Reliability Assessment of Structural Concrete with Special Reference to Shear Resistance

By Kenneth Kwesi Mensah

PhD Research Proposal

Promoters: Dr. C. Barnardo-Viljoen & Prof. J.V. Retief

March 2012

Table of Contents
1. INTRODUCTORY COMMENTS 1

2. INTRODUCTION 2.1 Concept of basis of design and codification 2.2 Uncertainties in structural design and structural reliability 2.2.1 Aleatoric and Epistemic uncertainties 2.2.2 Quality control 2.3 Research motivation and significance 2.4 Research objectives 2.5 Work done in M-thesis 2.6 Expected outcomes of the research (Hypothesis) 2.7 Research contribution

2 2 2 4 4 5 6 8 10 12

3. LITERATURE OVERVIEW OF THE STUDY 3.1 Background of the reliability basis for structural design 3.2 The Problem of shear and its reliability basis 3.3 EC 2s variable strut inclination design method for shear 3.4 The MCFT 3.5 Reliability analysis of EC 2s variable strut inclination method for shear 3.6 Parametric analyses

14 14 18 20 21 22 23

4. RESEARCH METHODOLOGY 4.1 General Methodology 4.2 Methodology for reliability analysis and calibration of variable strut inclination method

24 24

26

5. PROPOSED RESEARCH PROGRAM

28

REFERENCES

30

ii

1. INTRODUCTORY COMMENTS

This research represents a continuation and extension of reliability-based investigations assessing the appropriate application of modern basis of design formats in deriving design guidelines for structural concrete resistance. The study began at masters level, from which a masters dissertation was submitted in December 2011 and has been recommended by the reviewers for an upgrade to a PhD at the University of Stellenbosch. For all purposes in this proposal, the masters dissertation will be referred to as the M-thesis. The objectives of the PhD research are an extension and continuation of those considered in the M-thesis to further harmonise the application of reliability principles in deriving design guidelines for structural concrete, particularly in South Africa. However, following

preliminary reliability analyses conducted in the M-thesis, the specific objective for this research of calibrating the EC 2 design method for the shear of members requiring stirrups arises. Assessments of the applications of the principles of structural reliability continue in an attempt to identify ways in which the process can be advanced, not only for South Africa but on an international platform as well. In the M-thesis, the terms EC 0 and EC 2 were established as convenient short forms for the Eurocode Basis of Structural Design Standard EN 1990 (EN 1990, 2002) and the Eurocode Standard for the Design of Concrete Structures EN 1992-1-1 (EN 1992-1-1, 2004), respectively. These conventions are maintained for use in this proposal.

2. INTRODUCTION

2.1 Concept of basis of design and codification The structural engineering fraternity has the social responsibility to ensure that all structures designed and constructed are safe and, further, perform as expected during service. To be able to achieve safe and durable structures, design and construction professionals require a system of verifying adequate structural performance of buildings and structures by applying rational and safe procedures through the different stages of a project: planning, design, analysis, detailing, construction, and maintenance of structures. This is typically done

through a code of practice with a well-established design basis specifying procedures and guidelines that enable assessments of structural performance and safety. Guidelines are therein also given to achieve certain levels of performance through detailing rules and quality management provisions for design and construction. A design code, or design provisions in general, should represent sound and well-established methods of engineering practice that have been thoroughly researched and validated by relevant experience (Ellingwood, 1994). In deriving code provisions, they should be

calibrated extensively to validate their use across the field of application in practice. Provisions should however not be too complex to use by design engineers in practice who do not always have time to study innovative trends in research, and are often under time demands of projects. A code is therefore a platform of disseminating efficient and current methods of design and construction between research and practice.

2.2 Uncertainties in structural design and structural reliability In general, problems of structural design must be resolved in the face of various uncertainties. Uncertainties arise not only in the assessment of actions which the structure has to sustain, and from the occasional lack of control during the production processes of the materials and components required, but also from incomplete knowledge about the mechanical formulations describing the response of the structure and its capacity to sustain those actions. Structural reliability techniques, compared to other basis of design formats, are aimed at rationally quantifying and assessing the effects of uncertainties associated with all aspects of
2

structural design. The uncertainties in the design and construction process are represented by way of mathematical statistics and the assessment of structural performance is conducted through probabilistic concepts and analyses. Such treatment of uncertainties gives a rational scientific (decision tool) approach to the calibration of structural design provisions.

Modern and technologically advanced design codes adopt the Partial Factor Limit States Design Method as their basis for design. This method applies partial factors, the vector ,

to increase action values as well as reduce material property and resistance values to generate their design values for use in a limit state assessment. Characteristic values, the vector , are also introduced into limit state functions where partial factors are applied to The governing

make an economically but safe assessment of structural performance.

condition of a limit state assessment is that the action effects should be less than the available resistance. In this method, dimensions are generally implemented at nominal values, but in some cases (second-order effects, geometrical imperfections, buckling) can assume design values by applying some tolerance limit. This method can account for the variability of materials by applying partial safety factors to the material properties. Further, it can also be used for safety verification of cross-sections and members as well, since the action effects and resistance force of cross-sections are calculated for use in the limit state verification. Until recently, partial safety factors used in limit state design verifications were derived mainly by expert judgement and by reference to sound traditional designs, thereby lacking the appropriate rational and scientific treatment they require. Structural reliability techniques arise as an attempt or method to represent variability and performance of physical models of structural systems by taking account of the distributions of the basic variables in mechanical formulations used for limit state verifications. Basic variables are the most fundamental quantities the designer has to consider in mechanical formulations. Structural reliability techniques are consistent with the Partial Factor Limit States Design format in the sense that partial factors can be derived from reliability analyses and calibration exercises and then applied in limit state verifications. The application of structural reliability as the theoretical basis for limit states design ensures that improved economic performance is achieved together with improved safety performance across a wide range of practical design situations. The design provisions of the suite of structural Eurocodes are formulated on reliability principles.
3

