processes

Article
Characterization of Oxidation-Reduction Potential Variations in
Biological Wastewater Treatment Processes: A Study from
Mechanism to Application
Xiaodong Wang 1, * , Yuxing Wu 1,2 , Ning Chen 1 , Heng Piao 2 , Delin Sun 3 , Harsha Ratnaweera 4 ,
Zakhar Maletskyi 4 and Xuejun Bi 1

1 School of Environmental and Municipal Engineering, Qingdao University of Technology, Jialingjiang East 777,
Huangdao, Qingdao 266520, China
2 Jiangsu Haixia Environmental Protection Technology Development, Ltd., Nanjing 210000, China
3 Shandong Oubeier Software Technology Co., Ltd., Jinan 250021, China
4 Faculty of Science and Technology, Norwegian University of Life Sciences, P.O. Box 5003, 1432 Aas, Norway
* Correspondence: [email protected]

Abstract: Oxidation-reduction potential (ORP) sensors would constitute a robust surveillance and
control solution for aeration and external carbon dosing in wastewater biological treatment processes
if a clear correlation exists between the ORP values and process variables (e.g., dissolved oxygen
(DO), nitrate, and chemical oxygen demand (COD). In this study, ORP values and other water quality
variables were analyzed, and principal component analysis (PCA) and analysis of variance were
used to study the relationships between ORP and main reactive substances under anoxic conditions.
Mathematical models were then established using multiple regression analysis. The results showed
that under anoxic conditions, ORP was positively correlated with nitrate, DO, and COD and nega-
Citation: Wang, X.; Wu, Y.; Chen, N.;
tively correlated with ammonia nitrogen, phosphate, and pH. COD had a low correlation with the
Piao, H.; Sun, D.; Ratnaweera, H.;
ORP value change. PCA showed that the mathematical model of ORP can be established by using
Maletskyi, Z.; Bi, X. Characterization
DO, nitrate, and phosphate, for which the adjusted R2 value was 0.7195. The numeric relationships
of Oxidation-Reduction Potential
Variations in Biological Wastewater
among ORP, COD, and nitrate were clearly established and applied to control external carbon dosing.
Treatment Processes: A Study from A precise and clear relationship between ORP and COD offers the possibility to substitute COD
Mechanism to Application. Processes monitoring for process control.
2022, 10, 2607. https://doi.org/
10.3390/pr10122607 Keywords: oxidation-reduction potential; process monitoring; wastewater treatment; process control;
principal component analysis
Academic Editor: Avelino
Núñez-Delgado

Received: 9 November 2022

Accepted: 28 November 2022 1. Introduction
Published: 6 December 2022
Oxidation-reduction potential (ORP) is a water quality variable employed to character-
Publisher’s Note: MDPI stays neutral ize the redox capacity of the whole system, which is represented by the ORP value. Wastew-
with regard to jurisdictional claims in ater biological treatment processes have many types of oxidative substances (e.g., dissolved
published maps and institutional affil- oxygen (DO), nitrate, and nitrite) and reductive substances (e.g., ammonium and organic
iations. matter); therefore, they lead to series of oxidation-reduction reactions. The ORP of the
system is determined by the interaction of various reductive and oxidative substances in
the whole system [1]. When the whole system showed oxidability, the ORP value was
lower, and vice versa [2]. Initially of interest, the ORP value was supposed to indicate the
Copyright: © 2022 by the authors.
DO level and control the aeration, although the function of ORP sensors was substituted
Licensee MDPI, Basel, Switzerland.
with more reliable optical DO sensors [3,4]; however, ORP sensors have been used for water
This article is an open access article
and wastewater treatment process monitoring [5] and chemical dosing control [6–8].
distributed under the terms and
In recent years, improving effluent quality and reducing operational costs have driven
conditions of the Creative Commons
Attribution (CC BY) license (https://
the development of instrumentation, control, and automation in wastewater treatment
creativecommons.org/licenses/by/
systems [9]. Various sensors that monitor the wastewater treatment processes have been
4.0/). installed in wastewater treatment plants (WWTPs). Automatic control and automation have

Processes 2022, 10, 2607. https://doi.org/10.3390/pr10122607 https://www.mdpi.com/journal/processes

