Development of an automatic test bench to assess sprinkler irrigation uniformity in different wind conditions. Computers and Electronics in Agriculture

1 Development of an automatic test bench to assess sprinkler 1 irrigation uniformity in different wind conditions 2 B. Salvatierra-Bellido a, J. Montero-Martínez b, L. Pérez-Urrestarazu c, * 3 aSistema de Asistencia al Regante, Instituto de Investigación y Formación Agraria y Pesquera 4 de Andalucía, Camino de Esparragosa s/n, 11550 Chipiona (Cadiz), Spain. E-mail address: 5 [email protected] 6 bDepartamento de Producción Vegetal y Tecnología Agraria. Universidad de Castilla La 7 Mancha. Campus Universitario, s/n.02071 Albacete. Spain. E-mail address: 8 [email protected] 9 c Urban Greening & Biosystems Engineering Research Group. Area of Agro-Forestry 10 Engineering. Universidad de Sevilla. ETSIA Ctra. Utrera km.1, 41013. Seville. Spain. E-mail 11 address: [email protected] 12 *Corresponding author 13 Abstract 14 In sprinkler irrigation the water distribution uniformity in field conditions is not 15 always a known factor, mainly due to the many variables involved, especially the wind. 16 The main objective of this study was to design, install and test an automatic sprinkler 17 bench to measure the irrigation uniformity of solid set systems for multiple wind 18 conditions in real time. The system developed measures the different wind speeds and 19 directions while simultaneously recording the rainfall distribution automatically. 20 Consequently, the system requires little manual intervention, thus reducing the 21 operating costs. All the information generated is stored in a database, obtaining multiple 22 results of irrigation uniformity for each stable wind regime. As a second step, 23 uniformities in different situations (layouts and wind directions) were studied. In 24 addition, this study shows the potential for assessing the influence of different variables 25 2 on irrigation uniformity for several sets of sprinklers. As an example of possible 26 applications, 12150 results of uniformity coefficients for conventional impact rotary 27 head sprinklers with hexagonal nozzles in windy conditions were generated. These data 28 were used to establish comparisons between different sprinklers. To do this, a multiple 29 linear regression methodology was applied in order to analyse the influence of the 30 different contour variables on the irrigation uniformity. The test bench presented along 31 with the methodology to simulate and generate multiple scenarios constitutes a powerful 32 tool for designers, farmers and technicians both for the improvement of existing 33 installations and for future designs. The generation of a large amount of irrigation 34 uniformity results for sprinkler irrigation in different wind conditions will lead to a large 35 database with the potential to be able to determine the irrigation uniformity in all 36 common scenarios. 37 Keywords: Solid set sprinkler systems; irrigation evaluation; automation 38 39 1. Introduction 40 The modernization processes carried out in the irrigated areas have led to the 41 installation of pressurized networks. This fact has induced a change from surface 42 irrigation to sprinkler or trickle irrigation, which are purported to have higher water 43 application efficiencies, better control of the water depth applied and enable automation. 44 However, sprinkler systems have certain disadvantages with respect to trickle irrigation. 45 The most important is the poor uniformity of irrigation in wind conditions and the 46 means to determine it. Water application uniformity is the main indicator of irrigation 47 quality. It can be expressed through different parameters or coefficients, such as the 48 Distribution Uniformity (DU) (Merriam and Keller, 1978) or the Christiansen's 49 3 Uniformity coefficient (CU) (Christiansen, 1942). Irrigation management with sprinkler 50 irrigation systems would benefit from site-specific, comprehensive and accurate 51 information about irrigation uniformity, especially in windy conditions. With one study 52 using a well-executed irrigation schedule based on crop requirements, yield increased 53 with a higher irrigation uniformity (Li., 1998). According to Keller and Bliesner (1990), 54 most irrigation sprinkler systems require a minimum CU value greater than 80%. Bralts 55 et al. (1994) indicated that a 5-12% increase in CU could lead to 3-17% more yield in 56 wheat grain. Moreover, according to Tarjuelo et al. (1999b), low CU values generally 57 indicate a faulty combination of the number and size of nozzles, pressure and spacing of 58 sprinklers. 59 Many factors affect the performance of sprinkler irrigation. However, the wind 60 is an uncontrollable variable and has a decisive influence on sprinkler irrigation 61 efficiency and uniformity (Tarjuelo et al., 1999b). Therefore, knowing the DU for each 62 irrigation scenario and possible wind regime is desirable. This allows for determining 63 the optimal timing for irrigation in order to minimize the effects due to wind (Sánchez 64 et al., 2011). . 