Professor Robert Melchers

Phenomenological modeling of the long term general corrosion of mild and low alloy steel specimens under marine conditions is considered, using weight loss as a function of time. A conceptual model for immersion corrosion, tidal corrosion and atmospheric corrosion under marine conditions is proposed. The model uses accepted theories for short term surface corrosion and employs modern understanding of the action of bacterial colonisation of the surfaces of specimens, including the development of anaerobic conditions. Kinetic, diffusion, nutrient and anaerobic components of the model are identified and mathematical descriptions given. The model is compared to some data available in the literature. Some observations are made about data requirements for further development of models of the type proposed.

Experiences are described of the use and application of quantified risk analysis (QRA) for medium size liquefied petroleum gas (LPG) facilities associated with automotive service (gas) stations in suburban and rural areas. The facilities consist primarily of one or more 7.5 kilolitre pressure vessels, pumps, pipework, dispensing facilities and various types of safety equipment. QRA was required to show compliance with recently introduced Australian national regulatory risk criteria for land use planning purposes. A description is given of the risk analysis procedure, the various hazard scenarios considered, the determination and application of appropriate data for initiating events and for rates of failure of mechanical components, the pressure vessel and various human operations. Various tests were performed to estimate matters such as flame length and flame impingement on pressure vessels. Some observations are made regarding the standard of conventional industry quantified risk analysis techniques and the way they are often applied. An area of particular concern is the use of 'conservative best estimates' widely used in the industry but inconsistent with the interpretation of risk criteria derived from expected (observed) risks.

Although the risk of damage to buildings and structures and the risk of personal death and injury due to earthquakes in Australia is relatively low, the 1989 Newcastle earthquake did point to the need to reexamine existing structural design and construction standards. There is a danger, however, that tough regulations will affect adversely the continued viability of existing building stock, in particular historical buildings. Commercial interests also may be affected. For buildings which have a projected life which is short relative to the recurrence rate of serious ground shaking, it is argued that the total structural life-time risk is low and might be ignored. This would appear to be the situation under Australian intra-plate earthquake conditions. The conclusion is relevant for structurally adequate existing buildings and for new buildings constructed in the first few years after the occurrence of major ground shaking.

A corrosion process acting on an individual structural member may be modelled as a time-dependent resistance deterioration mechanism. This leads to a deteriorating structural system, which may be subjected to one or more stochastic load processes. When resistances are modelled as Gaussian random variables, the loads as Gaussian random processes and deterioration is considered to be a simple function of time, some progress can be made in evaluation of structural system reliability using an outcrossing formulation. This was checked using numerical integration. The results are compared to purely simulation results both for the outcrossing rate and for various sensitivity results. Good agreement is demonstrated.

Quantitative risk analysis procedures increasingly are being applied to potentially hazardous industrial operations and facilities. A review of QRA analyses carried out for land-based LPG facilities has shown that the failure probabilities obtained for specific events may differ considerably between risk analyses and between analists. The present paper reviews possible reasons for the disparities and what might be done to produce more consistent results. It is necessary that the limitations of current QRA techniques be recognized and be made known. This is important because there is a distinct danger that quantified risk analysis in general may be discredited as a result of unfavourable experiences by industry, decision-makers and legislators. This may have implications also for the offshore industry.

The details of footing design, the soil stiffness and the soil resistance all affect the rotation characteristics of column bases of structural frames. This can have a considerable influence on structural economics (strength design), structural deflections (serviceability design) and on structural stability. Of particular and increasing importance in structural design is the ability to predict structural deformations under serviceability load level conditions. However, only limited theoretical and experimental attention has been given to understanding the footing-soil interface behaviour of shallow footings for realistic foundation conditions. In this paper, a set of experimental results for shallow footing behaviour under load reversal is presented. Based on this work and on previous modelling attempts, a relatively simple theoretical model for predicting rotation behaviour of simple pad footings under the first loading condition is described. © 1992.

