The results of the physicochemical analysis for water with three urea concentrations are presented in Table 3. Emitter clogging has a direct impact on the performance and lifespan of a drip irrigation system (Zhang et al. 2017). Thus, the presence of ions, iron and manganese in high concentrations is the main determinant, since it reduces the contact area of the water, increasing the head loss and causing a loss of pressure in the system; consequently, there is a reduction in the flow in the emitters.
Observing Table 3, it can be seen that only manganese presented a moderate risk of clogging for both concentrations analyzed. According to Nakayama and Bucks (1986), Capra and Scicolone (1998), all other physical-chemical parameters analyzed present a low risk of clogging.
The average values obtained for the pH of clean water and water with fertilizer presented results above the predicted limits. According to (Nakayama et al. 2007), pH values above 7.2 favor the precipitation of elements such as calcium and magnesium in filters, pipes and emitters, contributing to the clogging of the dripper. Therefore, it appears that the increase in the concentration of urea in 6 g L-¹ influenced the increase in pH when presenting 7.75. The electrical conductivity measured in the tests remains within the values indicated by Nakayama and Bucks (1986), who do not advise using it for irrigation when the EC is above 3.0 dS.m¹.
Table 3 - Physicochemical parameters for clean water and water with urea in different concentrations.
Parameters
|
Clean water
|
Urea
2 gL-1
|
Urea
4 gL-1
|
Urea
6 gL-1
|
Total Iron (mg L-1)
|
1,02*
|
0*
|
0*
|
0*
|
Manganese (mg L-1)
|
-0,06*
|
0,02**
|
0,02**
|
0,04**
|
Suspended solids (mg L-1)
Electric conductivity (dS m-1)
pH
Calcium (mg L-1)
Magnesium (mg L-1)
Total Nitrogen (mg L-1)
|
40*
0,058
7,41*
2,95*
0,77*
0,06
|
236
0,070
7,51*
0
0,78*
8,77
|
60
0,072
7,43*
0
0,79*
18,19
|
292
0,076
7,75*
0
0,80*
25,52
|
Total Iron (mg L ¹): Low (< 0.2 mg L ¹); Medium (0.2-1.5 mg L ¹); High (> 1.5 mg L¹). Source: NAKAYAMA & BUCKS (1986). *Low risk of clogging, **Moderate risk of clogging and ***Severe risk of clogging.
The evaluation of the normality of the data distribution is essential for the adequate description of the sample and its subsequent analysis. Therefore, Anderson Darling normality tests were performed. The tests were conduten out for the parameters UC (%) and UD (%), for all concentrations applied on the slope of the lateral line.
For the data set evaluated in Table 4, in relation to the parameters of CUD and CUC, in different slopes, it appears that at least one has a normal distribution, however, in the same data set, groups with p- value below 0.05. Therefore, Kruskal-Wallis was applied for non-parametric data analysis.
Table 3 shows the p-value <0.05 by the Kruskal-Wallis Test at all slope levels for both uniformity coefficients, that is, there was a statistical difference between concentrations and slope levels. Thus, it is demonstrated that alterations in the urea concentration modify the uniformity of distribution of the nitrogen solution in the fertigation system. Thus, multiple comparison tests were applied using the Mann-Whitney test to compare non-parametric data at a 5% significance level, which are shown in Table 4.
Table 4 - Descriptive analysis of the UD and UC values for the different urea concentrations on the slope of the lateral line.