 used for water and wastewater treatment process monitoring [5] and chemical dosing
control [6–8].
In recent years, improving effluent quality and reducing operational costs have
driven the development of instrumentation, control, and automation in wastewater treat-
Processes 2022, 10, 2607 2 of 11
ment systems [9]. Various sensors that monitor the wastewater treatment processes have
been installed in wastewater treatment plants (WWTPs). Automatic control and automa-
tion have the potential to reduce labor cost and energy consumption [10], and online sen-
the
sorspotential
providetoreal-time
reduce labor
data cost and energy
to enable consumption
automatic process [10], and[11].
control online sensors provide
However, nitrate,
real-time data to enable automatic process control [11]. However, nitrate,
ammonia nitrogen (NH4-N), and other nutrient online monitoring sensors are not reliableammonia nitrogen
(NH4-N), and control
for real-time other nutrient
due to online
frequent monitoring sensors
sensor fault, aredrifting,
e.g., not reliable
but for real-time
building the control
sensor
due to frequent sensor fault, e.g., drifting, but building the sensor network also
network also requires a high financial cost, which results in higher initial and operational requires a
high financialfor
investments cost, which results
WWTPs. Moreover,in higher
higherinitial and operational
maintenance frequencyinvestments
is requiredfordue
WWTPs.
to the
Moreover, higher maintenance frequency is required due to the harsh
harsh environment of WWTPs, which is labor intensive. Figure 1 shows the appearance environment of
WWTPs, which isonline
of an integrated labor intensive.
sensor beforeFigure
and1 after
shows the appearance
manual cleaning, of an integrated
which online
was installed to
sensor
measure nitrate and ammonium of raw wastewater before biological treatment. The and
before and after manual cleaning, which was installed to measure nitrate ion-
ammonium of raw wastewater
selective membrane before
of the sensor biological
would treatment.
be covered The ion-selective
by biofilm within 1 daymembrane
exposure of to
the sensor would be covered by biofilm within 1 day exposure to raw wastewater.
raw wastewater. Previous studies reported that ion-selective electrodes (e.g., ammonium, Previous
studies reported that ion-selective electrodes (e.g., ammonium, nitrate, and DO) need to be
nitrate, and DO) need to be maintained at least twice per week [12]. Despite the extensive
maintained at least twice per week [12]. Despite the extensive cost of sensor maintenance,
cost of sensor maintenance, sensor failure is another unavoidable challenge. Ohmura et
sensor failure is another unavoidable challenge. Ohmura et al. demonstrated two important
al. demonstrated two important assumptions for sensors based on the principle of ion
assumptions for sensors based on the principle of ion selectivity measurement that are not
selectivity measurement that are not valid: (i) sensor faults appear at distinct times in dif-
valid: (i) sensor faults appear at distinct times in different sensors, and (ii) any given sensor
ferent sensors, and (ii) any given sensor functions near-perfectly for a significant time pe-
functions near-perfectly for a significant time period following installation [13]. Sensor
riod following installation [13]. Sensor failure can occur at any time, and drift may happen
failure can occur at any time, and drift may happen immediately after calibration. ORP
immediately after calibration. ORP sensors have the advantages of being low cost, reliable,
sensors have the advantages of being low cost, reliable, and robust since the mechanism of
and robust since the mechanism of ORP measurements is based on electro-potential. The
ORP measurements is based on electro-potential. The relationship between the ORP value
relationship between the ORP value and the other water quality variables in the process
and the other water quality variables in the process of wastewater treatment has not been
of wastewater treatment has not been well-investigated yet. Applications of ORP as an
well-investigated yet. Applications of ORP as an indirect water quality variable and an
indirect water
alternative quality
to direct variable
water and
quality an alternative
variables to direct
and sensor water
anomaly quality are
detection variables and
attracting
sensor anomaly
increasing detection
research aresince
interest attracting increasing
the initial researchand
investment interest since the initial
maintenance invest-
cost may be
ment and maintenance cost may be significantly reduced compared
significantly reduced compared with direct measurement sensors of various wastewaterwith direct measure-
ment sensors
quality sensors.of various wastewater quality sensors.

Figure 1.
Figure 1. The
The integrated
integratedammonium
ammoniumand andnitrate
nitratesensors.
sensors.(a)(a)Being
Being used
used forfor raw
raw wastewater
wastewater moni-
monitor-
toring
ing in influent
in influent channels
channels for for
24 h24without
h without cleaning;
cleaning; (b) (b) after
after cleaning.
cleaning.

Previous studies have

Previous have indicated
indicatedthat
thatORP
ORPmay maybebe a useful water
a useful waterquality indicator
quality for
indicator
wastewater
for wastewatertreatment process
treatment control.
process Wang
control. et al.etstudied
Wang the relationships
al. studied between
the relationships ORP,
between
nitrate,
ORP, and phosphate
nitrate, in anin
and phosphate activated sludge
an activated system
sludge withwith
system a denitrification enhanced
a denitrification bio-
enhanced
biological phosphorus
logical phosphorus removal
removal process,who
process, whosuggested
suggestedthat thatORP
ORPcancanbe
be used
used to
to control the
the
process operation in the anaerobic zone for phosphorus removal [14]. Su studied the ORP
variation characteristics of high concentration organic wastewater treated by the sequencing
batch reactor (SBR), which demonstrated that the ORP had strong correlations with NH4 -
N, chemical oxygen demand (COD), DO, temperature, and pH [15]. These studies have
indicated the correlation between ORP, process operational variables, and environmental
variables. For process surveillance and control purpose, further investigation is necessary.
Processes 2022, 10, 2607 3 of 11

