Wind-Induced errors in rain gauge and disdrometer measurements

We are performing LES of flow around typical rain gauges and disdrometers who are the basic instruments for the in-situ rainfall measurements.  However, wind can cause severe undercatch errors in these measurements.  The wind-induced underestimation of the rain gauge rainfall is on the average about 5%, whereas in windy conditions it can be several times higher.   In a first phase LES is used to investigate the details of turbulent air flow around isolated and grouped rain gauges devices.  In a second phase two phase flow simulations will be performed to model raindrop behavior in the turbulent field using an Eulerian-Lagrangian approach that is based on a point particle model with two-way coupling between the flow and the particles.  As the final goal is to predict the raindrop dynamics, capturing the coherent structures and eddies around these gauges as well as the associated turbulence fluctuations is essential.  The forces acting on these particles will be the gravity and the drag.  Drag corrections for non-spherical particles will also be implemented.  The distributions of the particle diameter in the free stream (inlet plane in the computational domain) will be specified according to experimental data.

Two test cases are first considered.  In both cases the geometry consisted of two identical gauges mounted on a flat base (Figure 1).  The top of the gauge is an open cylinder of 30 cm diameter with flat bottom.  In the first case the direction of the wind is perpendicular to this axis, meaning that there is very little interaction between the flow around the two gauges (the wakes are parallel).  In the second case the wind direction is parallel to the axis defined by the two gauges, and the wake which develops behind the first gauge hits the gauge situated downstream.  In this case the interaction between the two gauges is the highest.  While the flow around the first gauge is not very different to the one observed around any of the gauges in the first test case, the flow around the second gauge is strongly modified.  The Reynolds number, defined with the upstream wind velocity and gauge diameter, is around 30,000 in both cases. 

Figure: View of the computational domain including the two rain gauges.  (RAIN6_3Dstream.tif)

Case 1.  Wind is perpendicular to axis of the gauges.

Movie: Animation (LES) showing instantaneous vorticity contours in a plane parallel to wind direction cutting through the center of one of the gauges.  (vorticity(x=0).avi)

Movie: Animation (LES) showing instantaneous pressure contours and 2D streamlines in a plane parallel to wind direction cutting through the center of one of the gauges.  (pressurestr(x=0).avi)

Movie: Animation (LES) showing vorticity contours in a horizontal plane situated at y/D=1.5 from the ground cutting through the cylindrical base of the gauges.  (vorticity(y=1p5).avi)

Movie: Animation (LES) showing vorticity contours in a horizontal plane situated at y/D=2 from the ground cutting through the open (top) part of the gauges.  (vorticity(y=2).avi)

Figure: Comparison of LES and RANS (SST model) solutions in a plane parallel to wind direction cutting through the center of one of the gauges.  (Rain6X=0.tif)

Figure: Comparison of LES and RANS (SST model) solutions in a horizontal plane situated at y/D=1.5 from the ground cutting through the cylindrical base of the gauges. (Rain6Y=1P5.tif)

Figure: Comparison of LES and RANS (SST model) solutions in a horizontal plane situated at y/D=2. from the ground cutting through the open (top) part of the gauges. (Rain6Y=2.tif)

Case 2.  Wind is parallel to axis of the gauges.

Figure: Comparison of LES and RANS (SST model) solutions in a plane parallel to wind direction cutting through the center of both gauges.  (Rain7X=0.tif)

Figure: Comparison of LES and RANS (SST model) solutions in a horizontal plane situated at y/D=2. from the ground cutting through the open (top) part of the gauges. (Rain7Y=2.tif)