The black-box nature of deep generative models like Variational Autoencoders (VAEs) limits their practical application in engineering design, particularly in aerospace, where interpretability is crucial for reliability and safety. This work investigates the use of Symbolic Regression (SR) to improve the interpretability of VAE latent spaces for airfoil shape optimization. Two approaches are developed: a latent analysis of a trained β-VAE and a novel SR-VAE model that integrates SR directly into VAE training.

The first approach shows that SR can approximate the decoder via analytical equations linking latent variables to geometric airfoil features. This even enabled parametric reconstruction independent of the decoder, though accuracy was limited for airfoils with extreme thickness or uncommon trailing edges.

The second approach investigates several SR integration strategies, with a per-batch method followed by retraining with fixed equations achieving the best balance between generalizability and reconstruction accuracy. The parametric equations of this final SR-VAE show an improvement over those from the latent analysis while preserving, and in some aspects even improving, the generative capability of the decoder. The latent space itself showed limited change due to the use of warm starts, suggesting that interpretability through SR is improved primarily at the output level.

The practical applicability of the decoders and equations obtained in this work are tested and compared to CST parameterization, using inverse design tests, as well as constrained and unconstrained optimization cases. The SR-VAE decoder consistently showed the highest reconstruction fidelity for inverse design, but limited use in practical optimizations due to its poor generalizability far from the training set mean. Although SR-based parameterizations show limited reconstruction fidelity in inverse design, they demonstrate competitive performance and the fastest convergence in optimization tasks.

Overall, this work demonstrates that SR can bridge the gap between black-box generative models and interpretable, equation-based design, opening new pathways for explainable AI in engineering contexts and beyond.

A rotary-wing aircraft, or helicopter, is crucial for maritime operations due to its Vertical Take-Off and Landing (VTOL) capabilities, making it essential for small-deck Naval ships. Helicopters serve multiple roles, enhancing the Royal Netherlands Navy’s (RNLN) missions. However, the rise of Unmanned Aerial Vehicles (UAVs) in the past decade has introduced a cheaper, expendable, and more efficient alternative for certain operations, as UAVs require no pilot and can operate continuously. UAVs are ideal for tasks related to improving situational awareness, thereby increasing mission effectiveness, reducing pilot risk, and lowering fuel consumption. The RNLN aims to utilize UAVs alongside crewed helicopters on new ships to maximize operational efficiency. Both helicopters and UAVs follow three mission phases: launch, mission, and recovery, with the recovery phase being the most challenging. During the recovery phase, a crewed helicopter approaches the sailing vessel in a standardized way for which operational limits are established. These limits are essential for rough weather situations and are determined through many flight tests, which is a time-consuming, expensive and hazardous task. In addition, these operational limits only apply to a standardized approach manoeuvre. No approaches for UAVs are prescribed yet, making it difficult to determine such limits, but opens the possibility to standardize newly optimal UAV-ship approaches for which operational limits can be established. Moreover, the ship needs to manoeuvre to obtain optimal wind conditions for an approaching helicopter, which is not desired for UAVs. This means that UAVs should be able to approach and land on the ship in any wind condition. This study investigated the effect of wind on optimized UAV-ship approaches, which can be used to aid and standardize maritime UAV-ship approaches. For this objective, a trajectory optimization framework was created that combined helicopter dynamics and performance, a ship, three wind models of increasing levels of accuracy and a solver to find optimal wind-affected approaches. Trajectories were optimized to be smooth and fast because this is most important for operational purposes. Experiments involved several wind directions, wind speeds and helicopter starting positions, affected by three wind models. This work shows that the created framework functions properly and can be used in a variety of situations. The optimized trajectories were evaluated by the Longest Common SubSequence (LCSS) similarity measure to investigate the effect of both wind direction, speed and model fidelity. With an output between 0 and 1, the LCSS algorithm provided intuitive results and clear trends. A conclusion is that the helicopter wants to exploit the wind as much as possible. It diverts its path to acquire a higher ground speed at the start and a stronger headwind to decelerate at the end of the approach. When introducing a wind gradient, it was observed that the helicopter adjusted its vertical manoeuvre with the same objective. In some wind cases, constraint limits were reached or violated, marking preliminary operational limits. Results were also interpreted with a different perspective to find the best starting positions for the helicopter to approach the vessel in a certain wind condition. These results showed a variety of optimal starting locations, based on the total objective function value, total energy consumption and by distance normalized quantities. Wind model fidelity did not influence this outcome significantly, but it showed that it is important to estimate performance parameters with higher-fidelity models for increased wind speeds. This could also be seen from the downdraft behind the hangar; while the optimized approaches did not deviate geometrically, performance indicators such as the power required were affected significantly.