Process control based on the ORP and other water quality variables (such as pH and
DO), or with the first derivative of ORP as one of the main control parameters has been
widely studied, and some progress has been made [16–21]. ORP and its first derivative
play an indicator role to identify the change in nitrate concentration. Fox et al. used an
online control system with ORP sensors involved for aeration control in a wastewater
treatment system, which saved more than 40% of energy consumption [20]. These studies
show the potential and advantages of ORP sensors in the wastewater treatment process
and proved that ORP can be a reliable process control indicator. However, the pattern of
ORP sensor faults and its mathematical relationship with other water quality indicators is
unclear [22]. To make better use of the indicator function of the ORP sensors in the future,
a more comprehensive mechanism study on the complex relationship between ORP and
water quality indicators should be conducted, and the ORP variation features caused by
water quality variation should be clarified. Though ORP sensors have demonstrated their
ability as alternative sensors to reflect NH4 -N and nitrate concentration, the means by
which to reverse reflect the changes in key water quality indicators in process control by the
ORP value remains lacking in scientific basis, and no substantial research breakthroughs
have been made. The application potential of carbon addition control in the anoxic zone of
the wastewater treatment process remains to be explored.
The primary goal of this research was to clarify the relationship between ORP and the
main reactions in the process of anoxic reactions by data analysis. After establishing the
direct relationship between ORP and the denitrification degree, the secondary goal was to
explore the method of using ORP to control the dosing of carbon source for denitrification
in wastewater treatment processes.

2. Materials and Methods

2.1. Experimental Setup
As shown in Figure 2, an acrylic sealed reactor with a volume of 20 L volume was
used to perform the mechanistic investigation of ORP and other water quality variables. A
rubber pad was placed between the top cover plate and reactor to ensure a sealing effect.
A hole was opened on the cover plate for passing through the sensor probes. DO and
multi-parameter water quality analyzer probes were used in the experiment to measure
DO, pH, and total dissolved solids (TDS) of the mixed liquid. Returned sludge from the
secondary sedimentation tank and effluent from the primary setting tank of the domestic
WWTP were mixed in the reactor to simulate an anoxic environmental of wastewater
treatment processes. The nitrified mixed liquid was also used by sampling from the aerobic
zone of the biological treatment tank to mix with the liquid in the reactor to create different
oxidative or reductive levels. Moreover, biocarriers of moving bed biofilm reactors (MBBRs)
from a well-operated pilot MBBR system were used to test the biochemical reaction effect
caused by biofilm and the corresponding variation in ORP values.

2.2. Sensors
Temperature, pH, ORP, and TDS were measured using a multi-parameter analyzer
(Lei-Ci DZS-706, Shanghai, China). DO was measured using the Hach HQ40d portable
multi-parameter water quality detector with a Hach LDO 10110 fluorescence electrode
(Hach, Loveland, CO, USA). Other water quality data were obtained by sampling and
analyzed based on standard methods [23]. The two-point calibration method was applied
for ORP sensors calibration, using the standard solution of 86 mV and 256 mV.

2.3. ORP and Water Quality Correlation Investigation Method

The purpose of the water quality analysis was to study the correlations between ORP
values and other water quality variables in an anoxic environment and obtain sufficient data
for analysis. Batch tests were carried out in a completed mixed reactor, and the volumetric
ratios of recycled sludge/settled raw wastewater were 50, 66, 100, 200, 400, and 600% to
simulate the external recycling rate of the activated sludge system. DO, pH, and ORP were
Processes 2022, 10, 2607 4 of 11
Processes 2022, 10, x FOR PEER REVIEW 4 of 11

continuously measured by sensors during the reaction time. Mix liquid samples from this
analyzed
reactor based
were on standard
immediately methods
filtered [23].0.45
through Theµm
two-point calibration
filter paper methodaccording
and analyzed was applied
to
for ORP methods
standard sensors calibration,
[23]. using the standard solution of 86 mV and 256 mV.

Figure2.2.Schematic
Figure Schematicdiagram
diagramofofthe
theexperimental
experimentalsetup
setupfor
forthe
thestudy
studyofofORP
ORPcorrelation
correlationwith
withwater
water
quality.
quality.

