## Moderately Conservative Ec7

Bl SUMMARY OF MAIN CONCEPTS 17

Bl.l Assumptions 17

B1.2 Geotechnical categories 17

B1.3 Safety format 17

B1.4 Geotechnical investigation 19

B1.5 Design procedures 19

B2 LIMIT STATE DESIGN 20

B2.1 Definitions 20

B2.2 Basis of the method 20

B2.3 Design procedures 21

B2.4 Calculations 21

B3 DESIGN BY CALCULATION, PRESCRIPTIVE MEASURES,

TESTING AND THE OBSERVATIONAL METHOD 23

B4 CHARACTERISTIC VALUES 24

B4.1 Significance 24

B4.2 Characteristic values in Eurocode 1 and in structural design 24

B4.3 Characteristic values used in geotechnical design 25

B4.4 Characteristic values dependent on failure mode 26

B4.5 Which value - peak, critical state, residual, mobilised...? 27

B4.6 Relationship to other texts and practices 27

B4.7 Why are structural and geotechnical characteristic values different? 28

B4.8 Relationship to mean values 28

B4.9 Significance of statistical methods 29

B4.10 A London example 30

B4.ll The Johannesburg experiment 31

B4.12 Characteristic values of stiffness and unit weight 33

B5.1 Background 34

B5.2 The concept of'cases' 35

B5.3 Current requirements ofECl and EC7 35

B5.4 Reasons for the requirements 36

B5.5 Problems caused by the requirements 38

B5.6 Steel sheet pile walls 38

B5.7 Case A 39

B6 TEMPORARY WORKS AND THE OBSERVATIONAL

METHOD 40

B6.1 Consequences of failure 40

B6.2 The observational method 40

### B1 SUMMARY OF MAIN CONCEPTS Bl.l Assumptions

Although it may well be applied elsewhere, it is important to realise that EC7-1 is drafted for use in Western Europe. Assumptions which follow from this are listed in Clause 1.4, which sets the standards for good practice to be followed on every project. The factors of safety used in the code are based on this good practice. Where standards of data collection, analysis, design, construction and maintenance fall below those standards, a more conservative design approach should be followed. Continuity between each of the stages of the project is also assumed, with a free flow of information and data between the ground investigators, designers and constructors.

The design process should therefore be an unbroken continuous process, with appropriately qualified and experienced personnel carrying out each stage.

Some problems in the interpretation of these assumptions are noted in C1.4. B1.2 Geotechnical categories

A flow diagram illustrating the recommended route through geotechnical design to EC7 is shown on Figure Bl.l.

Clause 2.1 introduces the concept of Geotechnical Categories. EC7 divides structures into Geotechnical Categories 1,2 or 3 according to a number of geotechnical design requirements, principally related to the complexity of the structure and previous experience of the particular ground conditions. Most engineered structures will fall in category 2, whilst very simple designs may be in Category 1 and complex problems fall into Category 3; Figure B1.2 is a flow diagram showing the decisions required in categorisation. The categories are used in the code to indicate the degree of effort required in site investigation and design. In C2.1, it is suggested that the categories also indicate the qualifications of the personnel required for the work.

Categorisation is not a mandatory part of the code, all reference to it being in application rules rather than principles. Concern has been expressed, particularly by foundation contractors, about its legal implications.

The intention is that a preliminary classification of a structure according to geotechnical category should normally be made prior to the geotechnical investigations. The category should be checked and possibly changed at each stage of the design and construction process, as indicated by the asterisks in Figure Bl.l. The procedures of higher categories may be used to justify more economic designs, or where the designer considers them to be appropriate.

### B1.3 Safety format

In common with all the Eurocodes, EC7 is based on the principles of limit state design. The application of this in geotechnical engineering is discussed in B2. Calculations are principally to be carried out by applying partial safety factors to characteristic values of soil parameters, which are discussed further in B4. Design is not entirely based on calculation, however, and use of observation and testing is also encouraged, as noted in B3.

Figure Bl.l The EC7 design process

Figure B1.2 Geotechnical categorisation

Job Title

'New start housing development Structure Reference:

Strip foundations

Job No.

Reports used:

Ground Investigation report (give ref. date) Factual: .