2.2.1 Aleatoric and Epistemic uncertainties Other uncertainties, apart from those associated with the prediction of action effects and resistance, affect structural performance. Aleatoric uncertainties arise due to the natural or inherent variability in a physical process which may never be determined with accuracy. It is simply a random uncertainty we may have to deal with but try to control through efficient design practice. Epistemic uncertainties are more systematic. They are due to one of two reasons: 1. Either due to insufficient knowledge or lack of understanding that causes some aspect to be constantly overlooked, 2. Or due to conservative assumptions and simplifications made, which are derived from extensive research to make final design equations manageable and relatively quick to manipulate for safe and efficient design practice. Regardless of any of the sources of epistemic uncertainties bulleted above, they can be quantified and subsequently calibrated against to build sufficient conservatism into design procedures. Model uncertainty is an epistemic uncertainty.

2.2.2 Quality control Structural failures are not only caused by the unfavourable uncertainties that affect a limit state assessment. Gross errors are, in fact, found to be the major cause of structural failures. Table 2.1 shows the origins and causes of structural failures.

Table 2.1. Origin and causes of structural failure (ISO workshop, 2011) Design Execution Use Origin 20% 50% 15% Others 15%

Causes

Gross errors 80%

Adv. cond. 20%

Gross errors can be limited by quality control during design and construction, as well as through routine maintenance. An important part of assuring reliability is to give guidelines
4

on quality management that aim primarily to reduce gross errors in design and increase the quality and integrity of constructions.

2.3 Research motivation and significance The European Commissions initiative to harmonise technical barriers between EU member states to allow exchange of information and intensify trade relations has caused each member states national structural design standards to be replaced by a unified set of Eurocodes. In this transformation, The British Standard The Structural Use of Concrete BS 8110-1 (BS 8110-1, 1997) on which the currently operational South African standard The Structural Use of Concrete SABS 0100-1 (SABS 0100-1, 2000) is based, is being withdrawn and replaced by a new operational Eurocode Standard for Design of Concrete Structures EN 1992-1-1 (EN 1992-1-1, 2004). For the on-going revision of South Africas standard for the design of concrete structures, which will be newly referred to as SANS 10100-1, the South African Concrete Code Committee has chosen to adopt EN 1992-1-1 as reference. The application of the principles of structural reliability to establish a standardised basis for structural design using partial factor limit states design procedures is done in the European Standard for the Basis of Structural Design EN 1990 from which it is adapted to the South African Basis of Design Standard for Building and Industrial Structures SANS 10160-1 (SANS 10160-1, 2010). The basis of design requirements stipulated in EN 1990 and SANS 10160-1 apply to all aspects of structural design: This includes reliability levels of structural performance and their differentiation and management; identification of various limit states and design situations; the specification of all the basic variables; separate treatment of actions and material-based resistance. However, application of these requirements is then primarily focused on actions whilst the provision for structural concrete is then left to the materials based design standards. The Eurocodes can be viewed as a general set of reference standards which need to be made operational as national standards through the selection of Nationally Determined Parameters in National Annexes. A key parameter for which national choice is allowed and has grave effect on matters concerning reliability is the selection of the target level of reliability, of which the Eurocode recommends a value of 3.0. Retief and Dunaiski (2009) propose that the reliability assessment of a future South 3.8 and South Africa uses a value of

African concrete standard could therefore consist firstly of reviewing the degree to which EN 1992 complies with and applies reliability principles as set out in EN 1990; and secondly to calibrate it in accordance with SANS 10160-1 requirements, including required levels and classes of reliability for the restricted scope of building structures.

2.4 Research objectives The general objective of this research, advancing from the masters, is to systematically study and trace the extent that EN 1990 and EN 1992-1-1 are harmonised in terms of the reliability based framework. To continuously and systematically achieve such objective, the reliability framework and its requirements are first identified in EN 1990 and other basis of design documents such as the JCSS Probabilistic Model Code (JCSS, 2001) and the Draft 2010 fib Model Code (fib, 2010). Thereafter, the nature and extent of the implementation of the reliability framework for structural concrete resistance is traced by studying the provisions of EN 1992-1-1 as well as relevant background documentation. The background documentation concerning the reliability basis of EC 2s design provisions may be found in documents such as the Eurocode 2 Commentary and Worked Examples (European Concrete Platform, 2008) and various fib and CEB-bulletins. During the harmonisation process, it is imperative that an assessment is made of the implications for South Africa where national choice is allowed. Further, where abstraction or incompleteness in the implementation of the reliability framework is identified in structural concrete provisions, improvements or suggested actions to harmonise design practice are recommended. Such efforts were made in the M-thesis concerning some quality aspect of reliability management. In the M-thesis, requirements of the framework were also exercised through extensive assessment of the model factor and reliability performance of the provisions for members requiring design shear reinforcement. objectives of the PhD research can be outlined as: Therefore, the specific

1.

To continually map out and study the reliability framework and requirements as presented in EN 1990.

2. To continually trace the extent to which the reliability framework is implemented in deriving the EN 1992-1-1 design provisions by use of relevant references and background documents.

3. To extend the reliability framework where abstraction or incompleteness is found in provisions for structural concrete resistance.