2.3. Sensor
ORP and Water Quality
stability Correlation
tests were Investigation
also carried out in theMethod
same reactor to investigate the time
required
Thefor the portable
purpose ORP sensor
of the water qualitytoanalysis
stabilize its to
was readings.
study the ORP and DO sensors
correlations betweenwere
ORP
applied to monitor the mixed liquid in the reactor where primary settled
values and other water quality variables in an anoxic environment and obtain sufficient raw wastewater
were
datafilled. The sensor
for analysis. Batchreadings
tests werewere recorded
carried out every 30 s.
in a completed mixed reactor, and the vol-
Moreover, to investigate the potential of
umetric ratios of recycled sludge/settled raw wastewaterORP to function
wereas50,an66,
indicating
100, 200,variable
400, and
of600%
the end
to simulate the external recycling rate of the activated sludge system. DO,nitrogen
of denitrification reaction in anoxic environment of the biological pH, and
removal
ORP were process, we investigated
continuously measured the
bycorrelation between
sensors during a low nitrate
the reaction time.concentration 0020
Mix liquid samples
(0–3
from mg/L) and ORP
this reactor under
were anoxic conditions.
immediately Biocarriers
filtered through 0.45 μmfrom a MBBR
filter paper system were
and analyzed
added into the reactor with a filling
according to standard methods [23]. rate of 50% to perform the denitrification reaction. The
nitrified liquid was mixed with the primary settled raw wastewater to achieve
Sensor stability tests were also carried out in the same reactor to investigate the time an initial
nitrate
requiredconcentration of approximately
for the portable ORP sensor to3stabilize
mg/L. ORP and DO ORP
its readings. sensors
andwere appliedwere
DO sensors for
data acquisition, and sodium acetate was added to provide sufficient carbon source for the
applied to monitor the mixed liquid in the reactor where primary settled raw wastewater
denitrification. The sensor readings were recorded every 1 min, and water samples were
were filled. The sensor readings were recorded every 30 s.
taken into measurement of nitrate and COD with the same sampling frequency.
Moreover, to investigate the potential of ORP to function as an indicating variable of
the end of denitrification
2.4. Statistical Methods reaction in anoxic environment of the biological nitrogen re-
moval process, we investigated the correlation between a low nitrate concentration 0020
Two multivariate statistical methods, i.e., principal component analysis (PCA), and
(0–3 mg/L) and ORP under anoxic conditions. Biocarriers from a MBBR system were
multiple regression, were used to analyze the experimental results in R software. The
added into the reactor with a filling rate of 50% to perform the denitrification reaction.
principle and implementation procedures of ANOVA were conducted according to Far-
The nitrified liquid was mixed with the primary settled raw wastewater to achieve an
away et al. [24]. PCA was used to analyze the collinearity and correlation between different
initial nitrate concentration of approximately 3 mg/L. ORP and DO sensors were applied
water quality variables. Previous researchers have studied how to use PCA for math-
for data acquisition, and sodium acetate was added to provide sufficient carbon source
ematical analysis of wastewater treatment processes [25]. The study of PCA results is
for the denitrification. The sensor readings were recorded every 1 min, and water samples
mainly presented by two charts, i.e., cumulative variance and loading plot. The cumulative
were taken
variance plotinto measurement
shows of nitrate
the proportion andvariance
of total COD with thedifferent
that same sampling
PCs canfrequency.
interpreted,
while the load plot is used to observe the correlation between different variables and group
2.4. Statistical Methods
them. If a small number of PCs representing the majority of variation in a large number of
original Two multivariate
variables, statistical
the original methods,
variables i.e., principal
are considered componentand
as collinearity analysis
can be(PCA), and
screened
tomultiple
simplifyregression, were used
the mathematical to analyze
model. the experimental
The relationship between results
ORP in R software.
and The prin-
multiple variables
ciple
was and implementation
established by applyingprocedures of ANOVA
multiple regression werebased
analysis conducted
on theaccording to method.
least square Faraway
et al. [24]. PCA was used to analyze the collinearity and correlation between different
The details of the principle and applications of the least square estimation-based multiple wa-
ter qualitywas
regression variables. Previous researchers
well-demonstrated have studied
in the literature [26]. how to use PCA for mathematical
analysis of wastewater treatment processes [25]. The study of PCA results is mainly pre-
sented by two charts, i.e., cumulative variance and loading plot. The cumulative variance
Processes 2022, 10, 2607 5 of 11

3. Results and Discussion

3.1. Variation in ORP during Denitrification under Anoxic Conditions
Figure 3 shows the variation in the ORP profile and various water quality indicators
during the denitrification under anoxic conditions. At the beginning of the experiment, the
ORP value of the system was 42 mV. In the first 20 min of the experiment, the ORP value
declined sharply, ranging from 42 mV to −160 mV. The ORP value decreased slightly
Processes 2022, 10, x FOR PEER REVIEW
and
6 of 11
entered the plateau stage thereafter. At this stage, the ORP value changed from −168 mV
to −191 mV and from 25 to 105 min. The ORP value decreased from −191 mV to −228 mV
after 105 min. Obviously, an ORP breakpoint appeared, which has been studied by many
sharp decrease
researchers in ORP
[27,28], andcoupled with negative
the breakpoint ORP the
may affect values
datamay perform
logging as an indicator
for real-time of
process
nitrate exhaustion [4,31].
control significantly.

Figure
Figure 3.
3. Variation in the
Variation in theORP
ORPvalue
valuetogether
togetherwith
withwater
waterquality
qualityvariables.
variables.(a)(a) DO;
DO; (b)(b)
pH;pH;
(c)(c)
NH NH 4-
4 -N;
N; (d) NO 3-N; (e) chemical oxygen demand (COD); (f) PO4-P.
(d) NO3 -N; (e) chemical oxygen demand (COD); (f) PO4 -P.

