Thesis

Novel representations for modelling free-form objects under constraints : a parametric modelling tool for computer-aided ship design

Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2021
Thesis identifier
  • T15893
Person Identifier (Local)
  • 201860169
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • Digital design of engineering systems, using Computer-Aided Design (CAD) tools fordeveloping and optimising products, has been used in the manufacturing process since1957, when Patrick Hanratty built the first commercial numerical-control programmingsystem known as PRONTO, while he was working for General Electric. One year later,the French physicist and mathematician Paul de Casteljau invented a system basedon the use of Bernstein polynomials, during his employment with the Frenchautomobile manufacturer Citro¨en. In 1968, another French engineer named PierreB´ezier, launched on behalf of the automobile manufacturer Renault the CAD/CAMUNISURF system for surface design, which was fully in use by 1975. The steppingstone of Computer-Aided Ship Design (CASD) was set in 1963, when the NorwegianCAD/CAM software Autokon was first used, developed by Trygve Reenskaug. It was 14years later when parametric modelling in ship design introduced by Prof. H. Nowackiwho pioneered in CASD via coupling form parameters with the then novel technologyof B-splines. Since then, Parametric Modellers (PM) play a crucial role in the development and -most importantly- shape optimisation of free-form objects with increasedcomplexity such as ship-hulls, for they have to represent robustly and efficiently everysolid object.The purpose of the current work is three-fold:Firstly, it is meant to develop and present the corresponding methodology of twotools:a) an innovative, robust, and cost-efficient parametric modelling tool for containerand tanker ship-hulls, which are hull forms of increased complexity. The tool is named TshipPM and is an extension of the work presented in,b) a ship-hull shape optimisation tool based on geometric criteria, namely volumemoments up to 2nd order.Secondly, to evaluate the performance of TshipPM by comparing it with a wellestablished, commercial parametric modelling tool, CAESES®1, opting for itsNURBS functionality.Finally, was to provide an easy to follow ”handbook” for developers and users onthe development of parametric modelling tools, illustrating as well how a PM canbe utilised for the construction of a given ship-hull.The key findings of the current work which also coincide with the main objectives, are:A method of constructing a T-splines-based parametric modelling tool (TshipPM)for complex ship-hull design, the development of which is taking into consideration the detailed characteristics of two different ship-hull types, i.e., tankers andcontainers.A set of geometric and design constraints imposed to TshipPM to tackle theintricate and complex issue of Geometric Validity, ascertaining the production ofvalid models for the whole design space the modeller covers.A step-by-step method to remodel (using TshipPM) a great variety of complexhulls with complexity not-higher than that of tankers and containers.The exceptional performance of TshipPM. A comparative study is conductedto evaluate the performance of TshipPM against CAESES, a well-establishedcommercial parametric modeller for ship design. This work is published in theOcean Engineering2international journal, titled ”A T-splines-based parametricmodeller for computer-aided ship design”.A method for ship-hull shape optimisation with respect to geometric criteriaand especially volume moments up to 2nd order, utilising TshipPM and a multiobjective Teaching-Learning-based optimisation tool. This part of the work alsostresses the insufficiency of 2nd order moments to perform as complex-shape optimisers.TshipPM is a parametric modelling tool for ship-hull design under geometric and designconstraints. It uses as a mathematical representation of surfaces -and to some extendsolids- the fairly new technology of T-splines, introduced by Sederberg et al. in 2003, which constitutes a generalisation of NURBS, exhibiting several advantages overthe latter. T-splines technology aids towards the development of modellers for complexship forms, by providing geometrically valid objects with smooth surfaces and increasedfairness throughout the model’s surface, and much lower complexity in comparison withPMs employing the standard NURBS technology. TshipPM uses 27 external (or input)parameters to produce the control cage of the ship-hull, 3 of which are global anddimensional, while 24 are non-dimensional and of local nature. It employs Autodesk®T-splines plug-in® v.4.0 for Rhino5® 3D to create the final, smooth T-splines