Bloggs Investigations Ltd report ABC/123 dated 21 Feb 95

Interpretation:

Codes and standards used (level of acceptable risk)

Eurocode 7 Local building regs

Description of site surroundings:

Formerly agricultural land. Gently sloping (4°)

Calculations (or index to calculations)

Characteristic load 60 kN/m. Local experience plus Local

Building Regulations (ref )

indicates working bearing pressure of 100 kPa acceptable. Therefore adopt footings 0.6 m wide, minimum depth 0.5 m (Building Regs) but depth varies to reach cu 60 kPa - test on site.

### B1.4 Geotechnical investigation

In Section 3, EC7-1 provides outline requirements for geotechnical site investigation, dividing the activities into preliminary and design investigations. The preliminary investigation (EC7,3.1 and 3.2.2) corresponds to what is traditionally referred to in the UK as a 'Desk Study'. It identifies the ground related hazards which the structure will face both during construction (eg the need for dewatering) and in the permanent condition (eg the need to resist whatever onerous combination of loads is most critical). The design investigation (EC7,3.2.3) corresponds to what used to be called in the UK 'Site Investigation' and is now more properly referred to as the 'Ground Investigation'. It investigates the hazards identified in the preliminary investigation and produces design parameters which are appropriate for the geotechnical category. This process is illustrated in Figure Bl.l.

The code requires that the final design is accompanied by formal reports, both factual and interpretative, of the investigations on which it is based. The

______contractual issues which this raises are discussed in C3.4. The factual report on the design investigation is incorporated in the 'Ground Investigation Report' (EC7,3.4) which also includes evaluation and interpretation of the data. The Ground Investigation Report can be included in the 'Geotechnical Design Report' (EC7,2.8) which also includes design assumptions, design calculations and the plan for site supervision and monitoring required by the design.

Clause 3.3 also gives requirements for the process of evaluating the main ground parameters used in calculations (generally characteristic values for EC7). Methods of carrying out field and laboratory tests are not described in EC7 Part 1, but in Parts 2 and 3 (see A2.5).

Made by:

Date.

Checked by:

Date.

Approved by:

Date.

Section through structure showing actions:

Assumed stratigraphy used in design with properties:

Topsoil and very weathered glacial till up to 1 m thick, overlying firm to stiff glacial till (c 60 kPa on pocket penetrometer).

Information to be verified during construction. Notes on maintenance and monitoring.

Concrete cast on un-softened glacial till with cu 60 kPa (pocket penetrometer)

Figure B1.3 Single page geotechnical design report

### B1.5 Design procedures

The later sections ofEC7-l consider the design of some specific types of foundations and other geotechnical structures. The code generally does not specify the precise form of calculations to be used, but states what criteria are to be checked by the calculations.

Clause 2.8 requires that assumptions, data, calculations and results of the verification of safety and serviceability must be recorded in a Geotechnical Design Report. The level of detail of Geotechnical Design Reports will vary greatly, depending on the type of design. For simple designs, a single sheet may be sufficient. An example of such a sheet is given in Figure B1.3.

The scope of the report includes a plan of supervision and monitoring, as appropriate, and the clause requires that relevant parts of the report must be provided to the client.

### B2 LIMIT STATE DESIGN B2.1 Definitions

Limit state design is a procedure in which attention is concentrated on avoidance of limit states. Limit states are defined as 'states beyond which the structure no longer satisfies the design performance requirements' (EC1, 3.1 (1)P). Strictly, it is the exceedence of a limit state which is not acceptable, though EC7 often refers to avoiding the occurrence of a limit state.

This definition of limit states is essentially practical and relates to the possibility of damage, economic loss or unsafe situations. It is not directly concerned with states of stress in materials or distinctions between elastic and plastic behaviour, though designers may need to consider these in order to demonstrate that limit states will not be exceeded.

Limit state design is concerned with any state in which a structure does not satisfy the design performance requirements. For example, cracking or distortion which has no more consequence than giving a disappointing appearance constitutes a limit state, just as does a catastrophic collapse. The severities of these two limit states are obviously very different.

It has been found convenient to categorise limit states as ultimate or serviceability limit states. EC1 defines ultimate limit states as those associated with collapse or with other similarforms ofstructural failure (3.2 (1) P). Serviceability limit states correspond to conditions beyond which specified service requirements for a structure or structural element are no longer met (3.3 (1) P). The serviceability requirements should generally be determined in contracts and/or in the design (3.3(4)).

