Comparison between indirect and direct spray drift assessment methods

biosystems engineering 105 (2010) 2–12

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/issn/15375110

Research Paper

Comparison between indirect and direct spray drift assessment methods D. Nuyttens a,*, M. De Schampheleire b, P. Verboven c, B. Sonck a a

Institute for Agricultural and Fisheries Research (ILVO), Technology and Food Sciences Unit, Agricultural Engineering, Burg. Van Gansberghelaan 115, bus 1, 9820 Merelbeke, Belgium b University Ghent, Department of Crop Protection, Coupure links 653, 9000 Ghent, Belgium c Catholic University of Leuven, Department Biosystems, MeBioS, De Croylaan 42, 3001 Leuven, Belgium

article info The drift characteristics of 10 different spray nozzles were tested using three contrasting Article history:

drift risk assessment means namely; phase Doppler particle analyser (PDPA) laser

Received 12 March 2009

measurements, wind tunnel measurements (both indirect drift risk assessments) and field

Received in revised form

drift experiments (direct drift risk assessments). The effect of nozzle size (ISO 02, 03 04 and

1 August 2009

06) and nozzle type (standard flat-fan, pre-orifice flat-fan, air-induction) on droplet char-

Accepted 13 August 2009

acteristics, drift potential and field drift were studied. A comparison was made between the

Published online 31 October 2009

results from the indirect and direct measurements to evaluate their potential for predicting the losses occurring from pesticide drift from field crop sprayers. In total, 90 PDPA laser measurements, 46 wind tunnel experiments and 61 field drift experiments were carried out with 10 different spray nozzles at a pressure of 300 kPa. The reference spray application for the field measurements was defined as a Hardi ISO F 110 03 standard flat-fan nozzle at a pressure of 300 kPa with a nozzle or boom height of 0.50 m and a driving speed of 8 km h1; conditions that were used for each of the comparative assessments of the different nozzlepressure combinations. Results showed that with the indirect risk assessments (wind tunnel and PDPA laser measurement), driftability experiments can be made with different spraying systems under directly comparable and repeatable conditions and that both methods are suitable for relative assessments of drift risk. Measuring the proportion of the total volume of droplets smaller than 75 mm diameter was best suited to represent the drift reduction potential in the field with different nozzle-pressure combinations. This was followed by the wind tunnel approach numerically integrating the measured fallout deposit curve. Both wind tunnel approaches for measuring airborne spray gave inferior results. Based on these indirect drift measurements and a statistical drift prediction equation for the reference spraying, it was possible to come to a realistic estimate of field drift data at a driving speed of 8 km h1 and a boom height of 0.50 m. ª 2009 IAgrE. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. E-mail addresses: [email protected] (D. Nuyttens), [email protected] (M. De Schampheleire), [email protected] (P. Verboven). 1537-5110/$ – see front matter ª 2009 IAgrE. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.biosystemseng.2009.08.004

biosystems engineering 105 (2010) 2–12

Nomenclature spray drift percentage in the field expressed as a proportion of the application rate, % drift_dist drift distance parallel with wind direction, m arithmetic mean diameter, mm D10 surface mean diameter, mm D20 volume mean diameter, mm D30 sauter mean diameter, mm D32 total drift reduction potential based on field drift DRPt measurements up to 20 m drift distance from the directly sprayed zone, % DPRP drift potential reduction percentage, based on wind tunnel measurements, % DPRPV1 drift potential reduction percentage based on the calculation of the first moment of the wind tunnel airborne deposit profiles measured at a distance of 2.0 m downwind of the nozzle at distances of 0.10, 0.20, 0.30, 0.40 and 0.50 m below the nozzle, % DPRPV2 drift potential reduction percentage based on numerical integration of the wind tunnel airborne deposit profiles measured at a distance of 2.0 m downwind of the nozzle at distances of 0.10, 0.20, 0.30, 0.40 and 0.50 m below the nozzle, %

DPRPH

drift%

1.

Introduction

Within the framework of a research project on spray drift from field sprayers (Nuyttens, 2007), different spray nozzles were tested using three different risk assessment means; namely phase Doppler particle analyser (PDPA) laser measurements (Nuyttens et al., 2006, 2007a, 2009a), wind tunnel measurements (Nuyttens et al., 2009b) and field drift measurements (Nuyttens et al., 2007b). The results of the experiments have been used to perform a spray drift risk assessment (De Schampheleire et al., 2006) and to develop a CFD drift model for field sprayers (Baetens et al., 2007, 2008). PDPA laser measurements were used to measure droplet sizes and droplet velocities, both characteristics which affect driftability of the droplets (Satow et al., 1993; Bird et al., 1996; Carlsen et al., 2006). Different researchers such as Fietsam et al. (2004), Bayat and Bozdogan (2005), Guler et al. (2007) and Qi et al. (2008) have used wind tunnel measurements for relative studies of spray drift and to classify spray equipment under controlled conditions using various measurement protocols. In this study, three different wind tunnel protocols were considered. Drift values, under realistic working conditions, can only be obtained by means of field drift experiments which are time consuming (Philips and Miller, 1999; Ravier et al., 2005; Carlsen et al., 2006; De Schampheleire et al., 2008; Rimmer et al., 2009). Moreover, field experiments with different spraying systems cannot be made under directly comparable and repeatable conditions. In this paper, a comparison is made between the results obtained by indirect drift assessment methods (PDPA laser and wind tunnel) and the direct drift assessment method,

Dv

x

NMD

RSF

T VSF

vavg vvol x Vx V3.25m XH2 O

3

drift potential reduction percentage based on numerical integration of the fallout deposit curves measured at distances of 2.0, 3.0, 4.0, 5.0, 6.0 and 7.0 m downwind of the nozzle and 0.50 m below the nozzle, % volume diameter below which smaller droplets constitute x % of the total spray volume, mm number median diameter below which the droplet diameter for 50% of the number of drops are smaller, mm relative span factor, a dimensionless parameter indicative of the uniformity of the drop size distribution average temperature,  C velocity span factor, a dimensionless parameter indicative of the uniformity of the drop size velocity distribution¼vvol0:9  vvol0:1 =vvol0:5 arithmetic average droplet velocity, m s-1 droplet velocity below which slower droplets constitute x% of the total spray volume, m s1 proportion of total volume of droplets smaller than x mm in diameter, % average wind speed at a height of 3.25 m, m s1 absolute humidity expressed in grams of water vapour per unit mass of dry air, g kg1

(field drift measurements), to evaluate their potential for predicting the absolute losses of pesticide drift sedimenting from field crop sprayers.