There is lack of evidence that the variable strut inclination method for the shear design of members requiring stirrups adopted in EC 2 is properly calibrated. A preliminary reliability investigation for shear was conducted in the M-thesis. It was motivated by the fact that the modelling factor associated with the shear prediction model is excessively conservative, coupled with some very inconsistent behaviour at varied amounts of shear reinforcement provided in design. The results of the

preliminary reliability investigation indicate that further characterisation and subsequent calibration of EC 2s shear design method is necessary. The primary objective of the PhD research is to fully calibrate the variable strut inclination method to both SANS and EC 0 reliability requirements.

The Partial factor modification scheme, particularly reduction, prescribed for use in an EC 2 Annex, has not been harmonised with the defined differentiation scheme warranting such action as set out in EC 0. In the M-thesis, a link was established between the two and a reduction scheme has been proposed for use in materials codes reflecting the requirements set out in EC 0. Further developments and examples of how this framework could be applied in practice are warranted.

4. To present and publish the research innovations and findings at conferences and in journals. This will serve to disseminate and impart advancements and motivate similar such action for other modes of resistance and materials, particularly in South Africa. Table 2.2 below shows a list of papers that are currently being conceptualised and planned for submission as Journal publications. Sufficient research has thus far been undertaken to publish most of the papers outlined in Table 2.2. These papers are therein indicated to be based on the M-thesis, whilst the fourth paper will require completion of the PhD study.
7

Table 2.2. List of possible Journal publications No. 1 Title of paper Review of Reliability Basis of Structural Design (RBoSD) and its application in South Africa 2 Model uncertainties: Characterisation and implications for reliability modelling 3 Reliability analysis EC 2's variable strut inclination design method for shear of members requiring stirrups 4 Reliability calibration of EC 2's design method for members with stirrups PhD M-thesis M-thesis Based on: M-thesis

2.5 Work done in M-thesis The thesis studied the principles of reliability presented in EC 0 and more importantly, reviewed their level of implementation in deriving EC 2s guidelines. The investigation identified that: 1. Model uncertainties are important and deserve proper treatment and characterisation in reliability modeling

2. Annex A in EC 2 which allows for partial factor reduction given certain quality requirements is consistent with the reliability differentiation framework in EC 0 that allows and guides such reduction. Action was taken in the thesis to: 1. First determine the model factor of the variable strut inclination method for shear to a compiled database of 222 tests. The statistics of the model factor were then

determined to ready its use for reliability modeling. The model factor associated with the Modified Compression Field Theorys (MCFT) prediction of shear resistance for members requiring design shear reinforcement was also determined by comparison to a subset of 116 tests. The prediction quality of MCFT was not a

primary objective, but was applied in independent reliability modelling of design situations as step of validating previous obtained results.

2. To generate the common reliability model where the design method is converted for use as the general probabilistic model (gpm). In reliability modeling, the gpm serves as the true descriptor of shear resistance where any conservative bias incorporated for use in design are omitted such as the use of partial factors, characteristic values and some simplifications or modifications. Mean values of the basic variables are used in gpms. An effective tool of comprehensive calculation steps set to characterise the reliability of EC 2s calibrated variable strut inclination design method were established. First, extensive reliability modelling of EC 2s design method for shear is considered paying attention to all basic variables that contribute to shear reliability performance. Further analysis then

condensed the process to highlight the most dominant, hence important, basic variables affecting reliability performance. Reliability models provide resourceful insights that aid decision making and ultimately calibration. However, European models of basic variables published in literature (JCSS, 2001; Holick, 2009) are mostly used in the thesis due to lack of availability of similar models based on South African practice and standards of workmanship and quality. Due to time constraint, reliability modelling was conducted for only two important and critical design situations. In one instance, a section representative of a design situation with low amounts of shear reinforcement (Test Case 1) was investigated whilst, conversely, the other contained a relatively high amount of shear reinforcement (Test Case 2). It was found that: 1. In general, the reliability of the design situations considered was

acceptable/satisfactory, particularly according to SANS 10160-1 requirements

2. The reliability of test case 2 did not meet EC 0s reliability requirements although it did not fall alarmingly below the threshold value of 3.04

3. In order to validate the initial reliability model, the more accurate and rationalscientific MCFT was applied in reliability modelling. Using the program Response2000, the MF associated with the MCFT was determined by comparing its predictions to a subset of 116 tests.

4. It was found that a more severe assessment of reliability results from the conventional process of using the design prediction model as gpm as opposed to using the more accurate MCFT.

2.6 Expected outcomes of the research (Hypothesis) The expected outcomes of this research are discussed as bulleted arguments below. 1. More extensive reliability investigation for shear considering more design situations is warranted for PhD study. For calibration purposes, more insight is required into shear reliability performance over a wide range of design situations.

2. To fully calibrate the model for shear. At current, tools for analysis have been developed and have merely been used to determine the reliability of two test cases in the M-thesis. This is not nearly enough to consider the use of partial factors and 1.15 as an economical set of factors used over a wide range of design 1.5

situations. Already for Eurocode requirements, the aforementioned partial factor set fails to meet the basic requirement of safety; failing to achieve target reliability for test case 2. Full calibration of partial factors for design resulting from extensive parametric analyses are required, to find an economic and safe combination for EC 2 requirements and to optimise the combinations according to SANS requirements. There is need for comprehensive parametric analyses that explore reliability trends across a wide range of design situations and across various plausible partial factor schemes. The partial factor scheme that leads to the most economic reliability

performance as long as minimum reliability requirements are satisfied across a wide range of design situations should be an outcome of this study.

