The Netherlands boasts an extensive road network that requires meticulous maintenance and preservation. Dutch asphalt pavements are assessed through functional properties such as mix stiffness and resistance to fatigue. However, current testing practice for these properties is intensive in time and resources, leading to the exploration of alternative methods for performance prediction. Recently, Artificial Intelligence (AI) has emerged as a tool for performance prediction in pavement engineering. Despite its potential, the application of AI is constrained by its limited interpretability and inconsistency with known physical laws. To enhance consistency and interpretability in AI predictive models, Physics Informed AI (PIAI) emerges as a promising approach.

This research develops a PIAI framework for physics infusion in pavement performance predictions. This infusion is accomplished through a Physics-Informed Loss Function balancing data and physics components in model training. The data component assures model predictions approximate the targets, whereas the physics component enforces a subset of features to follow a preset physical model. These components are also present during feature selection, where the physical model is used to guide the inclusion of important features in a PIAI model.

Using the developed framework, this research presents two PIAI prediction models based on the NL-LAB datasets. These models infuse homogenization theory and energy dissipation theory to enhance interpretability and consistency in stiffness and fatigue predictions. The results obtained on both models suggest that physics infusion is feasible without compromising prediction accuracy, balancing physical and statistical knowledge when predicting pavement performance. These findings also indicate that the PIAI framework is a promising approach for infusing physics into AI prediction models. Physics infusion can potentially enhance the acceptance and trust of AI within the pavement engineering community. Furthermore, the developed framework has the potential to accelerate pavement performance assessments by reducing the need for extensive material testing. Its flexibility also supports the incorporation of new physical models, fostering innovation and sustainability in pavement engineering.

Smartphones have become widely accessible and advanced, pocket-sized computers capable of sensing, sending, and processing an ever-increasing variety of data. Modern phones are, for example, equipped with sensors such as accelerometers and GPS-sensors. Meanwhile, an increase in rail traffic leads to an increase in the need of having a reliable and comfortable railway system, an objective which can be achieved through additional monitoring. Through mobile crowdsensing, the smartphone data from passengers could potentially provide an additional and almost continuous information source on the railway system health state.

This thesis explores to what extent smartphone accelerometer measurements on board of in-service passenger trains can be used to monitor railway track quality. The thesis contains a literature review and a survey with European infrastructure managers. They identify that repeatability, as a result of operational variance, is a key research gap and challenge in the implementation of this technology. The thesis assesses accelerometer characteristics from a set of contemporary smartphones through laboratory tests. The tests uncover limitations and heterogeneity in the sensing capabilities such as sampling frequencies, low-pass filters and eigenfrequencies of the devices. The thesis also makes use of a case study, in which different smartphones are placed in various positions on different in-service passenger trains, running over the same tracks in the Netherlands. The vertical acceleration signals are analysed and compared. The vehicle speed and the position of the smartphone within the car body are found to significantly influence both the frequency content and vertical acceleration amplitudes measured. Finally, the different signals are related back to a Track Quality Index based on the standard deviation of the Longitudinal Level D1 track parameter. A degree of consistency in identifying locations with a low relative track quality is displayed, albeit with a significant degree of variance due to the previously mentioned factors.

This research addresses the challenge of standardising and integrating information from laboratory tests in pavement engineering to enhance lifetime predictions. Data from laboratory tests is often scattered across various platforms, stored in disparate formats, and described using inconsistent terminologies. This fragmentation poses significant challenges in data sharing, retrieval, and interpretation, leading to inefficiencies and potential errors. The objective is to develop a comprehensive ontology that organises and connects extensive laboratory data from the Knowledge-based Pavement Engineering (KPE) project. This framework aims to facilitate enhanced data interoperability, retrieval, and application, thereby contributing to more efficient and effective pavement design, construction, and maintenance practices. This objective leads to the main research question: “How to effectively standardise and integrate laboratory information in pavement engineering to enhance lifetime predictions?”

The development of the ontology follows a structured methodology involving several steps. The process begins with identifying relevant stakeholders, such as pavement researchers, road owners, contractors, and third parties, and to understand their specific objectives and requirements. Following this, competency questions are formulated to define the scope and functional requirements of the ontology. These questions help in identifying the necessary classes, attributes, and relations within the ontology. The subsequent step involves specifying the structural requirements, followed by showing the alignments with the existing Dutch standard. This structured methodology ensures that the developed ontology is comprehensive, addresses the stakeholders' needs, and utilises existing knowledge.

The Lab Ontology focuses on standardising and structuring laboratory data related to pavement engineering. It categorises key concepts such as laboratory instrument, testing conditions, sample properties, and their relations. By establishing clear definitions and relations, the ontology enhances data consistency, comparability, and interoperability. Ultimately, the Lab Ontology answers the main research question by providing a standardised framework that improves data sharing and enhances decision-making processes in pavement engineering. This framework can significantly contribute to improving the lifetime predictions of pavement materials.