EC1, Section 3 adds more detail to this description of ultimate and serviceability limit states. For geotechnical design, it is important to note that ultimate limit states includefailure by excessive deformation,... loss of stability of the structure or any part ofit. Hence, a state in which part of a structure becomes unsafe because of foundation settlement or other ground movements should be regarded as an ultimate limit state, even if the ground itself has not reached the limits of its strength, to form a plastic failure mechanism. For example, large amounts of heave of plastic, over-consolidated London Clay have occurred over long periods following the removal of trees. While there is no question of the ground strength having reduced, to the degree that bearing capacity failure is approached, the movements have been large enough to induce collapse in a building, following loss of bearing in lintels over window and door openings. The application of this concept to retaining walls is noted in EC7,8.4(2).

Limit states are generally checked by considering design situations, in which adverse conditions apply; design values, which are deliberately pessimistic, are used for both loads and material strengths. Design values are used in calculations for both ultimate and serviceability limit states, though the values will usually be different for the two states. The Eurocodes specify how design values are to be derived. The design values required for serviceability limit states are often equal to the characteristic values of parameters (formally, a partial factor of 1.0 is applied), but there is no fundamental reason why this must always be so. EC7 states that SLS design values will normally equal characteristic values for actions in 2.4.2(18) and for materials in 2.4.3(13).

### B2.2 Basis of the method

The definitions considered above show that limit state design is concerned with what might go wrong. Attention is concentrated on states which it is intended will not occur rather than on what, it is hoped, will actually happen. Occasional exceedence of serviceability limit states might be economically tolerable, but generally ultimate limit states must be avoided. Thus the design should have appropriate degrees of reliability (EC1,2.1(1)P).

The aim of limit state design is to avoid limit states in general, and to make very remote any possibility of an ultimate limit state. Ultimate limit states are intended to be unrealistic possibilities. Hence, in calculations, the codes sometimes require the adoption of design values for parameters which are unrealistically pessimistic.

It may be questioned whether there is anything distinctive about limit state design, or whether the definitions are so broad that they incorporate all design processes. This is particularly relevant in geotechnical design, where, historically, there has been more consideration of plastic failure mechanisms -undesired states - than on working states of elastic stress. The distinctive feature of limit state design is essentially one of emphasis, with attention concentrated on what might go wrong.

Limit state design is sometimes contrasted with permissible stress design in which attention is concentrated on prediction of the stresses in materials in the intended working state. This terminology becomes confused if permissible stresses at the limit states are considered - and there is no logical reason why these should not be used. Hence it is preferable to contrast limit state design with working state design.

Some of the pros and cons of limit state design have been discussed by Simpson (1997).

B2.3 Design procedures

The basic limit state design procedure has two stages: a set up design situations;

b show that limit states will not be exceeded in the design situations.

EC1, Clause 2.3 states: The selected design situations shall be sufficiently severe and so varied as to encompass all conditions which can reasonably be foreseen to occur during the execution and use of the structure. Design situations are categorised as persistent, transient and accidental situations, and the limit states relevant to the various situations may vary. For example, for an accidental situation, which involves exceptional conditions, the structure may be required merely to survive without collapse; in this case serviceability limit states would not be relevant. More information on design situations may be found in EC1,2.3 and EC7,2.2.

The limit state method does not restrict the means by which it may be demonstrated that limit states,will not be exceeded in the design situations. Often, calculations will be used for this purpose, but other approaches provide alternatives or supplements to design by calculation. These include load testing at full scale or on models, which is particularly relevant to design of piles and ground anchors (EC1, Section 8 and EC7,2.6); prescriptive measures, in which well-established details are adopted without calculation (EC7,2.5); and the Observational Method (EC7,2.7). These methods are discussed further under the relevant clauses in Part C.

The definition of serviceability limit states often requires the specification of a limiting value of displacements or strain. It is essential that this is realistically assessed as values representing an unacceptable condition. Unnecessarily severe values may lead to highly uneconomic design.

