Unsteady Fluid Mechanics Benchmark Dataset - Imposed Cylinder Motion at Re=100

This dataset contains computational fluid dynamics simulation data for the two-dimensional flow past a circular cylinder undergoing an imposed oscillatory motion perpendicular to a uniform incoming cross-flow. All simulations are performed at Reynolds number 100, a regime in which the wake is laminar and remains, to a very good approximation, two-dimensional. The dataset is intended as a benchmark for nonlinear system identification, reduced-order modelling, and data-driven analysis of unsteady fluid mechanics.<br>The system represents a canonical fluid-structure interaction problem in which the cylinder motion is prescribed rather than determined by structural dynamics. This choice makes the problem easier to control computationally and allows systematic exploration of the input space through selected forcing amplitudes and frequencies. The imposed motion acts in the cross-flow direction, while the oncoming flow remains uniform in the streamwise direction.<br>The dataset includes four classes of imposed cylinder motion. The first class consists of swept-sine excitations with linearly varying frequency and constant amplitude. These signals sweep from 0 Hz to 1.5 times the natural shedding frequency and then sweep back down, after which the cylinder is held stationary for several seconds so that the natural vortex shedding can re-establish itself. The second class consists of full random-phase multisines with flat amplitude spectra over the same frequency range, using a frequency resolution of 0.1 Hz and two random-phase realisations for each amplitude level. The third class consists of monosine experiments. These are generated using a sweep-and-hold procedure: the excitation frequency first ramps up linearly and is then held constant for the remainder of the simulation. These monosine cases were chosen to cover combinations of frequency and amplitude both inside and outside the lock-in region. The fourth class comprises a stationary-cylinder experiment.<br>For the swept-sine experiments, the imposed displacement amplitude ranges from 0.025 to 0.3 times the cylinder diameter, in steps of 0.025 times the cylinder diameter. The sweep period for a single sweep direction is 16 s. For the multisine experiments, the amplitude ranges from 0.05 to 0.3 times the cylinder diameter, in steps of 0.05 times the cylinder diameter. For the monosine experiments, the forcing frequencies are 0.50, 0.70, 0.90, 1.10, and 1.30 times the natural shedding frequency, combined with amplitudes of 0.10 and 0.20 times the cylinder diameter. The cylinder diameter is 0.1 m, and the natural shedding frequency of the stationary cylinder is taken as 3 Hz for the description of the excitation signals.<br>All variables are sampled at 40 Hz, which provides sufficient bandwidth to resolve up to the fourth harmonic of the highest excitation frequency. Each simulation is initialised from the final state of a long stationary-cylinder computation, so that the flow starts from a fully developed and stable natural vortex shedding regime.For every simulation, the dataset provides time-resolved measurements of the cylinder motion and the fluid response. The available variables are:<br>cylinder position in the cross-flow directiondrag coefficientlift coefficientkinematic pressure along the cylinder surfacestreamwise velocity in the wakecross-flow velocity in the wakekinematic pressure in the wakespanwise vorticity in the wake<br>The measurement locations are also specified. Surface pressure is recorded along the cylinder surface. Wake variables are collected on a fixed dense measurement grid downstream of the cylinder, obtained by interpolation from the CFD mesh. This wake measurement domain extends from 0.06 m to 0.76 m in the streamwise direction and from −0.20 m to 0.20 m in the cross-flow direction. The grid contains 105 points in the streamwise direction and 61 points in the cross-flow direction.The dataset therefore includes both the prescribed input signal, given by the imposed cylinder displacement, and multiple possible output quantities ranging from integrated force coefficients to densely sampled wake-field variables. This makes it suitable for single-input single-output, single-input multiple-output, and larger multiple-input multiple-output benchmark studies. It is particularly useful for studying nonlinear phenomena such as autonomous vortex shedding, synchronisation, lock-in, hysteresis, and harmonic generation in forced bluff-body wakes.