The aerodynamic interaction between a lifting surface and a ground plane presents a highly non-linear landscape that is critical to the performance and safety of racing vehicles, wing-in-ground (WIG) craft, and aircraft during take-off and landing. While steady-state wing-ground interaction is well-documented, the coupling of unsteady kinematics with ground proximity remains a frontier in aerodynamics, particularly regarding the evolution of the dynamic stall vortex (DSV). This thesis presents a comprehensive experimental investigation into the steady and unsteady aerodynamics of an infinite NACA0015 airfoil operating in ground effect at a Reynolds number of Re≈2×10⁵. By combining high-frequency surface pressure measurements with phase-resolved 3D Particle Tracking Velocimetry (3D−PTV) using the Shake-the-Box algorithm, the study resolves both surface-level load transients and off-body wake and vorticity dynamics.

The investigation initially establishes a steady-state baseline, revealing that ground influence is non-monotonic. At large clearances (H>0.8), the flow is largely unaffected by the presence of the ground. At moderate clearances (0.4<H≤0.8), lift enhancement is driven by the Venturi effect, where geometric confinement accelerates the underbody flow. In extreme ground effect ($H≤0.4$), a viscous-confinement domain is observed. In this state, the interaction between the airfoil and ground-plane boundary layers induces a pressure-side blockage that caps circulation growth, a phenomenon that contradicts classical inviscid predictions.

In the unsteady domain, the airfoil was subjected to sinusoidal pitching across a matrix of reduced frequencies (k=0.05 to 0.20) and ground clearances (H=0.1 to ∞). In attached-flow conditions, ground proximity modulates the convective time lag of the wake. At intermediate heights, the accelerated gap flow reduces the time required for circulation to equilibrate, effectively compressing the lift hysteresis loops. At H=0.2, physical blockage and air-column stiffening significantly widen the hysteresis loops, indicating a substantial increase in phase lag relative to freestream conditions.

The most significant findings emerge in the dynamic stall domain. The presence of the ground plane serves as a powerful stabilising mechanism. By imposing a vertical geometric constraint, the wall suppresses the coherent roll-up and subsequent shedding of the Leading-Edge Vortex (LEV). This suppression of discrete vortex shedding replaces abrupt load collapses characteristic of classical dynamic stall with geometrically constrained separation. This transition results in a marked increase in aerodynamic damping, suggesting that ground proximity inherently mitigates the energy extraction mechanisms that typically trigger instabilities.

This work demonstrates that the non-dimensional height must be considered as a primary kinematic parameter, and reduced frequency alone is insufficient to characterise unsteady stall in ground effect. These insights provide a quantitative foundation for the development of active aerodynamic control systems and offer a physical explanation for the suppression of porpoising in high-downforce racing applications and low-flying vehicles.

Aerodynamic drag reduction of heavy-duty vehicles is essential for lowering fuel consumption, with the front wheels contributing significantly to overall drag. Truck aerodynamics is typically investigated using Computational Fluid Dynamics (CFD) and small-scale wind tunnel testing. However, validating CFD results in the highly complex and unsteady front wheel wake region remains challenging. Conventional validation methods, such as wind tunnel measurements or wake-rake techniques, are often limited to simplified models or planar data.

This study addresses this limitation by applying the non-intrusive “Ring of Fire” measurement technique to reconstruct the three-dimensional front wheel wake of a full-scale truck. The objective was to design and implement a Ring of Fire setup capable of resolving the wheel wake region to support CFD validation, while also evaluating the conventional wake-rake method.

The Ring of Fire measurement technique is based on Particle Image Velocimetry (PIV). Helium-Filled Soap Bubbles (HFSBs) served as tracer particles, illuminated by high-power LEDs and recorded by high-speed cameras. A Shake-The-Box (STB) Lagrangian Particle Tracking (LPT) algorithm reconstructed three-dimensional particle trajectories, from which velocity fields were derived. For comparison, a wake-rake equipped with Kiel probes measured a two-dimensional total pressure field in the wheel wake.

Experiments were conducted on a test track using a full-scale European cab-over-engine tractor–trailer combination. Two configurations were tested: a baseline (Variant A) and a reduced-aero configuration (Variant B). Containing the HFSBs in an outdoor environment was a major challenge and was addressed using a foldable dome tent housing three vertically oriented LED units and a seeding rake. Four high speed cameras recorded the motion of the HFSBs within a measurement domain extending 1 meter from the truck surface and up to 1 meter in height, covering the longitudinal distance of 2.2 meter. A dedicated run procedure ensured synchronization between dome opening and data acquisition during truck passage.

A total of 100 runs were performed over two days. 8 valid runs for Variant A and 21 for Variant B were processed after excluding invalid measurements. Crosswind effects were assessed using conditional averaging and were found to be negligible within measurement uncertainty. Convergence analysis indicated that at least 16 combined runs were required to achieve a stable, time-averaged flow field with over 95% spatial data coverage.

For both configurations, the wheel wake region, characterized by significant reduced velocity magnitude, emerged downstream of the footstep and expanded further downstream. Overall, Variant A showed a larger wake region.