This thesis investigates a novel load alleviation technique known as mini-tabs, small devices placed on the upper surface of wings to reduce lift. Reynolds-Averaged Navier-Stokes (RANS) simulations were conducted to analyze the impact of geometrical properties on mini-tab performance in reducing the load on a wing.

The initial simulations focused on a single mini-tab, varying parameters such as height, aspect ratio, chordwise, and spanwise positions. Results indicate that increasing height or aspect ratio leads to decreased lift due to flow deceleration at the leading edge and expansion of the separated flow region downstream of the mini-tab. This will also increase the pressure drag. Increasing the height or aspect ratio also accelerates the flow at the lower surface, further reducing the lift generated by the wing.

Placing mini-tabs closer to the wingtip exhibits similar effects on lift reduction as increasing the height of the mini-tabs. This is because the relative height of the mini-tab with the local chord length increases when the mini-tab is positioned closer to the wingtip. A maximum lift reduction is also achieved when the mini-tab is placed at a 60% chordwise position.

Subsequent simulations explored the performance of multiple mini-tabs, revealing that orienting them orthogonal to the wind direction may not be optimal due to gap formation between the mini-tabs that energizes the wake downstream of the mini-tabs and reduces the reduction in lift. Conversely, aligning the mini-tabs with the leading edge has a higher efficiency in the reduction of the lift.

Overall, the findings suggest that mini-tabs offer a viable method for reducing wing forces and for load alleviation

A recent trend in the Aerospace industry is the miniaturisation of spacecraft. Historically these could not be controlled actively, limiting their operations and lifespan. TU Delft has already committed numerous research towards the development of miniaturised propulsion system using different technologies. This thesis continues work on the Vaporising Liquid Micro-resistojet (VLM) developed by Versteeg. Already three other master students have worked with this thruster and encountered several issues. Water tests had been attempted before but were unsuccessful due to the used feed system. Moreover, the latest experiments showed a significant degradation in terms of nozzle geometry and leak rate, which likely led to a decreased thruster performance. Lastly, the change to the TB-50m thrust bench resulted in unexpected drift behaviour.

This thesis addresses these challenges by implementing changes to the thruster en test set-up. The thruster’s sealing surface has been restored to its original state using a CNC-machine. This was done to counter degradation of the nozzle geometry and reduce the VLM’s leak rate. Additionally, a sealing gasket was developed for the same reason. Unfortunately, its implementation resulted in an undesirable deviation of the nozzle’s geometry and was thus left out. A new water feed system is developed, based on the use of a pressurised tank. This reduced unwanted effects, such as bubble formation, that were encountered with the previously used syringe-based feed system. With the implementation of a liquid mass flow meter, the flow rate can be measured more accurately, thus reducing uncertainty in the VLM’s performance. While these improvements proved to be successful initially, several issues encountered during the work have resulted in a worse overall performance. The nozzle geometry already showed signs of deformation after just two months. Furthermore, a major leak was identified which could not be addressed on time. Despite this, tests were executed with chamber temperatures up to 400 °C using both nitrogen and water as propellants. For nitrogen, the measured thrust varied between 4.1 and 8.6 mN for a throat Reynolds range of 1250 to 3300. The specific impulse increased from 34 to 37 s, while the specific impulse efficiency increased from 0.33 to 0.52. Discharge coefficients between 0.66 and 0.72 were measured. Although the errors are comparable (± 15% at most), this performance is significantly worse compared to earlier results. The water experiments were not successful due to an uncontrollable thruster operation. Regardless of this, the obtained test results showed an equally degraded thruster performance. These were executed at chamber pressures of ∼0.85 bar and at various chamber temperatures. Thrust and specific impulse levels of 4.2 mN and 67 s were achieved, with errors as large as ± 34%. The corresponding specific impulse efficiency and discharge coefficient were 0.56 and 0.39, respectively. The throat Reynolds number decreased from 1550 to 1000. Because of these unsuccessful tests, the VLM performance could not be verified against earlier results, let alone used for validating the developed analytical model. Recommendations are provided to address the encountered issues and improve the thruster’s performance. This should enable successful nitrogen and water tests in the future.