3. Once sufficient calibration has been achieved, and giving the results obtained, judgement based approaches should be exercised in specifying how effective shear
10

design using the variable strut inclination method should be applied in practice. The results may comprise either of or any combination of the following:

a. Set of partial factors to be used b. Range of design situations over which design method can be applied effectively or limit of its use c. Aspects of quality management and perhaps detailing can be considered over general design scenarios as well as critical design situations, to specify effective use of these measures to ensure that acceptable reliability is achieved in practice. Also, poor or highly uncertain quality control expected in practice could warrant slightly conservative partial factors to be prescribed for design as compared to those deemed sufficient by calibration studies. This may be the case when calibrating the design method according to SANS requirements, considering the fact that at current reliability modelling has been based on some European models of basic variables. European levels of quality control in production and level of workmanship are generally perceived to be higher or stricter than those in Africa. This gives rise to the next expected outcome of PhD research.

4. To consider, in so far as is possible, how best SA models of basic variables can be derived and used in actual reliability modelling. This would better reflect the

reliability performance of the variable strut inclination design method in an assessment of South African design requirements, conditions and practice.

5. An important step, already partially achieved in the M-thesis, is to validate the reliability models used in the thesis. This has been achieved through the use of the MCFT in independent reliability analyses to check the validity of the results when the design prediction model is converted for use as the gpm in reliability modelling.

11

2.7 Research contribution This research adds to existing studies that have dealt, in various ways, with the effective application of reliability techniques and principles in deriving structural design guidelines. However, the study of applying these principles to properly calibrate the EC 2 variable strut inclination method for shear is not only unique but important as well. This design method is used in practice to provide shear reinforcement for reinforced concrete members. Most beams in practice are provided with shear reinforcement, save for those of minor structural importance. Shear reinforcement is provided in design to avoid and limit cracks due to shear, thereby avoiding sudden and brittle shear failures. More importantly, effective control and avoidance of shear failures allows members to reach their full flexural capacity which, unlike shear failure, provides enough warning about impending failure through excessive cracking and deflections. Proper calibration of the variable strut inclination method is therefore

essential to achieve economic and sufficiently safe designs over a specified and suitable range of design situations. Comparison of the variable strut inclination prediction model to an experimental database of lab tests indicates that the unbiased model is generally very conservative, coupled with some unconservative predictions when relatively high amounts of shear reinforcement are provided in design. The object of good design practice is to ensure that economic designs are

implemented (therefore not excessively conservative) and that they meet set safety requirements, currently reflected internationally by in modern calibration procedures. MS

Excel tools that enable the reliability analysis, based on the FORM method, for design situations for shear according to the EC 2 design method were developed in the M-thesis. During the PhD research, these tools will be used to conduct the full parametric investigation to determine the combination of partial safety factors that best achieve economic and safe performance for shear. To achieve such objective, plausible partial factor combinations will be investigated across a wide range of design situations, particularly at varied amounts of shear reinforcement, in which the code may be applied. The variable strut inclination

prediction model for shear will therefore be calibrated in this PhD research, in an attempt to assess and characterise the reliability performance of members designed for shear against Eurocode and South African requirements. The use of the MCFT in the process of reliability validation also presents a unique and thorough approach to the investigation of shear reliability performance, giving credibility to
12

obtained results. The explicit derivation of the statistical properties of the modelling factors from a compiled database for the two shear prediction models applied in reliability modelling presents a thorough treatment of the basic variables of concern and is also a unique research contribution. Efficient representation of the South African quality requirements and practice will be implemented, specifically in terms of promoting the use of South African models of basic variables used in reliability modelling. This would imply that a thorough assessment of the performance and applicability of the model for shear for South African design practice will be obtained. Annex A in EC 2 allows partial factor reduction based on the control of geometry of critical sections and concrete strength. However, no clear guidance is given in EC 2 on how partial factors can be reduced for a specified reliability class, whereas the principle for such action is given in EC 0. A detailed assessment of how the reliability differentiation framework can be used to effect partial factor reductions for specified reliability classes was performed in the M-thesis. A more detailed assessment of how such procedures could be practically applied in design, particularly for shear, should be demonstrated and would provide insight and initiative in applying it efficiently in design; across all other materials and modes of resistance. The key is to emphasise that quality control is essential to reliability management. Prescribing reliability levels would, in effect, give the designer an incentive and opportunity to demand certain quality requirements, motivated by his use of reduced partial factors in design.

13

3. LITERATURE OVERVIEW OF THE STUDY

3.1 Background of the reliability basis for structural design This research is mainly centred on the effective application of the more advanced calibration methods described in Annex C of EC 0: Basis for Partial Factor Design and Reliability Analysis. Particular interest is taken in the application of these methods to assess the

reliability performance and calibrate the variable strut inclination method for members requiring stirrups. The recommendations for management of structural reliability for

construction works in Annex B of EC 0 have been implemented in the M-thesis to complement partial factor reduction. Reduction is allowed for a given reliability class

provided levels of standard design supervision and site inspection are increased. Applying the differentiation format from EC 0 gives a formal approach, consistent with basis of design requirements, that introduces the need and benefit of better quality at all phases and levels of construction. Annex C in EC 0 gives recommendations on code calibration methods for structural models used in design, highlighting methods of partial factor calibration. Probabilistic methods that incorporate levels of structural performance Figure 3.1 below depicts the different in partial factor determination are outlined. methods of partial factor calibration.

