Dimitris Diamantidis

Safety of long railway tunnels

Abstract Planning and designing railway tunnels with an explicit reference to safety issues is becoming of utmost importance since the combination of high speed, mixed goods–passenger traffic and extreme length of the new tunnels under design or concept evaluation, have sensitively modified the inherent safety of the railway tunnel. Although the probability of occurrence of accidental events may still be considered rather low, the possible consequences of such events in long tunnels can be catastrophic, therefore raising the overall risk to levels that may be no more acceptable. The scope of this paper is to illustrate the state-of-practice related to risk analysis of long railway tunnels. First, ambitious tunnel projects are briefly reviewed. The applicable risk-analysis procedures are then described and discussed. The problem of risk appraisal is addressed and quantitative target safety levels are proposed. Safety systems for risk reduction are outlined.

Reliability assessment of existing structures

Abstract This paper discusses the implementation of modern probabilistic methods in the reliability evaluation of existing structures. The difference between prior uncertainty modelling and reliability updating based on available information is emphasized. Possibilities for applications of first-order reliability methods are given. Practical examples dealing with realistic risk scenarios in offshore structures illustrate the methodological aspects.

Target reliability for existing structures considering economic and societal aspects

Abstract Specification of target reliability levels is one of the key issues of the assessment of existing structures. ISO 13822:2010 and ISO 2394:2015 indicate procedures for specification of target reliability levels by optimisation of the total cost related to an assumed remaining working life of a structure. These approaches are critically compared with human safety criteria, with target levels based on a marginal life-saving costs principle, and with recommendations of present standards. Optimal target reliability levels are then derived in the representative case study for an existing structural member. It appears that the requirement to reach the same target reliability levels for existing and new structures is uneconomical. Decisions made in the assessment can result in the acceptance of the actual state or in the upgrade of an existing structure. Two reliability levels are thus needed – the minimum level below which the structure is unreliable and should be upgraded, and the target level indicating an optimum upgrade strategy. It is recommended that these levels be established using economic optimisation and the marginal life-saving costs principle, as both the approaches take into account the costs of safety measures and the failure consequences.

Bayesian network application for the risk assessment of existing energy production units

A Bayesian network is applied in this contribution in order to assess the risks of a selected production unit in a fossil power station. A general framework for the risk assessment of production units of a power station is presented first by implementing statistical methods and Bayesian networks. Special emphasis is given to the input data consisting of failure rates which are obtained on the basis of recorded data and expert judgements. The consequences of failure are divided into economical and human (societal): economic consequences include outages of key technological devices, societal consequences cover potential injuries and fatalities. Probabilistic risk assessment methods are applied to the selected production unit of a power station. The influence of the uncertainties in the considered technical parameters on the availability of the unit is assessed and the acceptance of the calculated availability represented through the mean value and the standard deviation is discussed. Societal risks given in terms of weighted injuries and fatalities are obtained and respective risk acceptance criteria are presented. Uncertainties affecting the risks are discussed. It appears that the proposed framework provides a valuable assessment of the influence individual devices and their components on availability and societal risk. For that purpose the used methodology, intentionally simplified for operational applications, includes important factors affecting risks of production units. It is concluded that Bayesian networks are a transparent method for the probabilistic risk assessment of complex technological systems. The results of the performed analyses can be easily updated when additional information becomes available as illustrated in characteristic examples.

Bayesian Network Application for the Risk Assessment of Existing Energy Production Units

A Bayesian network is applied in this contribution in order to assess the risks of a selected production unit in a fossil power station. A general framework for the risk assessment of production units of a power station is presented first by implementing statistical methods and Bayesian networks. Special emphasis is given to the input data consisting of failure rates which are obtained on the basis of recorded data and expert judgements. The consequences of failure are divided into economical and human (societal): economic consequences include outages of key technological devices, societal consequences cover potential injuries and fatalities. Probabilistic risk assessment methods are applied to the selected production unit of a power station. The influence of the uncertainties in the considered technical parameters on the availability of the unit is assessed and the acceptance of the calculated availability represented through the mean value and the standard deviation is discussed. Societal risks given in terms of weighted injuries and fatalities are obtained and respective risk acceptance criteria are presented. Uncertainties affecting the risks are discussed. It appears that the proposed framework provides a valuable assessment of the influence individual devices and their components on availability and societal risk. For that purpose the used methodology, intentionally simplified for operational applications, includes important factors affecting risks of production units. It is concluded that Bayesian networks are a transparent method for the probabilistic risk assessment of complex technological systems. The results of the performed analyses can be easily updated when additional information becomes available as illustrated in characteristic examples.

Reliability and Risk Acceptance Criteria for Civil Engineering Structures

Abstract The specification of risk and reliability acceptance criteria is a key issue of reliability verifications of new and existing structures. Current target reliability levels in standards appear to have considerable scatter. Critical review of risk acceptance approaches to societal, economic and environmental risk indicates that an optimal design strategy is mostly dominated by economic aspects while human safety aspects need to be verified only in special cases. It is recommended to specify the target levels considering economic optimisation and the marginal life-saving costs principle, as both these approaches take into account the failure consequences and costs of safety measures.

Determination of target safety for structures

Codes of practice aim at guaranteeing structures having the risks acceptable to the public and the minimum total costs over a design working life. However, current criteria for structural design provide a broad range of target reliability indices, specified often for different reference periods even though their recalculation for different reference periods is indeterminate due to mutual dependence of failure events. General approaches for selecting target reliability levels are discussed in view of costbenefit optimisation and human safety aspects. Design strategies seem to be driven by economic arguments rather than by human safety criteria.

Crack Offset Corrections to Post-crack Performance Parameters Obtained from Third-point Loaded Fiber Reinforced Shotcrete Beams

Fiber Reinforced Shotcrete (FRS) is used as an economically competitive method of ground support in tunnel excavations throughout the world. Fibres are incorporated into this material to impart post-crack load capacity. Beam tests are commonly used to assess the post-crack performance of FRS but high within-set variability has hindered the usefulness of beam test data. A method of correcting performance data to account for the offset between the position of the crack and the center of the beam has been suggested as a means of improving within-batch variability for post-crack performance parameters obtained from third-point loaded beam tests. In this paper, an offset correction method has been applied to a large set of beam test data obtained as part of Quality Control testing for a tunnel-lining project to determine whether the method is effective in reducing within-batch variability.

Applicability of First Order Reliability Methods A State of the Art

This paper introduces First-Order Reliability Methods (FORM) and presents in a condensed form examples from recent applications of FORM in structural analyses.

Maximum loss prediction for earthquake insurance

Insurance is one of the most important elements of earthquake risk management. In this paper an attempt is presented to formulate a conceptual basis for maximum loss prediction to be used in earthquake insurance policies. Due to the complexity and randomness of the physical phenomena such as damage prediction and seismic hazard a full probabilistic approach combined with well established models is proposed. Modern reliability methods are thereby implemented, which are able to take all influencing uncertainties into consideration. The method is illustrated in an example application.