Research & Development


Mechanical-System Design Automation using Multiple Product-Models and Assembly Feature Technology
  • Knowledge Based Engineering systems (KBE) offer an efficient and flexible environment to store and manage Design & Engineering Knowledge independently of technical constraints implied by available CAD and CAE methods and tools.
  • A new definition and implementation of the concept of “Assembly Feature” has been developed maximizing preprocessing of assembly information at part level.
  • The Automatic Assembly Synthesis Model (AASM) serves direct translation of “Product Configuration” (described using various IT tools, e.g., a KBE system) into a complete 3D-CAD assembly model, to be used by downstream engineering-analysis applications.
  • The AASM includes two major components: The Schematic Assembly Model (SAM) and the Intermediate Assembly Model (IAM); see Fig. 1. The SAM is a preliminary model that converts the “Product Configuration” into an object-oriented assembly-structure form that functions as a configuration rule guiding automatic (or interactive) assembly synthesis. The IAM fills the informational gap between SAM and 3D-CAD assembly modeling. The IAM is created in four steps [C’17] leading to an informationally-complete description of the mechanical assembly so that a 3D-CAD model is automatically created.

Figure 1. The Automatic Assembly Synthesis Model (AASM)

Sample of Related Publications:

[CS’17] I. Chatziparasidis, N. Sapidis, “Framework to Automate Mechanical-System Design using Multiple Product-Models and Assembly Feature Technology”, International Journal of Product Lifecycle Management 10(2), pp. 124–150, 2017.

[C’17] I. Chatziparasidis, “Automatic Assembly-Model Synthesis in Mechanical Design using Simulated Dynamic Finite-Element Experiments”, Ph.D. Dissertation, Department of Mechanical Engineering, University of Western Macedonia, 2017.

[CS’16] I. Chatziparasidis, N. Sapidis, “Automatic Assembly Design for Engineering-To-Order Products based on Multiple Models and Assembly Features”, Proceedings of PLM’16: IFIP 13th International Conference on Product Lifecycle Management [in:  R. Harik et al. (eds.), Product Lifecycle Management for Digital Transformation of Industries, pp. 261-274, IFIP Advances in Information & Communication Technology -Vol. 492, Springer], Columbia SC, USA, July 11-13, 2016.

Figure 2. A Panoramic Elevator Car Automatically Generated by the CabinsKBE System Employing AASM [CS’17] [C’17].

Simulated Structural-Integrity Analysis and Automatic Assembly Synthesis for Mechanical Design Automation
  • Integrated methodology (see Fig. 3) where Finite-Element (FE) models are used to simulate experiments, analyzing newly defined product configurations. The results of these simulated experiments are used to deduct new design rules that are passed to the Knowledge Based Engineering (KBE) system supporting the work of a Company producing Engineering-To-Order (ETO) products.

Figure 3. Integrated Methodology for Automated Product-Design

  • A Simulated Experiment Validation Method (SEVaM) has been developed for the validation of the initial experimental product-design and of the corresponding FE models; see Fig. 4. In SEVaM, the main experimental assembly is firstly divided into its major functional subsystems. For each subsystem, an initial FE model based on the corresponding 3D CAD model is constructed. Then, the corresponding experimental structure for each of subsystem is built. For each subsystem of the experimental product, acceleration values are measured experimentally and are passed as excitation forces to the FE model. Then, a dynamic analysis is performed. From the FE results, high stress areas are identified. In the experimental structure, strain cages are placed at the high stress areas (identified by FE analysis) and stresses are measured experimentally. If the calculated and experimental stress values are not in agreement, then FE-model updating methods are applied, otherwise the FE model is considered valid. After each subsystem FE-model is validated, the experimental structures are synthesized to form the complete assembly, and a complete FE model is constructed. Finally, experimental measurements, on the complete structure, are recorded and compared to the calculated FE values.

Figure 4. Simulated Experiment Validation Method (SEVaM)

Figure 5. Experimental Set Up for a Complete Elevator System [CGS’18].

Sample of Related Publications:

[GCS’18] D. Giagopoulos, I. Chatziparasidis, N. Sapidis, “Dynamic and Structural Integrity Analysis of a Complete Elevator System through a Mixed Computational-Experimental Finite Element Methodology”, Engineering Structures 160, pp. 473-487, 2018.

[CGS’18] I. Chatziparasidis, D. Giagopoulos, N. Sapidis, “Simulated Dynamic Finite-Element Experiments and Automatic Assembly Synthesis for Mechanical Design Automation”, International Journal of Product Lifecycle Management 11(1), pp. 19–46, 2018.

Figure 6. Maximum-Stress Areas on the Elevator System and Comparison with Experiment [CGS’18].

Shape-Preserving Interpolatory Curves for CAD and other Applications
  • There are industrial applications and modes of design work, where it is considered essential for a curve to exactly interpolate a set of given points. The employed curve-creation method must produce a polynomial spline which has a continuous first-derivative (class C1) and also mimics the shape of the data in the following sense: if the data values have locally constant sign and/or they are locally monotonic and/or they are locally convex, then the interpolating spline should share also the same sign, monotonicity and convexity in the corresponding segments. The problem turns out to be a constraint interpolation problem. Positivity of data should be preserved, for example, when interpolating measurements of volumes, density, pressure etc. In a similar manner, there are many physical quantities which are monotone, e.g., volume-height diagrams of a ship compartment, ship-resistance with respect to speed at constant displacement, etc. Convexity preservation, finally, prevents unwanted oscillations in the interpolatory curve, allowing the least number of inflection points. The reconstruction of the so-called ‘‘p–y’’ (soil resistance, p as a function of the lateral deflection y of the pile) and ‘‘t–z’’ (axial response z of a pile subjected to vertical load t) curves, used in the design of pile foundations for onshore and offshore structures, requires interpolatory curves preserving sign, monotonicity and convexity; see Fig. 7. This problem recently surfaced at DNV GL – Digital Solutions and it was the motivation for this research.

Figure 7. Two Examples: A ”p-y” curve and a ”t-z” curve. The C1 shape preserving curve produced by the method [GS’18] (blue) vs a C2 convexity preserving curve (red) which fails to preserve monotonicity of the data; the sources for these two examples (data sets) are given in [GS’18].

  • This research has produced a comprehensive analysis of the Sign/Monotonicity/Convexity preservation problem, and has combined it with the standard Variable-Degree Spline (VDS) and current methods for estimating nodal derivatives of splines.
  • A fully-automatic algorithm [GS’18], of linear time-complexity, has been proposed producing a C1 VDS solving the above Sign/Monotonicity/Convexity preservation problem.
  • The publication [GS’18] concludes with a “Numerical Results” section, where the new algorithm is applied on the curve-modeling problems of DNV GL – Digital Solutions, that triggered this research, as well as on 30 standard test-cases from the pertinent literature. The presented images of curves and the curvature numerical-results [GS’18] establish that the produced curves not only are shape-preserving but also are visually-pleasing (“fair”).

Figure 8. Two Examples: Rapidly-varying data points and the shape-preserving interpolants produced by the method [GS’18] using various methods to estimate first-derivatives at the data points; the sources for these two examples (data sets) are given in [GS’18].

  • Current research focuses on the issue that a VDS may include curve segments with a high polynomial degree, which is undesirable. Work in under way towards a variation of the algorithm [GS’18] that will produce a C1 cubic


Sample of Related Publications:

[GS’18] N. Gabrielides, N. Sapidis, “C1 Sign, Monotonicity and Convexity Preserving Hermite Polynomial Splines of Variable Degree”, Journal of Computational and Applied Mathematics 343, pp. 662–707, 2018.