Figure 3.1. Overview of Reliability Methods (EN 1990, 2002)

14

Annex C in EC 0 states that most partial factors and action combination factors proposed in the currently available Eurocodes are derived through Method a; that is, on the basis of judgement calibrated to a long experience of building tradition. This Research is aimed at the effective application of Method c in calibrating the partial factors used in the variable strut inclination method, according to the requirements of both EC 0 and SANS 10160-1. In the First Order Reliability Method (FORM), is the measure of structural performance whose

value can be adjusted in the process to reflect different levels of safety calibration of partial factors. The probability of failure , is related to the reference level of reliability , by:

[3.1]

Where is the cumulative distribution function of the standardised Normal distribution. Table 3.1 shows some numerical representation of the relation between and .

Table 3.1. Relation between 10-1 1.28 10-2 2.32 10-3 3.09 10-4 3.72

and

. 10-5 4.27 10-6 4.75 10-7 5.20

The objective of calibration for concrete resistance is to determine a set of partial factors , for use in design that attain acceptable reliability levels through a number of

analytical situations that are representative of all practical scenarios that the code may perceivably be applied in. Full probabilistic methods are possible but are rarely used in code calibration due to the frequent lack of statistical data. The FORM method is a convenient tool for calibration. It is a first order method because is evaluated at a linearised plane on

the failure surface. The method is compatible with the use of the first (mean) and second (variance) moments of basic variables. Full distribution statistics of basic variables are not necessarily required. Where distributions are available, the first and second moments of nonnormally distributed variables are expressed in the equivalent normal representation (Ang & Tang, 1984). This transformation is illustrated in Figure 3.2.

15

Figure 3.2. Schematic representation of the standard normal transformation process (Taken from Dithinde, 2007)

The FORM method gives insight into which basic variables mostly affect reliability performance, by determining each of their direction cosines or sensitivity factors. This makes it an effective decision and calibration tool, particularly regarding critical design situations where marginal reliability performance is prevalent. Design values should be based on the values of the basic variables at the FORM design point, which can be defined as the point on the failure surface closest to the average design point in the space of normalised variables. EC 0 allows separate calibration of action and resistance standards. This principle was and will continue to be used in the reliability analysis and calibration of the EC 2 design method for shear. The separation is achieved by the use of FORM sensitivity factors, and as shown in the following Equations:

Where variables. The design resistance is expressed in the following form: is negative for unfavourable actions and

[3.2] [3.3] is positive for resistance or resistance

16

[3.4]

Where

is the partial factor covering uncertainty in the resistance model, plus geometric
,

deviations if these are not modelled explicitly.

is the design value of material property .

,

is a conversion factor taking scale effects into account. unfavourable deviations of material properties into account.

is the partial factor taking refers to the design value of

geometrical quantities. Consistent with Equation [3.4] above, Figure 3.3 gives a schematic diagram that shows the elements to be calibrated for use as operational partial factors.

Figure 3.3. Relation between individual partial factors (EN 1990, 2002)

Taerwe (1993) states that special calibration of the model uncertainty as part of the global resistance factor is warranted for coefficients of variation of 20 % and above. Model

uncertainties should be taken into account. They are, however, usually treated nominally by use of recommended models from literature or through subjective professional judgement. Models found in literature are usually derived from and representative of European levels of quality control and workmanship, thus incorporating further uncertainty due to lack of knowledge of the influence of South African specific conditions on reliability performance. South African models of basic variables should be made available and be more readily
17

applied in reliability modelling for an assessment against South African performance requirements.

3.2 The Problem of shear and its reliability basis Literature and published research (Cladera & Mari, 2007; Huber, 2005) have made it evident that the variable strut inclination prediction model, in general or across board, yields a rather conservative estimate of shear resistance. The model factor ( ), or the ratio of the

experimentally determined shear resistance to the predicted shear resistance, was used to describe model uncertainty. An independent investigation was carried out in the M-thesis, finding that the unbiased prediction model has mean model factor ( ) 1.65 when compared to a carefully

compiled experimental database of 222 tests, as well as a large scatter associated with this result with a standard deviation 0.51. Further, inconsistent predictions were realised provided in design as shown in

with varying amounts of shear reinforcement (

Figure 3.4 below. Figure 3.4 is taken from Chapter 7 in the M-thesis, where the unbiased variable strut inclination methods ability to predict true shear resistance was investigated.

Figure 3.4. Logarithmic regression trendline fit to the scatter plot of the EC 2 model factor against the amount of shear reinforcement (taken from M-thesis)
18

From Figure 3.4, it can be observed that the 0.21

can be as high as 2.5 at equals 1 at about 1.9

, progressively decreasing logarithmically to values as low as 0.8 at 2.6 . Further, the , and progressively Design situations with

falls below 1 with increasing amounts of shear reinforcement.

relatively high amounts of shear reinforcement are clearly the more critical region of shear performance as conservatism in the EC 2s shear predictions reduces with increasing amounts of shear reinforcement. With these uncertainties in shear prediction, the question now is how to proceed to design for shear as it clearly is a phenomenon affecting structural performance? The answer is twofold: 1. By the use of structural reliability techniques, to appropriately calibrate shear models with their inherent uncertainties. This would build sufficient conservatism into the procedures by the use of partial factors and characteristic values to ensure that safe designs are achieved. Effort must be directed to achieve as uniform reliability

performance as possible across different design situations, to avoid excessively conservative and therefore expensive designs at lower amounts of shear reinforcement provided in design.

