Engineering design almost always involves conflicting objectives: lower cost against higher performance, lighter weight against greater strength, faster speed against lower fuel consumption. This course introduces multicriterion optimisation in engineering design through a marine case-study workflow. It starts by clarifying optimisation concepts and distinguishing multicriterion optimisation from multidisciplinary design optimisation, then explains Pareto optimality, feasible parameter and objective spaces and the weighted-sum method with normalisation. A spreadsheet-based workflow for a marine design case is set up, validated and tested with sensitivity studies. Finally, the course reviews advanced approaches described in published case studies, including dominance-based multi-objective genetic algorithms, CFD-coupled optimisation loops, surrogate and hybrid models and multi-criteria decision-making methods for choosing a final design.
Course at a glance: Duration 20 hours · Self-paced online · 6 sections, 18 lessons, a 69-page downloadable course book
What you will learn
- Distinguish multicriterion optimisation from multidisciplinary optimisation in engineering design.
- Define design variables, objectives, constraints, feasibility and the final decision in an optimisation problem.
- Explain Pareto optimality and identify non-dominated solutions in a constrained design problem.
- Map a problem from parameter space to objective space and interpret the feasible region.
- Apply the weighted-sum method with criterion normalisation to select a compromise design.
- Set up, validate and run sensitivity studies on a spreadsheet-based multicriterion optimisation.
- Evaluate when genetic algorithms, CFD-coupled loops, surrogate models and MCDM methods are more suitable.
Inside the course: a high-level summary, section by section.
Optimisation concepts
The course opens by establishing a clear vocabulary. Design variables are the quantities the designer can change, such as dimensions, materials or operating parameters. Objectives or criteria are the measures of merit to be minimised or maximised, such as mass, cost, resistance or emissions. Constraints are limits that a design must satisfy, and a design that breaks a constraint is infeasible. The final decision is the choice of one design. When criteria conflict, there is usually a set of good compromises rather than a single best design, and the final choice reflects project priorities.
An important distinction is made between multicriterion optimisation, a mathematical approach to problems with several objectives, and multidisciplinary design optimisation, a system-level philosophy for coordinating design decisions across disciplines such as hydrodynamics, structures and machinery. Confusing a mathematical method with a system-design philosophy can lead to choosing the wrong level of analysis or describing a workflow loosely in a design report, so precise language matters.
Pareto optimality and the weighted-sum method
A Pareto optimum is a feasible design that cannot be improved in one criterion without becoming worse in at least one other. The course unpacks this definition into a practical test and explains why multicriterion design usually produces a set of acceptable trade-offs, the Pareto front, rather than one universal winner.
A constrained design problem can be viewed in two layers. Parameter space contains the design variables themselves, with constraints defining a feasible region. Objective space shows the consequences of each design in terms of the criteria. Mapping feasible designs from parameter space into objective space exposes the non-dominated edge of the design possibilities, which is where good compromises lie.
The weighted-sum method combines criteria into a single score using weights that express their relative importance. Normalisation is essential: without it, the criterion with the largest raw numbers dominates the result simply because of its units. In normalised objective space, lines of equal weighted score have a slope determined by the weights, and the chosen design is where such a line touches the Pareto front. The course also explains a known limitation: the weighted-sum method can fail to find designs on non-convex parts of the front and may not sample the trade-off surface evenly.
Spreadsheet-based marine case study
The fourth section turns theory into a working tool. A marine design case is set up in a spreadsheet with clearly defined variables, objectives and constraints. The spreadsheet must handle both single-criterion optimisation, to establish the extreme points, and multicriterion optimisation using normalised weighted sums. Validation compares the spreadsheet results with known or published solutions to build confidence in the model. Sensitivity studies examine how the selected design changes as weights, constraints or input data vary, revealing whether the recommendation is robust or fragile. This practical approach shows that meaningful multicriterion optimisation does not require specialist software.
Advanced approaches and published case studies
The final technical section reviews methods suitable for larger or more complex problems, drawing on published case studies. Dominance-based multi-objective genetic algorithms evolve a population of designs and rank them by Pareto dominance, generating an approximation of the whole Pareto front in one run. CFD-coupled optimisation loops link parametric geometry, mesh generation and computational fluid dynamics solvers with optimisation algorithms, for example to refine hull forms for lower resistance. Because CFD is expensive, surrogate models, such as response surfaces or kriging, approximate the expensive analysis, and hybrid approaches combine global and local search. Multi-criteria decision-making methods help decision-makers choose a final design from the Pareto set in a transparent way, accounting for stakeholder preferences.