2.

Materials and methods

2.1.

Spray application techniques

In this study, 10 Hardi spray nozzles (Hardi International A/S, Taastrup, Denmark) were tested at a pressure of 300 kPa with three different risk assessment means (PDPA laser measurements, wind tunnel measurements and field drift measurements) as presented in Table 1. The reference spray application was defined as a standard horizontal spray boom operating without air assistance with a spray boom height and nozzle distance of 0.50 m, using ISO 03 standard flat-fan nozzles at a pressure of 300 kPa and a driving speed of 8 km h1, giving an application rate of approximately 180 l ha1. Driving speed is only applicable for field measurements. This reference spray application was also used for a comparative assessment of the different other spray nozzles.

2.2.

PDPA laser measurements

Droplet size and one-dimensional vertical droplet velocity characteristics of the different spray nozzles (Table 1) have been measured using an Aerometrics phase Doppler particle analyser (TSI, Minneapolis, MN, USA) at a distance of 0.50 m below the nozzle. The measurement set-up, protocol and

4

biosystems engineering 105 (2010) 2–12

Table 1 – Overview of the tested spray nozzles at a pressure of 300 kPa (and the number of repetitions) with the PDPA laser, in the wind tunnel and in the field. Nozzle

Nozzle type

ISO nozzle size

Hardi ISO F 110 02 Hardi ISO F 110 03a Hardi ISO F 110 04 Hardi ISO F 110 06 Hardi ISO LD 110 02 Hardi ISO LD 110 03 Hardi ISO LD 110 04 Hardi ISO Injet 110 02 Hardi ISO Injet 110 03

F F F F LD LD LD Injet Injet

02 03 04 06 02 03 04 02 03

Hardi ISO Injet 110 04

Injet

04

BCPC nozzle class

EC

Nominal flow rate, l min1

PDPA laser

Wind tunnel

Field

0.80 1.20 1.60 2.40 0.80 1.20 1.60 0.80 1.20

9 9 9 9 9 9 9 9 9

2 18 3 2 4 3 3 5 3

3 32 4 3 3 3 3 3 3

1.60

9

3

4

90

46

61

Total F, Standard flat-fan nozzles; LD, Pre-orifice flat-fan nozzles; Injet, Air-induction flat-fan nozzle. a Reference spray application.

results were described in detail by Nuyttens et al. (2006, 2007a, 2009a).

2.3.

Wind tunnel measurements

Wind tunnel experiments were carried out in the Silsoe Research Institute wind tunnel facility (Wrest Park, Silsoe, Bedford, UK) to measure fallout and airborne drift. The measuring set-up, protocol and results were described by Nuyttens et al. (2009b). Drift potential values for the different spray nozzles were compared with the equivalent results obtained from the reference spraying by calculating their drift potential reduction percentage (DPRP, %) following three different approaches. DPRPV1 was calculated based on the first moment of the airborne deposit profiles, DPRPV2 based on numerical integration of the airborne deposit curves and DPRPH based on numerical integration of the fallout deposit curves.

2.4.

Field drift measurements

Field drift measurements to measure absolute drift values under field conditions for the different techniques were carried out according to ISO 22866 (2005). The measurement set-up, the protocol used and data processing were described in detail by Nuyttens et al. (2007b). Based on 32 drift experiments with the reference spray application, the following validated drift prediction equation was used to predict the amount of spray drift for this reference spraying at various meteorological conditions (Nuyttens et al., 2007b) 1:05

 13:00 þ 0:50:V3:25m þ 0:40 drift% ¼ ðdrift distÞ
 T  1:74  XH2 O

ð1Þ

where drift% is the spray drift percentage expressed as a proportion of the application rate (%), drift_dist is the drift distance parallel with wind direction (m), V3.25m is the average wind speed at a height of 3.25 m (m s1), T is the average ambient temperature ( C), XH2 O is the absolute humidity expressed in g of water vapour per unit mass of dry air (g kg1). Using Eq. 1, drift results for other spray nozzles were compared with the reference spraying, taking into account

variations in weather conditions. The total drift reduction potential (DRPt, %), as described by Nuyttens et al (2007b) for other spray nozzles was calculated by comparing their integrated measured drift curves with the integrated predicted drift curve of the reference spraying, for the same weather conditions to measure absolute drift values under field conditions for the different techniques. All field drift experiments were carried out on a flat meadow using a driving speed of 8 km h1 and with a nozzle height of 0.50 m for distances up to 20 m downwind from the directly sprayed zone.

3.

