This study presents a probabilistic analyses of suction bucket installation in cohesionless soils. The spatial variability of soil properties is quantified using a representative survey dataset from practice. Vertical random field modelling is used to model cone resistance variability and probability density functions are fitted using drained parameter estimates. Conventional installation analysis methods are adapted for layered soil deposits. The effect of varying permeability is included through the incorporation of a finite element seepage model. Parametric uncertainties are considered through a Monte Carlo analysis and the results are interrogated through a variety of feasibility and sensitivity studies. Performing such an analysis allows a designer to gain insights into governing failure mechanisms and objectively quantify the impact of uncertainty regarding parameter estimates.

Knowledge of the fines content is necessary for all soil classification systems and an important factor in the evaluation of soil strength in liquefaction and seismic settlement analysis. This paper presents the application of cone penetration test, CPT data for estimating the soil fines content. The correlation can be used either as a first estimate of fines content (for example in the offshore environment) or to provide statistical information on the variation of fines content within a given area of interest (e.g. for a regional liquefaction study). The paper shows how field and laboratory test data were used with a neural network to correlate the CPT results and the fines content. Data from five site investigation locations across Northern Croatia were utilised. Verification of the approach is performed using field and lab test data from the Veliki vrh landslide.

Soil classification is a means of grouping soils into categories according to a shared set of properties or characteristics that will exhibit similar engineering behaviour under loading. Correctly classifying site conditions is an important, costly, and time-consuming process which needs to be carried out at every building site prior to the commencement of construction or the design of foundation systems. This paper presents a means of automating classification for fine-grained soils, using a feed-forward ANN (Artificial Neural Networks) and CPT (Cone Penetration Test) measurements. Thus representing a significant saving of both time and money streamlining the construction process. 216 pairs of laboratory results and CPT tests were gathered from five locations across Northern Croatia and were used to train, test, and validate the ANN models. The resultant Neural Networks were saved and were subjected to a further external verification using CPT data from the Veliki vrh landslide. A test site, which the model had not previously been exposed to. The neural network approach proved extremely adept at predicting both ESCS (European Soil Classification System) and USCS (Unified Soil Classification System) soil classifications, correctly classifying almost 90% of soils. While the soils that were incorrectly classified were only partially misclassified. The model was compared to a previously published model, which was compiled using accepted industry standard soil parameter correlations and was shown to be a substantial improvement, in terms of correlation coefficient, absolute average error, and the accuracy of soil classification according to both USCS and ESCS guidelines. The study confirms the functional link between CPT results, the percentage of fine particles FC, the liquid limit w_L and the plasticity index I_P. As the training database grows in size, the approach should make soil classification cheaper, faster and less labour intensive.

Rainfall thresholds express the minimum levels of rainfall that need to be reached or exceeded in order for landslides to occur in a particular area. They are a common tool in expressing the temporal portion of landslide hazard analysis. Numerous rainfall thresholds have been developed for different areas worldwide, however none of these are focused on landslides occurring on the engineered slopes on transport infrastructure networks. This paper uses empirical method to develop the rainfall thresholds for landslides on the Irish Rail network earthworks. For comparison, rainfall thresholds are also developed for natural terrain in Ireland. The results show that particular thresholds involving relatively low rainfall intensities are applicable for Ireland, owing to the specific climate. Furthermore, the comparison shows that rainfall thresholds for engineered slopes are lower than those for landslides occurring on the natural terrain. This has severe implications as it indicates that there is a significant risk involved when using generic weather alerts (developed largely for natural terrain) for infrastructure management, and showcases the need for developing railway and road specific rainfall thresholds for landslides.