65 Wind speed and direction are the main parameters that have a greater impact on 66 the water distribution model (Tarjuelo et al., 1999b) and play an important role in drift 67 and evaporation losses (Tarjuelo et al., 2000; Keller and Bliesner, 1990). Many authors 68 indicate that the influence of the wind depends greatly on system design parameters, 69 such as working pressure, spacing, nozzle size or type of sprinkler (Keller and Bliesner, 70 1990). 71 Different methods can be used to determine sprinkler irrigation uniformity. Each 72 procedure is adapted to information requirements, with a more or less limited scope of 73 4 results. In a Radial model (Vories and Von Bernuth, 1986), an isolated and windless 74 evaluation of the sprinkler is performed, using a certain nozzle and a specific operating 75 pressure. It is basically an evaluation with a row of rain gauges along the radius of the 76 wet area of the sprinkler. The results obtained are used to measure the irrigation 77 uniformity that the entire wet area in the field would have. It is mainly used to 78 characterize the sprinklers and nozzles in ideal conditions without wind (Tarjuelo et al., 79 1999a). It is required information for sprinkler manufacturers which offer basic data for 80 the irrigation design. 81 The Matrix model (ISO 15886-3:2012, 2012) is also an evaluation of an isolated 82 sprinkler, but having the advantage of knowing the complete water distribution pattern 83 of the sprinkler in the whole wet area. It is mainly used to characterize the sprinkler and 84 the nozzles in windy conditions. It consists in setting a network of rain gauges covering 85 the wetted surface of an isolated sprinkler. This will allow for overlapping data 86 according to the operation layout. This procedure has three disadvantages: (1) the 87 variability of the climatic conditions during the test, (2) the different evaporation rate in 88 the peripheral collectors with respect to the central ones and (3) the high manpower 89 requirements for each test. It is mainly used in research centres dedicated to the study of 90 sprinkle irrigation. 91 Lastly, the evaluation of the system (Merriam and Keller, 1978; Merriam et al., 92 1980) consists of the actual field evaluation of an existing irrigation facility. It is 93 performed in a sample area of the installation and by the provision of a network of 94 collectors. It is ideal to determine the quality of irrigation in specific conditions (wind, 95 pressure, etc.) in which the evaluation is done. 96 97 5 However, in recent decades, many simulation models for irrigation have been 98 developed with different theories, in order to avoid the problems of experimental field 99 tests. Ballistic models are based on simulating the trajectory of drops of water in the air 100 when they come out of the sprinkler and are distorted by the action of wind (Seginer et 101 al., 1991; Carrion et al., 2001; Montero et al., 2001; Playan et al., 2006; Li et al, 2015 102 and Yongchong et al., 2015). Semi-empirical models simulate the shape of water 103 distribution distorted by the wind, starting from results in windless conditions (Richards 104 and Weatherhead, 1993; Han et al., 1994; Molle and Le Gat, 2000; Granier et al., 2003 105 and Oliveira et al., 2013). Other models use mathematical techniques of artificial neural 106 networks, simulating the effect of the wind on the sprinkler water distribution pattern 107 (Lazarovitch et al., 2009; Hinnell et al., 2010; Sayyadi et al., 2012). In each case, the 108 simulation models should be calibrated and validated through experimental tests. 109 Depending on the chosen method, the quantity and quality of information will 110 vary. In ballistic models, a large database that characterizes all the sprinklers can be 111 obtained but only for the conditions of operation without wind. Semi-empirical models 112 can be considered the most accurate and their results can be easily extrapolated. 113 However, this evaluation is costly both in time and resources, and requires a specific 114 infrastructure. In the third case, many field evaluations can be performed but in such 115 specific conditions that they will not readily adapt to other circumstances. 116 For obtaining water distribution data from isolated sprinklers, radial or matrix 117 models can be used. Both can be automated to avoid labour costs, while they have the 118 advantage of reducing the error due to evaporation from the collectors during the test. 119 Hodges et al. (1990) used the matrix system in an automated test facility programmed to 120 operate unattended when wind speed exceeded 2.2 m·s-1. Although, as has been stated, 121 previous attempts have been made, it is necessary to consolidate a system in order to 122 6 have an operational tool that addresses the lack of knowledge about irrigation 123 uniformity under multiple real operating conditions. 