The setting of structural engineering design standards for earthquakes in Australia is particularly difficult. The country lies within a ``continental'' plate and evidence from other such regions uggests that Australia is therefore a relatively low risk country with only rare occurrences of major earthquakes. By comparison to other countries Australia also has a very short recorded history relative to the frequency of significant earthquakes. A discussion is given herein of factors which influence seismic zoning, the underlying assumptions and limitations which apply to the methods used and how these matters should relate to the development of design codes for earthquake forces. The discussion should be seen as a contribution towards understanding the background and assumptions and not as criticism of present code-development efforts.

Mathematical models for creep and shrinkage of concrete are used in the paper to present a whole-process analysis of creep and shrinkage effects in reinforced concrete structures, allowing in particular for various uncertainties. A simulation method is proposed to predict the time-dependent load effect due to creep and shrinkage under stochastic loads. It is shown through the formulation and by example that the method has the advantage that all the random variability can be included in the load effect at any point in time. Because of the uncertainties of concrete material properties and the applied loads, all parameters that describe the properties of the structures and the nature of load at any point in time are treated as random variables.

In determining the force effects due to wave action on offshore structures it is conventional to use an equivalent linearisation to convert wave velocities to forces via Morison's equation. Different linearisations have been suggested. However, it is not, a priori, clear which is appropriate when structural fatigue is to be estimated. A (theoretical) criterion is introduced herein and examined in relation to a simple offshore structure.

Time-variant structural reliability problems arise when the structural resistance may change with time (such as in structural deterioration) or when the system is acted upon by two or more stochastic load processes. This paper reviews current trends in using sampling based techniques to solve such problems.

The force effects due to wave action on offshore structures are conventionally determined using Morison's equation. This is a nonlinear relation in velocity, conventionally represented by an equivalent linearisation for converting wave velocities to forces, necessary to allow structural dynamic analysis by frequency domain methods (the most commonly applied approach). The suggested approaches for the linearisation can be shown to be of essentially two classes; those equivalent to Borgman's approach and those equivalent to Bolotin's. In general, it is not clear, a priori, which is more the appropriate. Nor is it clear whether these classical linearisations are relevant, particularly when structural fatigue is to be estimated (e.g., through the Palmgren-Miner rule). A criterion for derivation of a linearisation constant under fatigue considerations can be derived relatively simply. Unfortunately the criterion is not easily applied except by iteration. A sensitivity study has been performed for a simple but realistic offshore structure to estimate the relative importance of correctly estimating the linearisation constant.

While there has been significant progress in the assessment of structural reliability for complex structures (e.g., offshore oil platforms) in the time independent domain, significant problems remain in dealing with structures subject to time varying loading. The crux of the problem lies in the fact that the loads are time dependent (e.g., wave loading, wind loading, etc.) but the material properties can be considered to be time independent. A procedure utilizing refined Monte Carlo simulation in the hyper-polar co-ordinate space of the load processes is described herein.

The reliability of offshore structures under ultimate load conditions is considered under a potentially diverse load process scenario. An 'outcrossing' formulation in which the loads are modelled as processes rather than only as random variables is developed in the (hyper-) polar co-ordinate space. This has distinct advantages for problems, such as involving offshore structures, for which the structural analysis for safety checking is complex (e.g., requiring a finite element analysis or dynamic analysis etc.).

Considerable progress has been made in the theory of structural reliability in the last 15-20 years, particularly in relation to the analytical treatment of information. Much of this has gone relatively unnoticed by the engineering profession, even though the potential for structural reliability assessments in certain practical problems is now significant. Structural reliability theory has been applied, of course, to structural design code calibration, but has a much wider potential. Particularly in the offshore industry and in large projects it is being used where risk must be specifically considered and where to do so may yield considerable design and construction cost savings. In this paper the basic concepts of structural reliability theory will be reviewed and related to the standard and type of information usually availabel in engineering studies. An overview will also be given of the various calculation technique now available.

