Now Online: COMPSs Workflows for the CAELESTIS ISE
Aviation accounts for ~2.5% of global CO₂ emissions, a figure projected to rise without intervention. The EU’s Flightpath 2050 targets demand a 75% reduction in CO₂ per passenger kilometre by 2050 compared to 2000 levels. Traditional aircraft development relies on physical prototypes and fragmented simulations, which are time-consuming, costly, and limit innovation. CAELESTIS addresses these challenges by creating a hyperconnected digital ecosystem that bridges design, manufacturing, and quality assurance through probabilistic modeling and predictive analytics.
It is in this context that the project has recently published COMPSs Workflows – an integral part of the CAELESTIS project. This system is designed to manage and execute complex workflows for aircraft manufacturing processes using high-performance computing (HPC).
CAELESTIS Strategy
At the heart of our workflow management system lies COMP superscalar (COMPSs), a powerful parallel programming model developed by the Barcelona Supercomputing Center (BSC). COMPSs simplifies the development and execution of complex applications on distributed computing infrastructures, such as supercomputers and cloud environments. It allows engineers to define tasks as independent units of work, and then automatically schedules and executes these tasks in parallel, optimizing resource utilization and minimizing execution time. By abstracting away the complexities of parallel programming, COMPSs enables our team to focus on the core simulation logic, accelerating the design and optimization of next-generation aircraft.
So what are the technical main aims of CAELESTIS? Firstly, there’s digital thread integration. This will aim to link design, simulation, and manufacturing workflows across the aircraft value chain (airframes, engines, etc.) to enable seamless data exchange between stakeholders. For example, engineers modifying a wing design will be able to instantly see how changes affect manufacturing tolerances or engine compatibility. Next, there’s the advent of probabilistic digital twins, virtual replicas of aircraft components that simulate real-world performance under varying conditions (e.g., stress, temperature) and manufacturing uncertainties (e.g., material defects). Unlike static models, these twins use machine learning to predict outcomes like fatigue life or fuel efficiency with quantified confidence intervals. The use of High-Performance Computing (HPC) analytics, stemming from thousands of parallel simulations, will reduce design-test cycles from months to days. For instance, optimizing a turbine blade’s aerodynamics might involve 10,000+ computational fluid dynamics (CFD) runs. This will cycle back to Smart Manufacturing Strategies, able to implement real-time defect detection using sensors and inline quality checks. This reduces waste by identifying flawed components early, such as detecting microcracks in composite materials during layup.
Key Components of the Workflow
The workflow system consists of three main components:
- CAELESTIS Simulation Service: Facilitates interaction between engineers and HPC systems for defining and submitting simulations.
- CAELESTIS Repositories: Store simulation codes, datasets, and metadata descriptions.
- Workflow Management System: Orchestrates the execution of simulation workflows on HPC systems using PyCOMPSs, a task-based parallel programming model.
This integrated system allows for seamless data flow between different simulation stages, such as automated fiber placement (AFP) and resin transfer molding (RTM), ultimately feeding into mechanical performance simulations.
By leveraging high-performance computing and advanced simulation techniques, the CAELESTIS workflow system is poised to play a crucial role in achieving Europe’s 2050 climate goals while enhancing the competitiveness of the EU aeronautics industry.
The Workflow Engine: How CAELESTIS Functions
The project’s backbone is its Interoperable Simulation Ecosystem (ISE), a three-layered framework:
1. Workflow Design & Templates
Engineers use predefined workflow templates (stored in GitHub repositories) to automate multi-step processes like sensitivity analyses or uncertainty quantification. Templates are customizable: A composite wing analysis might swap CFD tools or adjust defect thresholds based on material specs.
2. High-Performance Execution via PyCOMPSs
The PyCOMPSs framework breaks workflows into parallel tasks. For example, a Monte Carlo simulation analyzing 50,000 design variants distributes batches across HPC nodes, automatically managing dependencies.
Key features:
Automatic parallelization: Tasks like mesh generation or stress analysis run concurrently.
Fault tolerance: Failed tasks are rerouted without halting the entire workflow.
3. Digital Twin Feedback Loops
Manufacturing data (e.g., 3D printer temperatures, CNC machine tolerances) feeds back into design twins, refining predictions. Machine learning models trained on HPC output predict outcomes like:
Defect propagation risks in engine blades
Optimal wing shapes for minimal drag
Impact & Industry Transformation
By 2025, CAELESTIS aims to demonstrate:
- 30% reduction in fuel burn through lighter, aerodynamically optimized designs.
- 50% faster certification of new aircraft by replacing physical tests with validated digital prototypes.
- 20% cost savings in manufacturing via defect reduction and material efficiency.
A case study involving additive manufacturing illustrates this:
- A turbine bracket design is simulated under 15,000 thermal and mechanical load scenarios.
- ML algorithms identify optimal support structures to prevent warping during 3D printing.
- Real-time sensors during printing validate predictions, adjusting laser power if deviations occur.
Challenges & Future Horizons
While promising, CAELESTIS faces myriad hurdles ahead before its innovations can be widely adopted. Data interoperability poses a significant barrier, as integrating legacy CAD systems from Airbus, Safran, etc., requires standardized APIs to be in use across the value chain. Additionally, HPC accessibility – i.e., viable connections with supercomputers – may prove challenging, as smaller suppliers may lack infrastructure to run resource-intensive workflows, necessitating cloud-based solutions. Looking ahead, the ISE could expand to maintenance (predicting part replacements) and supersonic transport. As noted by coordinators AIMEN:
“This isn’t just about building better planes—it’s about creating a living digital ecosystem that evolves with the industry.”
By bridging virtual and physical realms, CAELESTIS positions Europe at the forefront of sustainable aviation, proving that computational ambition can indeed take flight.
Find out more about RTDS Group's work with CAELESTIS
RTDS Association is supporting implementation and management of European research and innovation projects to optimise the use of results for maximum impact. RTDS has a proven track record in enabling the innovators to move research and technology from the lab to the market.