Results and discussion

3.1. Droplet characteristics, drift potential reduction percentages (DPRP) and total drift reduction potentials (DRPt) The most important droplet characteristics are presented in Table 2, together with the DPRP (DPRPV1, DPRPV2 and DPRPH) resulting from the wind tunnel measurements and the DRPt from the field drift measurements for the different nozzles. Other droplet characteristics such as Dv0.25, Dv0.75, NMD, V50, V150, V250, vavg, vvol10, vvol25, vvol75, vvol90 and the arithmetic (D10), surface (D20), volume (D30) and Sauter mean (D32) diameters were considered in the analysis but are not presented in this table. Sedimenting drift curves resulting from the field experiments at average meteorological conditions (XH2 O ¼ 8 g kg1 , V3.25m ¼ 3 m s1, T ¼ 16  C) are presented in Fig. 1a. For the same nozzle size, standard flat-fan nozzles produce the finest droplet size spectrum and the highest amount of airborne spray deposits in the wind tunnel and highest amount of sedimenting field drift. This is followed by pre-orifice flat-fan nozzles and air-induction nozzles. The larger the ISO nozzle size, the coarser the droplet size spectrum and the lower the amount of airborne deposits in the wind tunnel and sedimenting field drift. Because droplet size and droplet velocity spectra are correlated to each other, there are important differences between the three common wind tunnel measurement methods. An overview of all the results

5

biosystems engineering 105 (2010) 2–12

Table 2 – Droplet characteristics, DPRP (DPRPV1, DPRPV2 and DPRPH) and DRPt for different spray nozzles at 300 kPa (Nuyttens et al., 2006, 2007a, 2007b, 2009a, 2009b). Nozzle type F Fa F F LD LD LD Injet Injet Injet

ISO nozzle size 02 03 04 06 02 03 04 02 03 04

BCPC nozzle class

EC

Dv0.1, Dv0.5, Dv0.9, RSF V75, V100, V200, vvol10, mm mm mm % % % m s1 112.5 144.1 154.9 176.1 157.4 169.8 175.1 281.6 324.3 321.3

214.2 273.6 303.4 345.1 294.9 348.2 331.2 506.8 537.4 584.0

343.9 421.9 518.3 538.7 440.0 509.3 499.2 644.7 689.1 733.5

1.1 1.0 1.2 1.1 1.0 1.0 1.0 0.7 0.7 0.7

3.2 1.6 1.3 1.0 1.3 1.1 0.9 0.2 0.1 0.1

7.4 3.7 3.0 2.2 2.9 2.7 2.1 0.5 0.3 0.3

43.2 24.5 20.1 14.2 19.9 14.8 14.5 3.7 2.2 2.8

0.6 1.1 1.3 1.9 0.6 1.1 1.2 1.6 1.8 2.0

vvol50, VSF DPRPV1, DPRPV2, DPRPH, DRPt, m s1 % % % % 2.4 3.9 4.6 6.6 2.6 4.4 5.2 4.6 4.8 5.6

2.8 2.1 1.9 1.6 2.4 2.0 1.8 1.1 1.1 1.1

 57.9 0.0 58.9 74.8 52.5 10.6 6.5 80.3 89.4 86.6

 54.6 0.0 53.8 74.8 14.6 24.1 33.0 86.8 92.5 89.4

73.0 0.0 48.4 73.3 0.5 32.3 41.1 87.3 93.0 90.4

136.5 0 33.9 29.5 3.6 38.4 54.9 67.2 89.8 77.7

F, Standard flat-fan nozzles; LD, Pre-orifice flat-fan nozzles; Injet, Air-induction flat-fan nozzle; VF, very fine; F, fine; M, medium; VC, very coarse; EC, extremely coarse; Dv0.1, Dv0.5, Dv0.9, diameter below which smaller droplets constitute 10, 50 and 90% of the total volume; V100, V200, proportion of total volume of droplets smaller than 100, 200 mm diameter; vvol10, vvol50, droplet velocity below which slower droplets constitute 10, 50% of the total spray volume. a Reference spray application.

together with a detailed discussion can be found in Nuyttens et al. (2006, 2007a, 2007b, 2009a, 2009b).

3.2. Relationship between droplet characteristics and field DRPt values To investigate the importance of the droplet characteristics on DRPt values, first-order linear regressions were carried out with the different droplet size and velocity characteristics defined as the independent variable and DRPt defined as the dependent variable. In Table 3, the coefficients of determination (R2) are presented for the different droplet characteristics together with the intercepts and slopes in cases where the linear relation was significant at a level a ¼ 0.05 (F test). These results demonstrate that droplet size as well as droplet velocity characteristics are related with DRPt values. Only in some particular cases (Relative span factor RSF, vvol75, vvol90 and vavg) the linear relationship was not significant (a ¼ 0.05). DRPt values generally increase (b0 > 0) with increasing values of droplet diameter and droplet velocity characteristics and decrease (b0 < 0) with increasing percentages of fine droplets. From the different individual droplet characteristics, V75 was found to have the highest predictive power (R2 ¼ 0.94) with regard to DRPt which was also found by Hobson et al. (1990). In Fig. 2 measured DRPt values are compared with the predicted DRPt values using the simple first-order linear regression with V75 as the independent variable which is: DRPt ¼ 98:7  68:4:V75

(2)

where DRPt is the total drift reduction potential (%) and V75 is the proportion of total volume of droplets <75 mm in diameter (%). Using these predicted DRPt values in combination with the drift prediction equation (Eq. 1), the expected sedimenting drift values can be calculated, for average meteorological conditions, as shown in Fig. 1b. Besides V75, the droplet size characteristics V50, V100, V150, V200 and V250 also had statistically significant

