Laboratory dataset for a locally fabricated flat Coanda-effect screen: 52 clear-water runs with Ogee accelerator plates, and capacity model predictions under two discharge relations

Experimental dataset and analysis code for a laboratory study of a flatCoanda-effect screen intake fabricated from locally available steel bar andtested at the hydraulics laboratory of the Universidad Tecnica Particular deLoja (UTPL), Loja, Ecuador, in November 2014. SCREEN AND STRUCTUREThe wedge-wire panel comprises 32 wires machined to a triangular profile 7.6 mmwide, mounted with a 0.8 mm slot opening and a 4.75 degree downstream tilt,giving a shearing offset of 0.69 mm. The active panel is 0.18 m wide and 0.273 mlong, installed in a rectangular flume 0.20 m wide and 0.30 m deep. Fourinterchangeable Ogee accelerator plates combine vertical drop heights of 0.10 mand 0.15 m with screen tangency angles of 45 and 50 degrees (configurations A toD). The crest sits 0.100 m above the approach channel invert. MEASUREMENTSFifty-two clear-water runs cover inflows from 2.24 to 21.36 L/s. The inflow wasdetermined from the head over the accelerator-plate crest; the captureddischarge was gauged with a sharp-crested 60 degree V-notch weir installed in acollection channel 0.26 m wide, with the notch vertex 0.05 m above the bed. Inthe 19 runs where the screen captured the entire inflow the two gauging devicesmeasure the same discharge independently and agree to within 0.28 per cent(mean difference -0.05 per cent, standard deviation 0.12 per cent), whichverifies the whole gauging chain. FILES coanda_screen_dataset_52runs.csv - all 52 runs. Columns: configuration;plate_drop_height_m; tangency_angle_deg; head_ogee_crest_m; head_vnotch_m(the two raw stage readings); Q_in_crest_Ls; Q_R_weir_Ls (measured capture);Q_T_HPCES_Ls and Q_bypass_HPCES_Ls (2014 model output); deviation_Ls;ratio_QR_QT_pct; informative (1 for runs above full wetting of the panel, whichare the only ones that test the capture model; 0 otherwise). HPCES_reruns_results.csv - the 34 informative runs with predictions recomputedunder four combinations. Columns: config; hp_m; theta_deg; Qin_Ls;QR_measured_Ls; QT_2013_w018_Ls and QT_2013_wcorr_Ls (Wahl 2013 angle-of-attackrelation, inflow distributed over 0.18 m and over 0.20 m respectively);QT_2021_w018_Ls and QT_2021_wcorr_Ls (offset-based relation of Wahl et al. 2021,same two conventions); and the flow state computed at the leading edge of thescreen: V0_ms; d0_mm; Froude; delta_psi_deg (angle of attack); Reynolds; Weber;wetted_len_m. All discharges in litres per second. coanda.py - a Python reimplementation of the published Coanda screen capacitymodel, ported from the source of the Excel release dated 2020-12-16 (originalVisual Basic program by T. L. Wahl, US Bureau of Reclamation; Excel port byC. Prestwich, USDA-NRCS). It reproduces the ogee crest design parameters of the2014 analysis exactly and its predictions with a mean bias of +0.004 L/s and amaximum departure of 0.045 L/s over the 34 informative runs. All four coefficientoptions implemented in the source are available, so both the 2013 and the 2021relations can be evaluated. Internal units are English (feet, cfs per foot);inputs and outputs of the Screen class are SI. Requires only the Python standardlibrary. PROVENANCEThe experimental campaign forms part of the undergraduate civil engineeringthesis of R. J. Rivas Loaiza, completed at UTPL in January 2015 under thesupervision of E. F. Carrion Coronel. The dataset has not been publishedelsewhere. Reuse note: the runs marked informative = 0 agree with the model trivially,because the screen captures the entire inflow and mass conservation alone forcesthe agreement. They verify the gauging, not the capture model, and should beexcluded from any agreement metric.