Process capacity index in drip irrigation with cassava wastewater processing

The study aimed to evaluate the performance of drip irrigation systems with clean water application and cassava wastewater processing, by determining the distribution uniformity coefficient (CUD) and the process capability index (Cpl). The research was carried out in an agroindustrial area from western Paraná State, Brazil, where two irrigation systems were set and operated in different periods. After the essays implementation, the two irrigation systems were identified and stored in order to be retested after one year (2 year of collection). So, the same conditions that were established in the first year had still been considered. Treatments with effluent showed flow averages near to those ones obtained with the use of clean water. Only in the T4 treatment, CUD was classified as good, while the others were excellent. The process was rated as efficient only in T1 with 80% LCL, with a 2.04 Cpl. Thus, based on this technique, it was possible to check that the system is able to keep suitable levels of uniformity.


INTRODUCTION
Due to water scarcity in many parts of the world, drip irrigation is becoming a popular system (Sahin et al., 2005), for its lower water consumption compared to other systems.This method of irrigation has a better efficiency of water use (Basso et al., 2008), minimizing the negative environmental impacts and becoming viable alternative for sustainable irrigated agriculture (Valipour, 2012;Bhattarai et al., 2008).
The use of wastewater for irrigation has many advantages.The most important of them concerns recovering a resource of great importance to agriculture: water; besides, its use works as an extra source of nutrients for plants.It helps on reducing development costs and with the addition of nutrients such as nitrogen, phosphorus and potassium from chemical fertilizers (Wang and Huang, 2008;Sandri et al., 2009;Juchen et al., 2013).Paraná state is in the Southern Brazilian region and it is *Corresponding author.E-mail: elianehermes@yahoo.com.br,Tel: 55 4432118574.Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License the largest producer of cassava starch in Brazil, 374.300 tons production in 2012 (CEPEA, 2013).The vegetable or cassava wastewater, a liquid effluent from starch extraction, it has high COD and presents toxic substances, hindering oxygen transportation into living beings organisms (Campos et al., 2006).
For the use of this effluent in irrigation does not become a problem for the environment, several factors related to its application need to be taken so that your exploration is properly managed.Among them, Alvarez et al. (2009) emphasize the study of water quality, the nutritional needs of plants, soil requirements and quality of effluent application throughout the area, represented by the distribution uniformity.This has directly influenced agricultural yield, which tends to increase as uniformity is improved (Brauer et al., 2011).Li et al. (2006) highlight the importance of managing a system with high uniformity, since low uniformity decreases the quality of irrigation as well as causes contamination and soil degradation.Many coefficients are used to express the distribution variability of water in irrigation, as the distribution uniformity coefficient (CUD) suggested by Keller and Karmeli (1974).Bralts (1986) has classified their values as: excellent when they were superior to 90%; good -between 80 and 90%; regular -between 70 and 80%, and bad inferior to 70%.
The techniques of statistical control are important to evaluate the quality of processes (Montgomery, 2009), since graphing to control and determine the index of process capability are options that can be used.These tools aim at evaluating the variability of a certain process, such as irrigation over time, to correct and eliminate possible wastes and failures, in order to increase yield (Justi et al., 2010).
Thus, this study aimed at evaluating the performance of drip irrigation systems by using clean water and cassava wastewater processing, with determining the CUD and establishment of process capability index.

