Finland's reputation for advanced marine propulsion rests on a cluster of world-class manufacturers and intensive research capabilities. When extreme-condition vessels demand reliable, efficient and powerful systems, the industry turns to companies like Steerprop, which has built global reach through a distribution network spanning more than 50 countries. To maintain this edge, the firm has begun tapping into LUMI, Europe's leading supercomputer.

Steerprop, established in 2000 and headquartered in Rauma, specialises in bespoke marine propulsion solutions including azimuth thrusters—rotating units that deliver omnidirectional thrust by spinning around a vertical axis. These systems power icebreakers, ferries and offshore vessels, replacing conventional shaft-and-rudder arrangements with integrated propulsion and steering capability.

Mahish Mohan, the company's Chief Hydrodynamicist, oversees propeller design and computational fluid dynamics modelling. He arrived at Steerprop in 2018, drawn by Finland's concentration of propulsion expertise, and is simultaneously pursuing doctoral research at Aalto University.

From trial to production use

Steerprop encountered LUMI through participation in Business Finland's Necoprop research initiative, which prompted the team to evaluate high-performance computing options. A Try&Buy pilot scheme proved decisive in demonstrating LUMI's fit for the company's simulation workloads.

The free of charge Try & Buy project includes CPU, GPU, and storage resources for testing, as well as expert support for implementation

Morten Mathisen, CSC's Customer manager

Mathisen explained that after a 3–6 month evaluation window, organisations can either transition to a commercial service agreement or conclude their engagement. For Steerprop, the experience proved compelling.

I'm very happy with LUMI. It fits our use cases perfectly

Mahish Mohan

Mohan identified a specific technical advantage: LUMI's compute nodes contain substantial core counts, which substantially accelerates parallel processing—particularly with OpenFOAM, Steerprop's simulation software of choice. Fewer nodes translate to reduced inter-node communication overhead and faster overall execution.

The transition proceeded smoothly, aided by Mohan's existing HPC background. While system-specific learning was necessary, LUMI's support infrastructure delivered rapid assistance—typically within days—for software updates, technical troubleshooting and custom code compilation. Steerprop subsequently moved into commercial operation.

Simulation speed unlocks design quality

Steerprop maintains its own internal computing cluster, but capacity constraints force difficult trade-offs. Large simulations or concurrent workloads quickly exhaust available resources, requiring engineers to either extend timelines or reduce mesh resolution—both compromising accuracy and commercial outcomes. LUMI eliminates these constraints through elastic resource scaling, enabling large-scale runs and parallel execution without quality degradation.

The impact becomes most apparent in customer engagements. Propeller design projects typically operate within compressed schedules of 3–4 weeks. Individual simulations can span several days on Steerprop's internal systems, but LUMI dramatically compresses these windows.

With LUMI, simulation times are significantly reduced. For example, a job that would take 5–6 days on our own cluster can be completed in under two days on LUMI.

Mahish Mohan

This acceleration enables broader design exploration, superior optimisation and reliable project delivery. Customers ultimately receive higher-fidelity solutions within agreed timeframes.

Alignment with clean shipping ambitions

Steerprop and its Finnish peers are pursuing zero-emission shipping and sustaining Finland's export leadership in clean marine technology and Arctic shipbuilding. Advanced computational fluid dynamics sits at the centre of meeting market expectations.

State-of-the-art CFD capability is in the centre to meet the market expectations set for Steerprop. Using LUMI we ensure that we have high performance computational resources available even for the most demanding simulation cases

Tero Tamminen, Technology Manager at Steerprop

LUMI itself embodies sustainability principles. Ranked among the world's fastest and greenest supercomputers, it operates with minimal environmental footprint and cost-effective energy consumption. Its Kajaani data centre recycles excess cooling heat into district heating infrastructure, directly supporting the green transition that Steerprop's business depends upon.

Scaling challenges and future capacity

Steerprop intends to expand LUMI utilisation for increasingly complex and computationally intensive projects. Acoustic modelling represents a critical growth area, responding to mounting environmental and regulatory requirements around underwater noise.

Acoustic simulations are significantly more demanding than traditional CFD. They require additional solver stages and heavy post-processing.

Mahish Mohan

Mohan anticipates future acoustic work may demand 1,000 cores or more—a scale LUMI can accommodate. However, expansion brings organisational challenges.

Currently, only one user at Steerprop – myself – uses LUMI. We are actively evaluating whether LUMI usage should be expanded to support multiple simultaneous users and projects at Steerprop. I see this as a challenge

Mahish Mohan

The core issue centres on resource availability. Following intensive computational periods, users encounter queue delays when seeking renewed access. LUMI operates Slurm (Simple Linux Utility for Resource Management), a standard workload and resource management system that schedules jobs, allocates resources and ensures fair multi-user operation across computing units.

We fully understand the fairness of the Slurm-based queuing system, but it can make it difficult to use LUMI when schedules are tight and predictability is critical. The challenge can be accentuated if multiple users from the same organization are simultaneously dependent on the system.

Mahish Mohan

LUMI's support team recognises the constraint and is investigating whether the national Roihu supercomputer—launching in May 2026—or the more powerful EuroHPC system LUMI-AI, scheduled for late 2027, could address industrial computing capacity needs.