linear relationships with DRPt at a level a ¼ 0.001. This means that the proportion of the total volume of fine droplets is the best indicator for the amount of sedimenting spray drift found in the field experiments since it can explain up to 94% of the total variation in DRPt values. Other studies also found that droplet size was one of the most influential factors related to spray drift (Bird et al., 1996; Carlsen et al., 2006) and different researchers have considered other droplet sizes than 75 mm diameter as indicators of drift. For example droplets <100 mm diameter (Bode, 1984), <150 mm (Combellack et al., 1996) and <200 mm (Bouse et al., 1990; Baetens et al., 2008) have been considered to be the most drift-prone. On the other hand, Butler Ellis and Bradley (2002) concluded that there was a poor correlation between spray volume contained in droplets <100 mm diameter and drift which is in contrast with the results from this study. From the other droplet size characteristics, a statistically significant linear relation at a level a ¼ 0.01 was obtained for Dv0.5, Dv0.75, Dv0.9. Using Eq. 2, the effect of the proportion of the total volume of droplets <75 mm diameter (V75) on DRPt can be calculated. For example, an increase of V75 from 0.5 to 2% resulted in a decrease of DRPt from about 64.5% to 38.1% corresponding to an increase in the total amount of near-field sedimenting spray drift with a factor of about 4. From Fig. 2, it is clear that the small deviation of the first-order regression line from the bisector is again mainly caused by the leverage effect of the F 110 02 nozzle with its relatively high predicted value of DRPt compared with the measured DRPt. In general, there was a good agreement between measured and predicted DRPt values was found although there are some small deviations which are also reflected in the corresponding predicted drift curves presented in Fig. 1a, b. For example, for the standard flat-fan nozzles, the predicted DRPt based on V75 is higher, corresponding with lower drift values (Fig 1b), for the F 110 06 compared to the F 110 04 while for field measurements the opposite results were found (Fig. 1a). The air-induction nozzle (Injet 03) had the highest DRPt and the lowest drift values based on droplet size characteristics as well as on the field drift experiments.

6

biosystems engineering 105 (2010) 2–12

a 30

Drift, %

25

20

15

10

5

0 0,5

1

2

3

5

10

15

20

10

15

20

10

15

20

Drift distance, m

b 30

Drift, %

25

20

15

10

5

0 0,5

1

2

3

5

Drift distance, m

c 30

Drift, %

25

20

15

10

5

0 0,5

1

2

3

5

Drift distance, m Fig. 1 – Predicted sedimenting drift curves for different nozzle types at average meteorological conditions (T [ 16 8C, V3.25m [ 3 m sL1 and XH2 O [8 g kgL1 ) at a driving speed of 8 km. hL1, a boom height of 0.50 m on a flat meadow based on (a) field drift measurements (Nuyttens et al., 2007a); (b) Eq. 2 and V75 from the PDPA laser measurements and (c) DPRPH from the , Hardi ISO F 110 03; , Hardi ISO F 110 04; , wind tunnel measurements: –––––––, Hardi ISO F 110 02; , Hardi ISO LD 110 02; , Hardi ISO LD 110 03; , Hardi ISO LD 110 04; , Hardi Hardi ISO F 110 06; , Hardi ISO Injet 03; , Hardi ISO Injet 04. ISO Injet 02;

Table 3 – Characteristics of first-order linear regressions of the form DRPt [ a0 D b0.X with the different droplet characteristics as the independent variable (X ) and DRPt, DPRPV1, DPRPV2, DPRPH as the dependent variables. X

2

R

a0

b0

R

0.54* 0.56* 0.59** 0.65** 0.71** 0.49* 0.51* 0.54* 0.58* 0.89*** 0.94*** 0.94*** 0.93*** 0.92*** 0.88*** 0.41* 0.35 0.54* 0.51* 0.49* 0.19 0.03 0.18 0.76**

99.1 109.9 124.4 15.2 210.6 101.1 101.0 107.4 120.0 89.4 98.7 99.6 101.5 105.3 110.7 97.0

0.62 0.48 0.40 0.39 0.40 0.95 0.73 0.63 0.47 247.3 68.1 29.7 10.2 5.02 3.19 1.33

0.58* 0.60** 0.62** 0.70** 0.78** 0.50* 0.53* 0.56* 0.62** 0.55* 0.66** 0.66** 0.67** 0.68** 0.69** 0.43* 0.25 0.92*** 0.86*** 0.72** 0.27 0.06 0.48* 0.89***

98.3 101.3 132.3

93.4 49.0 35.2

204.7

100.1

a0 79.5 88.3 100.9 130.1 181.7 77.9 79.4 85.5 97.9 75.2 85.0 85.7 87.6 91.2 97.1 73.7

DPRPV2, 2

b0

R

0.55 0.43 0.36 0.35 0.4 41.0 0.64 0.56 0.42 167.3 49.4 21.5 7.47 3.72 2.44 1.15

0.65** 0.67** 0.70** 0.78** 0.85*** 0.57* 0.60** 0.63** 0.69** 0.72** 0.83*** 0.83*** 0.84*** 0.85*** 0.85*** 0.49* 0.33 0.87*** 0.82*** 0.67** 0.23 0.04 0.36 0.94***

108,2 109.0 132.3

105.8 54.5 36.7

120.8 199.7

92.3 93,7

DPRPH 2

a0

b0

R

a0

b0

72.9 81.8 94.5 122.9 171.9 72.2 73.3 79.2 91.1 84.5 93.6 94.5 96.4 99.8 105.4 68.0

0.55 0.43 0.35 0.35 0.40 0.83 0.64 0.55 0.41 180.6 51.9 22.7 7.87 3.90 2.53 1.15

0.65** 0.67** 0.71** 0.78** 0.85*** 0.57* 0.61** 0.64** 0.69** 0.81*** 0.91*** 0.91*** 0.91*** 0.91*** 0.90*** 0.50* 0.36 0.81*** 0.77** 0.65** 0.23 0.03 0.31 0.92***

71.8 81.1 94.0 122.5 171.0 71.6 72.4 78.4 90.3 89.2 98.0 98.9 100.9 104.4 109.8 67.3