Choosing the right level of method
A recurring theme is matching the method to the problem. For a design with a handful of variables, well-understood relationships and cheap analysis, a spreadsheet with normalised weighted sums and a careful sensitivity study may be all that is needed, and it has the advantage of being transparent to reviewers. As the number of variables grows, as objectives become non-convex or as each evaluation requires an expensive simulation, population-based methods and surrogate models become more attractive. The course encourages learners to ask what question the optimisation must answer, how much the analysis costs, how reliable the underlying models are and who will need to understand and trust the result.
The course also highlights common pitfalls: optimising a poorly validated model, choosing weights after seeing the results to justify a preferred design, ignoring constraints that only appear later in the project, and presenting a single optimum without showing the trade-offs that surround it. Avoiding these pitfalls is as important as mastering the algorithms themselves.
Why this matters in practice
Engineering projects increasingly need to show that designs balance cost, performance, safety and environmental impact in a rational and transparent way. Multicriterion optimisation provides the tools to explore trade-offs systematically, communicate them to clients and decision-makers and avoid designs that are unnecessarily poor in one respect. Even a simple, well-validated spreadsheet model can reveal options that intuition would miss, while an understanding of advanced methods helps engineers judge when investment in genetic algorithms, CFD loops or surrogate models is justified. The ability to explain why a design was chosen, and how sensitive that choice is to assumptions, is also valuable when presenting recommendations to clients, investors and regulators.
Key takeaways
You leave able to define optimisation problems precisely, identify Pareto-optimal designs, apply the normalised weighted-sum method, build and validate a spreadsheet optimisation for a marine case and judge when advanced methods such as genetic algorithms, CFD-coupled loops, surrogates and MCDM are appropriate.
How you study
The course is delivered fully online and is self-paced, so you can fit roughly 20 hours of study around professional commitments. Lessons combine clear explanation with examples, diagrams and short interactive checks, and the main sections end with graded quizzes (4 in total) so you can confirm you can apply each method before moving on. A 69-page course book can be downloaded and printed for offline reading and later reference.
Course contents
- Module Context & Core Concepts: Module Context; Optimization Concepts
- Pareto and Weighted Sum: Pareto Optimum; Feasible Spaces; Weighted-Sum Method
- Bulk-Carrier Case Study: Problem Setup; Validation Results; Sensitivity Studies
- Advanced Optimization Methods: Dominance and MOGA; CFD-Coupled Loops; Surrogates and Hybrids; MCDM Selection
- Summary: Key Takeaways
Who this course is for
This course is for engineering graduates and practising engineers in naval architecture, marine, offshore and mechanical engineering who make design trade-offs, as well as researchers and students starting work on design optimisation. Familiarity with spreadsheets and basic engineering mathematics is assumed.
Frequently asked questions
How long does Optimization in Engineering Design take?
The course is designed for about 20 hours of study. It is self-paced and fully online, so you can spread it over a few weeks or complete it more quickly.
What is covered?
The course is organised into 6 sections with 18 lessons, moving from foundations to applied methods, as summarised above.
Is there material to keep after the course?
Yes. A downloadable 69-page course book accompanies the lessons for offline reading and later reference.
How is progress checked?
Each main section closes with a graded quiz (4 in total). The quizzes test understanding of the concepts and methods explained in the lessons rather than memorisation, and you can review the relevant lessons before attempting them.
Can I record this study for my own professional development?
Yes. The course notebook includes a self-assessment of the course objectives before and after study, a reflection page and a learning log, which you can use to keep your own record of professional development.
Duration: 20 hours
Related courses: A09 – Sustainable Ship Design; A01 – Principles of Naval Architecture: Support, Form, and Stability; C04 – Probabilistic Design; A17 – Advanced Structural Design and Analysis
Enrol
Study at your own pace and work through the full course material, worked examples and section quizzes. Enrol in C07 – Optimization in Engineering Design →
Recent Comments