Variable
|
Data normality (p-Value)
|
Mín
|
Q1
|
Median
|
Q1
|
Max
|
Kruskal Wallis (p-Value)
|
UD
|
Level system
|
0
|
0,009
|
93,39
|
95,12
|
96,31
|
96,78
|
97,47
|
<0,005
|
2
|
0,008
|
78,95
|
84,73
|
86,56
|
87,59
|
96,85
|
4
|
0,756
|
79,89
|
83,21
|
84,49
|
85,80
|
87,40
|
6
|
0,089
|
63,11
|
76,98
|
80,13
|
85,28
|
90,83
|
UC
|
Level system
|
0
|
0,007
|
93,39
|
95,11
|
96,27
|
96,80
|
97,47
|
<0,005
|
2
|
0,007
|
88,41
|
91,97
|
93,15
|
93,51
|
95,11
|
4
|
0,560
|
89,76
|
91,57
|
92,20
|
92,95
|
93,63
|
6
|
<0,005
|
64,65
|
88,29
|
89,75
|
92,16
|
95,07
|
UD
|
Uphill system
|
0
|
<0,005
|
95,51
|
96,66
|
97,26
|
97,44
|
97,70
|
<0,005
|
2
|
<0,005
|
71,88
|
86,57
|
88,38
|
90,38
|
93,04
|
4
|
0,171
|
86,70
|
89,16
|
89,91
|
92,43
|
93,73
|
6
|
0,358
|
91,80
|
93,51
|
94,18
|
95,07
|
96,54
|
UC
|
Uphill system
|
0
|
<0,005
|
92,97
|
93,99
|
95,50
|
95,93
|
96,32
|
<0,005
|
2
|
<0,005
|
85,94
|
92,86
|
93,94
|
94,72
|
95,66
|
4
|
0,075
|
86,70
|
89,16
|
89,91
|
92,43
|
93,73
|
6
|
0,879
|
91,80
|
93,51
|
94,18
|
95,07
|
96,54
|
UD
|
Downhill
system
|
0
|
<0,005
|
83,59
|
92,46
|
93,47
|
93,78
|
96,13
|
<0,005
|
2
|
0,020
|
83,42
|
89,09
|
91,12
|
92,04
|
93,11
|
4
|
0,171
|
74,33
|
80,16
|
81,58
|
86,12
|
87,93
|
6
|
0,011
|
86,94
|
90,17
|
90,70
|
91,63
|
92,74
|
UC
|
Downhill
system
|
0
|
<0,005
|
91,70
|
95,37
|
96,04
|
96,26
|
97,17
|
<0,005
|
2
|
0,074
|
91,42
|
94,09
|
95,22
|
95,69
|
96,44
|
4
|
0,075
|
86,70
|
89,15
|
89,91
|
92,43
|
93,73
|
6
|
<0,005
|
93,16
|
94,97
|
95,26
|
95,49
|
96,19
|
By the multiple comparison test, the control treatment (without fertilizer) statistically differed from the other treatments for the level system. Thus, when analyzing Table 4, it can be seen that, in none of its comparisons, the p-value was greater than 0.05 for UD. In Figure 2, it is possible to see the results more clearly through the box plot, as the uniformity for UD is superior, applying only water, being the only treatment with a median above 90%, so that it maintains excellence, according to (Frizzone 2012).
Similar results were presented by (Lopes et al. 2019), when analyzing the uniformity of water application in a drip irrigation system. Therefore, under different slopes, they obtained results of 99.03% for UC and 98.45% for UD.
Considering the median values, presented in Table 3, for UD, a decreasing behavior in uniformity is observed, according to the change in urea concentration and slope of the lateral line, which can reach 16.18% for the level system (96.31% to 80.13%), being almost sixteen times lower for the uphill system, with 2.77% (93.47% to 90.70%). This is because it showed greater similarity to the uphill system with 3.08% (97.26% to 94.18%), being more sensitive in the analysis, compared to the UC, which showed less variation. Therefore, the increase in the concentration of fertilizer affected the uniformity, showing greater variation in uniformity for the system at level.
In Table 5, for the multiple comparison of data, for CUD, the variation in urea concentration is highlighted, which was statistically different for the level system, not showing the same difference when analyzing the downhill system. When comparing 2 g L-¹ and 6 g L-¹, the uniformity did not differ when presenting p-value >0.05, confirming the results presented in Table 4.
(Szekut et al. 2018) and (Lopes et al. 2019), studying the uniformity of irrigation application distribution, observed greater excellence in operation for lines arranged on a downhill, compared to level and uphill.
Table 5 - Multiple data comparison for UC and UD
Comparation
|
UC LEVEL
|
UD LEVEL
|
UC UPHILL
|
UD UPHILL
|
UC DOWNHILL
|
UD DOWNHILL
|
P – Value
|
0
|
2
|
<0,005
|
<0,005
|
<0,005
|
<0,005
|
0,001
|
<0,005
|
0
|
4
|
<0,005
|
<0,005
|
<0,005
|
<0,005
|
<0,005
|
<0,005
|
0
|
6
|
<0,005
|
<0,005
|
<0,005
|
<0,005
|
0,001
|
<0,005
|
2
|
4
|
<0,005
|
<0,005
|
0,050
|
0,005
|
<0,005
|
<0,005
|
2
|
6
|
0,332
|
<0,005
|
<0,005
|
<0,005
|
0,938
|
0,613
|
4
|
6
|
<0,005
|
<0,005
|
0,001
|
0,018
|
<0,005
|
<0,005
|
In general, the best performance was observed for the downhill system at concentrations of 2 gL-1 and 6 gL-1, with uniformity values >90%; however, when applying concentrations of 4 gL-1, the system showed a change in uniformity with values between 80 and 85%. (Lopes et al. 2021a) when applying MPK fertilizer obtained results >90% for all slopes.
