Electrical Conductivity and pH Prediction in a Recirculated Nutrient Solution of a Greenhouse Soilless Rose Crop
ABSTRACT The influence of nutrient solution (1) mixing rate, and (2) time of use on pH and elec-trical conductivity (EC) of a recirculated nutrient solution used for the irrigation of a greenhouse soilless rose crop was studied. Measurements of microclimate variables, pH, and EC of nutrient solutions and crop transpiration were conducted. The measurements of pH and EC values of nutrient solutions mixed with different mixing rates and applied for crop fertigation were used to develop and calibrate a model for pH and EC prediction in relation to nutrient-solution mixing rate and time of use. Application of the calibrated model gave satisfactory results. It was found that nutrient solutions with high mixing rates or volume equal to or double that of the total water consumed by the canopy during the conservation period had the most stable EC evolution and minimal pH changes.
- [Show abstract] [Hide abstract]
ABSTRACT: vegetable growth and nutrient solution, thus it is a challenge to achieve optimal control of nutrient solution. In this work, based on Q-Iearning, we first propose an in-situ optimal control method of nutrient solution compositions for greenhouse vegetable. Instead of modeling the correlations between greenhouse vegetable growth and nutrient solution, this method searches for optimal control policy through systematic interaction with the environment. The effect of nutrient solution compositions on photosynthetic rate of greenhouse vegetable is experimentally investigated, and on this basis reward function is designed. The experimental results show our method is effective and practical.01/2011;
- [Show abstract] [Hide abstract]
ABSTRACT: A simplified four-electrode multisensor prototype incorporating a homemade sensor based on chitosan-clay nanocomposite, and making use of Case-Based Reasoning (CBR) as pattern recognition technique, is effective for discerning among complex solutions of similar composition. The long-term stability of the sensors array together with the possibility of applying drift and weight corrections to the CBR database increase the efficiency of the CBR-based system, enhancing its ability to detect variations in the composition of fertigation solutions used in greenhouse crops.Sensors and Actuators B: Chemical. 01/2009;
Journal of Plant Nutrition, 29: 1585–1599, 2006
Copyright ©Taylor & Francis Group, LLC
ISSN: 0190-4167 print / 1532-4087 online
Electrical Conductivity and pH Prediction in a
Recirculated Nutrient Solution of a Greenhouse
Soilless Rose Crop
Christos Lykas, Nikolaos Katsoulas, Panagiotis Giaglaras,
and Constantinos Kittas
Department of Agriculture, Crop Production and Rural Environment, School of
Agricultural Sciences, University of Thessaly, Magnisia, Greece
The influence of nutrient solution (1) mixing rate, and (2) time of use on pH and elec-
trical conductivity (EC) of a recirculated nutrient solution used for the irrigation of a
and EC of nutrient solutions and crop transpiration were conducted. The measurements
of pH and EC values of nutrient solutions mixed with different mixing rates and applied
in relation to nutrient-solution mixing rate and time of use. Application of the calibrated
model gave satisfactory results. It was found that nutrient solutions with high mixing
the conservation period had the most stable EC evolution and minimal pH changes.
Keywords: hydroponic crop, closed hydroponic systems, mixing rate, irrigation man-
Closed hydroponic systems and reuse of drained nutrient solution are becom-
ing common practices for many greenhouse crops in Mediterranean countries,
as they enable economical use of water and fertilizers and provide an ample
water supply to the crop (Sonneveld and Welles, 1984). In closed hydroponic
Received 14 October 2003; accepted 19 May 2006.
Phytokou St., N. Ionia, GR-38446, Magnisia, Greece. E-mail: email@example.com
1586 C. Lykas et al.
systems, the plants are fed by recirculating nutrient solutions in which electri-
cal conductivity (EC) and pH are monitored and adjusted to a desired level by
injecting water or stock solutions and acid.
However, long-term recirculation results in accumulation of ions, such as
sodium (Na) and chloride (Cl), that will be added refill water or remain in the
used nutrient solution and that are not absorbed by the plants. Sooner or later,
ion accumulation makes the replacement of the recirculating nutrient solution
necessary. In commercial systems after a period of use, the nutrient solution
will have elevated EC and pH values that can no longer be corrected with
water or stock-solution injections because there may be nutrient imbalances,
indicating that the nutrient solution must be replaced. Thus, the prediction of
the spontaneous time evolution of EC and pH of a recirculated nutrient solution
may provide information useful for the maximization of the time a nutrient
solution can be used without correction before it needs to be replaced.