Transition zones in railway tracks are locations with an abrupt change in stiffness in the vertical rail supporting structure. These locations are typically located as approaches near engineering structures where there is a sudden change in track substructure. Due to these abrupt changes the dynamic forces of the wheel-rail interaction on the railway track are significantly amplified resulting in the deterioration of track geometry. With higher operational velocities, the degradation process of the track is accelerated. The goal of this research is the investigation of the relationship between vertical acceleration from the axle box of the vehicle bogie (axle box accelerations - ABA) and track geometry changes at transition zones. While most studies investigate the behavior of railway transition zones making use of finite element models, this study focuses on the use of a more computational efficient multi body simulation (MBS) software (VI-Rail flextrack).
After the software is tested and the results validated, different stages of the service life of a transition zone (new - used - heavily used) are simulated. From the results two main conclusion can be drawn. First, the vertical acceleration is able to show the changes in differential settlement and stiffness that are occurring at transition zones. The indicators for these changes are the frequency responses with a wavelength between 1.2 < 𝜆 < 5 meters. This shows that ABA is a powerful non-invasive monitoring technique for long wavelength track irregularities. Second, the multi body simulation software is able to model complex railway tracks. The software shows the same characteristic frequencies as the measurement data does.
The results of this investigation could especially be of interest for asset owners and contractors. By showing that ABA measurements tends to be an effective way of monitoring transition zones, predictive maintenance could be implemented which saves time and high costs.

This research project aims to implement an elasto-plastic Neo-Hookean Alpha material with a Desai flow surface with hardening and softening in Abaqus via the Continuum Mechanics framework. This is done by first defining the elasto-plastic potato diagram and deriving the corresponding stress tensor definition and dissipation inequality. The elastic deformation gradient is needed in the definition of the stress tensor, which is found by using the Newton-Raphson procedure for the maximization of the dissipation inequality subjected to the plastic flow surface. The Neo-Hookean Alpha elastic material, the Desai plastic flow surface and the definition of the hardening and softening curve are given and are substituted in the stress tensor definition and the corresponding Newton-Raphson procedure, which finishes the derivation of the theory needed to implement this model. This model is implemented in both Abaqus and Python, where the latter is used for rapid prototyping of the model. The methodology of verifying both implementations is presented. The results of the verification of the implementations are given and show that the behaviour of both implementations is as expected and that both implementations correspond to each other; although there are a few small issues. In the end, these issues are discussed, and future recommendations are given to solve them.

For railway travel the Insulated Rail Joint (IRJ) is a critical part in most railway safety systems but at the same time also considered a weak link. The IRJ creates a discontinuity in stiffness and geometry that leads to wheel-rail impact forces. The angled IRJ is a proposal to reduce these impact forces. Despite the real-world experiences with angled IRJs the research on the topic is marginal. In this thesis a study is presented into the dynamic behaviour of the wheel-rail impact occurring at IRJs. Numerical models are established to simulate a wheel rolling over IRJs with angles of 0, 15, 30 and 45 degrees, using the implicit-explicit sequential finite element method with the software ANSYS/LS-DYNA. The impact forces are the main output and analysed. Three variations of the IRJ models were used for simulations. The basic version IRJ model is built up with a fully constrained rail foot to simulate an infinite support stiffness condition. The second version is supported by spring and damper elements to simulate supports such as ballast and rail pads. Height differences between the rail ends are generated when a wheel passes the joint, which corresponds to a real-world scenario. Different degrees of support degradation are simulated by varying the support stiffness, using the spring and damper element parameters. The third version also incorporates the spring and damper elements as support but couples some nodes between the fish plate and rail web in vertical displacement to reduce the rail height difference caused by wheel pass-by, aiming to simulate a ‘factory-new’ condition joint. The established models were validated in terms of wheel-rail contact solution and contact force amplitude. In comparison with existing FE wheel-rail impact models the proposed models in this thesis are less time consuming and more flexible for joint angle adjustments. The second version simulating degraded joint conditions provided the most interesting findings and was used in a sensitivity analysis by varying the velocity and wheel load. The simulation results show that angled IRJs are advantageous when degradation is present, whereas in ‘new joint’ conditions the gap width plays a significant role and angled IRJs with larger gaps produce higher impact forces. Based on the results the recommendation is given to consider angled IRJs on tracks with a low maintenance scheme and further research is suggested.