124 The main objective of this work is the design of an automatic test bench for the 125 study of uniformity in solid set sprinklers systems in different wind conditions. The 126 main differences with respect to the bench developed by Hodges et al. (1990) are: 127 - Increasing the surface area of the bench so as to permit rain collection with 128 winds greater than 2.2 m·s-1. 129 - Improving measurement precision by increasing collector size. 130 - Having a more accurate and efficient data acquisition system which permits 131 real-time data analysis. This allows for instantaneous extraction of data 132 concerning water distribution and wind speed and direction, therefore 133 permitting the execution of more trials per working day thus increasing 134 bench performance. 135 The development of this equipment involves a complex data acquisition and 136 processing system which allows further analysis. This tool will not only generate a large 137 amount of experimental data, but coupled with the simulation method proposed by Han 138 et al. (1994) will be able to recreate solutions for any real situation that may occur in the 139 field. 140 141 2. Materials and methods 142 2.1. Design requirements 143 7 The test bench was located at the Agrarian Research and Training Centre of 144 Chipiona in Cadiz, Spain (Geographical coordinates: 36.751351,-6.4003860). The 145 following requirements were contemplated: 146 1. Isolated sprinkler test following the method proposed in ISO 15886-3:2012 147 (2012). 148 2. Instant and continuous measurement of the temporal and spatial water 149 distribution, using automatic rain gauges with a tipping bucket, that register the 150 amount and the time in which water is collected at each sampling point. The 151 sprinkler is located in the centre of the grid of rain gauges, which are 152 electronically interconnected, with a spacing of 2 by 2 meters. 153 3. Instantaneous measurement of the wind speed and direction at all times with an 154 automatic wind sensor. The relative wind directions are standardized in a later 155 simulation, with respect to the irrigation lateral (Norenberg et al., 2017), in three 156 directions: parallel, oblique and perpendicular, independent of the wind direction 157 (Figure 1). This allows for the organization and simplification of the substantial 158 amount of results obtained. 159 8 160 Figure 1. Standardization of relative wind directions with respect to the lateral line. 161 4. Obtaining and processing of data: All values of instant rainfall at each sampling 162 point and wind speed and direction were stored in a data acquisition and storage 163 system. They were processed for all results of irrigation uniformity with respect 164 to the recorded wind in the different situations. 165 The equipment was prepared to keep working non-stop, in order to acquire data 166 from different wind regimes throughout the trial period. 167 2.2. System Architecture 168 The bench consists of six functional units (Figure 2) each of which is described 169 below. 170 9 171 Figure 2. (a) In-field test and (b) components of test bench. (1) sprinkler, (2) catch can, 172 (3) automated rain gauges, (4) wind sensor, (5) Data Acquisition System (DAS), and (6) 173 data acquisition tower. 174 2.2.1. Hydraulic system 175 Water was supplied from a well by means of a 2.2 kW pump Prisma 35 N 176 (ESPA 2025 S.L., Banyoles, Spain). A polyethylene 90 mm pipe carries water from the 177 pump to the sprinkler (Figure 3). 178 According to the standard ISO 15886-3:2012 (2012), changes of pressure must 179 not be over 2% throughout the trial. Therefore, a pressure regulator is placed 180 downstream of the pump in order to ensure the exact required pressure at all times. Two 181 flowmeters (ARAD, model M25, one inch in diameter) were installed to register the 182 16 maintenance plan included two fundamental tasks: cleaning the rain gauges and its 287 subsequent calibration. 288 The cleaning operations consisted of removing the collector and the rain gauge 289 cover and washing with a pressure hose. When the presence of weeds or snails was 290 detected, herbicides and helicides were applied to keep the mechanism of the rain gauge 291 free of obstacles. 292 A recalibration of the rain gauges was carried out in each maintenance task. In it, 293 the value measured by each rain gauge is corrected with respect to a known rainfall 294 value. This generated a correction coefficient for each rain gauge. 295 The calibration consisted of the following steps: activation of the bench without 296 sprinkler, pouring 0.25 L of water into each rain gauge during 2.5 minutes (3 s per 297 pulse), simulating the maximum flow rate that a sprinkler can give and not exceeding 298 the range of measuring capability provided by the manufacturer of the rain gauge. Once 299 provided the correct amount of water in all gauges, data from the DAS is downloaded 300 and the correction coefficient is calculated for each rain gauge to be subsequently 301 applied to all measurements. 