The time course of EC and pH of a nutrient solution is the result of crop
water and nutrient absorption and chemical reactions that have nothing to do
with the crop and occur in the nutrient solution itself. In any nutrient solution,
some components such as CaCO3, CaHPO4, MgHPO4, and FePO4KSO4can
precipitate and dissolve at any time, depending on the concentration and pH
of the solution (De Rijck and Schrevens, 1998a, 1998b; Lykas et al., 2001).
A common practice to minimize the precipitates (maximize solubilization) in
order to maintain the nutrient solution in a steady state in regard to ion concen-
tration, EC, and pH, is to mix the solution in a conservation tank (Lykas et al.,
The objective of this work was to provide insight into the relative im-
portance of chemical processes, including salt precipitation and solubilization,
under various mixing regimes and bio-physical processes in order to determine
the time course of the EC and pH of recirculated nutrient solutions.
MATERIALS AND METHODS
(Rosa hybrida cv. ‘First Red’), installed in a 200 m2glass-covered greenhouse
at the University of Thessaly, situated near Volos, (39◦44?,22◦79?, 85 m) on
the coastal area of eastern Greece, was used. The experiments were performed
× 0.3 m width × 0.16 m depth) filled with perlite (six plants per container, 8 L
perlite per plant). There were six 21 m long crop lines spaced at a distance of
1 m between lines, with 126 plants per line (Figure 1). This planting scheme
Electrical Conductivity and pH Prediction1587
Figure 1. Experimental greenhouse, crop and instrument arrangements; 1: ventilated
psychrometer, 2: pyranometer.
resulted in a plant density of six plants m−2. The plants were grown following
the “bending” technique. During the period of measurements, the leaf-area
index of the rose crop was about 2.
Greenhouse air temperature and relative humidity were measured using a
ventilated psychrometer (wet and dry bulb) located at the height of the upper
level of the flower stems, and solar radiation incident above the rose crop was
measured using a solar pyranometer (model CM-6, Kipp and Zonen, Delft, the
Netherlands). Climate data were measured every 30 sec and 10 min average
values were recorded using a data-logger system (Model DL3000, Delta-T
Devices, Cambridge, UK).
The Hydroponic System
The experiments were conducted using two identical closed hydroponic sys-
tems. Each system was composed of an 800 L nutrient solution tank, a 3 m3
h−1irrigation pump, a drip irrigation pipe network, a recycling pump bringing
drained solution back to the main tank, and a mixing pump with a constant flow
rate of Q = 4800 L h−1(Figures 1 and 2). Mixing rates were modulated at 48,
24, 12, and 6 changes h−1by changing the volume of the nutrient solution in the
tank (V) at 100, 200, 400, and 800 L, respectively. However, this setup implies
that in nutrient solutions with low volume (high mixing rate) the total amount
1588 C. Lykas et al.
Figure 2. Arrangement of the closed hydroponic system.
of dissolved salts was significantly lower than in the high volume (low mixing
rate) nutrient solutions. This mixing-rate modulation technique was selected
for its similarity to common greenhouse practice.
The above volumes of the nutrient solutions were chosen to represent one,
two, four, and eight times the maximum volume of the total water expected
to be evapo-transpired by the crop during an experimental period (7 to 8 d).
Mixing rate (mr in changes h−1) was estimated using the following equation:
the greenhouse (one line per system) (Figure 1). Nutrient solution was supplied
for 2 min every hour from 6 a.m. to 9 p.m. via the drip system. The drainage
rate was maintained near 60% in order to obtain an ample water supply to the
plants and to avoid ion accumulation in the root zone.
(Ca), 160; magnesium (Mg), 24; iron (Fe), 1.3, boron (B), 0.28; copper (Cu),
fertilizer was added in the nutrient solution after solubilization by mixing in a
pH and EC values were 5.5 and 1.7 dS m−1, respectively. The tap water used
for nutrient solution preparation had a pH of 7.0 and an electrical conductivity
of 0.7 dS m−1.