The exponential increase in the number of microsatellite launches over the past decade has generated a research gap within the realm of satellite propulsion, needed to maintain the orbit of such satellites and hence increase their lifetime. The research towards the miniaturisation of propulsion systems has hence steadily gained traction within universities and other educational institutions. With this in mind, in collaboration with the Else Kooi Laboratory (EKL), the TU Delft is developing a novel resistojet thruster with integrated Titanium heaters based on Micro-Electromechanical Systems (MEMS) technology, called the Low Pressure Micro-Resistojet (LPM). Using their resistance, the newly fabricated thruster chips were characterised as temperature sensors for the range of 40-140 C. A new interface was designed and manufactured out of Teflon, allowing for the necessary mechanical and electrical connections to the thruster and adhering to Delfi-PQ system requirements. A maximum thrust of 0.37 mN was found during preliminary testing.

This study analyses the feasibility of an engine architecture using bypass cooled cooling air as a method for reducing specific fuel consumption of current and future high bypass turbofan engines. As cooled cooling air reduces required turbine cooling massflow, a major source of loss in modern aeroengines, it is believed that the concept can improve engine efficiency. No extensive or explicit study on the merits of this concept has been performed, however.
Using the 0-D engine analysis software GSP, engine performance parameters at Take-Off and Cruise, such as, specific fuel consumption and (specific) thrust were evaluated against altering feed cooling air temperature by 0 to -300K and heat exchanger induced pressure losses of 0 to 8% in the bypass. This was done for two test cases; the Leap-1A engine (TOC OPR 50, CRZ BPR 11.1, TO FN 121kN) representing current conventional technology and GTF2050, a geared turbofan engine (TOC OPR 75, CRZ BPR 17.1, TO FN 174kN) with entry into service of 2050. For the Leap-1A engine the specific fuel consumption increased up to 1.9% at cruise and for GTF2050, the increase was up to 4.1%. It is shown that the change in fuel consumption is linear with the decrease in cooling fraction, as for both engines the specific fuel consumption increases by +0.34% per percent cooling air reduction.
With a reduction in cooling fraction, the exhaust pressure of the core increases, resulting in a higher thrust (up to 2.3%). It is shown that the optimal take-off bypass ratio for minimum fuel consumption shifts up, for the GTF2050 engine from 16.4 to 17.5.
Furthermore, scenario tests and an exergy analysis were performed to find that the impact of different mechanisms; the pressure loss in heat exchanger has minor impact on performance; change in turbine efficiency has small impact; the effect of changed turbine massflow and heat rejection into the bypass are the dominant phenomena.
It is concluded that as a method for reducing fuel consumption, cooled cooling air with the bypass as a heat sink is not feasible for conventional architectures even with extreme design parameters such as high OPR and BPR. Heat rejection to the bypass cannot be effectively utilised and no improvement in core efficiency from reduced cooling air can compensate. In the edge case where there is no heat transfer in cruise, there is an observed specific fuel consumption benefit of up to 4%. The achievable benefit will be lower, however, when accounting for installation penalties.
It is recommended to focus turbine cooling studies on the impact and feasibility of active cooling massflow control without heat rejection, investigate alternative heat sinks for cooled cooling air (e.g. cryogenic fuel), and only consider bypass cooled cooling air as last resort when up-flowing cooling
schemes is not permissible or ineffective.

Wind farm layouts in industry show a range of patterns with an overall trend from regular to irregular patterns over time. Wind farm layout optimisation studies in literature generally result in irregular wind turbine patterns. Review of existing literature shows that the performance of irregular and regular wind farm layouts has not been compared in a consistent way. Although there are indications that irregular layouts outperform regular layouts, it is yet to be determined if this is the case for the overall performance and if so, to what degree. In this research the effect of regular and irregular wind farm layouts on selected performance indicators is quantified. This quantification is performed through means of a comparative case study.
The performance of both regular and irregular wind farm layouts is assessed on the basis of three performance indicator groups: (1) power performance; (2) wake-induced tower fatigue; and (3) inter-array cabling system. The performance indicators in these groups are affected by a change in wind farm layout, feasible, site independent, and technical as concluded from a multi-criteria decision analysis.
The irregular wind farm layout has a higher annual energy production and a higher persistence to wind direction. The net present value of this increase in cash flow over the lifetime of the wind farm is estimated at 10 million Euros. A higher persistence to wind direction means that the power output is less sensitive to fluctuations in wind direction. This characteristic increases the predictability of the wind farm power, which can indirectly lead to a decrease in imbalance cost on the electricity market.
The wake induced tower fatigue is found to be negatively impacted by an irregular wind farm layout. Implementation of the Frandsen model shows that the maximum effective turbulence of the irregular wind farm is 23.8% higher than that of the regular wind farm layout. For fatigue-driven tower design, this leads to an increase in tower wall thickness, which in turn results in an increase in tower material consumption. The increase in tower cost in the wind farm is estimated at 4 million Euros. Application of a minimum inter-turbine spacing to ameliorate the negative effect on effective turbulence is can lead to a decrease of 20 % (as compared to the case study).
The inter-array cable design results show a marginal increase in cable cost of 1.15 % for the irregular wind farm. The analysis reveals that this performance indicator is strongly dependent on site-specific input data. Due to this marginal change and dependency on site specific input data, this performance indicator is omitted from further conclusions.
Comparing the negative effect of the tower cost and the increase in revenue due to the higher AEP, the net present value is computed. With a discount rate of 5 %, the net present value of the AEP reduced by the increase in tower cost results in an increase of 6 million Euros.
For improved performance in future wind farm layouts, the implementation of irregular wind turbine patterns is advisable. That is, with the proviso that the minimum inter-turbine spacing is taken into consideration with respect to wake-added turbulence levels.