surfacesusing as input the control cage created by TshipPM.The PhD thesis consists of 6 Chapters, structured in the following way:The Novelties of the current work and its contribution to the field of parametricmodelling in CASD.Chapter 1, after a brief discussion on the historical background of splines, it delivers acomparison between T-splines, the representation underlying TshipPM, and NURBS,the industrial standard in CAD, which is also used in the context of CAESES (§1.1).In addition, it introduces the reader to parametric modelling, reviewing in brief themain advancements in the field (§1.2).Chapter 2, after the necessary introduction to TshipPM’s features and characteristics(§2.1), analyses the structure of TshipPM regarding the parameters involved (§2.2),making the distinction into internal and external, dimensional (or physical) and nondimensional. §2.3 demonstrates the process of creating a ship-hull model, starting frombuilding the control cage along with its link to external parameters, which are used asinput for TshipPM, up to its conversion into the resulting ship-hull surface with theaid of T-splines plug-in.Chapter 3, after a brief introduction on the importance of robust PMs (§3.1), definesthe constraints encountered in ship-hull design and introduces the reader to the concept of geometric validity, illustrating examples of invalid ship-hull models (§3.2). §3.3presents the geometric and design constraints imposed to TshipPM to ascertain itsrobustness as far as the production of geometrically valid models is concerned. In §3.4we provide an experimental indication of the robustness of TshipPM in its response tostrong parameter values variation. Finally, in the last section (§3.5), we present theoutput of a Monte Carlo sampling of 200,000 ship-hull instances produced by TshipPMfor both containers and tankers, and we analyse its output information to determine thedesign space of the PM, by measuring TshipPM’s flexibility with regards to geometriccharacteristics, namely volume centroid and moments of inertia.Chapter 4 begins by stressing the need of remodelling a given ship-hull with PMs touse it as basis for optimisation (§4.1), while §4.2 refers to the remodelling evaluationcriteria. The remodelling process is thoroughly described in §4.3 using as a case studythe MOERI KCS container ship-hull, and aims to the reconstruction of a parentship-hull CAD model; we use TshipPM for the construction of model’s control cageand we feed it into Rhino5 to create the corresponding surfaces with the aid of theT-splines plug-in. Finally, in §4.4 the evaluation of the remodelled MOERI KVLCCtanker [10] ship-hull is conducted, under a predefined set of criteria, providing feedbackto users and developers for any adjustments and their specific locations. The evaluationof MOERI KCS container hull is conducted in Chapter 5.In Chapter 5, TshipPM is compared against CAESES with regards to their outputsagainst a parent hull, the KCS container ship-hull, which has been extensively usedby the research community for CAD and Computational Fluid Dynamics (CFD) benchmarking purposes. The comparison criteria are described in §5.1. §5.2 presents the basiccharacteristics of the CAESES parametric modelling tool, as well as a description of thephysical (or dimensional) parameters CAESES uses to remodel KCS ship-hull and thecorresponding external non-dimensional parameters, while §5.3 refers to the commonexternal parameters both PMs use. Finally, in §§5.4 - 5.8 the comparison of both PMsis conducted.Lastly, Chapter 6 presents an in-house, C#-built, shape-optimisation tool usingTshipPM and an adjusted to the needs of the current work Multi-objective TeachingLearning-based Optimisation (MO-TLBO) method. TshipPM MO-TLBO is optimisingthe shape of a given hull (MOERI KCS) against a set of ship-design criteria, and especially volume moments up to 2nd order. After the required introduction to the chapter(§6.1), a brief presentation of the TLBO method is conducted (§6.2). The set of theobjective functions against which the optimisation is conducted is provided in §6.3,while in §6.4 the method of building TshipPM MO-TLBO, its features and functionality are presented in detail. The chapter concludes with §6.5, illustrating the resultedKCS-ship-hull instances produced by TshipPM MO-TLBO.The thesis concludes with the Discussion and Summary of the current work, highlighting its key points, stressing the delivered novelties and main contributions to thefield of parametric modelling in CASD.
Advisor / supervisor
  • Kaklis, Panagiotis
Resource Type
Note
  • Previously held under moratorium from 23rd June 2021 until 3rd July 2023.
DOI
Date Created
  • 2021
Former identifier
  • 9912989592402996

Las relaciones

Elementos