On 28 December, 1989 the city of Newcastle was struck by a magnitude 5.6 earthquake killing 12 people and causing losses estimated at $1000 million. It was the first time a major Australian city has been so extensively damaged as a result of an earthquake, and the first time that deaths have occurred. With a few exceptions, most of the damage was to older loadbearing masonry construction or to infill masonry in modern framed construction. There was relatively little structural damage to modern buildings, industrial facilities and lifelines. This paper reviews the damage to structures and brings out the main points to be considered in future design and construction.

The process of structural design has received very little attention from the research community despite its centralness in day-to-day structural engineering. Much of the research which has been done has been carried out in Australia over a number of years and with the co-operation of many structural engineers. The present paper reviews the reasons for being engaged in this relatively unusual research activity and provides an overview of the work to date. A review is given of the incidence and type of errors made in basic design tasks such as calculations, table look-up and table interpolation, and how these errors relate to results obtained for more comprehensive tasks such as loading determination and member design. One of the outcomes of the work has been to suggest areas in certain current design codes which are prone to misinterpretation by practitioners. Preliminary results related to design computation checking and inspection are also reviewed. This work potentially has major implications for quality assurance programmes.

Overview checking occurs whenever an engineering design is reviewed by another engineer without specifically carrying out checking calculations or following through the design process in detail. Results are reported herein of a survey examining the effectiveness of overview checking. The respondents were requested to assess whether a structural design should be considered "undersized", "correct" or "oversized". Using the responses as a basis, a decision model is proposed for the effectiveness of overview checking as a function of error magnitude and of checker's experience. It is concluded that correct assessment of a design as "safe" is not a function of experience. However, it appears that more experience is beneficial in selecting economical designs. © 1989.

Absolute structural optimization problems such as leastweight and layout problems become both more realistic and more complex when their parameters are described probabilistically. The theoretical framework for absolute optimal design of sandwich beams, grillages and reinforced slabs is extended herein to deal with probabilistic parameters. It is shown that using a simplified probabilistic framework (socalled First Order Second Moment), some classical solutions for optimal layout remain valid. Remarks about the more general problem and the difficulty of its solution close the paper. © 1989 Springer-Verlag.

Improvements can be made to importance sampling by recognising some of the essential features of the time-independent structural reliability problem. Thus non-interesting regions in the integration domain can be deleted from consideration using a priori information and updating techniques. These can also be used to obtain the point(s) of maximum likelihood, a central parameter for importance sampling in cartesian co-ordinates.

Elementary Monte-Carlo sampling as a technique to artificially replicate an experimental program has been used successfully for relatively simple problems, provided the failure probability to be calculated is relatively large. The present paper reviews the elementary Monte-Carlo method and describes little known, but improved approach known as Importance Sampling.

The modelling of applied wave loading for steel tubular offshore platforms and appropriate statistical models are discussed. For the purposes of reliability analysis, the fact that wave trains reach the various structural components at different times, and the effect this may have on the sequence of most likely member failure, is considered.

A mathematical model, based on the physical properties of nominally 'pinned' column-bases and intended to describe the moment-rotation characteristics for given eccentricity of axial load, is presented. The model uses three main parameters to describe the characteristic: elastic stiffness, strain hardening stiffness, and plastic moment capacity. In addition, an empirically determined 'shape' factor allows for the transition from elastic to inelastic behavior. The model is compared to a series of full-sized test results for a range of practical column-base connection. The behavior of soil under the footing is not considered.

The reliability analysis of multi-membered structures such as steel tubular offshore platforms has been largely confined to considerations of structural member behavior and given loading uncertainty. However realistic modelling of wave forces may require account to be taken of the effect of wave forces varying with the incident wave location as it passes through the structure. Such an analysis is described in the present paper, using the Truncated Enumeration technique previously described. Main emphasis is given to structures composed of elastic-brittle members: the analysis for elastic-plastic structures degenerates to that of a probabilistic adaptation problem.

