Thesis supervisor: Katalin Bagi
Location of studies (in Hungarian): Tartószerkezetek Mechanikája Tanszék Abbreviation of location of studies: BMETM
Description of the research topic:
Regolith, a granular assembly that consists of zigzag-shaped sharply concave particles on scales ranging from very fine powder to pebbles and boulders, covers the surface of several celestial bodies (solid planets, moons, asteroids). In lack of atmosphere and surface fluid flows, the shape of particles remains fragmented and sharp. Due to the special particle shapes, to the very wide size distribution, the clinching of the concave grains and to the relative importance of atomic forces like van der Waals effects, macro-level regolith behaviour strongly differs from what engineers are used to in soil mechanics. However, in order to allow for safe engineering design (e.g. to quantify angle of repose in slopes, resistance between rover wheels and regolith, forces expressed by the regolith on structural members of habitats etc.) reliable mechanical modelling is indispensable. The aim of the proposed PhD research is to contribute to these efforts. The goal is to reconnoitre how the following effects influence the behaviour of regolith:
1. Particle shape
The particles in regolith are zig-zag shaped, sharp and strongly concave, hence the usual shape quantifiers of the engineering practice (e.g. roundness, flatness; surface roughness) are insufficient. Alternative or further descriptors are to be suggested, and analysing how they affect the behaviour, the suggestions should be supported with discrete element simulations.
2. Gravity
Gravity on surface level varies from celestial body to celestial body. Hence, according to results in the literature, those laboratory experiments done on Earth on regolith simulants are not always reliable. The effect of the strength of gravity on the macro-level behaviour is to be analysed with DEM simulations.
3. Particle breakage
Concave grains with sharp needle-like corners are particularly exposed to grain breakage, which leads to the modification of particle shapes and size distribution. As a result, the macro-level behaviour changes in comparison to the intact undisturbed regolith. In order to suitable simulate this phenomenon, discrete element modelling requires the – possibly simple but still acceptably realistic – mechanical modelling of particle breakage on the level of individual grains. The aim is to propose such model(s), and validate it (them) with experimentally calibrated discrete element simulations.
All the above tasks are to be done with discrete element models whose material parameters should be calibrated according to the results of real measurements taken from the literature and from international partners. Such a calibration is a very lengthy and troublesome task, and today this is a crucial obstacle against the widespread application of DEM in everyday engineering practice. In a DEM model approximately half a dozen or a dozen material parameters have to be calibrated properly, in such a way that a large number of computer simulations are executed that are to be compared to a low number of real experiments. This is done today manually. Only a few initial attempts exist today to algorithmize this task, though according to private discussions, such attempts are planned in international teams recently. Hence, our aim is to take advantage of the already existing methods in artificial intelligence. The peculiarity of the expected calibration algorithm for regolith modelling is that a relatively low number of parameters should be found, while the evaluation of the fit is computationally very expensive. Hence, special goal-oriented machine learning algorithm(s) should be developed for the task.
Number of students who can be accepted: 1
Deadline for application: 2022-12-20
2024. IV. 17. ODT ülés Az ODT következő ülésére 2024. június 14-én, pénteken 10.00 órakor kerül sor a Semmelweis Egyetem Szenátusi termében (Bp. Üllői út 26. I. emelet).
Login