As shown in Figure 2, the greatest decrease in uniformity was observed when applying 4 gL-1 uphill when the system presented uniformity values <80% for the 1 quartile, which can be observed in Table 2. While the control treatment (water) achieved performance >90% at all slope levels for CUD, demonstrating that the increase in nitrogen fertilizer concentration changes the performance conditions of the fertigation system.
To verify if the process is under statistical control, Shewhart charts were made for individual measurements in relation to CUC (%) and CUD (%), in order to identify the variability of distribution uniformity during the 25 tests for level, uphill and downhill at different fertilizer concentrations. According to (Montgomery 2009), to apply the Shewhart Graphs, as well as the Process Capability Index, the data must meet the specifications of normality and not show autocorrelation between them. As shown in Table 6, only treatments with p-value > 0.05 were considered in the analysis, according to the Anderson Darling test.
Considering the autocorrelation between them, only the 2 g L-¹ concentration for sloping CUD was not analyzed. With this, the Shewhart Control Charts and Process Capacity Index were determined for the treatments, following (Montgomery 2009).
Table 6 - Normality and Autocorrelation of data
Variable
|
Data noarmality (p-Value)
|
Autocorrelation
|
UD
|
LEVEL SYSTEM
|
4
|
0,756
|
No
|
6
|
0,089
|
No
|
UC
|
LEVEL SYSTEM
|
4
|
0,560
|
No
|
UD
|
UPHILL SYSTEM
|
4
|
0,171
|
No
|
6
|
0,358
|
No
|
UC
|
UPHILL SYSTEM
|
4
|
0,075
|
No
|
6
|
0,879
|
No
|
UD
|
DOWNHILL SYSTEM
|
4
|
0,075
|
Não
|
UC
|
DOWNHILL SYSTEM
|
2
|
0,074
|
Yes
|
4
|
0,075
|
No
|
Thus, Figure 3 presented the individual application of the Shewhart graphs, for the analysis of the CUC uniformity indices (%), only for the downhill system with 2 g L-¹.
The graph allows the presentation of sample points within the control limits, which is equivalent to not rejecting the hypothesis that the process is under statistical control. In this context, points outside the limits are equivalent to rejecting this hypothesis, according to (Montgomery 2016).
The downhill system presents a sequence of events above the average with a significant drop in flow (13 to 18), exceeding the lower control limit, so that isolated points can be false alarms. (Andrade 2017), when evaluating the influence of the topographic slope on the fertigation process, obtained similar results when observing isolated points outside the control limits.
Analyzing the Shewhart Graph for CUC (%), at level for 4 g L-¹ concentration (Figure 4), there is a point outside the control limits, non-randomness around the mean, as well as increasing trend lines between the evaluations (11 to 15) and decreasing from (5 to 9), characteristics that configure a system outside the control limits.
However, for (Zhang et al. 2017), the clogging of each emitter may have randomness, in addition to the variation in flow rate of a single emitter, which may not fully represent the complete picture of the degree of emitter clogging of the irrigation system, which is of interest. Also, for the level system, although the points do not comply with the quality control characteristics, all points showed high uniformity rates >90%.
As for the uphill system with the same concentration, the points remained between the control limits, in addition to remaining in randomness around the mean, over the first 8 points evaluated. However, there is a decrease in uniformity between points (9 to 14) and (16 to 21) and trend line (21 to 25). Compared to the level system, its performance was not satisfactory, since, between the tests (12 to 19), there is the biggest drop in uniformity.
(Lopes et al. 2021b) when applying liquid peat observed that level fertigation had a better performance. Discussing drip irrigation using the Shewhart Graph (Hermes et al. 2013; 2014; 2015) also observed events below the lower control limit.
For level and uphill systems, consecutive points below or above the midline are observed; with this, a data trend is verified. For (Montgomery 2009), when seven values present this behavior, by assumption, there is a probable lack of statistical control. Only the downhill system remained within the control limits, however, a decreasing sequence of (8 to 12) was identified. (Szekut et al. 2018) report that lines positioned on a slope indicate better working conditions compared to a level and slope system.
Regarding uniformity values, all slope levels presented values <90%, when compared to the results found for CUC (%), shown in Figure 5. (Ribeiro et al. 2012) also obtained CUD values lower than the CUC, justifying that this fact is inherent to the variables of the equations for the calculation of the CUD, in which 25% of the area that received the smallest blades is considered, but it is not a rule for a set of data Zhang and Merkley (2012). The worst performance was verified for the uphill system with values lower than 75%, for CUD, classified as regular, according to (Frizzone et al. 2012).
Analyzing the data distribution in relation to the 70% value line (Figure 5), according to (Frizzone et al. 2012), minimum irrigation uniformity value is classified as poor. In this way, points 9, 17 and 23 present values below this line. Thus, the spacings are less adequate in relation to the uniformity of distribution for the CUD parameter (%), according to the control chart. Increasing the urea concentration by 6 g L-¹ affected the uniformity of the system in level, with values for CUD<70%, observed in the Shewhart Control Chart (Figure 6).