### B2.4 Calculations

Historically, the limit state method became popular at about the time that partial safety factors began to be adopted. The two are therefore often linked, though there is no fundamental connection between them. A calculation using a global factor of safety or directly assessed pessimistic design values could be sufficient to demonstrate that limit states will not occur. It was noted above that the limit state method does not necessarily require calculations as the basis of design.

Limit state calculations are usually carried out by showing that the design properties of materials are sufficient to withstand the design values of all applied actions (ie loads - see A2.8). The design values generally incorporate all the required safety elements, with no further overall factor of safety. Generally, design values of parameters, Xd, are derived from characteristic values, Xk, by applying partial factors y: for actions: Fd = Fk x y for materials: Xd = Xk / y

The derivation of design values by applying partial factors to less pessimistic characteristic values provides a means by which codes of practice can exert some influence over the degree of pessimism of the design values. The concept of'characteristic' values in geotechnical engineering is discussed in B4.

The limit state design method requires that all possible limit states are considered and eliminated, with 'appropriate degrees of reliability'. In general, this will at least mean that ultimate and serviceability limit states must be considered. For geotechnical design, this puts an increased emphasis on the need to consider deformations, but EC7 aims to discourage excessive or spurious attempts to calculate displacements.

The purpose of partial factors is generally stated to be to allow for uncertainties and inaccuracies in the values of the parameters (EC1,9.3.1 and 9.3.3). Some authorities deduce from this that the values of the partial factors may be derived from statistical studies of these uncertainties. In this approach, the factors used for ULS design have no bearing on the serviceability of the structure. This contrasts with the use in BS 8002 of a 'mobilisation factor', which is effectively a partial factor, but its stated purpose is to prevent stress levels in materials reaching a point at which displacements become unacceptable; that is, the factor's role is mainly in serviceability.

Referring to partial factors on actions, Eurocode 1 takes a broader view in 9.4.3 (3): The values have been based on theoretical considerations, experience and back calculations on existing designs. The calibration of the ULS factors based on experience and back calculations will necessarily mean that their values make some provision for serviceability as well as ultimate requirements. EC7 notes in 2.4.1(7) that in some situations it is necessary to use factors applied in the analysis of one limit state in order to cover another, for which calculations are not reliable.

In the authors' opinion, this pragmatic approach to the use of partial factors is realistic. The factors adopted are inevitably calibrated against previous designs and therefore make some provision for serviceability as well as ultimate safety. Where EC7 makes additional requirements for checks on serviceability, these should be followed, however.

B3 DESIGN BY CALCULATION, PRESCRIPTIVE MEASURES, TESTING AND THE OBSERVATIONAL METHOD

The fundamental design requirements for limit state design are set out in EC7, 2.1. In 2.1(7), it is stated that these requirements may be achieved by: a use of calculations; b adoption of prescriptive measures; c experimental models and load tests; or d an observational method.

It is clear that design based on calculation is not the only process envisaged. The same paragraph says that these four approaches may be used in combination. This often forms the basis of good geotechnical engineering.

Prescriptive measures (EC7,2.5) involve conventional and generally conservative details in the design, and attention to specification and control of materials, workmanship, protection and maintenance. They may be used when calculations are not available or not necessary. They could be used for design for durability, for example, and will often be based on the observed performance of existing structures. More generally, they might be used to make a quick, conservative design in cases where the cost of extensive site investigation and analysis cannot be justified. In Hong Kong, the Geotechnical Engineering Office is preparing to publish a series of recognised prescriptive measures for stabilisation of small slopes, for example.

Design of piles and ground anchors has traditionally been based very largely on load testing. This is in the category design by experimental models and load tests, in which confidence in the safety of the design depends on test results, either in place of or in combination with calculations. This use of test results in design is discussed further in EC7,2.6,7.5-7.7 and 8.8, and in E7, E8 and E12. EC7 mentions the use of model testing in 2.6, but does not enlarge on this.

EC7,2.7 is a specific clause about the Observational Method, which has received a great deal of support from the geotechnical community. It is used in recent publications Safety of New Austrian Tunnelling Method (NATM) Tunnels, by HSE and the CIRIA report on the Observational Method (Nicholson et al (1997)). Since the Observational Method relates mainly to the design of temporary works, it is considered further in B6.