2. To apply rational scientific methods such as non-linear analyses through the use of finite elements and use of the MCFT to improve on the model uncertainty inherent in the model itself. In any case, as a requirement of design bases, the model would still have to be calibrated to achieve sufficient conservatism that accounts for other uncertainties. Considering the fact that the EC 2 design method for shear for members with stirrups is currently applied in most countries where the Eurocodes are currently operational, and will most likely be used in South Africas revised SANS 10100-1, proper calibration of the method is warranted. The European Concrete Platform (2008) presents some reliability based verification of the shear procedures for members not requiring stirrups but with no similar justification conducted for members subjected to shear that require design shear reinforcement. Most beams in practice are designed promoting ductile failure as extensive warning (cracks and deflections) is given before failure. Shear failures are brittle and failure occurs suddenly. The provision of shear reinforcement is therefore an important situation that is conducted to limit shear failures and allow members to reach their full flexural
19

capacity.

Extensive reliability assessments that properly calibrate the partial factor

requirements for the EC 2 variable strut inclination design method are essential.

3.3 EC 2s variable strut inclination design method for shear In the variable strut inclination method all the shear force will be resisted by the provision of stirrups with no direct contribution from the shear capacity of the concrete itself. Crushing of the inclined concrete struts is checked to avoid situations where premature web crushing may occur. In design situations where the web crushing strength is predicted to be lower than the yield strength of the stirrups, the width of the section is normally increased to an extent that the web crushing strength exceeds or at least equals the yield strength of the stirrups. The performance function for the reliability analysis is, therefore, based on the steel contribution of the stirrups provided during the design of a section or member. The shear resistance provided by the stirrups is determined by:

cot
,

[3.5] is the cross-sectional is the

Where

is the design resistance force provided by the stirrups,

area of 2-legs of the links,

is the spacing of the links, z is the internal lever arm, is the angle of inclination of the concrete struts.

design yield strength of the links and The angle

increases with the magnitude of the maximum shear force on the beam and to occur

hence the compression forces in the diagonal concrete members. EC 2 limits between 21.8 cot 2.5 and 45 cot 1 .

For most cases of predominately

uniformly distributed loading the angle

will be 21.8 but for heavy and concentrated loads

it can be higher in order to resist crushing of the concrete diagonal members (Mosley et al., 2007). The limits placed on , which affect the quality and performance of the models

predictions, are set from applying the plasticity theory to the truss model. EC 2 provides an upper limit,
,

, on design shear force that is limited by the ultimate

crushing strength of the diagonal concrete strut in the analogous truss, where its vertical component is given by:

20

cot

tan is a coefficient due to prestress.

,

[3.6]

Where

is a concrete effectiveness factor and

The minimum amount of shear reinforcement,

,

, is given in EC 2 as: [3.7]

0.08

An additional requirement for links, as set by EC 2, is that the stirrup spacing must not exceed, in any direction, the lesser of 75 % of the effective member depth, , and 600 .

3.4 The MCFT The Modified Compression Field Theory (MCFT) has been adopted in this Research to use as a validation reliability model that aims to check the results obtained by the conventional or more routine method used in reliability analyses. The MCFT has an extended rational base and has been shown in a wealth of literature and research to make better predictions of shear resistance than most prediction methods available and in use today. The MCFT, unlike conventional truss models, does not just consider equilibrium, but additionally treats compatability as well as more general stress-strain relationships of the steel and concrete, all of which are formulated in terms of average stresses and average strains. The angle of inclination of the compressive struts, , is determined by considering the cross-sectional

dimensions of a member and its deformations, caused by bending moments concomitant with shear at the studied section, of the transverse reinforcement, the longitudinal reinforcement and the diagonally stressed concrete (Cladera & Mari, 2007). With these methods alongside equilibrium conditions, compatability conditions, and stress-strain relationships for both the reinforcement and the diagonally cracked concrete, the load deformation response of a member subjected to shear can be determined. The MCFT may be explained as a truss model in which the shear strength is the sum of the steel and concrete contribution. As such, it provides itself as a general model for the load-deformation behaviour of two-dimensional cracked reinforced concrete subjected to shear. In this Research, the MCFT is implemented by the use of Response-2000. Response-2000 is a non-linear sectional analysis program for the analysis of reinforced concrete elements subjected to shear according to the MCFT. The Program was developed at the University of
21

Toronto

by

Evan

Bentz

in

2000

and

is

available

for

free

download

at:

http://www.ecf.utoronto.ca/~bentz/r2k.htm

3.5 Reliability analysis of EC 2s variable strut inclination method for shear Consistent with the FORM method, the performance function for shear is described by:

, Where

[3.8]

represents the distribution of true shear resistance, based on EC 2s

variable strut inclination method for shear or the MCFT, determined using unbiased values of the basic deterministic and random variables and neglecting the use of partial factors in the resistance model. , and is the single deterministic value of shear resistance as

would be determined for a practical design situation in accordance with the stipulations in EC 2. The vectors imply that the single deterministic value of shear resistance is

calculated using appropriate characteristic representative values of all the basic variables, which are all treated deterministically when the code method is applied for design. Figure 3.5 shows schematically the probabilistic representation of the performance function.

Figure 3.5. Probabilistic representation of the performance function for shear

22

The FORM method is used to evaluate

for a given design situation. The goal is to use

FORM to conduct parametric analyses to assess partial factor requirements that achieve acceptable reliability performance according to EC 0 and SANS 10160-1 requirements.