3.2. Correlation
DO decreased Analysis
to 0.1between ORP and
mg/L within Water
5 min Quality
of the Variables
experiment and subsequently remained
at a lower level of <0.1 mg/L. The mixture contained a large
PCA was used to analyze the collinearity and correlation between amount of DO at the
water beginning
quality var-
of the reaction. The initial phase of the anoxic reaction process
iables and ORP. Principal components (PCs) were selected according to the principle consumed DO fromthat
the
raweigenvalue
the water, andofdenitrification
PCs should beusing nitrate
greater than as the electron
1. Figure acceptor
4a presents the resulted
scree plotinshowing
a sharp
decline
the in the ORP
eigenvalue value
of each PC,[29]. The indicated
which ORP sensor obtained
that the firsta three
high reading
PCs are from the protective
qualified to repre-
fluidthe
sent andcharacteristics
required a certain
of the time to stabilize
original data framethe (ORP
reading,andwhich
water also
qualityledvariables).
to a deviation
The
between the reading and the actual values. The consumption of alkalinity
eigenvalue, variance, and cumulative variance contribution rates of each PC are shown in due to oxygen
utilization
Table 1, andled
thetofactor
a negligible
loads of pH decrease.
each PC are Then,
shownthe in system
Table 2.entered
The firstthe
PChypoxic state and
(PC1) explained
33.27% of the total variance of the data, and the first three PCs explained 77.03%pH;
commenced denitrification to produce alkalinity. This led to a gradual increase in the
of the
pH stabilized at 7.50 after approximately 60 min of reaction
total cumulative variance. There was strong collinearity among different time. The NH 4 -N concentration
water quality
was always
indices, and maintained
three PCs couldat approximately 24 mg/L
be used to replace during
all seven the entire
original experiment and no
variables.
obvious nitrification
As shown reaction
in Table occurred
2, DO, ORP, and inPOthe anoxic condition. Moreover, phosphorus-
4-P had large weights on PC1, which indicated
accumulating bacteria released phosphate,
their strong correlations to PC1. PC2 explained 24.61% which increased
of thethe phosphate
variance and concentration
mainly repre-
of the mixed solution. The phosphate concentration increased
sented pH. NH4-N and COD were mainly represented by PC3, which explained to a peak of 7.36 mg/L
19.15%at
40 min after
of the variance.the reaction. Then, the phosphate release basically ceased, and the phosphate
concentration decreased slowly due to the simultaneous removal of phosphate along with
denitrification
Table [1]. variances, and cumulative variances.
1. Eigenvalues,

Variance Contribution Rate Cumulative Variance

Principal Component Eigenvalue
(%) (%)
1 2.33 33.27 33.27
2 1.72 24.61 57.88
 Processes 2022, 10, 2607 6 of 11

Nitrate and COD decreased synchronously during the whole reaction process, which
proved that the system had undergone an obvious denitrification reaction. The denitrifi-
cation reaction was not obvious in the first 5 min of the reaction because the DO brought
by raw streams inhibited the occurrence of denitrification. Denitrification happened after
oxygen depletion and the nitrate and COD started decreasing simultaneously. Theoreti-
Processes 2022, 10, x FOR PEER REVIEW 7 of 11
cally, denitrifying 1 g of nitrate requires 2.86 g of COD, based on stoichiometry. However,
considering microbial assimilation in the process, the theoretical carbon/nitrogen ratio is
approximately 3.7. In the practice of wastewater biological treatment process operation,
4 g ORP
of COD is dosed to 0.57371
remove 1 g of NO3 -N as the carbon source. In the lab-scale
−0.27286 study,
−0.07986
12 mg/L of NO3 -N was removed and approximately 50 mg/L of COD was consumed,
pH 8.84462 × 10−4 0.66654 −0.20011
which was in accordance with the theoretical value. Nitrate decreased significantly after
NH 4-N −0.15559 0.48268 0.54411
the ORP entered its plateau (Figure 3d), and the ORP value decreased significantly when
thePO 4-P
nitrification reaction−0.50098 −0.23805 reaction of nitrate0.33222
ended, indicating that the respiration or oxygen
the3-system
inNO N 0.40813 0.40849 0.10586
had finished and entered the anaerobic state [30]. This study indicated that
theCOD
progress of denitrification
0.19231process can be reflected by the ORP variation, and 0.67516
−0.14471 the second
sharp decrease in ORP coupled with negative ORP values may perform as an indicator of
nitrate exhaustion [4,31].
Figure 4b shows the loadings of at the PC1 and PC2 panels. The load vector of COD
was3.2.
obviously
Correlationsmaller
Analysisthan thatORP
between of other original
and Water variables,
Quality Variableswhich indicated that its influ-
ence onPCAthe PCA results was obviously smaller than that of other
was used to analyze the collinearity and correlation variables;
between watertherefore,
quality the
influence
variablesofand
COD could
ORP. be ignored.
Principal componentsAt the PC1
(PCs) level,
were PO4-P,
selected and NH
according -N principle
to 4the had negative
weights, whereas DO, ORP, and NO -N had positive weights. The
that the eigenvalue of PCs should be greater than 1. Figure 4a presents the scree
3 weight of pHplot
was al-
most zero, the
showing indicating
eigenvaluethatof pH
eachcontributes merely that
PC, which indicated to PC1 andthree
the first the pH
PCsvariation could
are qualified to not
represent the characteristics of the original data frame (ORP and
be explained by PC1. Regarding PC2, pH, NH4-N, and NO3-N had positive weights;water quality variables).
The eigenvalue,
whereas, variance,
other original and cumulative
variables varianceweights.
had negative contribution
DOratesandofORP,
each PC
whichare shown
were close
in Table 1, and the factor loads of each PC are shown in Table 2. The first PC (PC1) explained
to each other in the biplot (Figure 4b), indicating a similarity of variation and collinearity.
33.27% of the total variance of the data, and the first three PCs explained 77.03% of the total
The preliminary results of PCA showed that DO, NO3-N, and PO4-P are the most signifi-
cumulative variance. There was strong collinearity among different water quality indices,
cant variables
and three PCscorrelated withtoORP
could be used variation.
replace all seven original variables.