Electrical Conductivity and pH Prediction 1589
Three series of experiments were performed.
To study the time course of EC and pH without influencing the crop, the nu-
trient solution was mixed in the tank for 8 d after preparation with a con-
stant mixing rate and without its being used for crop fertigation (first series
of experiments). Electrical conductivity and pH were measured three times
per day, at 9.00 a.m., 2.00 p.m., and 17.00 p.m., using a conductivity meter
(Model HI 8633, Hanna Instruments) and a membrane pH meter (Model HI
8314, Hanna Instruments), respectively. Calculated daily values of EC (EC)
and pH (pH) are means of the three time-spaced measurements taken each
day. This experiment was repeated four times for each mixing rate (48, 24, 12,
and 6 changes h−1). These experiments were completed in 64 d from April
2001 to May 2001 (4 mixing rates × 8 d per mixing rate × 4 replications/2
tanks = 64 d).
To study the influence of the crop system on the evolution of pH and EC values
of the nutrient solution, the solution was mixed and applied to the crop for a pe-
riod of 8 d (second series of experiments). Two different treatments (two tanks
and nutrient solutions with different mixing rates) were applied simultaneously
in two different crop lines during the 8 d period. During this period, the mixing
rate gradually increased because the volume of the nutrient solution progres-
sively decreased due to crop transpiration. Table 1 shows the nutrient solution
mixing-rate range for nutrient solutions used for crop fertigation. Each 8 d
Initial, middle, and final nutrient-solution mixing rates applied during recirculation
for the four initial volumes of the nutrient solution
Mixing rate (in changes h−1)
initial volume (in L)
InitialMiddleFinal (during the
last day of use) (after preparation) (after 4 d of use)
1590 C. Lykas et al.
experiment was repeated six times by changing the hydroponic system–mixing
rate combination in order to ensure that the results were due to the mixing rate
and not to the hydroponic system. These experiments were completed in 96 d
from September 2001 to November 2001 (4 mixing rates × 8 d per mixing
rate × 6 replications/2 tanks = 96 d). The data obtained during this experiment
were used to develop a descriptive model to predict changes in EC and pH,
based on mixing rate.
The above series of experiments was repeated once again (during December
2001), using only one of the two hydroponic systems, in order to collect
pH and EC evolution data for validation of the model developed from the
data collected in experiment 2 (4 mixing rates × 8 d per mixing rate × 1
replication = 32 d).
It should be noted that pH and EC correction did not occur in any of the
above experiments during nutrient-solution use.
Data Fitting and Statistical Analysis
The volume of the nutrient solution was measured every morning and the mix-
ing rate of the nutrient solution was calculated using Equation 1. Using the
calculated values of the mixing rate of the nutrient solutions and the mean
daily pH (pH) and EC (EC) values measured in the nutrient solutions with
the different initial mixing rates (48, 24, 12, and 6 changes h−1) supplied to the
crop, the time courses of pH and EC were estimated and fitted to the following
pHi= pH0+ (apHmr + bpH)∗?1 − e−(cpHmr)i?
ECi= EC0+ (aECmr + bEC)∗i +?1 − e−[cECln(mr)]i?
where pHiand ECiare the mean pH and EC values of the nutrient solution
during the day i, respectively; pH0and EC0are the initial pH and EC values at
the beginning of the 8-d experiment, respectively; and apH, bpH, cpH, aEC, bEC,
and cECare constant model parameters, identified with nonlinear regression
using the SPSS package (SPSS 10.0 for Windows).
The predictive quality of the parameterized models was assessed using the
same statistical package to perform linear regressions between predicted and
measured values, looking at the intercept and the slope to not be significantly
different from zero and one, respectively (Confidence interval = 95%) (Gauch
et al., 2003).
Electrical Conductivity and pH Prediction1591
Mean monthly values of the greenhouse climate variables col-
RESULTS AND DISCUSSION
The mean monthly values of the greenhouse climate variables collected from
in Table 2.
Evolution of pH and EC in Nutrient Solutions Not Applied to the Crop
Figure 3 shows that for nutrient solutions not applied to the crop, mixing rate
and/or the volume of the nutrient solution strongly affected the time course of
pH. In nutrient solutions with mixing rates of 48 (V = 100 L) and 24 changes
h−1(V = 200 L), the pH value was relatively constant for the period from
24 h after preparation until the end, with the pH near the target value of 5.5.