0.55 0.43 0.36 0.35 0.40 0.84 0.64 0.56 0.42 192.1 54.4 23.8 8.25 4.07 2.63 1.16

84.0 86.5 107.9

93.20 48.71 32.9

98.9 91.6 111.5

96.4 50.0 33.3

86.4 199.1

75.0 90.1

200.5

biosystems engineering 105 (2010) 2–12

Dv0.1, mm Dv0.25, mm Dv0.5, mm Dv0.75, mm Dv0.9, mm D10, mm D20, mm D30, mm D32, mm V50, % V75, % V100, % V150, % V200, % V250, % NMD, mm RSF vvol10 (m s1) vvol25 (m s1) vvol50 (m s1) vvol75 (m s1) vvol90 (m s1) vavg (m s1) VSF**

DPRPV1

DRPt 2

89.9

*/**/***, statistically significant linear relation at a level a of 0.05 (*), 0.01 (**) and 0.001 (***); R2, coefficient of determination; a0, b0, intercept and slope of the first-order linear regression; X, independent variable; Dv0.1, Dv0.25, Dv0.5, Dv0.75, Dv0.9, diameter below which smaller droplets constitute 10, 25, 50, 75 and 90% of the total volume; D10, D20, D30, D32, arithmetic, surface, volume and Sauter mean diameter; V50, V75, V100, V150, V200, V250, proportion of total volume of droplets smaller than 50, 75, 100, 150, 200 and 250 mm in diameter; NMD, number median diameter; RSF, relative span factor; vvol10, vvol25, vvol50, vvol75, vvol90, droplet velocity below which slower droplets constitute 10, 25, 50, 75 and 90% of the total spray volume; vavg, arithmetic average droplet velocity.

7

8

biosystems engineering 105 (2010) 2–12

-120

-100

-80

-60

-40

-20

0

20

F 110 04

0 -140 -20

40

Injet 110 03

LD110 04

LD110 03

LD 110 02

20

F 110 03

40

Predicted DRPt based on V75 , %

F 110 06

60

Injet 110 02

80

Injet 110 04

100

60

80

100

-40 -60

-100

F 110 02

-80

-120 X =Y

-140

DRPt, %

Fig. 2 – Comparison between predicted DRPt values based on the first-order linear regression DRPt [ 98.7 – 68.4.V75 and , predicted DRPt [0.93. DRPt D 1.75 measured DRPt values together with the corresponding first order linear regression: (R2[0.94).

Besides the different droplet size characteristics, the velocity span factor (VSF) was also statistically significant related to DRPt at a level a ¼ 0.01. The higher the value of VSF the less uniform the droplet velocity distribution and the lower the DRPt value. This can be explained by the fact that there is a clear relation between droplet size and velocity for hydraulic spray nozzles (Sidahmed, 1996; Sidahmed et al., 1999) which is reflected in the VSF values as described by Nuyttens et al. (2007a). From the other droplet velocity characteristics, a statistically significant linear relation at a level a of 0.05 was found for vvol10, vvol25, vvol50. The potential of a multiple linear regression to provide an improved prediction of DRPt based on the different droplet characteristics was investigated, but the simple first-order linear model using V75 was best suited to predict DRPt values for the different conventional spray nozzles tested in this research. This was caused by the strong interrelationship between droplet size and velocity characteristics for the tested nozzlepressure combinations mainly because larger droplets retain their momentum for longer. It can be assumed that in cases where droplet velocities and sizes are less well correlated, such as with air assistance (Nuyttens et al., 2007c), another approach, combining droplet velocity and droplet size characteristics, will be necessary to obtain a good prediction for DRPt.

3.3. Comparison between wind tunnel DPRP values and field DRPt values In Fig. 3, the DRPt values from the field drift experiments are compared with the corresponding DPRPV1, DPRPV2 and DPRPH values resulting from the wind tunnel measurements for the different tested spray application techniques. The simple linear regressions and their corresponding R2 values are presented.

It is clear that there is a fairly good linear relation between DRPt and DPRP values for the different spray nozzles with similar first-order relationships for the three wind tunnel approaches. Based on the R2 values of 0.66, 0.81 and 0.88, respectively for, DPRPV1, DPRPV2 and DPRPH, it can be concluded that the approach of numerically integrating the measured fallout deposit curve, is best suited to represent real near-field sedimenting drift characteristics. Predicted sedimenting drift values based on DPRPH values and Eq. 1 are presented in Fig. 1c for average meteorological conditions. From this point, mainly DPRPH values are taken into consideration. Despite the fairly good correlation between field drift DRPt and wind tunnel DPRP values, there are some important discrepancies which are important to keep in mind when interpreting wind tunnel results and are reflected in the corresponding drift curves (Fig. 1a, c). The deviation of the firstorder regression lines from the bisector was mainly caused by the leverage effect and the results of the F 110 02 nozzle with its relatively high DPRP values (ranging from 73.0 to 57.9%) compared with the corresponding DRPt value of 136.5%, although the differences are statistically non-significant (a ¼ 0.05) because of the high standard deviations. Among the other standard flat-fan nozzles, a considerable and statistically significant difference was also seen between DRPt and DPRP values for the F 110 06 nozzle with a DRPt of 29.5% and a DPRPH of 73.3%. Although for the ISO 03 and ISO 04 standard flat-fan nozzles, there was good agreement between DRPt and DPRP values, it can be seen that DPRP values were generally higher than DRPt values for the standard flat-fan nozzles. This means that based on the wind tunnel measurements, the driftability of this nozzle type is underestimated when compared with the results from the field measurements. Knowing that for the standard flat-fan nozzles, DPRPV1 values were the highest