Many of the earthworks assets on rail transport networks were constructed in the 1800s and have thus operated for periods far in excess of their expected service life. Incidences of failure — particularly shallow planar landslides — are increasing, in part due to the effect of more intense and longer duration rainfall events. Network owners have difficulty in targeting scarce resources to reduce risk across networks. This paper proposes a methodology for developing fragility curves for rainfallinduced landslides on transport networks. Fragility curves provide the probability of exceedance of different limit states for a given hazard considering a range of magnitudes. In this paper, the vulnerability of slopes as expressed by a loss of performance is quantified for rainfall events of various intensities and duration. The approach expands upon probabilistic slope stability analysis and provides a rational logical framework for considering how vulnerable a slope is to rainfall-induced failure.

The trend for development in the offshore wind sector is towards larger turbines in deeper water. This results in higher wind and wave loads on these dynamically sensitive structures. Monopiles are the preferred foundation solution for offshore wind structures and have a typical expected design life of 20 years. These foundations have strict serviceability tolerances (e.g. mudline rotation of less than 0.25° during operation). Accurate determination of the system frequency is critical in order to ensure satisfactory performance over the design life, yet determination of the system stiffness and in particular the operational soil stiffness remains a significant challenge. Offshore site investigations typically focus on the determination of the soil conditions using Cone Penetration Test (CPT) data. This test gives large volumes of high quality data on the soil conditions at the test location, which can be correlated to soil strength and stiffness parameters and used directly in pile capacity models. However, a combination of factors including; parameter transformation, natural variability, the relatively small volume of the overall sea bed tested and operational effects such as the potential for scour development during turbine operation lead to large uncertainties in the soil stiffness values used in design. In this paper, the effects of scour erosion around unprotected foundations on the design system frequencies of an offshore wind turbine is investigated numerically. To account for the uncertainty in soil-structure interaction stiffness for a given offshore site, a stochastic ground model is developed using the data resulting from CPTs as inputs. Results indicate that the greater the depth of scour, the less certain a frequency-based SHM technique would be in accurately assessing scour magnitude based solely on first natural frequency measurements. However, using Receiver Operating Characteristic (ROC) curve analysis, the chance of detecting the presence of scour from the output frequencies is improved significantly and even modest scour depths of 0.25 pile diameters can be detected.

Modern engineered slopes are designed to exceed certain safety targets set out in design codes. This is in stark contrast to earthen infrastructure inherited from the 18^th century which typically was constructed in a haphazard manner without design. This infrastructure seldom meets modern deterministic guidelines yet clearly exhibits some degree of safety, as a failure has not occurred in the intervening years. This paper highlights the use of reliability theory for evaluating the stability of existing engineered slopes. A comprehensive review of geotechnical uncertainty and existing reliability based techniques are outlined. Furthermore, the paper highlights the issue of finding the critical slip surface and gives a brief summary of the current state of the art. Finally a case study of an Irish railway embankment is presented and both a deterministic and reliability analysis is performed on it highlighting the benefits of probabilistic methods over traditional techniques.

Many engineered and natural slopes have complex geometries and are multi-layered. For these slopes traditional stability analyses will tend to predict critical failure surfaces in layers with the lowest mean strength. A move toward probabilistic analyses allows a designer to account for uncertainties with respect to input parameters that allow for a more complete understanding of risk. Railway slopes, which in some cases were built more than 150 years ago, form important assets on the European rail network. Many of these structures were built at slope angles significantly higher than those allowed in modern design codes. Depending on the local geotechnical conditions these slopes may be susceptible to deepseated failure; however, a significant number of failures each year occur as shallow translational slips that develop during periods of high rainfall. Thus, for a given slope, two potential failure mechanisms might exist with very similar probabilities of failure. In this paper a novel multimodal optimisation algorithm (‘Slips’) that is capable of detecting all feasible probabilistic slip surfaces simultaneously is presented. The system reliability analysis is applied using polar co-ordinates, as this approach has been shown to be less sensitive to local numerical instabilities, which can develop due to discontinuities on the limit state surface. The approach is applied to two example slopes where the complexity in terms of stratification and slope geometry is varied. In addition the methodology is validated using a real-life case study involving failure of a complex slope.

