.
23
EN1990. (2002) Eurocode: Basis of design. BSI,
London. (BS EN 1990:2002).
EN1997 1
(2004)
EC7.
Eurocode7:
Geotechnical design Part 1: General rules. BSI,
London. (BS EN 1997 1:2004).
BSI (2007) Eurocode 7 Geotechnical design
Part 2: Ground investigation and testing. BSI,
London (BS EN 1997 2:2007).
Foye, K.C. Salgado, R. & Scott, B. 2006.
Resistance factors for use in shallow foundation
LRFD. J. Geotech. Geoenviron. Eng., 132(9), 1208
1218.
Gulvanessian, H, Calgaro, J A, Holicky, M
(2002) Designers' Guide to EN 1990 Eurocode:
Basis of structural design. Thomas Telford.
Hocombe, T., Pellew, A., McBain, R., and Yeow,
H C. (2007) Design of a new deep underground
station structure in Florence. Proc. XIV ECSMGE,
Madrid, Vol 2, pp 1049 1054.
Pappin, J.W., Simpson, B., Felton, P.J., and
Raison, C. (1986) Numerical analysis of flexible
retaining walls. Symposium on computer
applications in geotechnical engineering. The
Midland Geotechnical Society, UK, April.
Schneider, H. R. 1997. Definition and
determination of characteristic soil properties.
Contribution to Discussion Session 2.3, XIV
ICSMFE, Hamburg. Balkema.
Schuppener B., Simpson B., Orr T. L. L., Frank
R. and Bond A. J. (2009). Loss of static equilibrium
of a structure definition and verification of limit
state EQU. Proc 2nd International Symposium on
Geotechnical Safety and Risk IS GIFU 2009, Gifu,
Japan, (editors: Y. Honjo, M. Suzuki, T. Hara & F.
Zhang, Taylor & Francis Group, London) 11 12 June,
pp 111 118.
Simpson, B and Hocombe, T (2010) Implications
of modern design codes for earth retaining structures.
Proc ER2010, ASCE Earth Retention Conference 3,
pp.786 803, Seattle, Aug 2010.
Simpson, B, Vogt, N & van Seters AJ (2011)
Geotechnical safety in relation to water pressures.
Proc 3rd Int Symp on Geotechnical Safety and Risk,
pp 501 517, Munich.
Simpson, B and Yazdchi, M (2003) Use of finite
element methods in geotechnical limit state design.
LSD2003: International Workshop on Limit State
Design in Geotechnical Engineering Practice.
8 CONCLUDING REMARKS
The paper has discussed some of the
current debates related to application of
Eurocode 7, providing tentative conclusions
where available.
For determination of
characteristic values of material properties
and of water pressures, some possible
processes have been proposed, but the need
to exploit engineering insight and expertise
has also been emphasised.
This is a
particular challenge for code drafters.
Problems affected by the loading of water
pressure, or with the balanced loads from a
single source considered by EQU, require
very careful assessment of safety. It has been
suggested that EQU should be regarded as
just another load case rather than a
different limit state.
The use of numerical analysis is expected
to become increasingly common in
geotechnical design, so it is important that
modern codes accommodate this. Current
discussions have been summarised and
approaches for analysing ultimate limit states
have been presented.
9 REFERENCES
Bauduin, C, De Vos, M & Simpson, B (2000).
Some Considerations on the Use of Finite Element
Methods in Ultimate Limit State Design. LSD2000:
Int. Workshop on Limit State Design in Geotechnical
Engineering, ISSMGE, TC23, Melbourne.
Becker, D.E. 1996. Eighteenth Canadian
Geotechnical Colloquium: Limit states design of
foundations. I: An overview of the foundation design
process. Canadian Geotech J, 33(6), 956 983.
K Cheung, K West, H C Yeow & B Simpson
(2009) Do Eurocodes make a difference? Geotechnics
and Tunnelling, Vol 3, No. 1, pp35 47. Wiley
Interscience. . (Report on Special Workshop on
Consequences of Eurocode 7 on the design of tunnels,
Austrian Society for Geomechanics, Salzburg,
Austria.)
Dahlberg, R and Ronold, KO (1993) Limit state
design of offshore foundations. Proc Int Symp Limit
state design in geotechnical engineering, Vol 2,
pp491 500. Danish Geotechnical Society.
Dahlberg, R and Ronold, KO (1993) Limit state
design of offshore foundations. Proc Int Symp Limit
state design in geotechnical engineering, Vol 2,
pp491 500. Danish Geotechnical Society.
Ellison, KC, Soga, K and Simpson, B (2012) A
strain space soil model with evolving stiffness
anisotropy. Gotechnique, in press.
24

EC7 Fundamental Issues and Its Implications On Users PDF

Uploaded by

Copyright:

Available Formats

EC7 Fundamental Issues and Its Implications On Users PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

EC7 Fundamental Issues and Its Implications On Users PDF

Uploaded by

Copyright:

Available Formats

What are the main issues discussed in the paper regarding Eurocode 7?

What are the main issues discussed in the paper regarding Eurocode 7?

What are the three alternative Design Approaches mentioned in Eurocode 7?

What are the three alternative Design Approaches mentioned in Eurocode 7?

Keynote lecture Eurocode 7 fundamental issues and some implications for users

Eurocode 7 fundamental issues and some implications for users

Simpson, B (2012) Eurocode 7 fundamental issues

The paper refers to previous publications

2 SAFETY FORMAT OF EC7

applied to resistances rather than to material

EC7 requires that both ultimate limit states

to determining characteristic properties of

3.1 Background in design practice

3.2 Definitions in EN 1990

where is a conversion factor relating

variation seen in tests on small specimens.

Figure 1. Building on estuarine beds.

Although the characteristic value is

The requirement to consider information

Figure 2. Derivations of characteristic values

Figure 3. Undrained shear strengths from

3.5 A London example

Figure 5. Comparison of results.

On the basis of these inconsistent data

Figure 4. SPT results from boreholes on site.

Figure 6. Chosen characteristic values.

Generally the strength to be used in

3.6 Which value of as characteristic?

overview will be provided here in an attempt

3.8 Further development by SC7

4.2 Requirements of EC7

4 DESIGNS DOMINATED BY WATER

situations. For serviceability limit states, the

4.4 Explicitly accommodate the worst water

4.3 Robustness allow for secondary

Figure 7. Gravity wall retaining free water.

Figure 7 shows a wall supporting water

Figure 8. Submerged anchor block.

illustrated in Figure 9. Methods 3 and 4

the bending moment in the wall is increased

Figure 9. Factored water pressures on anchor

In Figure 10, the allowable characteristic

indicating that for this problem it may not be

which the pressures above and below the

4.6 Partial factors on the density of water?

Figure 11. Gravity construction retaining

EC7 has no middle third rule for

within the middle two thirds, as illustrated in

4.7 Use of an offset in water level?

where P,t is a resistance factor on pile

Figure 12. Deep basement subject to uplift.

The uplift force beneath the basement is

where G,dst and G,stb are partial factors on the

It is also possible to consider the problem

The gradient of this line depends on the

application in a code of practice, however. If

included here because it has been a further

Figure 14. HYD in EC7

4.8 The star approach DA2* , DA1* ?