302 2.6. Example of use. 303 As an example of how the test bench can be used, multiple tests were performed 304 with different sprinklers in diverse conditions taking into account eight variables (Table 305 1). 306 Table 1. Summary of the number of values defined for each test variable. 307 Variable Number of values 17 Manufacturer 5 Sprinkler models per manufacturer 2 Number of nozzle combinations per model 1-4 Working pressures per combination 3 Wind speeds per combination 2 Wind directions per wind speed 3 Layouts per combination 81 Layout shapes per combination 2 308 The first three variables (manufacturer, sprinkler model and nozzle combination) 309 refer to the most common sprinklers used in Spain (Table 2). The following (pressure, 310 wind speed and direction of the wind and layout) correspond to the following values: 311 - Tested pressures: 200, 250, and 350 kPa. 312 - Wind speeds: 2 values preferably were processed only for each sprinkler 313 model, one with slight wind (less than 2 m·s-1) and the other with 314 moderate wind (between 2 and 4 m·s-1). 315 - Wind directions: parallel, oblique and perpendicular with respect to the 316 lateral, categorising the multiple results that could be obtained in those 317 three. 318 - The layouts studied correspond to all the combinations ranging from 10 x 319 10 m to 18 x 18 m in all possible configurations with an interval of 1 m 320 between each layout variation. 321 - The two shapes studied are rectangular and triangular. 322 Table 2. Manufacturer, sprinkler models and nozzle combinations considered. 323 18 Manufacturer Sprinkler model Nozzle combination Main nozzle (diameter in mm) Jetstraightening vane (Yes or No) Secondary nozzle (diameter in mm) Unirain F46 2.6 Y 3.6 Y 3.2 Y 4.0 Y Y 2.4 F46-PRO 4.4 Y 2.4 4.8 Y 2.4 3.6 Y Naandanjain 6025 SD 3.5 Y 5035 SD 3.5 Y 2.5 4.0 Y 2.5 Senninger 3023-2 3.2 Y 2.4 4.0 Y 2.4 Vyrsa Vyr 36 4.0 Y 2.4 3.6 Y 2.4 Vyr 37 4.0 Y 2.4 3.6 Y 2.4 Nelson R33-LP 4.0 N 4.8 N 4.4 N R2000-WF 3.6 N 3.2 N 324 Afterwards, the data obtained for an isolated sprinkler was used to simulate 325 multiple scenarios, calculating irrigation uniformity for different layouts. For the 326 analysis, three sets of sprinklers have been considered: 327 - Set 1: Conventional impact sprinklers with hexagonal nozzles. 328 - Set 2: Non-conventional impact rotary head sprinklers with bayonet nozzles 329 - Set 3: Rotary sprinklers. 330 331 3. Results and discussion 332 19 The test bench developed enables the performance analysis of sprinklers with 333 different operating characteristics and wind conditions. In order to show the potential of 334 the information generated, an example of application of the equipment was described. 335 Furthermore, analysis of the results and the potential to determine the influence of 336 different variables on the water uniformity distribution were presented. 337 3.1. Example of data extracted in a field test. 338 The results presented correspond to one in-field irrigation test extracted from the 339 total dataset, and are always characterized by a constant wind speed and direction (wind 340 speed standard deviation not exceeding 1 m·s-1 and 20º in direction). Figure 7 represents 341 an example of the data obtained from the DAS with the boundary conditions displayed 342 in Table 4. 343 344 Figure 7. Unirain F46 sprinkler water distribution pattern during one in-field test 345 Table 4. Variables involved in the test 346 Averagewind direction Sprinkler position Averagewind speed: 3.05 ms-1 20-25 mm 15-20 mm 10-15mm 5-10mm 0-5 mm 20 Variable Description Tested sprinkler Unirain F46 sprinkler (Unirain S.A., Spain) Nozzles diameter 3.17 mm (primary) and 2.38 mm with vane (secondary) Working pressure 300 kPa Test duration 60 min Sprinkler height 1.30 m Average wind speed 3.05 m·s-1 Average wind direction 5.99 rad with respect to geographic North Sprinkler flow rate 954 L·h-1 347 Afterwards, the numerical data of these results was introduced in the Space Pro 348 program to calculate CU and DU with different irrigation layouts and wind directions. 349 In this way, the water distribution for different irrigation layout configurations is 350 calculated for the three already defined wind directions. Figures 8 and 9 show the 351 results for the most common irrigation layouts in Spain. These results are only a sample 352 of the capacity for analysis of sprinkler irrigation uniformity. In addition, any other 353 layouts and wind direction can also be simulated. 354 21 355 Figure 8. Results of the trial for an example with three irrigation layouts and three wind 356 directions. 357 Figure 8 shows the results for 12 x 12, 12 x 15 and 12 x 18 m layouts in three 358 wind directions (parallel, oblique and perpendicular to the irrigation lateral). The best 359 values of uniformity indicators are obtained for the 12 x 12 m layout, which differ 360 depending on the wind direction. For all the layouts, a small improvement in DU is 361 observed with an oblique wind, compared to other conditions, while the perpendicular 362 wind direction offers the worst results. For the 12 x 15 m and 12 x 18 m layouts, the 363 coefficients obtained present lower values (the worst being the latter) and the 364 22 differences between the three types of wind direction are more pronounced. In the third 365 case, the differences between parallel and oblique types are attenuated, being the 366 perpendicular wind which leads to the worst coefficients. 