Five disciplines that overlap on almost every project. We pick the right
technique — classical or modern — and channel it into efficient software.
01
FE Analysis & Optimization
Bespoke software to automate complex finite-element analysis and structural optimisation — wrapped in tools your engineers can actually drive.
Raphos Design Aid: automated portal-frame analysis with embodied-energy metrics.
—Steel-member optimization that maximises utilisation across hyperstatic systems, re-solving internal forces at every iteration.
—Clustering of sections and parts to cut the number of distinct profiles — reducing variability, cost and waste while respecting constraints.
—NL gradient descent and stochastic gradient descent integrated with FE solvers.
—Bespoke heuristics: genetic algorithms, simulated annealing and multi-objective optimization.
—Custom plugins for the major FE and CAD packages, plus standalone optimization GUIs.
02
Machine Learning & Data Science
Machine learning, AI and state-of-the-art algorithms applied to engineering problems — from aerospace to civil to mechanical.
A neural network solving a combinatorial routing problem.
—Automatic design of steel-joint connections suggested with high accuracy by ML and optimization.
—Analysis of a wide range of medical scans to derive insight and support clinicians’ decisions.
—Statistical learning to build predictive models and expose the underlying phenomena, complementing deep learning.
—Deep learning in practice: RNN/GRU for sequence generation, CNNs to identify structural sections from point clouds, GANs to generate viable, optimized frame structures.
—Clustering across structural and architectural parts to reduce variability and cost.
03
Signal Processing & Dynamics
Signals, sensors and structural dynamics — research-grade methods turned into interactive, usable tools.
Across-wind force time histories from a high-rise wind-response study.
—Generation of spatially and temporally correlated signals from spectra, used to simulate wind turbulence on slender structures — skyscrapers, bridges and cable structures.
—Analysis of months of a high-rise building’s acceleration data and its correlation with wind speed.
—Frequency-sensitivity analysis of the highest response factors to quasi-random inputs.
—FE models and Matlab/Python scripts modelling response to wind, pedestrian movement and machinery.
—Field instrumentation: Arduino-based sensing with ~660M data points and bespoke apps for real-time transfer functions and correlations.
04
Computational Geometry
Making free-form geometry buildable: meshing, reparameterization and vector-field methods built on optimized C++ libraries.
A free-form surface panelised and rationalised for fabrication.
—Quad-only remeshing aligned to custom stress and curvature fields, with user-weighted field blending.
—State-of-the-art surface reparameterization and feature selection.
—Free-form glass-panel optimization: a two-stage algorithm that reduces panel curvature, then clusters panels to minimise the number of moulds — cutting cost and wastage.
—Surface rationalisation that reduces curvature and clusters panels while keeping edge continuity.
—Real-time geometry voxeling in C++ and differential computational geometry on discrete surfaces.
05
Software & Cloud Development
Full-stack web and standalone software — scalable, expandable, API-driven and easy to use.
A standalone Raphos application with interactive 3D structural visualisation.
—Standalone C#/C++ software, from simple GUIs to interactive 3D visualisation over your analysis backend.
—Large, scalable frameworks with clean APIs that are straightforward to expand.
—Cloud architecture on Azure — native apps with cloud web servers and databases, ML deployed on AzureML.
—Interoperability and database frameworks that move data cleanly between the many tools in an engineering stack.
—Add-ins for the Synera visual-programming platform, developed in close partnership.
Not sure which of these you need?
Describe the problem — we'll tell you how we'd approach it.