MATERIALS AND METHODS
The study was carried out in a flat area without green cover crop, in a producing starch industry in Western Paraná, in Terra Roxa -PR municipality.Two systems of drip irrigation of approximately 66 m 2 (6 × 11 m) each was set in different periods.The systems consisted of one drip pipe with 1.49 L h -1 flow (85 kPa) Streamline 16080, every 0.30 m.The area had seven lateral rows with a total of almost 373 drippers, which were equivalent to an average total flow of about 555 h -1 .
The two used reservoirs of 1,000 L were 1.5 m above the floor, where clean water and cassava effluent processing were stored.Thirty trials were carried out for each of the treatments with 1 h difference between each one.It was a 4-min flow collection time for the sampled drippers (ISO 9261, 2006).Subsequently, this flow was measured in 100 ml graduated beaker with the same operating characteristics of the systems maintained and kept for all treatments.
Pressure was measured by two digital gauges, ITMPD Instrutemp-15 Model 8215, whose accuracy varied from ±0.3% to 25°C.This measure was taken in duplicate for each assay at the beginning and end of the system.After tests accomplishment, two irrigation systems were identified and stored to be retested after one year (2 nd year of collection), and the same conditions were established in the first year.The data for the first year were collected from March to July 2012, while the one from the second year was obtained from March to July 2013.
Clean water was applied in the irrigation trials from an artesian well, installed near the studied area.The effluent was obtained from the last facultative pond, and this part of the treatment system was carried out by the agroindustry.Thus, the effluent was taken to the irrigation systems by dripping a CV 5 PPM 2900 pump AMP 15-85 220-330 VOLTS, installed near the facultative pond.Constitution of treatments can be seen on Table 1, while physicochemical effluents characterizations from cassava processing in the first and second collecting year are on Table 2.
The evaluation of irrigation systems was performed according to the methodology proposed by Keller and Karmeli (1974), using distribution uniformity coefficient (CUD), Equation 1.
Where, -average flow of 25% lowest values, L h -1 ; -average flow of all measurements in L h -1 .
The Shewhart chart was used to evaluate whether the tested irrigations were in accordance with the project specifications.The upper limits (UCL) and lower limits (LCL) of Shewhart charts for individual measures were calculated from Equations 2 and 3.

UCL = µ + 3σ
(2) Where, µ -average; σ -standard deviation Hence, only treatments that showed statistical control in Shewhart chart were considered, in order to determine process capability index (Cpl).The Cpl calculation was determined by Equations 4 and 5.
Where, µ-average.Montgomery (2009) describes that for new processes, the Cpl must be superior to 1.6, so that it can be classified as capable.The construction of control charts and calculating capability index were obtained by MINITAB 16 software.

RESULTS AND DISCUSSION
The lowest average flow (0.674 L h -1 ) was obtained in treatment T4, whereas the largest one (0.735L h -1 ) was obtained in treatment T3 (Table 3).Treatments that applied effluent showed flow averages very close to those ones obtained with clean water application.It is observed that there is a relationship between flow and pressure, because the treatment with the highest average  pressure also showed the highest average of flow, when compared to the other treatments.Based on linear regression, the following equation was determined, where: flow = 0.56 pressure -0.234 with R 2 = 99.70%, and the dependent variable was the flow rate (L h -1 ) and the independent variable was the system pressure (kPa).
It is observed that for all treatments there is a correlation between the variables pressure and flow (Figure 1), and the pressure increase (independent variable) results in increased flow (dependent variable).The highest coefficient of determination R 2 of 82.43% was obtained in the treatment T1 (clean water in first year of collecting).This result means that the fitted model explained 82.43% of the variation in the response variable Y (flow).That is, 82.43% of the variability of flow is explained by the regressor variable pressure.Ahmed et al. (2007) evaluated the emitter flow on irrigation tests using polluted water, obtained a coefficient of determination of 99.00%, much higher than reported here.
In general, the highest coefficients of determination between the flow and the pressure were obtained in the treatments that used clean water, which can be attributed to the better quality of the applied water.The use of water with high concentrations of suspended solids cause changes in the flow, a fact proven by Souza et al. (2005) that determined the pressure-flow equations for irrigation using clean water, wastewater from poultry and cattle.
For the CUD, only the treatment T4 was classified as good, while the other treatments showed excellent coefficients (Bralts, 1986) (Table 4).Borssoi et al. (2012) evaluated the water uniformity and fertilizer application with drip irrigation using the collecting methodologies of Keller Karmeli and Denículi.They also determined CUD values, classified as medium and excellent with averages ranging from 85. 8 to 91.7% fo r irrigation and 88.3 to y = 0.005x