With rapid industrialisation, the infrastructure sector has seen exponential growth in prefabricated concrete elements due to their speedy construction and efficient usage of material. Precast concrete elements have thus observed some deterioration due to increased internal temperature as a result of rapid curing, and also through deliberate heat curing techniques. This has led to researchers in the past to study the effect of such curing conditions on the durability aspect of the binders, especially their impact on Delayed Ettringite Formation. Precast elements such as railway sleepers, exposed to in the humid environment were thus prone to internal sulphate attack and needs to be investigated. Use of Ground Granulated Blast Furnace Slag (GGBFS) as a substitution for binder content in conventional portland cement in the Netherlands has been prevalent since the early 1900s primarily because of its abundant resource from iron industry. The benefits of slag have been then exploited as it was one of the supplementary cementitious systems which along with being a sustainable solution, provides good resistance to environmental degradation such as chloride penetration resistance. However, their advantages surrounding extreme curing conditions have to be studied, unless used at optimised quantities. The research focused on the potential of blast furnace slag systems to undergo internal sulphate attack due to high internal temperatures. The simulation of high internal temperature was done through heat curing inside an oven following which continuous storage under lime solution was carried out in order to saturate the system. Slag systems at low and high substitution levels (20% & 50%) were used along with a combination of coarser and finer surface areas, to investigate their subsequent influence was chosen for the study. Also, since infrastructure industry adopts CEM II & CEM III-A cement type, where the former was low slag concentration with moderate fineness and the latter with higher substitution level of slag in combination with high overall fineness, their potentials for DEF have also been studied. For all the mixes, the influence of high curing temperature and exposure to moisture was studied through microstructural changes, pore size variations and mineralogical composition effect along with fineness on paste specimens. All studies were compared to reference systems of CEM I, which was observed to be the most detrimental due to DEF. Test results indicate that at lower substitution levels of slag secondary ettringite forms in significant quantities in neat systems along with traces of carbo-aluminate phases in the case of slag systems. Also, higher substitution levels does not appear to completely suppress the formation of ettringite after exposure. Its formation in both the cases showed more or less no influence of fineness of slag added, except in the case of pore size distribution. The significant presence of carbo-aluminates was observed in the case of all slag systems that could prove to be beneficial as they do not translate to deteriorative expansion.

Porous asphalt resides on most top layers of Dutch roads. Scheduling maintenance for these roads is generally dependent on several factors, but ravelling, the loss of aggregates in the top layers, is the main reason for maintenance on Dutch roads. With the recent framework of the DOS-LCMS scheme generating values for aggregate loss in percentages, a prediction for the remaining lifetime of a road section surfaced with porous asphalt with respect to ravelling can be performed. The lifespan for porous asphalt layers is dependent on the most suffered 25% of the section on the respective 100 meter length.
The current threshold is set at 10%, implying that road sections of 100 meter need maintenance if more than 25% of the road (75^th percentile) measures aggregate loss over 10%. The present work approximates these 75^th percentiles throughout the years using parametric and non-parametric approaches, whereafter the estimates of the 75^th percentiles are used to construct smooth monotonic increasing convex curves. These curves, which are in fact functions built on P-splines, are then used to perform extrapolation and hence predict the dates on which the threshold is going to be surpassed. The study reveals problems in the raw data which is particularly prominent in the sequence of 75^th percentiles, frequently showing a lack of monotonicity and convexity. Putting the monotonicity and convexity constraints on a more granular level were found to be helpful for the predictions and improved the consistency of lifetime predictions over consecutive years.

The suitability of a cylindrical concrete sliding hinge connection in modern reinforced concrete structures has not been investigated yet. This type of connection is known for its large normal force capacity and a rotation limit only based on the geometry of the connection. For the case of underground constructions, large normal forces are often present at the wall-to-floor connections. Considering the risk of uneven settlements, such connections may also demand sufficient rotation capacity. This research is aimed at deriving the limit state of the connection and the surrounding concrete structure. A test program is carried out to support the research, where full scale experiments are performed at the Stevin II laboratory of the Delft University of Technology. The goal is to prove the functioning of this type of hinged connection in a system before it is implemented in practice. Different analytical contact models are investigated and implemented aiding in quantifying the performance of the connection. A finite element model is used to verify the analytical models and to obtain the stress distribution in the connection. Two different bearing types are investigated in its ability to aid the sliding in the connection: a bituminous and a PTFE-stainless steel interface. Both the PTFE-stainless steel and the bituminous interface are adequate to allow sliding with low rotational resistance. Significantly lower bending moments are obtained during testing compared to a conventional monolithic connection. From the testing project it is found that the capacity of the cylindrical hinge for the given reinforcement design is significant when loaded in the normal direction. The concave concrete element can fail in a splitting failure mode when loaded to the normal force limit. The bituminous interface shows the ability to spread contact stresses and allow for reduced cracking under large normal force loads. Large loads in shear, in combination with low levels of normal force is a point of attention to prevent shear cracks, irrespective of the type of bearing. It is found that the hinge functions fine with favourable frictional behaviour for large shear forces indicating the ability to cope with these forces when a sufficient reinforcement design is implemented.