### B4 CHARACTERISTIC VALUES B4.1 Significance

Characteristic values of geotechnical parameters are fundamental to all calculations carried out in accordance with the code. Their definition, in geotechnical terms, has been the most controversial topic in the whole process of drafting Eurocode 7. Some of the more difficult issues will be addressed here. More straightforward matters will be left to Part C. The most important text is in EC7,2.4.3.

Two factors underlie the importance and controversy of characteristic values.

a Calculations are to be carried out by applying partial safety factors to characteristic values in order to obtain design values of parameters. The partial factors are specified by the code, so the selection of characteristic values is the main point in calculations at which engineers are to apply their skills and judgment, with the possibility of dangerous mistakes, b Engineers have always had the responsibility for selecting values of material parameters for calculations. This process has sometimes been referred to as a 'black art', and it is difficult to find helpful advice on the thought processes necessary to derive appropriate values from site investigation and other information. In particular, the degree of conservatism necessary in choosing values for design purposes is rarely discussed.

Eurocode 7's definition of characteristic values is intended to make full use of the skills and judgment of experienced engineers, whilst helping less experienced engineers to choose values which are both reasonable and safe. This was, and remains, a major challenge.

B4.2 Characteristic values in Eurocode 1 and in structural design

Characteristic values, as used in Eurocode 7, are intended to comply with Eurocode 1 as far as possible, whilst remaining true to principles of sound geotechnical engineering. Although it arguably remains within the spirit of Eurocode 1, the definition adopted for geotechnical purposes differs from that of Eurocode 1 in some important respects. To understand this, it is necessary first to consider what Eurocode 1 says about characteristic values, Xk.

Eurocode 1, Subclause 9.3.3 states: The design value Xd of a material or product property is generally defined as: Xd = TlXk/YM»rXk/YM where:

yM is the partial factor for the material or product property, given in ENFs 1992 to 1999, which covers:

• unfavourable deviations from the characteristic values;

• inaccuracies in the conversion factors; and

• uncertainties in the geometric properties and the resistance model r| is the conversion factor taking into account the effect of the duration of the load, volume and scale effects, effects of moisture and temperature and so on. Characteristic values are introduced in Eurocode 1 Section 5 thus:

(1)P Properties ofmaterials (including soil and rock) or products are represented by characteristic values which correspond to the value ofthe properly having a prescribed probability of not being attained in a hypothetical unlimited test series. They generally correspondfor a particular property to a specifiedfractile ofthe assumed statistical distribution ofthe property ofthe material in the structure.

(2) Unless otherwise stated in ENFs 1992 to 1999, the characteristic values should be defined as the 5% fractile for strength parameters and as the mean value for stiffness parameters.

Note: For operational rules, see annex D,forfatigue, information is given in annex B.

(3)P Materialproperty values shall normally he determined for standardized tests performed under specified conditions. A conversionfactor shall be applied where it is necessary to convert the test results into values which can be assumed to represent the behaviour of the material in the structure or the ground (tee also ENVs 1992 to 1999).

This text specifies the following features:

a Characteristic values take account of the statistical distribution of the property. That is, the range of uncertainty of the property is relevant to their selection.

b They can normally be derived by a statistical process applied to a series of tests on specimens of the material. However, in principle they relate to a hypothetical, unlimited test series, so some correction may be required when test series are limited, c For strength properties, they are to correspond to the 5% low fractile of the test results; this is the strength below which 5% of test results fall, d Nevertheless, the characteristic values are said to represent the behaviour of the material in the structure or the ground, and corrections to test results may be needed in order to achieve this, e For stiffness, mean values are to be used. This is considered further in B4.12.

These definitions of characteristic value are clearly intended to be general. Eurocode 1 does not at this point mention the mode of failure or type of limit state being discussed, or the severity ofits consequences.

In structural design, characteristic values are generally defined using statistical procedures applied to the results of tests on material specimens. The specimen is generally not obtained from the structure and its relationship to material in the structure depends more on control of workmanship than on the designer's observation or judgement. In this respect, the definition of characteristic value for ground materials given in Eurocode 7 is distinctly different.