Realistic assessment of reliability must take account of the possibility of human error in the design and execution of the project of interest. This is also true for structural reliability assessment, even though for some applications, such as design code writing, a nominal reliability measure, largely ignoring human error, may well be sufficient. The present paper reviews, briefly, the principles by which human error information may be taken into account in reliability assessment, and, in a little more detail, both empirical models and data for some types of human error which may arise in structural design processes. Attention is also given to three forms of checking: independent, self- and overview checking. Only recently has some indicative data for checking processes become available; in all aspects the reported research must seem as preliminary rather than definitive.

Debate concerning the role, authority and complexity of contemporary structural codes has persisted in Australia and overseas for many years. The Report examines contemporary literature and assesses the results of a request for opinion. Conflicting views of different user groups are identified. The traditional roles of codes as multi-user documents, vehicles for communicating information, and educational resources are examined. The profession's view of codes is interpreted and a proposal put forward as to what should be the Institution of Engineers' position on the philosophy, form and drafting of codes.

Traditional design methods for offshore structures have, on the whole, worked well, but there are now moves to develop probabilistically-based design rules. This is a sound strategy since uncertainties in information can be rationally incorporated in the design and analysis process. The application of reliability theory extends, however, also to the sphere of analysis and decision making in the construction stages. The paper briefly reviews the essential concepts of structural reliability theory and then indicates the various ways in which it has been or can be applied in practical situations. This includes analysis for peak loading and for fatigue as well as other criteria. Some ongoing research areas will be noted.

Offshore structures are subject to poorly defined hazards such as storms, continued wave action, corrosion and ship impact and are themselves not well defined in terms of their strength or resistance. Traditional design methods for offshore structures have, on the whole, worked well, but there are now moves to develop probabilistically based design rules. This is a sound strategy since uncertainties in information can be rationally incorporated in the design and analysis process. The application of reliability theory extends, however, also to the sphere of analysis and decision making in the construction stages. The paper briefly reviews the essential concepts of structural reliability theory and then indicates the various ways in which it has been or can be applied in practical situations. This includes analysis for peak loading and for fatigue as well as other criteria. Some ongoing research areas will be noted. (A)

The effectiveness of quality assurance depends largely on the effectiveness of the facilitative measures it provides and the control measures it imposes. These are related to structural safety and structural performance through human error. Both empirical data about human error and models are selectively reviewed herein and directions of some ongoing research outlined.

A common feature of many steel building frames is the use of two bolts to anchor a baseplate welded to the column end to the concrete foundation. Such a connection is typically assumed to represent a pinned connection, meaning that it is assumed for design purposes that the connection is incapable of transferring bending moment. However, virtually no experimental observations about the behaviour of such connections have been reported. The paper presents some initial experimental findings for connection capacity under axial load and bending moment and reports on preliminary mathematical modelling of the behaviour of the connection using physical parameters such as endplate thickness and bolt characteristics as input.

The response of an initially curved cantilever strut with a lumped mass at the tip subject to arbitrary axial impact loads is described, as a model for the dynamic response of unrestrained piles during driving. The differential equation of motion defining the lateral movement of impact-loaded cantilever struts is derived. This equation includes the effect of inertial resistance to motion provided by both the strut self-mass and the mass of the pile hammer. The shape of force pulses produced during pile driving was approximated by an equivalent trapezoidal force pulse. The differential equation of motion was solved for this force pulse using the Runge-Kutta technique. The solution was found to agree with observed pile behavior during driving. Is is shown that the magnitude of lateral displacements reduces as the mass of the pile hammer increases, because of inertial effects.

For structural systems, calculation of the structural failure probability requires the determination of all the failure modes contributing significantly to the overall failure probability. For large systems it is not practical to enumerate fully all modes. The paper presents a procedure, termed the Truncated Enumeration Method, which has been shown to be derivable rigorously from exhaustive enumeration when truncation of solution possibilities, not significant for the reliability assessment, is imposed. The method has been able to reproduce independently all known solutions obtained by others, but usually in a more systematic manner. The relationship of this method to other system reliability calculation techniques is discussed, some limitations are noted and some calculation results are presented.