As for the uphill system (Figure 7), there was not the same behavior, since the uniformity values met the minimum (>90%). According to the classification by (Frizzone et al. 2012), for CUC (%), the points remained random in the mean, as well as in the graph for CUD (%), although the trials (17 to 25) have values below the mean line; this is not considered a trend, according to Montgomery (2009). Therefore, a similar result was obtained by (Andrade 2017), who observed points above the mean line when studying the use of Shewhart control charts in the spacing of microsprinklers.
Table 7 shows the results obtained for the flow and pressure uniformity parameters. The highest uniformities were attributed to the level system (99.18%) followed by the uphill system (99.04%) for the control treatment (0gL-1). Although the uniformities are excellent on all slopes, the small pressure variations that occur in the irrigation system are a consequence of the head loss and level variation (LIMA et al., 2003).
The increase in urea concentration decreased the CUp for all treatments compared to the control treatment, with the exception of the 6 gL-1 sloping concentration that presented (99.04%). The greatest pressure difference was observed in the system at the level with 2gL-1 (95.86%) and downhill (96.97%), similar results were observed by (Lopes et al. 2021b).
Table 7 - Flow and pressure uniformity
Treatment
|
Urea concentration
|
CUp
|
Level
|
|
99,18
|
Uphill
|
0gL-1
|
99,04
|
Downhill
|
|
98,64
|
Level
|
2gL-1
|
95,86
|
Uphil
|
98,66
|
Downhill
|
96,97
|
Level
Uphill
Downhill
|
4gL-1
|
98,54
|
98,93
|
96,97
|
Level
Uphill
Downhill
|
6gL-1
|
98,92
|
98,32
99,04
|
Regarding the process capability index (Table 8), only the lower control limit was considered. Thus, the value 80% was adopted as good and 90% as excellent, according to the classification by (Frizzone et al. 2012), for the parameters of CUC and CUD. It is noteworthy that, according to (Montgomery 2009), for existing processes and with unilateral specifications, values equal to or above 1.25 are able to be kept under control. Thus, in Table 7, the CUC parameter for LIC = 90% considers the system at an adequate level; when applied to LIC = 80%, it did not present averages within the pre-established limits; for CUD, there was no evaluation, as the values did not show normality and no correlation, following (Montgomery 2009).
(Hermes et al. 2015) obtained Cpl = 2.04, for the CUC, above 80%. (Juchen et al. 2013) also worked with drip irrigation, application of effluents from dairy and slaughterhouses, so they obtained 2.87 Cpl, indicating the capacity of the process. Vilas Boas and Sampaio (2010) concluded that the use of the process capacity index is a powerful tool to classify irrigation systems according to their distribution uniformity.
Table 8 - Process capability index for 2, 4 e 6 gL-¹
Treatment
|
Concentration
|
UC
|
UD
|
UC
|
UD
|
|
|
Cpl
LIC = 90%
|
Cpl
LIC = 80%
|
Cpl
LIC = 90%
|
Cpl
LIC = 80%
|
Level
|
2gL-1
|
2,86
|
*
|
*
|
*
|
Level
Uphill
Downhill
|
4gL-1
|
1,15
|
*
|
*
|
1,28
|
0,11
|
2,73
|
*
|
0,11
|
0,06
|
1,33
|
*
|
0,11
|
Level
Uphill
|
6gL-1
|
*
|
0,50
|
*
|
*
|
1,24
|
4,25
|
0,10
|
1,99
|
* It is not possible to calculate the process capacity index, they did not present averages within the pre-established limits.
Also for Table 8, for the 4gL-1 concentration, only the lower control limit was considered, so that the value of 80% was adopted as good and 90% as excellent, following the classification by (Frizzone et al. 2012). According to (Montgomery 2009) specifications, for LIC = 80%, both uphill and downhill systems are suitable for CUC, while for CUD, with LIC = 80%, only the level system is within specification. . Thus, the best performance was verified for the uphill system, however, no system was capable with LIC= 90%, minimum value of irrigation uniformity classified as excellent, according to (Frizzone et al. 2012).
Regarding the concentration of 6gL-1, only the lower control limit is considered, the value 80% as good and 90% as excellent, classification established by (Frizzone et al. 2012) for CUC and CUD. The CUC parameter, for LIC = 80%, only uphill, presented an average within the established limits and proved capable, with 4.25. Analyzing the CUD parameter for LIC = 80%, it is considered that the uphill is adequate, while, in relation to LIC = 90%, none of the levels proved to be adequate.