3.6 Parametric analyses Parametric investigations should be conducted across a range of factors that are known to affect shear resistance and its performance. Preliminary considerations from the M-thesis indicate that these should be: 1. The concrete strength, 2. The size of the cross section, & /

3. The amount of longitudinal tension reinforcement, 4. Maximum moment to shear ratio divided by the effective depth alternatively the / ratio 5. Amount of shear reinforcement, 6. Model Factor The preliminary reliability analysis presented in the M-thesis focused on isolating the main basic variables that affect shear reliability performance. The model factor was found to dominate. Therefore, any control or judgement concerning shear should be applied to its modelling ability, particularly at high amounts of shear reinforcement where its predictions are known to be unconservative. Either, larger partial factors are used to achieve acceptable reliability performance or a better model predicting shear should be used in such situations, thereby limiting the use of the conventional method. or

23

4. PROPOSED RESEARCH METHODOLOGY

4.1 General Methodology The proposed research methodology is based on a continuation and extension of the approach presented in the M-thesis. The approach taken to conduct the research is very much in line with the research objectives. First, a general survey of reliability principles governing basis of structural design are continually studied and reviewed. The main focus is, however, to determine the extent that EC 0 reliability principles have been applied in deriving EC 2 provisions, particularly for shear resistance and in terms of regulated quality control. EC 0 presents mature reliability concepts that should be effectively applied in achieving the guidelines for resistance. In order to achieve more innovative and unique application of reliability techniques as basis of design, general and modern documents as the JCSS Probabilistic Model Code (2001) and the Draft fib 2010 Model Code (fib, 2010) are reviewed. It has been found that Annex B and Annex C from EC 0 are not fully implemented as and where necessary in EC 2. Action is therefore stimulated to carry out reliability investigations and complementary assessments. Various motivations stimulate the application of Annex C to calibrate EC 2s variable strut inclination method for members requiring stirrups. Some contradiction exists between the reductions of partial factors allowed in EC 2 under conditions of increased quality control, particularly of deviations of geometry of critical sections as well as increased quality control of concrete production. The fib Model Code seems to prefer that partial factors are not adjusted given stricter quality control of concrete strength for a given reliability class whilst this is done in EC 2. An independent investigation in the M-thesis justifies that partial factor reduction for resistance is feasible and applies the reliability framework in EC 2 Annex B to partial factor modification in Annex A of EC 2. The general methodology is summarised schematically in Figure 4.1.

24

1. Map out Reliability Basis of Structural Concrete Design (RBoSD) done in accordance with modern
international standards e.g. EC 0, JCSS PMC, CEB documents, CEB-FIP and fib Model Codes

2. Identify deficient application of reliability principles in Structural Concrete Provisions - done using EC 2 and
relevant background documentation (EC 2 Commentary & Worked Examples, Papers (Cladera & Mari), Published papers and research applying reliability principles. Attention given to South African requirements and conditions

3a. Partial Factor reduction in EC 2 not synchronised with reliability differentiation principles warranting such reduction. Principles given in Annex B of EC 0. Action taken to formalise this procedure in M-thesis. Guidance given to European and South African requirements.

3b. Reliability Based Calibration in Annex C of EC 0 to be properly applied to EC 2 shear design method for members with stirrups. Explicit representation of modelling factor considered. Design method calibrated to both European and South African requirements

3. Identified issues warranting treatment/ proper implementation

Figure 4.1. General research methodology

25

4.2 Methodology for reliability analysis and calibration of variable strut inclination method The calibration of the variable strut inclination method for shear is central to this research. The manner in which the model is calibrated is therefore important and should be in line with the basic provisions given in EC 0 Annex C and other relevant documents where this issue is dealt with. Figure 4.2 (shown overleaf) shows a detailed breakdown of reliability analysis, up to the next step of parametric analysis, that was conducted in an effort to calibrate the EC 2 design method for members with stirrups. The extension of the parametric range of the representative cases is an important element of the extended investigation. From the parametric analysis, regions can be identified with sufficient reliability to excessive conservatism; transitional conditions with marginal reliability; conditions of insufficient reliability requiring modification in the design procedures.

26

1a. GENERAL PROBABILISTIC MODEL (GPM) FOR SHEAR, 2 cases: 1. 2. EC 2 Strut inclination method converted to gpm MCFT used as gpm more rational case

1b. EC 2 DETERMINISTIC VALUE FOR SHEAR RESISTANCE, 3 analyses: 1. 2. 3. Design values Characteristic values only Partial Factor only (PFs applied to mean values of basic variables)

2a. Allowance for Basic Variables to be randomly distributed: 1. 2. MF and other basic random variables e.g. , , etc. Quantities can also be deterministic where applicable or when justified

2b. Deterministic design (characteristic) input variables applied with partial factors; 1. 2. Vector ( , ) Appropriate bias also expressed to obtain mean values of basic variables

2. BASIC VARIABLES

3. LIMIT STATE FUNCTION (LSF) 1. 2. Analytical partial differentiation applied to LSF when EC 2 used as gpm When MCFT is used as gpm, numerical differentiation applied using Response-2000 + explicit analytical differentiation for MF only

4. REPRESENTATIVE CASES (Limited Parametric study) 4a. CASE 1: Low shear rnft 0.45 4b. CASE 1: High shear rnft 1.80

5. RELIABILITY ANALYSES VaP check 5a. Full prob. reliability model All basic variables considered as random variables for representative cases 1 & 2 5b. Simplified reliability model Only MF considered as basic random variable for representative cases 1 & 2 VaP check

6. PARAMETRIC ANALYSES 1. 2. 3. Assessment of reliability given different design situations and partial factor schemes Final critical judgements and decisions based on trends of parametric analyses Formalised design rules for application of EC 2 shear design method in SA; assessment also done of performance to Eurocode requirements and conditions

Figure 4.2. Flowchart outlining procedure for reliability analysis and calibration of EC 2s design method for shear
27

5. PROPOSED RESEARCH PROGRAM

A breakdown of the research program is outlined below.