This paper presents an exploratory example of the way structural reliability may change with time, in this case due to the effect of low-cycle fatigue. This may be of interest for structures subject to the possibility of high overload during their design lives. Inelastic fracture mechanics theory is reviewed in the first part of the paper. This is followed by a discussion of distributions appropriate for use in a reliability analysis. Monte-Carlo simulation is then employed to show how, typically, the failure probability changes with the number of load applications to the structure, i. e. with time.

Structures are composed of many members and connections, each with individual properties. When the member properties and the loadings applied to the structure are specified in statistical terms, the determination of structural system reliability becomes a problem of multi-fold integration. Since direct integration is feasible only for extremely simple structures, several alternative approaches to the problem have been suggested; either through simplifying the integration process itself or through simplified reliability theory. These approaches are discussed in the context of use of computer programs and computer usage.

For structural systems composed of many individual members each perhaps having a nonlinear constitutive relationship, the calculation of the probability of structural failure under applied random loading requires a knowledge of all failure modes which contribute significantly to the failure probability. For large structural systems this may present a formidable problem, even with modern computers. A calculation procedure which systematically generates failure modes and eliminates those of little significance is described, together with appropriate procedures for calculating the necessary modal failure probabilities. (Author abstract. ) Refs.

Some dynamic elastic buckling formulations for driven piles are given and some solutions presented. These are compared to limited model tests. Suggestions for an improved formulation and for practical implications are made. (Author abstract. )

For large structural systems, calculation of the structural failure probability requires a knowledge of all relevant failure modes. The so-called 'truncated enumeration method' to systematically derive the dominant failure modes is described. The derivation of the limit state expressions for each mode is through an incremental load approach, which allows member behaviour including strain hardening, strain softening and residual strength to be handled. It is noted that elastic-plastic member behaviour does not necessarily converge to rigid-plastic behaviour. Finally, the numerical problems associated with the procedures outlined are discussed and two examples given.

To predict the reliability of structures it is necessary to make reasonable allowances in the assessment for the influence of human error in the design, construction and use of the structure. The present paper deals with an attempt to model mathematically the human errors introduced into the structural design process and the effect they may have on structural safety. Preliminary data collection for use in the model is reported, and an outline of the model being developed is given.

This paper presents an example of the way structural reliability may change with time, in this case due to the effect of low-cycle fatigue. This may be of interest for structures subject to the possibility of high overload during their design lives. Inelastic fracture mechanics theory is reviewed in the first part of the paper. This is followed by a discussion of distributions appropriate for use in a reliability analysis. Monte-Carlo simulation is then employed to show how, typically, the failure probability changes with the number of load applications to the structure, i. e. with time.

For structural systems, calculation of the structural failure probability requires the determination of all the failure modes contributing significantly to the overall failure probability. For large systems it is not practical to enumerate fully all modes. The paper will present a procedure, termed the 'Truncated Enumeration Method', which has been shown to be derivable rigorously from exhaustive enumeration when truncation of solution possibilities, not significant for the reliability assessment, is imposed.

For the reliability assessment of structures which may fail in one of several failure modes, bounds on the probability of failure of the structure have been developed. It is well known that even the best of these bounds are dependent on the correlation between the failure modes and that this correlation is, in turn, dependent on the correlation between member properties. Relatively little appears to be known about the importance of member property correlation effects in reliability analysis, nor about the actual values of correlation coefficients. Before attempts are made to increase our knowledge of member correlation, the importance of correlation effects in structural reliability calculations should be assessed. The paper describes a simple study of the sensitivity of structural reliability assessments to variations in correlation coefficients for parallel structures. It is demonstrated that the structural reliability for the structure considered is sensitive mainly to correlation between different locations within the same member or related members.