B4.3 Characteristic values used in geotechnical design

In Eurocode 7, the characteristic values of geotechnical material parameters are based on an assessment of the material actually in the ground and the way that material will affect the performance of the ground and structure in relation to a particular limit state (EC7,2.4.3(2,3,4)). Field and laboratory tests are to be used, but they are only one means of assessing what is in the ground; characteristic values are not derived directly or solely from the test results. Statistical manipulation of test results will generally have only a minor role in this process, if any. The resulting value is inevitably subjective to some extent, being influenced by the knowledge and experience of the designer. However, this is considered preferable to an alternative, mechanical approach which has arithmetic objectivity but jettisons established engineering knowledge.

In many situations, the known geology of a stratum, and existing experience of it give a fairly good indication ofits parameter values. Soil tests are used as a check. It is good practice to base the selection of characteristic values on a combination of well established experience and the test results (EC7,2.4.3 (2,4)). If unusually good test results are obtained, engineers will normally spot this and treat them with greater caution, unless further investigation is possible to establish that they are relevant. Unusually bad results may lead to further investigation, or may otherwise be taken at face value unless the evidence of other experience is overwhelming.

Construction activities may affect the properties of the ground, adversely or beneficially (EC7,2.4.3(4)). Common examples occur during boring or driving of piles, or excavating to a level on which concrete will be cast. In many cases this will occur after any investigation and testing are complete. Nevertheless, the characteristic value is to account for these construction effects. Information from previous experiences and publications will contribute to the selection of characteristic values in these circumstances.

Having reviewed these items, EC7 says that the characteristic value of a soil or rock parameter shall be selected as a cautious estimate ofthe value affecting the occurrence of the limit state (EC7,2.4.3(5)). This is standard engineering practice. The relationship of the 'cautious value' to mean values will be considered in B4.9 to B4.ll below.

B4.4 Characteristic values dependent on failure mode

The characteristic value of one parameter in one stratum is not necessarily the same for two different failure modes. It may depend on the extent to which a particular mode averages out the variable properties of the stratum (EC7, 2.4.3(4,6)).

Figure B4.1 shows a small industrial building, founded on pad footings near a long slope. The underlying materials are estuarine beds, mainly of sands with some impersistent lenses of clay occurring at random. In this type of situation, the design of the footings would probably assume that they might be founded on clay, the more adverse condition for foundation design.

Figure B4.1 Small building on estuarine beds near slope

(An alternative could be, in some cases, to require an inspection and probe at each footing, so avoiding this adverse condition.) On the other hand, when the possibility of a large slip surface is considered, it is inconceivable that this will lie entirely, or even mainly in clay. In this type of situation, the characteristic values for strength parameters of the beds would be different for the footing design and for the slip, though their safety is controlled by the same stratum in both cases.

Figure B4.2 shows results of a CPT test in a mixed, estuarine deposit which has been overconsolidated, variably, by desiccation. A piled foundation is to be constructed in this material. If the piles are of fixed length (perhaps limited by construction equipment), the characteristic values of soil strength for the base and shaft may be quite different. The shaft averages the properties of a large amount of material, from many periods of deposition, whilst the base could be formed in one of the weaker layers. In this case the characteristic values of soil strength for the shaft would be higher than that for the base, in the same deposit. On the other hand, if the construction process allows the base to be tested, by pile driving for example, the characteristic value for the base could be higher that the averaged value used for the shaft. This discussion must also be modified to take account of any systematic variation of strength with depth.

Cone Resistance (MPa)

Cone Resistance (MPa)

Figure B4.2 CPT results in variable deposit

B4.5 Which value - peak, critical state, residual, mobilised...?

The question has been asked: Which value is the characteristic value? It is sometimes necessary to chose from one of the following, depending on circumstances:

a peak, critical state or residual shear strength; b ultimate strength or a 'mobilised' value; c strength of intact material or strength on joints; d strength at first loading or after repeated loading; e stiffness of intact rock or of the jointed material; f stiffness on first loading, or on unload-reload.

In all cases, the answer ofEurocode 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 (EC7,8.5.1(4)). In considering rocks, a study of the joint patterns will determine whether intact or joint strength is relevant (EC7,3.3.9).

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 EC7,2.1(9) and C2.1.

Generally the strength to be used in Eurocode 7 is the maximum available to prevent collapse, not a mobilised value.