In this thesis research, dutch aircraft noise model (NRM) calculations are compared to measurements to analyze the aircraft classification used in the model. To do so, a new implementation of the model has been set up to calculate with LVNL trackdata and NOMOS measurement data. First, the classification approach is analyzed as prescribed, which shows clustering of aircraft types when compared to measurements. After, the theoretical input data as prescribed for the model is assessed and replaced by real data for each type specific. This leads to improvements of the model calculations when compared to measurements and shows that aircraft-specific flight-information would improve the model significantly. Concluding, it is shown that aircraft noise models strongly rely on the quality of the input data. Current input data shows certain mismatches with respect to reality, resulting in differences between calculations and measurements. However, the calculations can give valuable insights in the effect of operations on the overall noise exposure.

Looking at performance prediction of small coaxial rotorcraft, it was found that there are no fast methods available for designers. This research was aimed at using a blade element method in combination with the momentum theory to quickly predict the performance at small coaxial rotorcraft. These methods are successfully applied in larger rotorcraft. It will therefore be of interest to see how these methods perform when the aerodynamics changes due to scaling. Especially the Reynolds number effects are expected to influence and potentially limit the application of the modelling methods.

The new developed model was tested on the full-scale Kamov 32 helicopter as well as two small Unmanned Aerial Vehicles. These are the commercial available Walkera Lama and the Guardian Angle, which was developed by undergraduate students during a design synthesis exercise. The results show a successful implementation of the calculation methods, with some interesting insight in the limits imposed by them.

Airports are highly complex systems that can generate economic growth
on their own. Accordingly, airports should take proactive actions to
create a status quo between the stakeholders(the airport itself,
airlines, passengers) in the tactical planning of the aircraft stand
allocation. Namely, the harmonization between the stakeholders’
interests is either reactively or not at all considered, so one cannot
be certain that the objectives of the stakeholders are met. For that
reason, a methodology is developed using Weight Space Search on a
many-objective tactical stand allocation model to establish a
reference performance set from which decision alternatives are created
using the k-means clustering algorithm. Decision makers then can
proactively assess and choose decision alternatives on the performance
of the tactical stand allocation to identify how the different
stakeholders can achieve their goals in (partial) synergy. The airport
can also apply the concept of empathetic negotiation to establish a
favorable status quo.

The sound of fly-over aircraft is measured with the use of an acoustic camera to gain more information on aircraft noise. This camera can be optimised to gain data of a higher quality. Within the ANCE section this optimisation was required mainly to diminish the influence of the ground effect in the data and to increase the resolution of the eventual source plots. The increase in resolution was required to be able to separate engine and airframe noise also for the smaller aircraft. The research started with a quantification of the previous fly-over camera of the ANCE section. It appeared that the fringes from Lloyds’s mirror pattern were present, but different for each microphone. This was due to the edge diffraction and should be solved by placing the microphones at least 30 cm from the edge. The increase in resolution should be solved by doubling the array aperture to 3.4 m. These and many other findings were implemented in a new design. This design consists out of a de-attachable rigid structure of 4x4m placed on adjustable legs to which 64 microphones could be clipped in an Underbrink spiral array configuration. The design was built and tested for fly-over aircraft at Schiphol Airport. The measured data was analysed and it appeared that the fringes from the ground effect were gone. Furthermore, the resolution was improved and the engines and main landing could be separated for smaller aircraft as the B737 and A321.