Skid resistance is an important parameter for road safety and therefore it is essential to monitor the skid resistance of pavements. In the Netherlands, skid resistance is measured with either the RWS Skid Resistance Tester -measuring the longitudinal friction coefficient- or the Seitenkraft-Messverfahren (SKM) -measuring the sideway friction coefficient. The latter is nowadays the preferred measurement device by Rijkswaterstaat. Skid resistance depends much on the vehicle speed: the higher the speed, the lower the skid resistance. Furthermore, the texture of the surface influences the speed dependency. Because it is not always possible to measure the skid resistance at target speeds set by Rijkswaterstaat, there is a demand for a speed conversion model for the skid resistance measured with the SKM. The objective of this research is therefore formulated as follows: the development of a speed conversion model for the wet skid resistance, measured with the SKM at different speeds, taking into account the macrotexture of the road surface. The used dataset consists of 718 sections of 100 metre, measured at the Dutch road network. Measurements were performed at 10 different roads with different pavement layers: porous asphalt, concrete pavements, dense pavements and stone mastic asphalt. The mean profile depth of the pavements varies between 0.21 and 1.80 mm. The performed measuring speeds were 40, 60 and 80 km/h, and for few sections 30 km/h. Three regression methods were performed. Firstly, a multiple linear regression was performed. The datapoints consisted of combinations of two measurements at different speeds on identical 100 metre sections. The second method estimated per 100 metre section a zero speed intercept which was used as a reference point. The third method used multilevel modelling and includes a hierarchical structure. Concluded was that the multiple linear regression on speed combinations is inappropriate for the objective of this research, because of two reasons: the datapoints are dependent on each other, and information is lost by splitting the 100 metre sections into datapoints with combinations of two measurements. In the second method, problems arise with estimating the zero speed intercept. The third method is most appropriate for this research and the three-level structure fits best on the dataset. The first level contains the individual measurements on the 100 metre sections, performed at different measuring speeds. The second level consists of the 100 metre sections and the third level consist of the roads on which the measurements took place. The standard error of the model on the training data is 0.032 whereas the average change in skid resistance for two datapoints is 0.053. This average change includes conversions over a speed difference from 10 to 40 km/h. From a sensitivity analysis of the macrotexture it was concluded that if no macrotexture can be measured, it is advised to use a different model in which no macrotexture is included. Recommendations for further research include among others registering more accurately the type and age of the measured pavements and extending the dataset with measurements performed at low measuring speeds and on curved sections. Furthermore, it is recommended to perform a more comprehensive outlier analysis and to optimise the hierarchical structure.

Despite therelevance of the road infrastructure, the mechanisms governing the mechanicalproperties of asphalt concrete pavements, are currently not sufficientlyunderstood. Many empirical models of different complexity are proposed in theliterature; however, (i) they all have below high (R2 =0.85) predictiveaccuracy; (ii) they are inflexible; and (iii) their prediction uncertainty isseldom quantified. This M.Sc. thesis aspires to overcome these three majorchallenges. It focuses on the prediction of three important mechanicalproperties of asphalt mixes: stiffness, resistance to permanent deformation,and indirect tensile strength (ITS). The used data are part of the NL-Labproject and represent six road work projects for asphalt concrete bottom andintermediate layers. The number of data points available for the threeproperties ranges from 100 to 400. Two approaches are used to predict thefunctional properties: multiple linear regression (MLR), and a machine learningtechnique: gradient boosting (GB). For both approaches the root-mean-squareerror is used as loss function and 5-fold cross-validation is applied to ensurea balanced fit. It is demonstrated on the NL-Lab dataset that (i) GB canachieve high and very high predictive accuracy; (ii) it is sufficientlyflexible to capture complex non-linear relationships; and (iii) its predictionuncertainty is low and can be estimated at the same computational cost asfitting a GB model. The predictive accuracy of the GB model significantlyoutperforms that of the MLR. For example, for stiffness: R2GB = 0.96 vs. R2MLR= 0.62, and for ITS: R2GB = 0.82 vs. R2MLR = 0.72. It is shown that the averagestandard deviation of the prediction uncertainty of the GB model is less thanhalf than that of the MLR model. Based on the completed analyses GB is stronglyrecommended over MLR for modelling asphalt concrete functional properties. Tothe author’s knowledge, this work is the first application of GB to modelasphalt concrete functional properties and the first that developed high andvery high predictive accuracy models for these properties. The results arepromising and encouraging further research into this subject.