Comparison with wake-rake results revealed similar wake topology, but the Ring of Fire provided higher spatial resolution and full three-dimensional reconstruction. The study demonstrates that the Ring of Fire technique successfully captures the complete front wheel wake of a full-scale truck and offers substantial advantages over conventional planar wake-rake measurements.

Closer propulsion-airframe integration through the use of boundary layer ingesting promises significant improvements in aircraft fuel consumption. However, little research has been performed on how these intakes are to be designed for swept-wing aircraft. In this study, a trailing-edge-mounted, parametrically designed buried engine intake is designed for a highly swept transonic wing. To connect the wing to the nacelle, two diagonal ramps are created. These allow for a smooth transition between the two, while minimizing the wetted area and allowing more control over the bottom surface of the connection. The intake is initially designed as an axisymmetric body, after which the intake is reshaped to allow for elliptical highlight and throat profiles. The intake duct can also be angled left and right, up and down, to align it with the incoming flow. This is a key feature, as a highly swept wing causes a significant level of cross-flow. Since the engine is placed near the trailing edge, there is also a large downwashing component to the flow. By angling the intake, it can be aligned with the flow. The intake is designed and parameterized in 3DX’s CATIA to allow quantifiable changes to be made to the design. The designs are tested using a compressible RANS simulation with a 𝑘 − 𝜔 SST turbulence model and third-order MUSCL spatial discretization at 𝑀 = 0.85, 𝐶𝐿 = 0.1719, and 𝑅𝑒 = 104.8 million. The initial baseline design has a TPR of 0.973, despite a section of flow reversal at the crown of the intake, equal to 3.8% of the fan area. The ingested boundary layer is significantly thickened ahead of the intake due to excessive curvature. This causes a DC60 of 0.349, unacceptable for modern aero-engines. The lift coefficient has increased by 12.2%, mainly caused by a sectional lift increase outboard of the engine. The drag coefficient has increased by 4.8%, although the wetted area has increased by 8.8%, indicating an increase in aerodynamic efficiency. The highest pressure recovery found is 0.985 for version 1 and is acceptable compared to conventional podded engines. However, it still featured a large portion of separated and reversed flow at the crown. Therefore, 5 further iterations were made, finally arriving at an intake with a similar TPR of 0.983, but with almost no reversed flow. The remaining total pressure losses are predominantly caused by the ingestion of the wing’s boundary layer. The DC60 has reduced to 0.171, while the SC60 is 0.0874. The lift coefficient has remained the same w.r.t the baseline design, but the drag coefficient has significantly reduced, being 3.0% less than that of the clean wing. During the iteration process, the angling of the intake into the flow was found to be most critical when trying to improve performance. Specifically, the centerline toe angle, angle of incidence, and throat offset were found to have the greatest effect on the flow quality at the fan face.

Vortex particle methods (VPM) have historically faced computational bottlenecks caused by the pairwise interactions of particles in solving the vorticity-velocity coupling. Several algorithms have been developed to address this bottleneck, including the Fast Multipole Method (FMM). This work describes the development of a FMM solver which is integrated within an in-house developed open-source VPM solver. The FMM improves the computational efficiency of particle velocities and velocity gradients from order of N^2 to order of N. The formulation uses a full octree structure and a Cartesian formulation of multipole and local expansions based on Taylor series.
Inline with the increasing adoption of GPUs in performing scientific computations, the solver developed assigns computationally heavy tasks to parallelised execution on the CPU or GPU, using the taichi compiler. The accompanying introduction of reduced floating point accuracy introduces new considerations for the accuracy of the method when performed on the GPU. Error quantification is performed on a single timestep for a range of flow scenarios, with an emphasis on the effect of main FMM parameters on accuracy. Results indicate a positive impact of increasing the Taylor series truncation order on FMM accuracy, up to a plateau caused by the accumulation of rounding errors. The octree depth was found to have low impact on the accuracy. Validation of the solver is performed through underresolved direct numerical simulation (DNS) of Lamb-Oseen vortex, vortex ring, and colliding vortex ring test cases. The impact of FMM on particle velocities and positions throughout simulations is measured, and key flow diagnostics are provided, indicating minimal FMM impact on the physical validity of simulations.
More chaotic flows exhibited an unbounded nature of FMM error growth while low Reynolds number flows counteracted the FMM error. The performance of the main operations of the solver was profiled, highlighting trade-offs between speed and accuracy, and indicating areas for future performance improvements.

Conceptual design techniques are used to investigate what aircraft configuration is best suited for liquid-hydrogen based medium-range aircraft. The baseline aft tank layout is known to have several aspects of its design that detract from its performance, primarily stemming from its large CG range. Three alternative configurations are examined, and it is observed that while the weight and drag of these configurations change by a large margin, the net effect on fuel consumption is moderate. Among these, the top-tank concept provides the best block energy consumption, while also improving on qualitative features of the design. Unexpectedly, the application of operational limits with the design of the baseline aircraft is found to result in much larger performance increments; at the cost of reduced flexibility for off-design missions.

This thesis presents a comprehensive numerical investigation of wall interference effects in transonic wind tunnel testing of swept-wing configurations, by the example of the Cryogenic Ludwieg Tube Göttingen (KRG) at the German Aerospace Center (DLR). By employing Reynolds-Averaged Navier–Stokes (RANS) simulations using the DLR TAU code and the Spalart–Allmaras turbulence model, the study quantifies how test section walls affect the spanwise flow homogeneity around a swept wing at high Reynolds numbers. A detailed analysis was conducted to isolate the influence of key flow parameters, like angle of attack, sweep
angle, chord length, and Mach number, on wall-induced distortions.

