Num3DEx (2018-2022)
EVOLUTIONARY OPTIMAL SHAPE SYNTHESIS WITH HETEROGENEOUS INTEGRAL AND PARTITINED 3D PHENOTYPES, DYNAMIC PARAMETERIZATIONS AND MESHLESS MODELING
The project continues the former two-year CSF project which resulted in six publications in prestigious journals and aims at developing a generic platform for optimization-based synthesis of 3D shape based solely on given excellence criteria and constraints, involving:
- an abstract initiation phase (3D shape genesis) with elimination of bias towards any candidate shapes or particular compositions of geometric primitives
- inheriting and import of shapes of existing objects based on 3D scanning into point clouds,
- evolutionary shape optimization of populations of heterogeneous phenotypes converted into common genotypes based on parametric fitting
- dynamic parameterization to encapsulate even dramatically changing 3D shapes based on intelligent adaptation to current shape features
- 3D shapes conceptually modelled both as integral entities and sets of partitions with coding of mutual relationships
- options for feature-detection approaches (e.g. edges, peaks) in integral geometries towards shape decomposition of optimized shape
- conceptual development of post-shape-optimization processing of integral shapes in terms of potential representation as sets of partial shape entities,
- implementation of isogeometric analysis (IGA) in the field analysis stages towards constraints and objective functions (isogeometric and meshless procedures could efficiently support shape optimisation using the same infrastructure in constraint dependent field analysis and reshaping)
- application of IGA in optimisation strategies with the same basis involved in shape parameterization and field calculations, thereby reducing time consumption in data transfers.
The project will develop computational workflows which efficiently implement above elements and systematically test them. Several theoretical and actual test-cases will be studied, and the ad-hoc aerodynamic tunnel will be upgraded for comparative experimental verification of optimized shapes of vertical-axis wind turbines.
Link: https://www.croris.hr/projekti/projekt/3927
Optimal3D (2015 - 2017)
ADAPTIVE PARAMETERIZATION OF VARIABLE 3D GEOMETRIES IN SHAPE OPTIMIZATION AND MESHLESS NUMERICAL MODELING
The project focuses on developing numerically efficient and dynamically adaptive parametric representations of shape which autonomously provide increased modeling capacity for the 3D shape itself and for the shape functions in numerical methods. The initial shape iteration will start from 3D point clouds, possibly resulting from shape digitization (existing high-resolution optical scanning system). Mathematical parameterization of the point clouds will be based on parametric surfaces. Integral as well as chained piecewise surfaces will be explored and compared. Dynamic aspects of the parameterizations will be studied. The shape and possibly topology of the object will be changing in simulation/ optimization quasi-time iterations. During the process, edges and peaks may disappear and new ones may arise, whereby the parameterization must be autonomously adaptable to such changes. Strategies for controlled reductions of the parameterization data-sets based on permissible threshold values for peak error and cumulative error norms will also be studied. These strategies will include adaptive re-allocation of the control points using feature recognition procedures (edges, peaks) based on eigenvalue ratios and slope and curvature estimators. Adaptive partitioning of the overall object surface based on different criteria will be evaluated. The optimization process involves extensive field calculations i.e. solving field problems governed by partial differential equations. The above shape modeling aspects also apply with meshless methods where a function across the domain determines the numerical solution of a given field problem. The option applied here will be meshless methods which are based on a nodes “cloud” and are dramatically less sensitive to the change in a material configuration.
Link: https://www.croris.hr/projekti/projekt/105