Figure 4. 4.
Figure Principal
Principalcomponent
component analysis (PCA).(a)
analysis (PCA). (a)Scree
Scree plot.
plot. (b)(b) Biplot
Biplot on plane
on the the plane of principle
of principle
components 1 (PC1) and 2 (PC2); the blue vector represents the loadings of each original variables
components 1 (PC1) and 2 (PC2); the blue vector represents the loadings of each original variables
andand
thethe
redred
dots are PCA scores.
dots are PCA scores.
As shown in Table 2, DO, ORP, and PO4 -P had large weights on PC1, which indicated
3.3.their
Quantitative Correlation
strong correlations to Analysis
PC1. PC2ofexplained
ORP and24.61%
Water of
Quality Variation
the variance and mainly repre-
The anoxic
sented pH. NHzones ofCOD
4 -N and bioreactors at WWTPs
were mainly are adopted
represented to perform
by PC3, which denitrification
explained 19.15% of for
the variance.
nitrogen removal from wastewater. Therefore, it is of great practical significance to estab-
lish the relationship between the ORP value and nitrate for process surveillance purpose.
Hence, nitrate should be included as one of the explanatory variables for the ORP quanti-
tative model. Considering the difficulty of real-time phosphate monitoring in practice,
phosphate was screened out from the list of explanatory variables. COD was neither an
easy-to-measure variables nor a significant essential variable to ORP variation, COD was
Processes 2022, 10, 2607 7 of 11

Table 1. Eigenvalues, variances, and cumulative variances.

Principal Variance Contribution Cumulative Variance

Eigenvalue
Component Rate (%) (%)
1 2.33 33.27 33.27
2 1.72 24.61 57.88
3 1.34 19.15 77.03

Table 2. Principal component (PC) of each variable.

Indicator PC1 PC2 PC3

DO 0.4383 −0.06182 0.28301
ORP 0.57371 −0.27286 −0.07986
pH 8.84462 × 10−4 0.66654 −0.20011
NH4 -N −0.15559 0.48268 0.54411
PO4 -P −0.50098 −0.23805 0.33222
NO3 -N 0.40813 0.40849 0.10586
COD 0.19231 −0.14471 0.67516

Figure 4b shows the loadings of at the PC1 and PC2 panels. The load vector of
COD was obviously smaller than that of other original variables, which indicated that its
influence on the PCA results was obviously smaller than that of other variables; therefore,
the influence of COD could be ignored. At the PC1 level, PO4 -P, and NH4 -N had negative
weights, whereas DO, ORP, and NO3 -N had positive weights. The weight of pH was
almost zero, indicating that pH contributes merely to PC1 and the pH variation could
not be explained by PC1. Regarding PC2, pH, NH4 -N, and NO3 -N had positive weights;
whereas, other original variables had negative weights. DO and ORP, which were close to
each other in the biplot (Figure 4b), indicating a similarity of variation and collinearity. The
preliminary results of PCA showed that DO, NO3 -N, and PO4 -P are the most significant
variables correlated with ORP variation.

3.3. Quantitative Correlation Analysis of ORP and Water Quality Variation

The anoxic zones of bioreactors at WWTPs are adopted to perform denitrification for
nitrogen removal from wastewater. Therefore, it is of great practical significance to establish
the relationship between the ORP value and nitrate for process surveillance purpose. Hence,
nitrate should be included as one of the explanatory variables for the ORP quantitative
model. Considering the difficulty of real-time phosphate monitoring in practice, phosphate
was screened out from the list of explanatory variables. COD was neither an easy-to-
measure variables nor a significant essential variable to ORP variation, COD was also
excluded for ORP quantification. Thus, three original variables were selected to establish
the quantitative model for ORP quantitative analysis (Equation (1)). The adjected R2 of
this fitted model was 0.7195. The overfitted model with all the water quality variables
are also constructed (Equation (2)), which was slightly better in terms of or adjusted R2
(0.7592). The fitting results showed a close relationship between the change in the ORP
value and nitrate removal during denitrification. The ORP value was positively correlated
with DO, nitrate, and COD, and negatively correlated with pH, phosphate, and NH4 -N. The
coefficient before DO in the model implied that a small fluctuation in the DO concentration
may result in a large variation in the ORP value under anoxic conditions. Although NH4 -N
cannot be removed under anoxic conditions, the NH4 -N concentration had a significant
effect on the ORP value fluctuation.
Equation (1) is the simplified model with easy-to-measure variables and overfitting
was avoided, which make the model to be valid in most practical cases for the interpretation
of ORP and water quality variation in anoxic denitrification process.