9

60

LD110 04

F 110 04

80

-120

-100

-80

-60

-40

-20

F 110 03

LD110 02

DPRP , %

0 -140

LD110 03

40 20

0

20

Injet 110 04

F 110 06

Injet 110 02

100

Injet 110 03

biosystems engineering 105 (2010) 2–12

40

60

80

100

-20

-60

F 110 02

-40

-80 -100 -120

X =Y -140

DRPt , %

Fig. 3 – Comparison between DPRPV1 , DPRPV2 A, DPRPH and DRPt values for different Hardi ISO nozzle types (F, standard flat-fan; LD, pre-orifice; Injet, air-induction) and sizes (ISO 02, 03, 04 and 06) together with the corresponding first order linear , DPRPV2 [ 0.73. DRPt D 19.20 (R2[0.81); , DPRPH[0.76. regressions: –––––, DPRPV1[0.70. DRPt D 13.98 (R2 [ 0.66); DRPt D 20.01 (R2[0.88).

followed by DPRPV2 and DPRPH (Nuyttens et al., 2009b), it is clear that DPRPH corresponds best with DRPt results. For the different sizes of pre-orifice nozzles, good agreement between wind tunnel and field drift results was found with DRPt values of 3.6, 38.4 and 54.9% and DPRPH values of 0.5, 32.3 and 41.1%, respectively for, the LD 110 02, LD 110 03 and the LD 110 04 nozzles. In contrast, with the standard flatfan nozzles, DPRP values are generally lower than DRPt values. Because for this nozzle type, DPRPV1 values were the lowest followed by DPRPV2 and DPRPH (Nuyttens et al., 2009b), DPRPH again relates best with DRPt. For the Injet 02 and Injet 04 air-induction nozzles, the DPRP values are limited but significantly (a ¼ 0.05) higher than DRPt values. The statistical significance is mainly caused by the high repeatability of the wind tunnel and field measurements for this type of nozzle. On the other hand, DPRP and DRPt values were almost equal for the Injet 03 nozzles. The Injet 03 was also found to have the lowest driftability followed by the Injet 04 and the Injet 02 nozzles based on the wind tunnel as well as on the field measurements. It should be noted that in contrast with the flat-fan and the pre-orifice nozzles, DPRPV1 and DPRPV2 values correspond better with DRPt values than DPRPH values but the differences are very small. In general, investigating the effect of nozzle type and size, similar trends can be found from the DPRP and DRPt results although there are some important deviations in absolute terms. For the same nozzle size, air-induction nozzles have the highest DRPt and DPRP values followed by the pre-orifice nozzles and the standard flat-fan nozzles. Only for the LD 110 04 nozzles, DPRP values were lower than for the F 110 04 nozzles which was not the case for the DRPt values. For the standard and the pre-orifice flat-fan nozzles, larger ISO nozzle

sizes correspond with higher DRPt and DPRP values. Again, one exception was found namely, for the DRPt of the F 110 06 nozzle which was lower than the DRPt value of the F 110 04 nozzle type which was not the case for the DPRP values.

3.4. Relationship between droplet characteristics and wind tunnel DPRP values First-order linear regressions were performed on the results with the different droplet size and velocity characteristics as the independent variable and DPRPV1, DPRPV2 and DPRPH as the dependent variables to investigate the importance of the different individual droplet characteristics on DPRP values. The results of the regressions are presented in Table 3. Droplet size characteristics are generally best related with DPRPH followed by DPRPV2 and then DPRPV1. For example, for V200, R2 values of 0.91, 0.85 and 0.68 were found, respectively for, DPRPH, DPRPV2 and DPRPV1. Again, the DPRP values increased with increasing droplet diameter characteristics (b0 > 0) and with decreasing percentages of small droplets (b0 < 0). On the other hand, the individual droplet velocity characteristics were generally best related with DPRPV1 followed by DPRPV2 and DPRPH. For example, for vvol10, R2 values decrease from 0.92 to 0.87 and 0.81, respectively for, DPRPV1, DPRPV2 and DPRPH. DPRP values increased with increasing values of droplet velocity characteristics (b0 > 0). This indicates that droplet size characteristics are more related with fallout deposits compared with airborne deposits while the opposite is found for the droplet velocity characteristics. Looking in detail DPRPH, which corresponds best with field DRPt values, V75, V100, V150 and V200 have the highest predictive power with an R2 value of 0.91. The droplet size characteristics

10

biosystems engineering 105 (2010) 2–12

Dv0.9, V50 and V250 and the droplet velocity characteristics vvol10, vvol25 and VSF also have a statistically significant linear relation with DPRPH at a level a ¼ 0.001. V75 was also found to be the best indicator for DRPt representing the amount of spray drift sedimenting in the field with a comparable R2 value of 0.94. The following first-order linear regression between DPRPH and V75 was found. DPRPH ¼ 98:0  45:4:V75

droplet characteristics was investigated. For DPRPV1 and DPRPV2, it could be concluded that similarly to DRPt, the simple first-order linear equations described by Eqs. 4 and 5 are best suited to predict DPRPV1 and DPRPV2. For DPRPH, a forward stepwise regression procedure (probability of F  0.05) also retained vvol10 as a second independent variable besides V75. This resulted in the following second-order linear equation with an R2 of 0.97.

(3)

DPRPH ¼ 22:6  36:5:V75 þ 42:4:vvol10

The similarities between Eqs. 2 and 3 confirm the similarities between the DPRPH values resulting from wind tunnel measurements and the DRPt values from the field measurements. While DRPt and DPRPH were the most closely related with the droplet size characteristic V75, DPRPV1 and DPRPV2 were most closely related with one of the droplet velocity characteristics, namely vvol10 for DPRPV1 (R2 ¼ 0.92) and VSF for DPRPV2 (R2 ¼ 0.94) with the following regressions. DPRPV1 ¼ 108:2 þ 105:8:vvol10

(4)

DPRPV2 ¼ 199:1  90:1:VSF

(5)

(6)

Eq. 6 demonstrates that the amount of fallout deposits in the wind tunnel increases with an increase in the proportion of the total volume of droplets <75 mm diameter as well as with a decrease in the droplet velocity as expressed by vvol10.