The simulations reveal that vertical tunnel walls introduce significant spanwise pressure gradients, especially at higher sweep angles and larger chord lengths, i.e., aspect ratios. These effects are mitigated through a mesh deformation approach based on Radial Basis Functions, which enables the adaptive reshaping of tunnel walls to align with streamlines from an idealized, unbounded flowfield. The implementation of these deformed geometries resulted in substantial improvements in spanwise flow uniformity, reducing the root mean square errors in the pressure coefficient distribution by nearly 50% in critical test section regions. Not only is it possible to reduce spanwise gradients in the flowfield, but also to reduce the difference to infinitely swept conditions almost over the entire width of the test section.

Beyond identifying optimal strategies to reduce the influence of the walls on the flowfield, the thesis validates its numerical findings against existing experimental and numerical data. The proposed methodology offers a valuable toolset for pre-test planning and supports the development of more representative aerodynamic experiments, especially in the context of laminar-to-turbulent transition studies. These contributions provide a scientifically robust foundation for minimizing wall interference in transonic testing and support the advancement of research on laminar-to-turbulent transition.

The goal of the research is to improve the fidelity of the TUDelft HIL numerical simulation by developing a higher fidelity mooring line model, both quasi-static and dynamic formulations. Furthermore, second-order wave excitation forces are included using Newman's Approximation to more accurately model the sea state. This implementation is verified versus OpenFast simulations. The improved model is then tested in the OJF windtunnel to test the different models and compare between them. Here combined wind-wave action are tested for a range of operating conditions. The effects of second-order forces are quantified, as well as the effects of aerodynamic loads on the motion of the system. Finally, a fundamental study is done on the aerodynamic damping at a range of wind speeds, in surging, pitching and yawing motion.

The integration of Generative Artificial Intelligence (GenAI) in engineering presents both opportunities and challenges, particularly in complex, safety-critical systems. This thesis explores two key aspects: (1) developing an AI-powered reverse engineering tool for Knowledge-Based Engineering (KBE) applications and (2) designing a communication framework to enhance interdisciplinary collaboration in AI-driven projects. The aerospace engineering component introduces the REProcess tool, leveraging Large Language Models (LLMs) to extract structured process representations from KBE applications, improving traceability and automation. The science communication component investigates collaboration barriers between AI experts and domain specialists, leading to the development of a structured ideation matrix for GenAI-powered automation systems. Together, these contributions demonstrate how Explainable AI principles can facilitate the adoption of GenAI in engineering workflows while bridging communication gaps. This interdisciplinary approach provides practical methodologies to enhance AI-driven engineering processes, ensuring both technical feasibility and effective human-AI collaboration.

A proven concept of the past that yields great potential in the reduction of shipping GHG's, is wind assisted ship propulsion or for short WASP. In particular, the Magnus effect driven Flettner Rotor, a rotating cylinder concept placed upright on the deck of a ship, is on the forefront of promising candidates, mainly due to its substantial aerodynamic force producing capabilities and its relatively low power consumption compared to more conventional ship propulsion.

To accommodate aerodynamic rotating cylinder research at the Low Speed Laboratory of the TU Delft, this study presents the design, realisation and characterisation of an experimental two-dimensional rotating cylinder setup for the SLT wind tunnel facility. Additionally, the effects of testing at a high blockage ratio were investigated for this initial study using the new setup. As a result, the area model blockage ratio deliberately reached an unconventional high 33%, where an aspect ratio of 4.5 was chosen, comparable to full scale WASP applications. The aerodynamic characteristics for three test cases at subcritical Reynolds numbers equal to 62500, 125000 and 250000 with spin ratio ranges equal to 0 ≤ k ≤ 8, 0 ≤ k ≤ 4 and 0 ≤ k ≤ 2 respectively, were obtained by the utilisation of force measurements, particle image velocimetry flow visualisation and proper orthogonal decomposition techniques.

As with similar rotating cylinder experiments, the magnitudes of the presented force coefficients differ in comparison, owing to the large blockage ratio and variations in experimental setup, however the experimental trends in the force coefficient curves are fully captured up to spin ratios of 2.5. Between 0 ≤ k ≤ 0.8, the aerodynamic characteristics are affected by both the Reynolds number and spin ratio, where differences in boundary layer transition, laminar separation bubble formation, boundary layer lengths and vortex shedding behaviour are the underlying mechanisms causing a change in the obtained force coefficients. Transition effects are stronger at the higher subcritical Reynolds numbers, where reversal of the Magnus effect can occur within this range of spin ratios and Reynolds numbers. Beyond k > 0.8, advancing side boundary layer transition has taken place for all tested Reynolds numbers. The force coefficients and flow field topology showed great similarities between the different Reynolds numbers with a further increase of the spin ratio. Slight variations in turbulent kinetic energy production were observed in the retreating side boundary layer and shear layer behaviour, possibly owing to laminar to turbulent transition. For k > 2.0, the periodic shedding of vortical structures is no longer observed, which coincides with the formation of the highly turbulent rotating boundary layer, settling of the near-wake, and the knee in the force coefficient curves within the range of 2.2 < k < 2.6. For remainder of the spin ratios, k > 2.6, the flow field has completely reversed and is pointed in opposite direction of the incoming free stream, where large negative mean drag coefficients and instantaneous force fluctuations are recorded. This is believed to be the result of unsteady nozzle-model gap interactions with the separated region of the cylinder and is dictated by the large blockage ratio.