The so-called 'rigid' connections in real frames are seldom rigid, and instead contribute to the deflection behaviour of the frame. In conventional practice this is usually ignored. The present paper is a first step in the analysis of the significance of connection flexibility, allowing for both elastic and plastic behaviour of frame members. It is shown that existing formulae for prediction of connection behaviour of bolted end plated connections are inadequate and that connection flexibility can make a significant contribution to frame deflection. However, this contribution varies in magnitude with load level and frame flexibility. A wide spectrum of realistic frames needs to be examined to assess the importance of connection flexibility for designers.

Methods for determining the deflections at service loads in plstically designed structures are presented. A number of approaches to calculating these deflections are considered, since the usual method of performing a completely separate elastic analysis of the plastically designed structure is generally very cumbersome unless a computer is available. The alternative methods considered are: (a) direct estimation from plastic bending moment distribution; (b) correction approach; (c) exact calculation using a newly developed theorem; and (d) an approximate procedure using estimates of the bending moment diagrams for use in the unit dummy load method. Several examples are worked to illustrate the methods. It is shown that method (c) is a powerful direct and accurate technique for most simple structures.

Study of cases of structural failure show these to be inconsistent between various investigators. To resolve these problems, detailed studies of the causes and underlying mechanisms for structural failure are required, even though this severely limits the data that can be employed. Starting from the development of a quite general model of the causes of structural failure, implications are derived for the type of information that can be obtained using limited data. The study of well-reported cases of failure can be useful in identifying ¿error-causing-potentials¿ , but only socially weighted relative values can be established from such a study. Particular attention is given to the relative importance of so-called ¿uncertainty¿ type (or ¿gross¿ ) errors compared to errors arising from natural variability. It is suggested that the manipulation of control processes is likely to be an effective practical means of reducing structural failure.

The role of human error in the effective implementation and use of structural engineering projects is reviewed and shown to be underestimated in conventional treatments of probabilistic structural design. Both suitable organization and control are required for project success. The concept of maximizing total expected socio-economic utility is employed to derive criteria for selecting the levels of control required on the various tasks constituting the construction and use sequence of a structural engineering project. These criteria are applied to a hypothetical example using both the limited experimental evidence available and hypothesized data. Refs.

A sandwich plate is considered in which the reinforcement fibers are located entirely within the outer faces of the plate. The layout of the fibers is determined within the faces of the plate to minimize the total fiber consumption. The variation of fiber material within the optimal plate to resist the internal stress resultants due to a given loading may be obtained once the fiber layout has been determined. Only one loading case is considered for a simply supported plate, clamped plate, and simply supported annular plate.

Earlier papers have considered optimal solutions for various layout shapes of plate but with the edges supported in the classical manner-either simple supports, fixed edges or free edges in various combinations. In these cases there is no work done by supported boundary moments or edge normal reactions. The present study focuses attention on the optimal layouts appropriate to cases where the plate edge is supported on (deflecting) edge beams and hence work is done and energy is stored or dissipated in the boundary zone. © 1974.

The investigation of optimal fibre reinforcement layouts begun earlier1 is continued and extended to another important combination of support conditions. It is also shown that all the deflected shapes for the slabs discussed in the previous paper are everywhere both displacement and slope continuous, and hence are global optima. © 1973.

In the earlier papers in this series optimal fibre layouts have been given for a variety of domain shapes and boundary conditions. The general feature which most of these layouts possess is that the reinforcement fibres are parallel to one another within the various zones. In the present paper attention is focused on cases with non-parallel reinforcement. The most common occurrence of such cases is in slabs with a free edge meeting a supported edge at other than a right angle, and at obtuse corners. A number of results for skew slabs and corners are presented. © 1973.

A general optimality condition is indicated for laterally loaded plates reinforced with fibers in their outer faces only. This result is applied to two examples. The first is a simply supported circular plate for which the thickness variation is specified. It is found that an optimal design has only radial fibres. The second problem is a simply supported square plate under arbitrary loading, for which the reinforcement fibers have different internal lever-arms in bending for different orientations within the plan of the plate.