91.0% for fertigation.
There is some similarity with this trial concerning both values of pressure (12-18 kPa) and CUD.Carvalho et al. (2006) obtained 68% CUD in drip irrigation with three-year use of conventional water supply.The authors ascribed this result to some clogging caused by bad storage conditions as well as system operation and physical damage of the equipment due to the time of use.In this study, there was a slight decrease in CUD from the first to the second cropping year.
Treatments T1, T3 and T4 are under statistical control (Figure 2); that is, they showed essays distribution next to their respective averages and there are no points outside the lower control limit (LCL) and upper control limit of control (UCL).The variability of essays in these treatments remained within the control, thus, indicating that there was no special factor that brought forth a different behavior when compared to the ordinary one or that could result in a shift regarding its expected quality (Juchen et al., 2013).Only treatment T2, which used cassava wastewater, did not appear under statistical control because the essay 28 is out of LCL (81.5%).Hermes et al. (2013) compared the behavior of CUC on Shewhart control chart with clean water and diluted processing of cassava effluent.In both treatments, the values were out of control, with undesirable arrangements, plus an essay out of LCL to irrigate with clean water.Justi et al. (2010) used the Shewhart control chart for CUC in sprinkler irrigation and found out that one of the essays was above the UCL and none of the trials recorded CUC lower than LCL.The remaining values were within limits as well as under the control.
The values of process capability index (Table 5), which according to CUD, only LCL existence is considered, so that, 80 and 90% values were set.The first value is the minimum answer admitted in the irrigation system to a value classified as good (Bralts, 1986) and the second one was applied since the top most values were  superior to the first one.It can be observed that for 80% UCL, the process has been reported as capable since it was superior to 1.6 only in T1 (Montgomery, 2009), while it was considered acceptable in treatment T3.On the other hand, all other indexes were classified as incapable.There was a behavior directly proportional among CUD and Cpl values, ie as CUD increased, there was an increase on Cpl.There was also a relation between such variables expressed by the Equation CUD (%) = 88.7 + 10.2 Cpl, whose coefficient of determination was R 2 = 70.10%for an LCL = 90%.For 80% LCL, the obtained equation was CUD (%) = 86.4+ 2.86 Cpl, whose coefficient of determination was R 2 = 85.7%.Capability indices determined in the second year of collection for the CUD (T3 and T4) were lower than those obtained in the first year of collection (T1 and T2) and the two treatments of cassava processing wastewater (T2 and T4) had lower rates when compared to the two treatments of clean water (T1 and T3).These results may indicate that both the wear of the irrigation system and water quality influenced the CUD values and consequently the process capability indices.Justi et al. (2010) applied statistical techniques of control in a sprinkler irrigation system and determined values greater than 2.26 for Cpl, in such a way that as Christiansen's uniformity coefficient (CUC) increased, there was also an increase of capacity index.Thus, there was a relationship between these variables, which was expressed by CUC (%) = 46.07+ 10.55 Cpl, whose coefficient of determination was R² = 78%.Hermes et al. (2013) monitored the uniformity of a drip irrigation system with diluted processing of cassava effluent.When CUC value ranged between 85 and 87.5%, the authors registered a 4.13 process capability index; when CUC values ranged between 87.5 and 90%, process capability index was 4.19; and finally, when CUC value was superior to 90%, the obtained a process capability index was 5.50.Thus, as well as this research, there was a directly proportional behavior between the CUC and Cpl values.This relation is expressed by: CUC (%) = 79.46 + 1.925 Cpl, with a coefficient of determination R 2 = 61.0%,lower than those ones established in the present study.Juchen et al. (2013) have also worked with drip irrigation and applied effluents from agriculture, dairy and slaughtered industries and obtained a 2.87 Cpl, in order to indicate the process capability.

Conclusions
There were no significant changes in the flows of systems when clean water was applied or with cassava wastewater processing.Only one of the treatments with cassava wastewater processing showed a CUD classified as good, while the other treatments showed excellent CUD classification.Capability index that were determined in the second cropping year for CUD were inferior to those ones obtained in the first cropping year.Thus, in just one treatment with clean water, whose lower control limit was 80%, the process water was classified as capable.

Figure 2 .
Figure 2. Shewhart control charts for CUD in the four applied treatments.

Table 1 .
Constitution of treatments.

Table 2 .
Characterization of effluents used in treatments T2 and T4.

Table 3 .
Descriptive statistics for the flow and pressure in the applied treatments.

Table 4 .
Descriptive statistics for CUD values in the applied treatments.

Table 5 .
Process capability index of drip irrigation system with application of clean water and cassava effluent.