The Delft Laminar Hump (DeLaH), discovered by the Aerodynamics Department of the Faculty of Aerospace Engineering at the Delft University of Technology, is a symmetrical smooth hump that is placed at a set distance parallel to the leading edge of the wing that reduces skin friction drag as it attenuates the growth of the crossflow instabilities (CFI). With the hump, transition can be delayed up to 14%. The only requirements are that the hump needs an natural laminar flow (NLF) airfoil and a condition where CFI is the dominant transition mechanism to be effective. Generally, CFI dominates when the wing sweep is greater than 30 − 35 deg, making the vertical stabilizer the best candidate for implementing the hump.

This research implements the Delft Laminar Hump on the vertical stabilizer of subsonic transport aircraft by modeling the effect of the hump as a shift in transition location. By using a Quasi-3D aerodynamic analysis in combination with a transition location database, the effect of the hump on the lift and drag coefficient of the vertical stabilizer is analyzed. The transition location database is constructed by using the external velocity of airfoil sections of the vertical tailplane and the boundary layer solver and stability analysis developed by the Group of Flow Control and Stability within the Delft University of Technology. With this, the N-factor curves along the chord can be calculated. Knowing the respective N-factors of the clean and hump configuration, the associated transition locations can be determined, which are then used to calculate the lift and drag coefficients of the vertical tailplane. To evaluate the aerodynamic effect of the hump on the full aircraft directional and lateral stability a stability analysis based on a method by Fokker / Obert is performed and checked against the CS-25 for Large Aeroplanes regulations by European Aviation Safety Authority (EASA).

It was found that the Delft Laminar Hump (DeLaH) on the vertical stabilizer of a subsonic transport aircraft does not affect the vertical tailplane lift curve slope, thus not affecting the stability of the aircraft. In contrast, the vertical tailplane drag coefficient is reduced by the hump. Retrofitting the hump on Airbus A320 (conventional tail) and the Fokker F-28 Mk1000 (T-tail) results in a reduction of the vertical tailplane drag coefficient of 6.73% and 8.72%, respectively. Translating this vertical tail drag reduction to the full aircraft drag coefficient results in a reduction of 0.17% and 0.34%. To evaluate the effect of the hump on weight and fuel consumption, additional weight and mission analyses are performed. Evaluating the harmonic range, an fuel reduction due to the hump of 0.16% and 0.32% is established, for the Airbus A320 and Fokker F-28 Mk1000, respectively. The aircraft weight is reduced by the same percentage through the fuel reduction, as it is assumed that the added weight due to the hump itself is negligible.

Additionally, two sensitivity analyses were performed, namely, sweep angle variation and surface area scaling to analyze the effectiveness of the hump for different vertical tailplane geometries. The effect of sweep angle on hump effectiveness does not affect the vertical tailplane lift curve slope and thus also not the stability coefficients. For the full aircraft drag coefficient, the hump effectiveness has an exponential relation with sweep angle and is most effective at lower sweep angles. A maximum full aircraft drag reduction was found of 0.41% at 30 deg sweep with an equivalent fuel reduction of 0.39%. Overall it is concluded that lower sweep angles are beneficial as the hump is most effective and results in reduced vertical tailplane weight, less fuel weight as well as an increased stability margin. Analyzing the effect of surface area scaling, the hump has no effect on the vertical tailplane lift curve slope regardless of surface area, again retaining the aircraft’s stability. For the full aircraft drag coefficient, the hump effectiveness increases linearly for increasing surface area up to 0.37% at a surface scaling factor of 1.2 times the original vertical tailplane surface area. In terms of fuel reduction, a maximum value of 0.35% was found. There will be an optimal vertical tailplane surface area, since the hump effectiveness increases for increasing surface area, whilst for the full aircraft drag, vertical tailplane weight, and fuel weight a smaller surface area is preferred. The stability margin becomes the limiting factor as a minimum surface area is required for sufficient stability. Comparing the baseline aircraft, the hump is more effective for the Fokker F-28 Mk1000 over the entire range of scaling factors and sweep angles. This leads to the suspicion that taper- and aspect ratio, and cruise speed play an important role in the effectiveness of the hump, but more research is required.

This research shows that the Delft Laminar Hump has a significant drag-reducing effect on the overall aircraft, whilst not affecting the aircraft’s stability. Even though the fuel savings for an individual aircraft are not very large, on a fleet level this would be significant. The hump can be retrofitted on existing aircraft by gluing it on the outer skin, making it a relatively simple and cheap way to improve efficiency for aircraft manufacturers. Nevertheless, more research is required before implementing the hump on commercial aircraft as it is still unknown whether the hump also works on the suction side of the wing as well as whether the hump causes a shock at cruise Mach. Also, interaction effects with the horizontal stabilizer and fuselage need to be taken into account, to fully quantify the effectiveness of the hump. Dedicated wind tunnel experiments, flight tests, and/or CFD simulations are necessary.

