Quanscient and Haiqu run the most complex quantum fluid simulation yet, on IBM’s Heron R3

A new quantum algorithm ran a 15-step nonlinear fluid simulation around a solid obstacle on real quantum hardware, the most physically complex publicly documented demonstration of its kind. The technique reduces qubit requirements and circuit depth, bringing industrial CFD applications closer to feasibility.

Finnish simulation company Quanscient and quantum middleware developer Haiqu have demonstrated what they describe as the most physically complex quantum computational fluid dynamics simulation run to date on real hardware.

The two companies ran a 15-step nonlinear fluid simulation around a solid obstacle,  fluid flowing around a shape, the kind of problem relevant to aircraft wing design or vehicle aerodynamics, on IBM’s Heron R3 quantum computer, using a new algorithm they developed together called the One-Step Simplified Lattice Boltzmann Method (OSSLBM).

Computational fluid dynamics, or CFD, is one of the most resource-intensive branches of engineering simulation. Modelling how fluids behave around complex shapes requires enormous classical computing power, and the demands grow non-linearly as simulations become more detailed.

Quantum computing has long been theorised as a potential path to simulations beyond classical limits, but turning that potential into practice has been constrained by the sheer number of qubits and the circuit depth, the length of the quantum computation, required to run even moderately complex scenarios without the calculation being overwhelmed by errors.

The OSSLBM algorithm addresses this directly. Built on the quantum Lattice Boltzmann Method (QLBM), an established approach to mapping classical fluid equations onto quantum computation, the new framework reduces the computational overhead of each step, allowing a longer multi-step simulation to stay within what current quantum hardware can reliably execute.

Haiqu’s middleware layer was central to this: it reduced circuit depth, developed new algorithmic subroutines, and applied targeted error-reduction techniques that allowed the system to complete a workflow that would otherwise have been out of reach for today’s devices.

The significance of the result lies in the obstacle. Previous quantum CFD demonstrations have largely focused on simpler linear scenarios, fluid behaviour without the complications of interacting with a solid boundary.

Modelling how a fluid moves around an object is a prerequisite for any industrially meaningful application. Professor Oleksandr Kyriienko, Chair in Quantum Technologies at the University of Sheffield, described the work as “an interesting and timely contribution to quantum CFD,” adding that more research of this kind is needed to reach industrially relevant quantum solutions.

Quanscient and Haiqu have been collaborating on quantum CFD since at least 2024, when they were finalists in the Airbus and BMW Quantum Mobility Challenge, and have previously demonstrated work on IonQ hardware via Amazon Braket. Industrial applications remain years away; the current work is a research milestone establishing that the approach is feasible on current hardware at this level of complexity.