IODA · Project

Software That Automatically Optimizes Product Shapes Using CFD Simulation

manufacturingTestedTRL 4

ioda.sems.qmul.ac.uk

Imagine you're designing a car air duct and you need it to push air as efficiently as possible through a tight space. Today, engineers tweak the shape by hand, run a simulation, check if it's better, and repeat — sometimes for weeks. IODA built software that figures out the best shape automatically, like a GPS that finds the fastest route but for aerodynamic design. The clever part is it works directly with the CAD software engineers already use, so the optimized shape comes back ready to manufacture, not as an abstract math result that someone has to manually redraw.

By the numbers

consortium partners

countries in consortium

industry partners

50%

industry ratio in consortium

total deliverables produced

demonstrated software components

The business problem

What needed solving

Engineering companies spend weeks manually iterating product shapes — running CFD simulations, tweaking geometry, re-running — to optimize performance of components like air ducts, turbine blades, and flow channels. Each iteration is expensive, and converting computationally optimized shapes back into manufacturable CAD models still requires manual expert work, creating a bottleneck between simulation and production.

The solution

What was built

The project delivered 25 deliverables including 6 demonstrated software components: a differentiated Open Cascade CAD kernel (OCCT) with automatic differentiation for shape optimization, a parametric engine for shape optimization, and workflow integration tools that return optimized shapes directly into CAD-ready format — all tested on industrial cases including Rolls-Royce Deutschland robustness problems.

Audience

Who needs this

Automotive OEMs and Tier-1 suppliers optimizing aerodynamic componentsTurbomachinery and aerospace companies designing engine flow pathsMarine engineering firms optimizing hull and propeller shapesIndustrial ventilation and HVAC system designersCFD software vendors looking to add adjoint optimization capabilities

Business applications

Who can put this to work

Automotive

enterprise

Target: Vehicle OEMs and Tier-1 suppliers with in-house CFD teams

If you are an automotive supplier dealing with lengthy design iteration cycles for aerodynamic components like air ducts, cooling channels, or body panels — this project developed adjoint-based optimization software that automatically finds the best shape while respecting your manufacturing constraints. The tools were demonstrated on realistic vehicle climate duct problems with 5 industry partners involved in development.

Aerospace & Turbomachinery

enterprise

Target: Turbine and engine component manufacturers

If you are a turbomachinery company dealing with expensive manual optimization of blade profiles and engine flow paths — this project built and tested optimization workflows on Rolls-Royce Deutschland robustness cases. The software integrates directly with Open Cascade CAD, returning optimized geometry ready for further engineering analysis without manual shape reconstruction.

Marine & Naval Architecture

mid-size

Target: Shipbuilders and hull design consultancies

If you are a marine engineering firm dealing with hull shape optimization where each CFD run is computationally expensive — this project's adjoint methods compute design sensitivities at a fraction of the cost of traditional approaches. The CAD-free parametrization handles arbitrary shapes, and the constraint system ensures optimized hulls fit within build-space requirements.

Frequently asked

Quick answers

What would it cost to implement this optimization approach?

The project built on Open Cascade (OCCT), which is an open-source CAD kernel, meaning there are no CAD licensing fees for the core geometry engine. Integration costs would depend on your existing CFD setup. Based on available project data, the software components were delivered as source code, suggesting potential for custom integration rather than an off-the-shelf product.

Can this handle industrial-scale problems, not just academic benchmarks?

Yes. The project explicitly targeted mid-size and large-scale industrial optimization problems supplied by 5 industry partners. Deliverables include demonstration on Rolls-Royce Deutschland robustness cases and application to industrial problems, confirming it was tested beyond academic toy examples.

What is the IP situation — can we license or use this?

The project was funded as an MSCA training network (MSCA-ITN-ETN) coordinated by Queen Mary University of London. IP likely resides with the university and consortium partners. The core CAD component (Open Cascade) is open source, but the adjoint differentiation layers and parametric engines would need licensing discussions with the consortium.

Does the optimized shape come back in a format our engineers can actually use?

This was a core focus of the project. Multiple deliverables addressed returning CAD-free optimized shapes back into CAD format for further design and analysis. The parametric engine and differentiated OCCT components were specifically built to close this gap, which previously required manual expert interpretation.

How does this compare to existing commercial optimization tools?

The adjoint approach computes design sensitivities in a single additional simulation regardless of how many design parameters you have, unlike traditional methods where cost scales with parameter count. Based on the project objective, this addresses two specific bottlenecks — automatic shape parametrization and design constraint handling — that limit current commercial tools.

What is the timeline to get this into our design workflow?

The project ran from 2015 to 2018 and produced 25 deliverables including 6 demonstrated software components with source code. Integration into an existing CFD workflow would require adapting the tools to your specific solver and CAD environment. The consortium included 10 partners across 6 countries with deep implementation experience.

Consortium

Who built it

The IODA consortium is notably industry-heavy for a training network, with 5 out of 10 partners coming from industry — a 50% ratio that signals genuine commercial relevance. The 6-country spread across Belgium, Germany, Greece, France, Italy, and the UK covers Europe's main engineering hubs. Coordinated by Queen Mary University of London, the project built on earlier EC-funded work (FlowHead and AboutFlow), suggesting accumulated expertise rather than a cold start. The inclusion of 1 SME alongside larger industrial partners indicates awareness of different adoption pathways. For a business looking to engage, the industrial partners are the most likely entry point for technology transfer discussions.

QUEEN MARY UNIVERSITY OF LONDONCoordinator · UK
ROLLS-ROYCE DEUTSCHLAND LTD & CO KGparticipant · DE
ETHNICON METSOVION POLYTECHNIONparticipant · EL
OPEN CASCADEparticipant · FR
ESI GERMANY GMBHparticipant · DE
THE QUEEN'S UNIVERSITY OF BELFASTparticipant · UK
VON KARMAN INSTITUTE FOR FLUID DYNAMICSparticipant · BE
VOLKSWAGEN AKTIENGESELLSCHAFTparticipant · DE
UNIVERSITAET PADERBORNparticipant · DE

How to reach the team

Queen Mary University of London, School of Engineering and Materials Science — reach out to the CFD research group

Order a report on this consortium See QUEEN MARY UNIVERSITY OF LONDON profile →

Next steps

Talk to the team behind this work.

Want to connect with the IODA team to explore integrating adjoint optimization into your design workflow? SciTransfer can arrange a direct introduction and technical briefing.

Order a report on this topic More in Manufacturing & Industry 4.0