Over the past few decades, aircraft design priorities have shifted from maximising cruise speed to reducing fuel burn and emissions. Rising fuel costs, environmental regulations, and the emergence of electric and hybrid-electric propulsion have renewed interest in propeller-driven concepts, such as Distributed Electric Propulsion. These concepts often involve placing multiple propellers in proximity to each other or close to other aerodynamic surfaces. As such, there exists a need to better understand the behaviour of the boundary layer on downstream surfaces.

At the moderate chord Reynolds numbers relevant here, laminar–turbulent transition, boundary layer separation and the formation or suppression of LSBs can alter lift, drag, stall margins and noise. LSBs arise when a laminar layer separates, transitions within a free shear layer and then reattaches, resulting in a turbulent flow. A propeller modifies this flow field by generating a non-uniform, unsteady slipstream consisting of accelerated axial flow, swirl, and a helical system of tip and root vortices and blade wakes. The resulting spatially and temporally varying inflow alters local incidence, dynamic pressure and boundary layer dynamics over the wing. Existing studies rely primarily on time-averaged measurements and lack sufficient spatial and temporal resolution near the wall. As a result, the detailed mechanisms by which helical vortices and wake sheets interact with the downstream boundary layer, their spanwise variation within and around the slipstream core, and their influence on LSB dynamics remain only partially understood.

The present thesis aims to provide an experimental characterisation of how a propeller slipstream modifies a downstream laminar boundary layer behaviour. Two complementary campaigns have been used to answer the research objective: oil-flow visualisations and acoustic measurements on the surface of an airfoil placed under a propeller slipstream to understand the time-averaged boundary layer in a global sense, and a local, high-resolution investigation using phase-locked stereoscopic Particle Image Velocimetry (sPIV) to resolve the vortex structures within the slipstream and understand their effect on the boundary layer. Together, these experiments address the questions of (i) how the slipstream alters laminar–turbulent transition and LSB formation, and (ii) what role unsteady vortical structures within the slipstream play in triggering and modulating transition along the span.

The results show that the propeller slipstream strongly energises the downstream boundary layer and generally promotes earlier transition, but in a manner that is highly non-uniform in both span and blade-passage phase. The results indicate that under a propeller slipstream, transition is advanced and the boundary layer thickens on the inboard side, where inboard shearing and higher slipstream velocities amplify near-wall turbulence. The slipstream also modifies LSB behaviour, and depending upon slipstream amplitude and effective local incidence angle, the LSB may be suppressed entirely, shortened and displaced upstream, or persist within regions of weaker propeller loading. Phase-locked sPIV reveals that these behaviours are governed by an intrinsically three-dimensional, intermittent interaction between the boundary layer and a coupled system of primary tip vortices, secondary wall-originating vortices and blade wakes. Wake-impingement events generate short bursts of high shear and TKE separated by phases of weakly turbulent or nearly laminar near-wall flow. Comparable TKE levels for tripped and untripped cases at the same wall-normal position show that near-wall turbulence inside the slipstream is dominated by the imposed unsteady forcing rather than by the state of the boundary layer.

Developing Knowledge-Based Engineering (KBE) applications remains a significant challenge in high-tech industries like aerospace, where front-loaded product development demands extensive automation. The manual code completion phase of these applications is particularly problematic: it consumes considerable time, requires specialized expertise in proprietary frameworks, and creates steep learning curves that limit broader adoption. While recent advances in Artificial Intelligence (AI) have revolutionized software development assistance, commercial AI systems consistently fail when working with specialized KBE frameworks, like ParaPy. In the absence of sufficient training data on proprietary frameworks, these systems produce code that appears correct but is actually non-functional.

This research validates that retrieval-augmented approaches offer a practical alternative to retraining models for specialized domains with limited data, such as ParaPy. In these approaches, AI systems dynamically access domain-specific knowledge during operation rather than relying solely on their training. This has significant implications for industries or institutions using proprietary tools where comprehensive model retraining is economically infeasible.

The research implements this approach by developing and evaluating a dual-agent framework for AI-assisted KBE application development that operates within industrial privacy and security constraints. The framework comprises a Developer Agent optimized for code generation and debugging with the ParaPy SDK, and an Educational Agent focused on ParaPy learning support and documentation. Both agents access a knowledge infrastructure that uses semantic search over indexed ParaPy documentation, curated examples, and technical references. Additionally, the Developer Agent employs verification mechanisms that progressively check code at two core levels: syntax correctness (ensuring the code follows programming language rules) and successful execution (confirming the code runs without errors).

User testing revealed how different skill levels benefit from AI assistance. Intermediate users benefited most, showing dramatic improvements in productivity and performance. Novice users achieved substantial productivity gains and reduced framework-specific (ParaPy) errors significantly, with task completion rates approaching expert baseline performance. Expert users, however, experienced slight performance degradation due to reduced code review under time pressure. The framework successfully reduced knowledge barriers for novice and intermediate users, broadening access to specialized engineering tools.