In contrast, in nutrient solutions with mixing rates of 12 (V = 400 L) and 6
changes h−1(V = 800 L), pH was above the target value of 5.5, exceeding
the value of 7.0 after the second day following preparation, and holding a rel-
atively constant value of about 7.5 after the third day following preparation.
The above patterns of pH evolution and the differences observed between the
mixing regimes may be due to the volume of the nutrient solution and to the
different amounts of CO2dissolution caused by the different mixing rates. In
a nutrient solution, the pH value is determined mainly by the initial concentra-
tion of H2PO−
Adamidis, 1999). A nutrient solution that is not used for crop fertigation con-
mainly by the relative concentration of the three ions. In nutrient solutions that
were not applied to the crop, HCO−
dissolved in the nutrient solution. High mixing rates and low volume of the
nutrient solution, may have significantly increased the CO2dissolution. That
4, and HCO−
3(De Rijck and Schrevens, 1997b; Savvas and
4, and HCO−
3, and accordingly its pH would be determined
3concentration depended on CO2amounts
1592 C. Lykas et al.
Figure 3. Evolution over time of pH of nutrient solutions kept mixed in the tank, as
affected by the different mixing rates: mr = 48 (-•-), 24 (-◦-), 12 (—?—), and 6 (-?-)
changes h−1. Data points are mean values of four replications for each treatment (n =
4) and vertical bars represent standard deviation.
in turn would result in the formation of H+, lowering the pH. In addition, in
these nutrient solutions, high mixing rate seems to have been able to inhibit
complexation and to precipitation and to encourage dissociation reactions of
contrast, the high pH values observed in high-volume nutrient solutions, which
were mixed at a low rate, may have been due to the low CO2dissolution. In
addition, low mixing rates in these nutrient solutions seem to have been unable
to inhibit complexation and precipitation reactions of H2PO−
In nutrient solutions with a mixing rate of 48 changes h−1and low volume
(V = 100 L), a continuous increase in the EC was measured, reaching a value
close to 2 dS m−1at the end of the 8 d period (Figure 4). In nutrient solutions
with mixing rates of 24, 12, and 6 changes h−1and volume of 200, 400, and
800 L, respectively, an increase in EC was observed, starting after preparation
of the nutrient solution and lasting for about 3 d, depending on the mixing rate
and /or the volume of nutrient solution. This increase in EC was followed by a
again on the mixing rate and/or the volume of nutrient solution.
Different mixing rates and/or volumes of the nutrient solution may cause
differences in the chemical reactions (complexation, dissociation, and precipi-
tation) taking place in the nutrient solutions (De Rijck and Schrevens, 1997a,
1998a, 1998b) as well as changes in the CO2dissolution. A high mixing rate
4, which contributed to acidification of the nutrient solution. In
Electrical Conductivity and pH Prediction1593
Figure 4. Evolution over time of EC of nutrient solutions kept mixed in the tank, as
affected by the different mixing rates: mr = 48 (–•–), 24 (–◦–), 12 (—?—), and 6 (-?-)
changes h−1. Data points are mean values of four replications for each treatment (n =
4) and vertical bars represent standard deviation.
of low-volume nutrient solutions inhibits complexation and precipitation and
encourages dissociation reactions and, accordingly, prevents decreases in the
EC of the solution. In addition, in these nutrient solutions, high CO2dissolu-
tion would result in the formation of HCO−
increase in the EC during the period of use (Figure 4) using mixing rates of 48
high-volume nutrient solutions did not inhibit complexation and precipitation
reactions and consequently resulted in a decrease in the EC of the solution. As
a result, a low mixing rate in the nutrient solution, e.g., 6 changes h−1(V = 800