ORP = −63.92 + 65.57·DO + 8.93·NO3 –N − 6.20·NH4 –N (1)

 NH4-N cannot be removed under anoxic conditions, the NH4-N concentration had a sig-
nificant effect on the ORP value fluctuation.
Processes 2022, 10, 2607 Equation (1) is the simplified model with easy-to-measure variables and overfitting
8 of 11
was avoided, which make the model to be valid in most practical cases for the interpreta-
tion of ORP and water quality variation in anoxic denitrification process.
ORP = −63.92 + 65.57 ∙ DO + 8.93 ∙ NO – N − 6.20 ∙ NH – N (1)
ORP = 297.32 + 0.22 ∙ COD + 57.89 ∙ DO + 6.07 ∙ NO – N − 4.70 ∙ NH – N − 47.71 ∙ pH − 7.53 ∙ PO – P (2)
ORP = 297.32 + 0.22·COD + 57.89·DO + 6.07·NO3 –N − 4.70·NH4 –N − 47.71·pH − 7.53·PO4 –P (2)
ORP online monitoring sensors are reliable in terms of accuracy, drifting, and anom-
ality. ORP
However,
onlinethe most commonly
monitoring used
sensors are nitrate/ammonium
reliable sensorsdrifting,
in terms of accuracy, in wastewater bio-
and anomal-
logical treatment
ity. However, the system are onlineused
most commonly monitoring.
nitrate/ammonium sensors in wastewater biologi-
But the reliability
cal treatment system are of online
a nitrate online monitoring instrument can be verified theoreti-
monitoring.
cally But
according to the model; alternatively, the ORP
the reliability of a nitrate online monitoring value obtained
instrument by the ORP
can be verified online
theoretically
monitoring instrument can be used to substitute the measurement of the nitrate
according to the model; alternatively, the ORP value obtained by the ORP online monitoring concen-
tration in thecan
instrument anoxic zone.
be used to substitute the measurement of the nitrate concentration in the
anoxic zone.
3.4. Analysis of Stability Time of ORP Sensor Reading
3.4. Analysis
The ORPofsensor
Stability Time of locate
readings ORP Sensor
withinReading
the range of −220 mV to 300 mV when placed
The ORP
the sensor sensor readings
in wastewater locateprocesses.
treatment within theTherange of −220
correct mV toof300
reading themV when placed
measured data
thethe
by sensor in is
sensor wastewater
critical for treatment processes.
data application. The correct
As shown reading
in Figure of and
5, DO the measured
the ORP value data
by therecorded
were sensor isevery
critical
30for
s bydata application.
submerging theAs shown
sensors inin Figure
raw 5, DO and
wastewater. theORP
The ORPsensor
value
were recorded
reading was 93 every 30 s byand
mV initially submerging
droppedthe sensors in at
dramatically rawthewastewater. The itORP
first 100 s, and sensor
stabilized
reading was 93 mV initially and dropped dramatically at the first 100 s, and
to a negative value after 400 s. The continuous decline in the ORP value was related to the it stabilized to
a negative value after 400 s. The continuous decline in the ORP value
gradual decline in the DO concentration in the reactor. The slope of the declining ORP was related to the
gradual
curve declinecontinuously
changed in the DO concentration
before 360 ins. the
Thereactor.
purpleThelineslope of the5declining
in Figure started toORP curve
collinear
changed continuously before 360 s. The purple line in Figure 5 started
with ORP from the 360th second to 600th second, which indicates a first-order linear to collinear with
ORP fromofthe
changing ORP360th second
versus DO to 600th second,
changing which
after ORP indicates
values reached a first-order
minus. linear changing
of ORP versus DO changing after ORP values reached minus.

Figure
Figure 5.
5. The
The DO and ORP
DO and ORP profiles
profilesofofduring
duringa atest
test submerging
submerging thethe sensors
sensors intointo
rawraw wastewater.
wastewater. The
The
blueblue
curvecurve is the
is the DODO profile,
profile, andand
the the
redred curve
curve represents
represents thethe
ORP ORP variation
variation during
during denitrifica-
denitrification.
tion. The purple
The purple line
line is theistangent
the tangent
to thetoORP
the ORP
curvecurve at 600th
at 600th second.
second.