The Irish Rail network was largely constructed in the mid-1800s. As a result of this, a significant proportion of the network is comprised of aged cuttings and embankments the construction of which predate modern design standards. Although most of the networks have remained stable for over a hundred and fifty years, a significant proportion of the network has slope angles in excess of those recommended in current design standards. Climate change predictions expect increased rainfall levels across Europe that will deteriorate these slopes further and increase incidence of failure. Current practice when populating earthwork asset databases is to conduct a technical walkover survey. Data obtained in this way is susceptible to bias errors and involve subjective approximations. This is particularly evident in slope geometry attributes such as slope height and angle. Remote sensing data is able to improve precision while reducing bias substantially, while being a much faster alternative than visual assessments on a network scale. Having precise and reliable data over the entire network is a fundamental prerequisite when conducting relative risk assessments of assets. In this paper, post-processed findings from an airborne LiDAR survey of the entire Irish Rail network are presented and compared to walkover assessment data. The current state of assets will also be discussed in light of modern design codes, together with the implications on infrastructure performance. Slope vulnerability to shallow planar type failures is expected to increase with predicted changes in climate such as increased environmental loading (rainfall events are predicted to be more intense and of longer duration, with longer dry periods in between). This type of failure is already the dominant failure mode across Irish Rail network. Typically these failures are instigated by rainwater percolating into the slope to a given depth, filling available pore space thus reducing in-situ soil suctions. This in turn reduces the shear strength of the soil. When the percolating water reaches some critical depth failure occurs. Fragility curves, a particular type of asset vulnerability assessment, provide a connection between triggering actions (such as rainfall) and expected damage to infrastructure assets. They can therefore be potentially useful in estimating a slope's response to predicted future climate loading. An example of a fragility curve applied to a typical slope on the Irish Rail network is presented in this paper.

The high profile failure of the Malahide viaduct in Dublin in late 2009 was attributed to erosion of the supporting soils around the bridge piers, commonly referred to as foundation scour. This is a widespread geotechnical-structural problem, where foundation scour has been identified as the number one cause of bridge failure in the United States. Monitoring scour is of paramount importance to ensure the continued safe operation of the ageing bridge asset network. Most monitoring regimes rely on expensive underwater instrumentation that is often subject to damage during times of flooding, when scour risk is at its highest. Scour causes a rapid reduction in foundation stiffness and can lead to complete failure of one or more sub-structural components of a bridge. In this paper, a novel scour monitoring approach based on dynamic measurement techniques is described. The investigation is based on using accelerometers mounted on the structure of interest to detect losses in foundation stiffness due to scour, which manifest itself as a change in vibration characteristics. Experimental and numerical analyses were performed to validate the potential of this new monitoring framework. A significant advantage of this monitoring method over traditional approaches is that the structure itself is used to monitor the damage. Therefore, if failure is likely, it is assumed that the dynamic characteristics will indicate such and remediation works may be implemented.

Probabilistic slope stability analysis typically requires an optimisation technique to locate the most probable slip surface. However, for many slopes particularly those containing many different soil layers or benches several distinct critical slip surfaces may exist. Furthermore, in large slopes these critical slip surfaces may be located at significant distances from each other. In such circumstances, finding and rehabilitating the most probable failure surface is of little merit, as rehabilitating that surface does not improve the safety of the slope as a whole. Unfortunately, existing slip surface search techniques were developed to converge on one global minimum. Therefore, to implement such methods to evaluate the stability of a slope with multiple failure mechanisms requires the user to define probable slip locations prior to calculation. This requires extensive engineering experience and places undue responsibility on the engineer in question. This paper proposes the use of a locally informed particle swarm optimisation method which is able to simultaneously converge to multiple critical slip surfaces. This optimisation model when combined with a reliability analysis is able to define all areas of concern within a slope. A case study of a railway slope is presented which highlights the benefits of the model over single objective optimisation models. The approach is of particular benefit when evaluating the stability of large existing slopes with complicated stratigraphy as these slopes are likely to contain multiple viable slip surfaces.