Despite these successes, the framework has persistent limitations in understanding three-dimensional spatial relationships. This is critical for KBE applications where code must define the precise position, orientation, and assembly of physical components. The framework struggles to correctly place components in space or apply proper rotational transformations. This produces code that may be syntactically correct but results in misaligned parts or incorrectly oriented features. These geometric errors require iterative refinement with human guidance, representing a fundamental limitation of the current approach and language model architectures.

While geometric reasoning limitations suggest fundamental boundaries of current AI capabilities, the demonstrated productivity improvements and reduced knowledge requirements establish a foundation for broader AI adoption in knowledge-intensive engineering domains. The framework contributes a validated operational prototype that addresses critical gaps in AI-assisted KBE development: reducing the manual coding bottleneck, lowering knowledge barriers for new users, and providing privacy-compliant deployment options. The framework functions most effectively as a development accelerator requiring expert oversight, supporting human engineers rather than replacing them.

Keywords: Knowledge-Based Engineering, Large Language Models, Code Generation, ParaPy, Retrieval-Augmented Generation, AI-Assisted Development, Aerospace Engineering, Multi-Agent Systems

Recent advancements in the UAM (urban air mobility) sector pave the way for making urban eVTOL (electric vertical take-off and landing) flight a reality. Operating in urban environments, noise generated by aerial vehicles plays an imperative role in their certification. Urban flight is expected to subject aerial vehicles to a broad spectrum of large-scale turbulent structures generated by various sources such as flow around high-rise buildings, terrain-level obstacles, incoming ABL (atmospheric boundary layer)
characteristics, local windspeed variations, etc.. Unlike the well-documented effects of aerial vehicles flying under steady freestream or small-scale turbulence impingement, research on their performance when subjected to large-scale turbulence representative of an urban airspace is relatively scanty. This experimental work aims to generate large scale turbulence in the order of the rotor diameter and analyze its influence on propeller loading and noise emission.

Bluff body shedding from a cylinder is utilized to generate rotor scale turbulence in the inflow, with a variation in upstream cylinder placement tested to quantify the influence of it’s proximity to the propeller. The presence of an upstream obstruction results in severe penalties to the aerodynamic performance and thrust generation compared to an isolated propeller. An increase in rotor scale fluctuations in the inflow leads to an increased intensity of low-frequency loading fluctuations. The highly turbulent inflow has a severe impact on the noise emissions, with an overall increase in the broadband content and retention of discrete tones emitted at BPF (blade passing frequency) harmonics across all tested advance ratios; contrasting drastically with the emission of the isolated propeller. Further, with the cylinder obstruction placed closer to the propeller, the effect of haystacking is observed in the noise emissions, which dominates the high-frequency range and results in a spectrum comprising of purely broadband content above the 2nd
harmonic of the BPF. The findings highlight a strong influence of large-scale turbulence on propeller performance and noise emissions, while emphasizing the requirement for further investigation which isolates and analyzes each underlying effect to obtain a complete understanding of the identified influence.

With the integration of the Cessna Skymaster into the TU Delft research aircraft fleet, a mathematical model is required to quantify its aerodynamic and propulsion characteristics. Aerodynamic model identification is conducted using the Two Step method applied to 19 decoupled doublet and 3-2-1-1 flight test manoeuvres. The engine model is developed as a digitised representation of coupled engine performance and propeller efficiency maps, providing corrected engine horsepower, thrust, and total pressure gain across the propeller. To support system identification, the aircraft’s centre of gravity and inertial characteristics are determined. In the absence of dedicated flight instrumentation, a method for measuring control surface deflections is devised and tested in flight. The resulting aerodynamic model demonstrates robust performance in prediction of the C_X, C_Z and C_n dimensionless forces and moment, while the C_l moment estimation remains rather weak. The engine model power estimation closely follows software predicted values.
Related dataset 4TU.ResearchData:
https://doi.org/10.4121/a74d4324-ce07-45a9-9e9a-d4177e733a0b

Laminar separation bubbles (LSBs) play a crucial role in determining the aerodynamic performance of airfoils operating at low Reynolds numbers, conditions commonly encountered by unmanned aerial vehicles (UAVs), wind turbines, and small aircraft. Their inherently unsteady nature significantly influences lift, drag, and noise generation. This thesis investigates the dynamic behavior of LSBs that forms on the suction side of a NACA 0015 airfoil through surface pressure measurements obtained using piezoelectric sensors (”piezofoils”) integrated into a wing model tested in the Boundary Layer Wind Tunnel of the Technische Universität Berlin. The device are tested at chord-based Reynolds numbers of 144,000, 197,000 and
262,000 and at angles of attack comprised between 1 and 5 degrees, in 1 degree increments. Two different piezofoil designs are used, ”in-line”, spanning most of the airfoil chord and ”staggered”, shorter but with an improved spatial resolution. Complementary measurements are performed using surface hot wire anemometry, oil flow visualization, and surface pressure taps to provide a comprehensive characterization of both mean and fluctuating flow features.
The primary objective of the study is to relate the surface pressure fluctuations within the LSB to its dynamic phenomena, specifically, shear-layer flapping and vortex shedding, and to identify the optimal sensor placement for their detection. The existence and the characteristics of the laminar separation bubbles at the range of Reynolds number and angles of attack considered is confirmed through surface pressure measurements using pressure transducers and oil flow visualization. Analysis of the spectra and of the standard deviation in the separation and reattachment regions demonstrates that the piezofoil sensor successfully detects and accurately resolves vortex shedding, but is unable to capture the low-frequency flapping motion. The cross correlation analysis provided further insight into the vortex shedding process, enabling the estimation of the convective velocity and streamwise wavelength of the vortical structures. Heating of the piezofoil and appropriate signal filtering are found to be effective in enhancing the signal-to-noise ratio and mitigating electromagnetic interference effects.

