The role of turbulence in particle-fluid interaction as induced by the outer geometry of catching-type precipitation gauges

This thesis work investigates the particle-fluid interaction of hydrometeors along the terminal part of their fall trajectories, while approaching the collector of catching-type precipitation gauges in windy conditions. Both the turbulence generated by the bluff body aerodynamics of precipitation gauges when impacted by the wind and the free-stream turbulence inherent to the natural wind are addressed to assess their role in precipitation measurements. The bluff body aerodynamics of precipitation gauges induces deviations in the trajectories of the approaching hydrometeors due to the acceleration, updraft and turbulence development upstream and above the collector of the gauge. The resulting wind-induced errors were studied in the literature using different approaches – field measurement campaigns, numerical simulations and wind tunnel experiments. In this work, the numerical approach based on Computational Fluid Dynamic (CFD) simulation, which reduces, when compared with field observations, the time and resources needed to investigate different configurations by varying the wind speed, type of precipitation and gauge geometry, is employed. A Lagrangian Particle Tracking (LPT) model provides the catch ratios as a function of the particle size and wind speed. The LPT model, already available from the literature, was adapted to simulate the trajectories of water droplets when falling through the atmosphere and approaching the gauge collector by parameterizing liquid particles with spherical shape and using suitable drag coefficient equations. The first part of the work aims to validate the numerical approach against a dedicated, innovative and robust experimental campaign obtained by means of Wind Tunnel (WT) experiments (flow velocity measurements, Particle Image Velocimetry and video tracking of water drops) conducted in the wind tunnel facilities available at DICCA and at Politecnico di Milano (within the PRIN 20154WX5NA project). The video tracking experimental setup allowed to compare observed and simulated trajectories under various wind velocity and drop size conditions, and to validate the Lagrangian Particle Tracking model, here adapted to simulate particles falling at a different vertical velocity than the terminal one. Comparison and validation of numerical simulation results against field-measured data introduce the problem of confronting this simplified approach with the natural atmospheric conditions actually affecting operational instruments in the field. Natural wind fields are indeed characterized by turbulent fluctuations, especially near to the ground where precipitation gauges are located. Dedicated CFD simulations with various turbulence generating solutions, based on imposing specific boundary conditions or inserting suitable obstacles designed to achieve the desired level of free-stream turbulence upstream of the gauge, were performed. Wind tunnel measurements were performed in the DICCA facility using, as a turbulence-generating device, a fixed solid fence with a regular square mesh inserted upstream of a calyx shaped gauge. CFD simulations were performed reproducing the same conditions and results were validated by comparison with WT measurements. The comparison between the uniform and turbulent free-stream conditions showed that the normalized updraft in the upwind part, upstream of the centre of the collector, and the downdraft in the downwind part are less accentuated in the turbulent free-stream configuration than in uniform free-stream conditions. This is ascribable to the energy dissipation induced by turbulent fluctuations. The dissipative effect of the free-stream turbulence also has a damping role on the acceleration of the flow above the collector as demonstrated by CFD results. The overall free-stream turbulence effect on the collection performance of the gauges was quantified by computing and comparing the Collection Efficiency (CE) values in uniform and turbulent free-stream conditions. Results demonstrated that the CE values are higher in turbulent free-stream conditions. The effect of the free-stream turbulence on the collection efficiency of the Hotplate© snow gauge was investigated, and the literature turbulence intensity level (from Øistad, 2015) impacting on the gauge by was obtained in the simulation by imposing a constant turbulent kinetic energy value as a boundary condition upstream of the gauge. The calculated catch ratios are larger for the free-stream turbulence condition with respect to the uniform one for all characteristic sizes of snowflakes. Consequently, the same effect was observed in the calculated CE values. In addition, in order to introduce a realistic level of turbulence at the gauge collector elevation in the simulation, wind speed measurements obtained from a 3D ultrasonic anemometer in the Nafferton Farm site (UK), recorded at high frequency (20 Hz) and at the gauge elevation, were analysed to calculate the free-stream turbulence intensity values for various wind speeds. This was used to perform a CFD simulation on a chimney shaped gauge and to calculate its effect on the collection performance. To better reproduce the decay of the turbulence intensity in space and its effect on the gauge, Large Eddy Simulations (LES) were also performed in both uniform and turbulent free-stream conditions while simulating the trajectories of solid precipitation particles, which are more sensitive than raindrops to the turbulent fluctuations. Results, in terms of the catch ratio for each characteristic size of snowflakes, show a different behaviour when compared to the uniform conditions. A larger free-stream turbulence intensity induces a more pronounced undercatch for small size particles (less than 2 mm) with respect to the uniform case, while the undercatch is reduced for larger particles. This is due to the greater aptitude of the small size particles to follow the turbulent velocity fluctuations, while larger particles are more inertial, and to the reduced velocity components that particles cross in turbulent free-stream conditions near the gauge body. The obtained CE values are higher in turbulent free-stream conditions, confirming the observations already obtained for the airflow features, where a potential overestimation of the undercatch obtained in uniform free-stream conditions was hypothesized. Based on the CFD results and on the validation provided by wind tunnel observations it is possible to conclude that accounting for the free-stream airflow turbulence in the simulation is required to avoid underestimation of the collection efficiency of precipitation gauges. A turbulent free-stream is indeed the natural atmospheric condition of the wind impacting on operational precipitation gauges in the field. This work demonstrates that numerical derivation of correction curves for use in precipitation measurements as proposed hitherto in the literature is affected by a systematic overestimation of the wind-induced error due to the simplifying assumption of uniform free-stream conditions. Finally, in order to achieve results that can be used in an operational context, suitable Collection Efficiency (CE) curves and the associated adjustment curves, which directly provide the expected undercatch as a function of the wind speed and the measured precipitation intensity, were derived for two sample measurement instruments. The first one is best suited for rainfall measurements and is characterised by the common cylindrical shape of traditional catching type gauges, therefore a numerical formulation of the CE curves as a function of rainfall intensity is proposed. The second one, the Hotplate© gauge, is best suited for snowfall measurements and is characterised by an innovative measuring principle implying a dedicated geometry of the sensor. In this case, the numerically derived CE curves are expressed as a function of snowfall intensity. For the typical cylindrical gauge, the residual dependency of the CE curves on the rainfall intensity was investigated in order to obtain a single CE expression as a function of both the rainfall intensity and wind speed. The parameters of the Particle Size Distribution (PSD) for various classes of the RI were derived by literature data from the Italian territory. Then the variation of the PSD parameters as a function of the RI was obtained, and subsequently also the parameters of the sigmoidal curves, used to fit the numerical CE values, were parametrized with the RI. As a result, easy to use adjustment curves as a function of both the measured rainfall intensity and wind speed were derived. In the case of the Hotplate© snow gauge, the shape of the CE curves differs from the typical sigmoidal one due to its complex geometry. At low wind speed, the aerodynamic response of the gauge is predominant and CE values decrease with increasing the wind speed up to a wind threshold value beyond which the geometrical effect on the collection performance starts to be relevant and the CE increases. At very high wind speeds the geometrical contribution prevails and the CE becomes even larger than one.