### B4.6 Relationship to other texts and practices

With regard to characteristic values, the intention of the drafters of EC7 was to clarify existing practice, rather than to introduce something new. The main problem was the difficulty of defining existing practice. Nevertheless, some texts give helpful indications of the way in which parameter values are to be chosen and it is relevant to compare these with characteristic values.

CIRIA Report 104 suggests that design may be based on moderately conservative values of parameters. 'Moderately conservative' is defined (p 40) as meaning conservative best estimate. It could be objected that the latter term is contradictory, since a value cannot be both conservative and a best estimate simultaneously. CIRIA 104 states that this approach is used most often in practice by experienced engineers. The authors consider that the conservative best estimate values of CIRIA 104 are essentially the same as the characteristic values of EC7.

In BS 8002, design values of soil strength (ie values entered into calculations) are derived by factoring representative values. For effective stress parameters, there is a further requirement that the design value must not exceed the representative critical state value. A representative value is defined (1.3.17) to be a conservative estimate of the mass strength of the soil. 'Conservative values' are further defined (1.3.2) as values of soil parameters which are more adverse than the most likely values. They may be less (or greater) than the most likely values. They tend towards the limit of the credible range of values. The authors suggest that this definition makes representative values essentially the same as moderately conservative values in CIRIA 104 and characteristic values in EC7.

The Dutch standard NEN 6740 (in Dutch) provides a more statistical approach to derivation of characteristic values. German recommendations for waterfront structures (EAU (1980, p38)) discuss the statistical background to characteristic values, and also provide some more pragmatic suggestions: When a large number of shear parameters have been determined, the characteristic value can also be estimated as being that value which occurs immediately below the mean of all tests made... With only three determined values, which have been obtainedfrom three samples of the investigated layer taken at well separated locations, the lowest value may also be used as the characteristic value if the values do not differ too much from one another.

B4.7 Why are structural and geotechnical characteristic values different?

The designer of a structure is concerned with the properties of materials which generally do not exist at the time of design, but which can be specified with fair precision. The range of uncertainty of their properties is fairly well known, and, in many cases, may be better understood by the drafters of codes than by designers in practice. Hence, it is appropriate that codes give specific rules about the measurement of characteristic values and that the possible range of uncertainty is entirely accommodated in factors prescribed by the code writers.

In geotechnical design, however, the designer is in possession of information not available to the code drafters. He knows where the site is located, what is its geology, and he has test results, relevant publications, observations of nearby constructions, and so on. The designer is therefore in a much better position than the code drafter to make allowance for the range of uncertainty of the parameter values. It is this extra information which Eurocode 7 requires the designer to incorporate in his selection of characteristic values.

### B4.8 Relationship to mean values

EC7 says that the characteristic value of a soil or rock parameter shall be selected as a cautious estimate of the value affecting the occurrence of the limit state (EC7,2.4.3(5)). The probability that the characteristic value will, in fact, prevail in such a way as to govern the occurrence of a limit state is fairly remote, nominally 5%.

It has been suggested that the characteristic value should be defined to be a mean value. Unfortunately, there is some confusion about different meanings of the word 'mean'. For the purpose of this discussion, three mean values will be defined: statistical, spacial and probabilistic.

a A statistical mean will be taken to be the simple average of established data. These could typically be test results, adjusted where necessary to allow for differences between the test and field situation, b A spacial mean is the average of a parameter over some space. This could be the volume which is compressed under a load or the surface over which a slip might occur. Many limit modes are governed by the average performance of such a volume or surface, and for these a spacial mean of the parameter value is appropriate. The decision to use a spacial mean does not dictate the degree of pessimism which may be attached to the chosen value.

C A probabilistic mean is a value, taken from a range of uncertainty, such that the value which will actually be found to govern the limit mode has a 50% chance of being worse than the probabilistic mean. Most often, this probability must be assessed by the engineer in advance of the actual events. One advantage of using a probabilistic mean is that it is equal to the statistical mean value of a set of relevant test results, provided they have been adjusted for. any difference between the behaviour of the soil in test and in situ.

In many situations, the characteristic value required by EC7 should be a cautious assessment of a spacial mean. If there is to be, in fact, a 5% chance that a worse value will govern field behaviour, then the cautious spacial mean will be much less pessimistic than the 5% fractile of relevant, adjusted test results. This reflects the fact that many limit modes average out the variabilities of a lot of ground.

## Post a comment