The work of previous investigators in the field of absolute optimal design of laterally loaded fibre reinforced plates is briefly reviewed and a further approach to the problem is outlined. This further approach suggests that the problem is essentially one-dimensional and consists in determining the optimum load carrying (load-path) directions at all points in the plate. Optimal fibre layouts for square, rectangular and triangular plates with both simply supported and clamped edges are given and some errors in previous work are noted. © 1972.

Copyright © 2013 by the Australasian Corrosion Association.In 2010 the disused railway bridge at St Lawrence on the Queensland Pacific Ocean coast was demolished after 90 years service across the St Lawrence Creek, Queensland, some 6 km inland from the Pacific Ocean and subject large (up to 9 m) seawater tidal movements. Detailed examination of samples about 300×300 mm square and larger cut from several of the bridge piers at various elevations showed corrosion loss varying on average between 2-3 mm for the immersion and tidal exposure zones. This is remarkably little for such aggressive environment and for such long duration of exposure. There was negligible corrosion in the atmospheric zone. The multiple samples allowed statistical analysis of the variability of corrosion losses at different elevations. The statistical results are unique for such long-term exposures and are of major importance for structural reliability analyses of similar old bridges. Also, the corrosion loss results were compared with results reported in the literature for which the longest period of observation is only a few years. Observations show that cast iron is not free from localized corrosion and pitting as sometimes assumed. The corrosion loss results were compared to patterns previously observed for corrosion of steel through the tidal zone. A pattern consistent with so-called accelerated low water corrosion was observed, suggesting that at some period of time in the history of the bridge the tidal seawaters suffered some degree of water pollution.

Copyright © 2013 by the Australasian Corrosion Association.Water injection or water flooding is widely used as a secondary recovery method to increase the production of oil usually from mature oil wells. It requires the injection of water into the reservoir, typically using carbon steel pipelines. Their corrosion is a major issue for the oil and gas industry, involving high costs of periodic chemical treatment, including biocides, oxygen scavengers, corrosion and scale inhibitors, and pipeline pigging. There is also the high potential cost of premature failure of the pipelines relative to life-time predictions, and the associated potential for unpredicted costs and major environmental implications should the pipeline fail. The present paper presents a review of a pilot study supported by the oil industry and the European Community that aimed to improve understanding of the corrosion behaviour of these pipelines. It draws on data and observations for the long-term corrosion, water quality, intelligent pigging records and microbiological testing to gain a better understanding why some pipelines suffer mild corrosion yet others, apparently similar, are prone to highly aggressive channelling corrosion. Herein the documented experience of several pipelines is reviewed. Observation data taken into consideration include pigging operation information, composition of corrosion products and their amount, corrosion rates and pipe geometry, identified biological consortia, water chemistry and process parameters. Also, the distribution of corrosion along the pipeline as well as the localization of corrosion attack is noted. Preliminary correlations between and conclusions about observed corrosion, operational conditions and service history are made. A Joint Industry research project currently is being arranged by the offshore oil industry to extend this work.

Copyright © 2013 by the Australasian Corrosion Association.Atmospheric corrosion tests of mild steel have employed thin flat steel panels of various sizes, with low edge to main surface ratios. The corrosion loss results relate well to the corrosion of flat steel products and surfaces. The few results reported for circular coupons such as wires showed corrosion losses higher than for flat plates (Jeffrey et al). Herein new observations are reported for the development of rusts, delamination of rusts and of atmospheric corrosion losses for mild steel square, hexagonal and circular prisms exposed in the vertical orientation at Belmont Beach for up to 3 years. The corrosion loss results are compared to previously reported results for flat plate coupons at the same site. In general, smaller samples, including the flat coupons, corroded much more than larger samples. For the smaller prisms rusts delaminated within 12 months or less whereas delamination was not observed for the larger samples. It is proposed that this the result of radial or equivalent cracking permitting higher rates of oxygen diffusion to the corrosion interface.