This thesis investigates the role of winglet slots in reducing tip vortices, inspired by bird wingtips. The effects of slotting and wingtip flexibility were analyzed using multiple bio-inspired models tested in a low-speed wind tunnel at Reynolds numbers from 2.9 × 10⁴ to 8.8 × 10⁴. Three experimental techniques were employed: force balance for aerodynamic force estimation, motion tracking of wingtip movement, and stereo PIV analysis to measure cross-flow planes in the winglet wake. Non-intrusive methods were used to estimate drag components based on PIV data and wake integral approaches. The slots significantly reduced induced drag by breaking down the tip vortex into smaller, weaker vortices. This breakdown was visualized using s-PIV results and analyzed through velocity component contours. The streamwise vorticity breakdown was examined in relation to wingtip flexibility, and the vortex cores' location and characteristics were analyzed, along with the phenomenon of vortex wandering.

By the end of the thesis, it was concluded that although slotted winglets reduced induced drag, they led to a significant increase in profile drag, resulting in higher overall drag compared to unslotted winglets. Additionally, the slots caused a loss in lift. The overall aerodynamic performance, measured by the lift-to-drag ratio, was also higher for the
unslotted winglet. Among the slotted winglets, static bending was common, while wingtip vibrations were minimal. The most flexible slotted winglet had the lowest induced drag. However, the results indicate that at very low Reynolds numbers, slotted winglets deteriorate the wing’s aerodynamic performance compared to unslotted ones. Among the slotted winglets, those with intermediate flexibility demonstrated the best aerodynamic performance, highlighting the importance of optimizing wingtip flexibility.

To reduce greenhouse gas emissions in aviation, innovative propulsion systems such as battery-powered electric aviation are essential. These systems are carbon neutral when powered by 100% green electricity. However, widespread adoption requires overcoming challenges including improving energy density, lowering costs, and maintaining safe thermal operation (incl. thermal stability). This thesis addresses the challenge of maintaining battery thermal stability during flight by developing a combined electronic circuit and thermal network model for the NLR-owned Pipistrel Velis Electro aircraft. Using flight test data from this aircraft as a validation source, the model evaluates three battery thermal management strategies: two liquid cooling methods (ribbon and cold plate) and one gas cooling method (air cooling).
The electronic equivalent circuit model, used to simulate voltage characteristics and heat production in a single lithium ion battery cell, requires pulse current characterization tests for accurate parameter estimation.
Extensive testing has been conducted to gather these data. The model achieves accurate voltage modeling accuracy with a low root mean square error Adding more than one RC branch to the circuit did not significantly improve the accuracy of the model.
The electronic equivalent circuit and lumped parameter thermal network models were validated with two flight data sets. Both ribbon and cold plate cooling solutions effectively matched the validation temperatures, performing similarly. In contrast, the air cooling solution was less effective. In case of ribbon cooling, the maximum cell temperature was highly sensitive to its geometric parameters, specifically the angle and height of the ribbon. The sensitivity of the cold plate solution in terms of maximum battery temperature was influenced by the diameter of the cooling channel and the thickness of the plate. The air cooling showed sensitivity in terms of maximum battery temperature relative to the inter-cell gap width. For the ribbon model, varying the number of thermal nodes led to a convergence in the maximum battery temperature as the node count increased.
Using the validated ribbon cooling model, two operational scenarios were analyzed. The first scenario involved charging operations, where simulations closely matched temperature validation data, showing only a minor temperature rise in the battery pack. The second scenario tested cold weather operations with ambient temperatures reduced to approximately 0 ◦C. Here, two simulations were conducted: one with the battery preheated to 20 ◦C and another without preheating. Without preheating, the battery pack’s temperature neared the operational lower limit of 0 ◦C. Preheating prevented reaching this lower limit. It is recommended to preheat the battery pack using an external charger, as using the battery’s own energy for preheating is inefficient.
The thesis was concluded by using the developed modeling approach to size and model a battery pack for the Eviation Alice, a larger aircraft. The approach was successfully scaled to this large use case with a known power profile. Furthermore, due to significant ambient temperature effects and higher operational altitudes compared to the Pipistrel Velis Electro, thermal insulation will be necessary for the battery pack to maintain temperatures above the lower operational limit of 0 ◦C during typical missions.