FIRST SEMESTER 2012 (Period mid-March to early July) 1. Extend the reviewing of literature and planning of investigative techniques for parametric analysis - towards full calibration 2. Compile present results on reliability performance of EC 2 and provisions for shear resistance design as background material to the adoption of EC 2 as South African standard. 3. Publish journal papers based on findings from M-thesis 4. Planning of conference papers, submission of abstracts, followed by preparation of full papers (based on acceptance) 5. Begin Parametric investigation. {Graphs, trends, design situations, analysis}

SECOND SEMESTER 2012 (Period early August to mid-December) 6. Continue Parametric analyses {bear in mind that this is done concurrently with results analysis and validation of reliability modelling using MCFT} 7. Collection and review information or data of available models for basic variables produced from surveys of local practice. Bayesian updating may be possible (additional process). 8. Incorporate South African variables into analyses for South African requirements 9. Complete analyses 10. Continued assessment of review of reliability basis of design and its possible applications. The relevance of the results to the issue of punching shear is considered with the objective of giving guidance.

FIRST SEMESTER 2013 (Period mid-January to end of July) 11. Consideration and critical appraisal, including judgement-based arguments, of the results yielded by parametric analyses 12. Final recommendations on calibrated elements {partial factors, reliability performance levels, design rules including NDPs for SA practice of all basic variables describing shear} 13. Final thesis write-up and compilation of research (mid-Feb to end of July)
28

14. Possible conference attendance (ACCTA 2013) 15. Submission of PhD dissertation 16. Continued assessment of review of reliability basis of design and its possible applications.

SECOND SEMESTER 2013 (Period mid-August to mid-Dec) 17. Publication of paper on the calibration process of the variable strut inclination method for shear according to European and South African requirements. 18. Possible conference attendance (SEMC 2013) 19. Graduate Dec 2013

Table 5.1. Dates of some conferences for possible attendance Conference/ symposium Abstract The International Conference on Advances in cement and concrete in Africa, ACCTA 2013 The Fifth International Conference on Structural Engineering, Mechanics and Computation, SEMC 2013 2013 fib symposium 2 Apr 2012 ???? 22-24 Apr Tel-Aviv 2013 30 Sept 2012 01 Mar 2013 2 - 4 Sept Cape 2013 Town 31 Mar 2012 15 Aug 2012 Deadlines Full Paper Conference date 28 - 30 Jan Johannes 2013 burg City

29

REFERENCES

Ang, A.H-S., & Tang, W.H. (1984). Probability Concepts in Engineering Planning and design. Volume II - Decision, Risk and Reliability. Wiley, New York. BS 8110-1. (1997). Structural use of concrete - Part 1: Code of practice for design and construction. British Standards Institution, UK. Cladera, A., & Mari, A.R. (2007). Shear strength in the new Eurocode 2: A step forward?. Structural concrete, 8 (2), 57-66. Dithinde, M. (2007). Characterisation of Model Uncertainties for Reliaility Based Design of Pile Foundations. PhD Thesis, Department of Civil Engineering, 327 pp. Ellingwood, B.R. (1994). Probability-based codified design: past accomplishments and future challenges. Structural safety, 13, 159-176. EN 1990. (2002). Eurocode Basis of structural design. European Committee for

Standardisation (CEN), Brussels. EN 1992-1-1. (2004). Eurocode 2: Design of concrete structures Part 1-1: General rules and rules for buildings. European Committee for Standardisation (CEN), Brussels. European Concrete Platform. (2008). Eurocode 2 Commentary. Retrieved Dec 2008, from: http://www.ding.unisannio.it/ricerca/gruppi/ingciv/ceroni/commentario_EC2_2004.pdf European Concrete Platform. (2008). Eurocode 2 Worked Examples. Retrieved Dec 2008, from: http://www.europeanconcrete.eu/index.php?option=com_docman&task=doc_view&gid=28 fdration internationale du bton (fib). (2010). fib bulletin 56, Model Code 2010 First complete draft, Volume 1. fib, Lausanne, Switzerland. fdration internationale du bton (fib). (2010). fib bulletin 56, Model Code 2010 First complete draft, Volume 2. fib,Lausanne, Switzerland. Holick, M. (2009). Reliability analysis for structural design. ISBN 978-1-920338-11-4, SUN MeDIA, Stellenbosch, South Africa.

30

Huber, U.A. (2005). Reliability of Reinforced Concrete Shear Resistance. Masters Thesis, Department of Civil Engineering, University of Stellenbosch, 188 pp. ISO Workshop. (2011). Harmonised Structural Standards Workshop held at the University of Stellenbosch. JCSS. (2001). Probabilistic Model Code, Part 1 to 4, Basis of design, Load and resistance models, Examples. Joint Committee on Structural Safety. http://www.jcss.ethz.ch/ Mosley, B., Bungey, J., & Hulse, R. (2007). Reinforced Concrete Design to Eurocode 2 (6th ed.). Palgrave Macmillan, New York, USA. Retief, J.V., & Dunaiski P.E. (2009). The limit states basis of structural design for SANS 10160-1. In Retief, J.V., & Dunaiski, P.E. (Editors) Background to SANS 10160. SUN MeDIA, ISBN 978-1920338-10-7. SABS 0100-1. (2000). The structural use of concrete Part 1: Design. South African Bureau of Standards, Pretoria. SANS 10160-1. (2010). Basis of structural design and actions for buildings and industrial structures. Part 1: Basis of structural design. South African Bureau of Standards, Pretoria. Taerwe, L. (1993). Towards a Consistent Treatment of model uncertainties in reliability formats for concrete structures. CEB Bulletin no. 219: Safety and Performance Concepts. Comit Euro-International du Bton (CEB), Switzerland.

31