Copyright © 2013 by the Australasian Corrosion Association.The external corrosion of buried cast iron water mains is a significant problem for the Australian Water Industry. To improve maintenance and replacement strategies for these mains the industry requires a method to estimate their remaining service lives. Predictions of long-term corrosion losses, pit depths and pitting extents are therefore required. A conceptual model for the prediction of long-term corrosion has been developed and the external soil conditions influencing corrosion were identified in a previous study. To calibrate the model, long-term data of corrosion (average loss and maximum penetration) and associated soil parameters for pipes in service is required. Field data has been collected at 18 condition assessment sites and two pipe replacement sites within Hunter Water Corporation's network. This paper presents the field observations and measurements taken from these sites, the model calibration procedure and results of the calibration. The results show that the long-term maximum corrosion penetration increases with increasing soil wetness (measured using degree of saturation) over the observed range of degree of saturation (0.5 to 0.86). The initial model calibration produced model parameters for cast iron pipes buried at standard depths in relatively homogeneous, low-permeability, soils with degrees of saturation equal to 0.57, 0.66, and 0.76.

Copyright © 2013 by the Australasian Corrosion Association.The Hornibrook Highway Bridge was a 2684m long timber and concrete bridge spanning across Bramble Bay north of Brisbane, Queensland. Completed in 1935, it was the longest road viaduct across water in Australia at the time. It was demolished in 2011 after 76 years of service due to substantial timber degradation and the deterioration of the concrete due to corrosion. A number of piles were salvaged from demolition and studied by the University of Newcastle. This paper outlines some of the findings from this investigation. The piles and reinforcing bars were in relatively good condition with minimal cross-sectional area losses despite the age of the bridge and exposure to an aggressive marine environment. An exception was the severe corrosion of a bar in one pile with no obvious external visual evidence. A summary of petrographic and XRD analyses, and carbonation profiles are also presented and discussed.

Copyright © 2013 by the Australasian Corrosion Association.There is some evidence that the addition of small amount of aluminium can reduce the atmospheric corrosion of steel. However, systematic long-term investigations are lacking. There also appears to be no information about resistance of such alloys to marine immersion conditions. This paper reports on the marine environment corrosion performance of steel alloyed with aluminium additions of 0% (control), 2%, 4%, 6% and 8%. The cast steel was forged into billets and machined into coupons. Sets of coupons were exposed to three exposure conditions: (a) in temperate seawater for 1, 2 and 3 years, (b) submerged below the mud-line for 1, 2 and 3 years and (c) exposed at two extremely severe atmospheric marine locations for 12 months. The results show that the increase in Al content reduced corrosion loss in submerged conditions from 10.5% after one year to 2.8% after 3 years. In the coastal atmosphere, corrosion loss dropped from 5.5% after one year with no Al addition to 1.2% with 8% Al alloying. The corrosion resistance in marine mud buried conditions was less favourable. Corrosion loss reduced from 4.4% after one year to 2.9% after 2 years but in some instance the addition of Al had minimal or no effect on corrosion resistance. These observations add to the body of knowledge regarding the effect of aluminium on long-term corrosion loss.

An important part of the assessment of the reliability of offshore structures involves the estimation of the reliability of the structural system. Most offshore systems are subject to a number of load processes (e.g., wave load, wind load, earthquake load, etc.). It is clear that not all these processes will be at their peak value at the same time. Hence a representation is necessary which considers the temporal effect of the loading processes. Much conventional structural reliability theory has relied on the use of a 'load combination' approach to represent loads. However, such an approach is not valid unless structures behave linearly elastically to failure. The paper outlines some new directions in the assessment of structural reliability for structures subject to a number of loading processes and which may be deteriorating or changing their resisting values with time. In particular, attention is given to modern methods of simulation to solve the surface integral required for 'time-dependent' reliability evaluation. As is indicated, the computations involves are not necessarily significantly more complex than those which are currently required for so-called 'time-independent' reliability assessments.