The global emission of harmful greenhouse gasses has to be reduced. Commercial shipping contributes to these emissions by using fossil fuel as the only energy source to provide propulsion and power auxiliary systems. In order to reduce fuel consumption, much research is focused on increasing the efficiency of the propulsion system and reducing the required power consumption. In recent years an increasing amount of research focuses on utilizing wind energy as an alternative energy source. This is a renewable source of energy which is plentiful at sea. This thesis focuses on the novel concept of using a Vertical Axis Wind Turbine (VAWT) as a means of ship propulsion and power generation. Existing Wind Assisted Ship Propulsion (WASP) systems can only be utilized in conditions when the ship is in transit and when the apparent wind conditions are favorable. The VAWT concept has the ability to generate power in wind conditions where conventional sailing systems are not functional, including when the ship is in port. This system can also be used as a thrusting device when the wind conditions are favorable for sailing systems. Therefore this system can be used more frequently which is beneficial for the overall power savings. The use of a VAWT for WASP in combined operation as a thruster and a power generator does currently not exist. Therefore the design of the rotor has to be investigated and optimized in order to evaluate effective designs for different conditions and to compare this device to existing sailing concepts. The aerodynamic performance of the VAWT will be evaluated using numeric simulation models. This model will be coupled to an optimization algorithm in order to find the optimal design parameters for different wind conditions and modes of operation. This project will result in a better understanding of the working principles of VAWT used for WASP and the potential energy saving of optimized designs for the operational modes as a generator, thruster and combined-operation. There is a large opportunity for capturing wind energy at sea. This project has the potential to add a new method of ship propulsion to reduce shipping emissions and reducing the cost of ship operation. When one or multiple of these systems can be applied on a large share of the worldwide shipping fleet, this research can contribute to the reduction of commercial shipping emissions.

This thesis is performed with the aim to investigate whether split flaps can be a feasible solution for increasing the Flying V directional control power. In order to do so, a set of wind tunnel experiments have been concluded with a 4.6% half wing model in the Open-Jet-Facility at the Delft University of Technology. Here, several split flap geometries have been tested on the model outboard wing section to fully establish their behaviour in terms of yaw and other parameters of interest. Additionally, calculations have been performed to provide an adequate split flap design for certification of the lateral-directional specifications from CS-25.

From the wind tunnel experiments, it was found that split flaps on the outboard wing section can effectively increase the directional control power over angles of attack between 0 and 30 degrees. However, yaw control is lost when over negative angles of attack due to lower wing stall. The maximum decrease in split flap effectiveness over positive angles of attack is found around 17.5 degrees, where a leading edge vortex is present over the outboard wing. This decreases pressure over the upper split flap, resulting in a effectiveness decrease of around 41%. As the effectiveness does not decrease further, split flaps can be continuously effective at higher angles of attack when compared to the winglet rudders. The low pressure on the upper flap also causes large adverse coupled moments in pitch and roll. The effect of split flaps on the yawing moment is found to be linear with deflection angle, where their effect on other aerodynamic properties is found to be more non-linear.

The deflection of outboard split flaps do not have a significant interference effect on rudder yaw power, but can have some interference effects on adjacent main wing control surfaces. Differential deflection between the upper and lower flap has been shown to potentially decrease coupling in pitch and roll. The split flaps can also be globally rotated trailing edge down to mitigate adverse coupled moments while beneficially increasing the total created yaw.

From trim calculations, it was found that a maximum additional yaw coefficient of 2.4113e-3 has to be provided by the split flaps. Designs with a maximum deflection angle above 30 degrees are deemed feasible, as this would maximally require only the replacement of the current CS3 surfaces. When a maximum deflection of 60 degrees is set by the designer, a sub-scale split flap width of 111.87 mm is needed, which translates to a full-scale split flap of 2.431 m. Based on the sensitivity analysis and projected Reynolds effects, this is considered a conservative design for the full-scale Flying V. At 0 degrees angle of attack, the recommended geometry is projected to increase the maximum directional control power of the Flying V by 38.1%. At 27.5 degrees, the maximum directional control power can even be increased by 85.5%. Both increases come at the cost of a significant drag penalty.

A new aircraft configuration, the Delft University Unconventional Configuration (DUUC), was recently invented by the Flight Performance and Propulsion group at TU Delft. As a solution to the concerns regarding the environmental and health impact due to increasing demand of air travel. The DUUC has a characteristic feature which consist of two ducted propellers at the empennage of the aircraft with vertical and horizontal vanes mounted downstream of the duct, spanning the whole diameter. With this newly integrated architecture the DUUC combines thrust with stability and control of the aircraft. Furthermore, with the current trend of increasingly larger bypass ratios on turbofan engines, these will at a certain point not fit below the wings anymore. Therefore, further highlighting the importance of this research than the study on the DUUC alone. From previous research on the DUUC, where the propulsive stabilizer was in a fixed, aft-fuselage position with a fixed duct aspect ratio. It has been observed that the rear center of gravity (CG) position and a large shift in CG due to the propulsive stabilizer, poses a number of challenges. Therefore, research has to be done on the positioning of the propulsive stabilizer. The objective of this research is to study the positioning of the ducted fan propulsive stabilizer, considering its impact on stability and control by development of a method to predict aerodynamic/stability & control derivatives for variable position and duct/fan size. In this study, the aerodynamic analysis tool FlightStream is used to determine the aerodynamic characteristics of the ducted propeller. These results are then used in combination with AVL and an ESDU method to determine the influence of the positioning and duct size on the aircraft stability and control. Furthermore, in order to assess the validity of FlightStream, a validation study has been performed as well and compared to analytical methods which have been used on earlier studies on the DUUC.