3.5. Comparison between indirect and direct spray drift assessment means Comparing results from the PDPA laser and the wind tunnel measurements, it can be concluded that the indirect drift risk assessment method measuring V75 values (R2 ¼ 0.94) is best suited to represent near-field drift characteristics followed by the wind tunnel approach calculating DPRPH values (R2 ¼ 0.88). Both wind tunnel approaches measuring airborne spray, DPRPV1 and DPRPV2, gave inferior results with R2 values of 0.81 and 0.66, respectively. On the other hand, with the PDPA laser measurements, it is only possible to investigate the effect of nozzle type, size and spray pressure since the effect of nozzle height could only be investigated by means of wind tunnel measurements. With both indirect techniques, it is difficult to investigate other effects such as driving speed and air-assistance. Based on DPRPH or V75, both resulting from these indirect drift measurements, the DRPt of a particular technique can be

This means that droplet velocity characteristics have a slightly better predictive power than the droplet size characteristics with regard to DPRPV1 and DPRPV2 bearing in mind that droplet size and droplet velocity characteristics are also mutually correlated (Nuyttens et al., 2007a, 2009a). In Fig. 4, the measured DPRP values are compared with the predicted values based on Eqs. 3–5. There is a better correlation between the first-order regression lines and the ‘ideal’ bisector compared with Fig. 3. It should be noted that in some individual cases, predicted DPRP values are above 100% and hence, physically meaningless. The potential of a multiple linear regression to derive an improved prediction of DPRP values based on the different

Injet 110 04

120

Injet 110 02

100 F 110 06

80

Injet 110 03

Predicted DPRP, %

60 LD 110 03

40

LD 110 04

20 LD 110 02

0 -100

-80

-60

-40

F 110 04

F 110 03

-20

0

20

40

60

80

100

-20 -40

F 110 02

-60 -80

X =Y -100

DPRP, %

Fig. 4 – Comparison between measured and predicted DPRPV1 , DPRPV2 A, and DPRPH values based on equations 3, 4 and , predicted DPRPV1 [ 1.00. DPRPV1 D 12.3 (R2 [ 0.92); 5 together with the corresponding first order linear regressions: 2 , predicted DPRPV2 [ 0.94. DPRPV2 D 2.26 (R [ 0.94); , predicted DPRPH [ 0.90. DPRPH D 3.95 (R2 [ 0.91).

biosystems engineering 105 (2010) 2–12

determined to establish a realistic estimate of field drift data at a driving speed of 8 km h1, a boom height of 0.50 m on a flat meadow as described by Nuyttens et al. (2007b) and presented in Fig. 1a–c.

4.

Conclusions

The results of three contrasting drift risk assessment methods, utilising PDPA laser measurements, wind tunnel measurements and field drift experiments, have been compared for 10 different spray nozzles. Droplet size as well as droplet velocity characteristics are related with field DRPt values and wind tunnel DPRP values. DPRP and DRPt values increase with increasing values of droplet diameter and droplet velocity characteristics and decrease with increasing percentages of small droplets. The proportion of the total volume of droplets <75 mm diameter, was the best indicator for establishing the amount of sedimenting spray drift found in the field. This variable was able to explain about 94% of the total variation in DRPt values. From the three wind tunnel approaches, the approach calculating the surface under the measured fallout deposit curve was the best suited to represent near-field sedimenting drift characteristics and similar trends were found concerning the effect of nozzle type and size from the DPRP and DRPt results. Both wind tunnel approaches measuring airborne deposits gave inferior results. Droplet size characteristics are more related with fallout deposits compared to airborne deposits, while the opposite occurs for droplet velocity characteristics. Comparing both indirect drift assessment means, it can be concluded that measuring V75 values is at least as well suited to represent near-field drift characteristics as the wind tunnel approach measuring fallout deposits and better suited than both wind tunnel approaches measuring airborne deposits. Simple first-order linear regressions with one of the droplet characteristics as a predictor variable, were the best choice to predict DRPt, DPRPV1 and DPRPV2. Only in case of DPRPH, both droplet size (V75) as well as droplet velocity (vvol10) were included in the multiple linear regression model. With both indirect risk assessment means, driftability experiments can be made with different spraying systems under directly comparable and repeatable conditions and both methods are suited to permit relative studies of drift risk. Moreover, based on DPRPH or V75 resulting from these indirect drift measurements the DRPt of a particular technique can be determined to establish a realistic estimate of field drift data. Based on this information, indirect drift assessment methods are a valuable alternative to time consuming and expensive field drift experiments but it is difficult to investigate effects such as driving speed, boom height and air assistance.