367 On the other hand, the option of using similar layouts with different shapes has 368 been studied. Figure 9 shows a comparison between the 12 x 15 m rectangular 369 configuration analysed in the previous case with a triangular staggered 15 x 12 m 370 sprinkler layout. Similar values of the coefficients are observed with an oblique wind. 371 However, if the wind is parallel or perpendicular to the laterals, a considerable 372 improvement using a triangular layout is obtained. Therefore, in a situation where the 373 wind direction is not oblique or often changes, a triangular configuration would be 374 recommended for this layout. Thus, thanks to this type of analysis, an improvement in 375 uniformity can be achieved by simply changing the configuration without any additional 376 investment cost. 377 378 379 23 380 381 Figure 9. Results of the trial comparing the rectangular layout with the triangular 12 x 382 15 m layout. 383 Taking into account all the combinations, there were 29160 results generated 384 with the automatic bench. Therefore, the irrigation uniformity in most common 385 situations occurring in the field was analysed. In addition to this number of results, any 386 wind scenario could be extrapolated from existing experimental data using the 387 methodology proposed by Han et al. (1994). 388 With the design of this prototype of automatic test bench, and thanks to the 389 combination of methodologies for data treatment, overlapping layouts and simulation of 390 24 multiple wind directions, a vast amount of information is available for each one of the 391 tested sprinklers. Therefore, by means of a query tool in the results database, any 392 scenario can be assessed as desired. 393 3.2. Example study of the influence of several variables on the irrigation 394 uniformity for different types of sprinklers. 395 The set of results obtained with the test bench can be used for multiple 396 objectives with a strategic purpose. This section presents an example of use for the 397 generated results. In this case, a mathematical model for a global data analysis with both 398 dependent and independent variables is used. This technique allows researchers to 399 establish relationships among dataset variables. The results have been classified 400 according to the sprinkler type as determined in the Materials and methods section: 401 - Set 1: 12150 results of uniformity coefficients for conventional impact rotary 402 head sprinklers with hexagonal nozzles. 403 - Set 2:10206 results of uniformity coefficients for non-conventional impact 404 rotary head sprinklers with bayonet nozzles. 405 - Set 3: 6804 results of uniformity coefficients for rotary sprinklers. 406 Figure 10 shows the result of applying a multiple linear regression to the three 407 sets of sprinklers. The R2 values are: 0.5801 for all analysed sprinklers, 0.7392 for the 408 conventional impact sprinklers, 0.6017 for not conventional impact sprinklers, and 409 0.581 for rotary sprinklers. All are acceptable values for a multiple linear regression. 410 The influence of each variable on the irrigation uniformity for the different groups of 411 sprinklers considered is estimated in the graphic. 412 This analysis produces ample information for irrigation management and system 413 design decisions. For example, when using conventional impact sprinklers, the layout or 414 25 wind conditions had less influence. However, the working pressure and the existence of 415 a secondary nozzle were key factors. On the other hand, for non-conventional impact or 416 rotary sprinklers, the layout and the wind were the most important variables, based on a 417 large dataset. 418 419 Figure 10. Influence of each variable on the irrigation uniformity of in each set of 420 sprinklers tested with wind. 421 The scale of analysis can be reduced as shown in Figure 11, which compares the 422 set of conventional impact sprinklers with and without wind (7128 results obtained from 423 a conventional radial bench). The R2 values are: 0.6345 with no wind and 0.7392 in 424 windy conditions. In this case the number of results is lower and the objective of the 425 analysis is different. In this example, pressure and layout of the sprinklers are less 426 -5% 0% 5% 10% 15% 20% 25% 30% 35% Presence of secondary nozzle Flow rate Triangular arrangement Distance between sprinklers Distance between lateralsPressure Wind direction with respect to lateral Direction of wind perpendicular to lateral Wind speed All sprinklers tested Conventional Impact Sprinklers Not conventional Impact Sprinklers Rotary sprinklers