L), led to a decrease in EC value (Figure 4). The increase and decrease in EC
at the beginning and near the end of the period of use of the nutrient solution,
respectively, can be explained by the changes in the pH values during the same
periods, as the concentration of the free ions in the solution depended on the
3, increasing EC. Accordingly, the
Evolution of pH and EC in Nutrient Solutions Applied to the Crop
As shown in Figure 5, there were no significant differences between the pH
values of nutrient solutions with mixing rates higher than 24 and 48 changes
h−1(low-volume nutrient solutions). The pH values of these nutrient solu-
tions increased to 6.5 during the first day of the application, progressively
decreased after that time, and remained near 5.0 after the fourth day following
1594 C. Lykas et al.
Figure 5. Evolution over time of pH of nutrient solutions used for crop fertirrigation
during the period of recirculation, as affected by different mixing rates: mixing rates:
mr = 48 (-•-), 24 (– –◦– –), 12 (–?–), and 6 (- - ? - -) changes h−1(n = 6). Data points
are mean values of six replications for each treatment (n = 6) and vertical bars represent
h−1(high-volume nutrient solutions), pH exceeded 7.0 and 8.5 during the sec-
ond day after preparation, taking a relatively constant value close to 7.0 and
8.2, respectively, after the fourth day following preparation.
dS m−1. EC did not exceed the value of 2 dS m−1only in the case of nutrient
solutions with a mixing rate of 48 changes h−1. For the rest of the cases, EC
was always higher than 2 dS m−1during the 8 d period of nutrient solution use
Model Calibration and Validation
The measurements presented in Figures 5 and 6 (Experiment 2) were used to
develop and calibrate equations 2 and 3, which were used to predict the pH and
EC evolution in nutrient solutions used for crop fertigation. The mixing rate of
these solutions was not stable, but became progressively higher as the volume
of the solution decreased due to transpiration. The equations were calibrated
separately for each group of nutrient solutions with the same initial mixing
rate (but with a different mixing rate each day) the values of the estimated
parameters, along with their standard errors and the equations’ determination
coefficients (R2), are presented in Tables 3a and 3b.
Electrical Conductivity and pH Prediction1595
Figure 6. Evolution over time of EC of nutrient solutions used for crop fertirrigation
during the period of recirculation, as affected by different mixing rates: mr = 48 (–•–),
24 (−◦−), 12 (−?−), and 6 (- - ? - -) changes h−1. Data points are mean values of six
replications for each treatment (n = 6) and vertical bars represent standard deviation.
rate (data not shown), indicate that the above parameters have been correctly
taken into account by the models.
Estimated values of parameters of Equation 2 for pH prediction of nutrient
solutions with initial mixing rates of 48, 24, 12, and 6 changes h−1.
Initial mixing rate
Standard error of the estimated values is given in parenthes; n is the
number of measurements and R2is the determination coefficient.
1596 C. Lykas et al.
Estimated values of parameters of Equation 3 for EC prediction of nu-
trient solutions with initial mixing rates of 48, 24, 12, and 6 changes
Initial mixing rate
(h) (h dS m−1) (dS m−1)n
Standard error of the estimated values is given in parenthess; n is the
number of measurements and R2is the determination coefficient.
Figures 7 and 8 show the (interpolative) predictive capacity of Equations
2 and 3, tested against pH and EC values from experiments performed during
December 2001 (third series of measurements). These values were not used for
previous model calibration.
Figure 7. Comparison between measured and estimated pH values, using Equation 2,
of nutrient solutions with mixing rates of 48 (•), 24 (◦), 12 (?), and 6 (?) changes h−1.
Straight line represents the 1:1 line. The regression equation found was pHestimated=
0.971 pHmeasured+ 0.180.
Electrical Conductivity and pH Prediction 1597
Figure 8. Comparison between measured and estimated EC values, using the recali-
brated Equation 3, of nutrient solutions with mixing rates of 48 (•), 24 (◦), 12 (?), and
6 (?) changes h−1. Straight line represents the 1:1 line. The regression equation found
was ECestimated= 0.980 ECmeasured+ 0.025.
The linear regression obtained between the measured and estimated pH values
gave the following coefficients for 95% confidence: slope α = 0.971(±0.07),
interceptβ = 0.18(±0.44),determinationcoefficient R2= 0.98.Thestatistical
testperformedtocheckthenullhypothesisthattheslopeα = 1andtheintercept
β = 0 gave a P value of 0.21 and 0.23, respectively.