Acknowledgements This research was funded by the Flemish Government IWTVlaanderen. The authors wish to thank the technical staff of

11

ILVO-T&V-AT and the University of Ghent, Belgium for their assistance.

references

Baetens K; Nuyttens D; Verboven P; De Schampheleire M; Nicolaı¨ B; Ramon H (2007). Predicting drift from field spraying by means of a 3D computational fluid dynamics model. Computers and Electronics in Agriculture, 56(2), 161–173. doi:10.1016/j.compag.2007.01.009. Baetens K; Ho Q T; Nuyttens D; De Schampheleire M; Endalew A; Hertog M; Nicolaı¨ B; Ramon H; Verboven P (2008). Development of a 2-D-diffusion advection model for fast prediction of field drift. Atmospheric Environment, 43, 1674–1682. doi:10.1016/j.atmosenv.2008.12.047. Bayat A; Bozdogan N Y (2005). An air-assisted spinning disc nozzle and its performance on spray deposition and reduction of drift potential. Crop Protection, 24(11), 951–960. doi:10.1016/ j.cropro.2005.01.015. Bird S L; Esterly D M; Perry S G (1996). Off-target deposition of pesticides from agricultural aerial spray application. Journal of Environmental Quality, 25(5), 1095–1104. Bode L E (1984). Downwind drift deposits by ground applications. Proceedings Pesticide Drift Management Symposium, p 50. Bouse L F; Kirk I W; Bode L E (1990). Effect of spray mixture on droplet size. Transactions of the ASAE, 33(3), 783–788. Butler Ellis M C; Bradley A (2002). The influence of formulation on spray drift. Aspects of Applied Biology, 66, 251–258. Carlsen S C K; Spliid N H; Svensmark B (2006). Drift of 10 herbicides after tractor spray application. 2. Primary drift (droplet drift). Chemosphere, 64, 778–786. doi:10.1016/j. chemosphere.2005.10.060. Combellack J H; Western N M; Richardson R G (1996). A comparison of the drift potential of a novel twin fluid nozzle with conventional low volume flat fan nozzles when using a range of adjuvants. Crop Protection, 15(2), 147–152. De Schampheleire M; Nuyttens D; Steurbaut W; Baetens K; Verboven P; Nicolaı¨ B; Ramon H; Sonck B (2006). The assessment of particle spray drift damage for ten major crops in Belgium. Aspects of Applied Biology, 77(1), 275–280. De Schampheleire M; Baetens K; Nuyttens D; Spanoghe P (2008). Spray drift measurements to evaluate the Belgian drift mitigation measures in field crops. Crop Protection, 27, 577–589. doi:10.1016/j.cropro.2007.08.017. Fietsam J F W; Young B G; Steffen R W (2004). Differential response of herbicide drift reduction nozzles to drift control agents with glyphosate. Transactions of the ASAE, 47(5), 1405–1411. Guler H; Zhu H; Ozkan H E; Derksen R C; Yu Y; Krause C R (2007). Spray characteristics and drift reduction potential with air induction and conventional flat fan nozzles. Transactions of the ASABE, 50(3), 745–754. Hobson P A; Miller P C H; Walklate P J; Tuck C R; Western N M (1990). Spray drift from hydraulic spray nozzles: The use of a computer simulation model to examine factors influencing drift. Conference Proceedings AgEng 1990, Berlin, 1–11. ISO 22866 (2005). International Standard: Equipment for crop protection – Methods for field measurement of spray drift. Nuyttens D; De Schampheleire M; Baetens K; Sonck B (2006). PDPA Laser-based characterisation of agricultural spray nozzles. Manuscript PM 06 024. Agricultural Engineering International: the CIGR Ejournal, VIII. December, 2006. Nuyttens D (2007). Drift from field crop sprayers: The influence of spray application technology determined using indirect and direct drift assessment means. PhD thesis nr. 772. Katholieke Universiteit Leuven, ISBN 978-90-8826-039-1, available at: http://hdl.handle.net/1979/1047, 293 pp.

12

biosystems engineering 105 (2010) 2–12

Nuyttens D; Baetens K; De Schampheleire M; Sonck B (2007a). Effect of nozzle type, size and pressure on spray droplet characteristics. Biosystems Engineering, 97(3), 333–345. doi:10. 1016/j.biosystemseng.2007.03.001. Nuyttens D; De Schampheleire M; Baetens K; Sonck B (2007b). The influence of operator controlled variables on spray drift from field crop sprayers. Transactions of the ASABE, 50(4), 1129–1140. Nuyttens D; Dekeyser D; De Schampheleire M; Baetens K; Sonck B (2007c). The effect of air support on droplet characteristics and spray drift. Communications in Agricultural and Applied Biological Sciences, 72(2), 71–80. Nuyttens D; De Schampheleire M; Verboven P; Brusselman E; Dekeyser D (2009a). Droplet size-velocity characteristics of agricultural sprays. Transactions of the ASABE, 52(5), 1471–1480. Nuyttens D; Taylor W A; De Schampheleire M; Verboven P; Dekeyser D (2009b). Influence of nozzle type and size on drift potential by means of different wind tunnel evaluation methods. Biosystems Engineering, 103(3), 271–280. doi:10. 1016/j.biosystemseng.2009.04.001. Phillips J; Miller P C H (1999). Field and wind tunnel measurements of the airborne spray volume downwind of

single flat-fan nozzle. Journal of Agricultural Engineering Research, 72, 161–170. Qi L J; Miller P C H; Fu Z T (2008). The classification of the drift risk of sprays produced by spinning discs based on wind tunnel measurements. Biosystems Engineering, 100(1), 38–43. doi:10. 1016/j.biosystemseng.2008.01.007. Ravier I; Haouisee E; Cle´ment M; Seux R; Briand O (2005). Field experiments for the evaluation of pesticide spray-drift on arable crops. Pest Management Science, 61(8), 728–736. Rimmer D A; Johnson P D; Kelsey A; Warren N D (2009). Field experiments to assess approaches for spray drift incident investigation. Pest Management Science, 65(6), 665–671. Satow T; Miyamoto K; Matsuda K; Rambanda M (1993). Influence of droplet size of spray on drift characteristics. Research Bulletin of Obihiro University, 18(2), 97–104. Sidahmed M M (1996). A theory for predicting the size and velocity of droplets from pressure nozzles. Transactions of the ASAE, 40(3), 547–554. Sidahmed M M; Brown R B; Darvishand M (1999). Drop-size/ velocity correlations at formation of sprays from fan nozzles. Transactions of the ASAE, 42(6), 1557–1564.