Similarly, the linear regression obtained between the measured and esti-
mated EC values gave the following coefficients for 95% confidence: slope
α = 0.98(±0.06), intercept β = 0.025(±0.15), determination coefficient R2
0.98. The statistical test performed to check the null hypothesis that the slope
α = 1 and the intercept β = 0 gave a P value of 0.15 and 0.35, respectively.
Consequently, the model presented in this study could accurately predict
the nutrient solution EC and pH values, information that may prove useful for
the maximization of the time a nutrient solution could be used without correc-
tions. However, it is necessary to evaluate the above models under different
circumstances in order to use them with different nutrient solutions, crops, and
Changes of EC and pH occurred in a nutrient solution not applied to the
crop, indicating changes in the composition of the nutrient solution due to
complexation, dissociation, and precipitation reactions as well as CO2 go-
ing into solution, without the influence of the canopy (water and nutrient
1598 C. Lykas et al.
absorption). In this case, the time course of EC and pH seemed to be deter-
mined by the mixing rate or the volume of the nutrient solution and the time of
nutrient solution use.
In a nutrient solution applied to the rose crop, pH and EC evolution were
affected by nutrient solution mixing rate and plant processes (water and nutri-
ent absorption). The effect of nutrient solution, chemical processes, and plant
processes on pH and EC values were taken into account and pH and EC of
nutrient solutions were estimated.
It was found that a nutrient solution with volume less than twice that of the
higher than 24 changes h−1) had the best EC evolution (almost stable value)
and the minimal pH changes. In addition, high mixing rate seemed to be an
alternative to acid addition for keeping the pH of the nutrient solution stable.
Accordingly, from the growers’ point of view, it could be recommended to pre-
pare nutrient solutions with a volume double that of the total water consumed
by the canopy during the period of use and to keep mixing rates higher than
24 changes h−1, using the appropriate pumps, in order to obtain the maximum
period of use without any corrections of pH and EC of the nutrient solution.
In view of this recommendation, the results presented in this study could help
the grower in the selection of the proper equipment (tanks and pumps) of the
hydroponic system in order to obtain the maximum conservation time of the
controlled using a mixing pump with constant flow rate and different nutrient-
solution volumes, and therefore it was not possible to distinguish between
the effects of volume and mixing ratio on pH and EC evolution. Therefore, it
between the effects of volume and mixing ratio on pH and EC evolution, con-
structing experiments with constant-volume nutrient solutions and regulating
the mixing rate using different mixing pumps. Furthermore, experiments with
different nutrient solutions and on different crops would aid in the evaluation
of the models presented.
Finally, the mathematical expressions developed in this paper for the esti-
mation of the time course of pH and EC values of a nutrient solution used for
rose crop fertigation gave satisfactory results and probably could be used for
the prediction of pH and EC in nutrient solutions with various mixing rates, an
ability that will provide valuable information for the maximization of the time
from in which a nutrient solution could be used.
De Rijck, G., and E. Schrevens. 1997a. Elemental bioavailability in nutrient
solutions in relation to dissociation reactions. Journal of Plant Nutrition
20 (7&8): 907–910.
Electrical Conductivity and pH Prediction1599
De Rijck, G., and E. Schrevens. 1997b. pH influenced by the elemental compo-
sition of nutrient solutions. Journal of Plant Nutrition 20 (7&8): 911–923.
De Rijck, G., and E. Schrevens. 1998a. Elemental bioavailability in nutrient
solutions in relation to complexation reactions. Journal of Plant Nutrition
21 (5): 849–859.
De Rijck, G., and E. Schrevens. 1998b. Elemental bioavailability in nutrient
solutions in relation to precipitation reactions. Journal of Plant Nutrition
21 (10): 2103–2113.
ison of model-based prediction and measured values. Agronomy Journal
Lykas, Ch., P. Giaglaras, and C. Kittas. 2001. Availability of iron in hydro-
ponic nutrient solutions for rose crop. Journal of Horticultural Science &
Biotechnology 76 (3): 350–352.
Journal of Plant Nutrition 22 (9): 1415–1432.
Sonneveld, C., and W. H. Welles. 1984. Growing vegetables in substrates in
the Netherlands. In Proceedings of the Sixth International Congress on
Soilless Culture, 613–632. Wageningen, Netherlands: ISOSC.