3.5. Correlation Analysis of Low Nitrate Concentration and ORP

Figure 6 shows the variation trends of nitrate and ORP values during the denitrification
process initialized with low nitrate concentration. By mixing the effluent from the primary
settling tank with the recycled nitrified solution, the nitrate concentration was 2.2 mg/L
at the beginning of the reaction when the corresponding ORP value was −59 mV. The
nitrate concentration decreased by approximately 0.2 mg/L after 1 min of reaction. Nitrate
was exhausted when the reaction continued for 13 min. In this test, the ORP breakpoint
 3.5. Correlation Analysis of Low Nitrate Concentration and ORP
Figure 6 shows the variation trends of nitrate and ORP values during the denitrifica-
tion process initialized with low nitrate concentration. By mixing the effluent from the
primary settling tank with the recycled nitrified solution, the nitrate concentration was 2.2
Processes 2022, 10, 2607 mg/L at the beginning of the reaction when the corresponding ORP value was −599 of mV.11
The nitrate concentration decreased by approximately 0.2 mg/L after 1 min of reaction.
Nitrate was exhausted when the reaction continued for 13 min. In this test, the ORP break-
point appeared
appeared and theand the corresponding
corresponding ORP valueORP value
was was
−138 mV.−138 mV.
It was It was consistent
consistent with
with the ORP
the
of ORP ofexperiments
previous previous experiments in ORP
in which the whichvalue
the ORP value significantly
decreased decreased significantly when
when the nitrate
the nitrate concentration
concentration wasproving
was depleted, depleted,
thatproving
the ORPthat the can
value ORPplay
value
ancan play an
indicator indicator
role in the
role inconcentration
nitrate the nitrate concentration
depletion. depletion.

Figure6.6.Variation
Figure Variationtrends
trendsof
ofthe
thenitrate
nitrateconcentration
concentrationand
andORP
ORPvalue.
value.The
Thehistogram
histogramrepresents
representsthe
the
nitrate concentration, and the blue curve represents the ORP value change.
nitrate concentration, and the blue curve represents the ORP value change.

Generally,
Generally,aacarbon
carbonsource
sourceisisdosed
dosedtotothe
theanoxic
anoxiczones
zonesof ofthe
thebiological
biologicalwastewater
wastewater
treatment processes to ensure the denitrification effect. If the relationship
treatment processes ensure the denitrification effect. If the relationship between between
a lowa
low nitrate
nitrate concentration,
concentration, COD,COD,
andandthe the
ORPORP value
value cancan
be be established,
established, then
then thethe amount
amount of
of carbon
carbon source
source dosage
dosage cancan be determined.
be determined. A regression
A regression modelmodel fitted on
fitted based based on the
the denitri-
denitrification withnitrate
fication with low low nitrate
loadingloading was presented
was presented as Equation
as Equation (3). (3).
ORP = −83.3944 − 1.0221 ∙ COD + 41.4904 ∙ NO -N (3)
ORP = −83.3944 − 1.0221·COD + 41.4904·NO3 –N (3)
The adjusted R was 0.5737 for the model in Equation (3). The established model can
2

be rewritten according
The adjusted R2 was to0.5737
Equation (4): model in Equation (3). The established model can
for the
be rewritten according to Equation 1
NO -N(4):
+ 5e + 6H = N + 3H O (4)
2
5 mol e− is needed for denitrification to remove1 of 1 mol nitrate nitrogen, i.e., 2.5 mol
NO –N + 5e− + 6H+ = N + 3H2 O (4)
of COD is needed to provide3 electrons. Therefore,2the2 amount of carbon source required
is 2.5 × 16/14 = 2.86 g COD/g nitrogen, based on COD. According to the activated sludge
− is needed for denitrification to remove of 1 mol nitrate nitrogen, i.e., 2.5 mol
5 mol
model, thee sludge yield coefficient YH = 0.67, i.e., the proportion of COD assimilated by
of COD is needed
microorganisms to theto provide
total electrons.
COD intake, Therefore,
was 67%.theTherefore,
amount of the
carbon
totalsource
COD required
demand is of
2.5 × 16/14 = 2.86 g COD/g nitrogen, based on COD. According to the activated sludge
microorganisms with denitrification of 1 mol nitrate was 2.86/(1 − 0.67) = 8.67 g COD/g
model, the sludge yield coefficient YH = 0.67, i.e., the proportion of COD assimilated by
nitrogen. The nitrogen consumed by assimilation was 0.07 g nitrogen/g COD, and the ni-
microorganisms to the total COD intake, was 67%. Therefore, the total COD demand of
trogen removed by microbial assimilation was 8.67 × 0.67 × 0.07 = 0.41 g nitrogen. There-
microorganisms with denitrification of 1 mol nitrate was 2.86/(1 − 0.67) = 8.67 g COD/g
fore, (1 + 0.41) g nitrogen was removed, when a total of 8.37 g COD was removed. The
nitrogen. The nitrogen consumed by assimilation was 0.07 g nitrogen/g COD, and the
COD required to remove 1 g nitrogen was 8.67/(1 + 0.41) = 6.15 g COD/g nitrogen, or 1.39
nitrogen removed by microbial assimilation was 8.67 × 0.67 × 0.07 = 0.41 g nitrogen.
g COD/g NO3-N. The model formula was rewritten according to the above derivation as
Therefore, (1 + 0.41) g nitrogen was removed, when a total of 8.37 g COD was removed.
Equation (5):
The COD required to remove 1 g nitrogen was 8.67/(1 + 0.41) = 6.15 g COD/g nitrogen, or
COD = 1.4721 + 0.0177ORP. (5)
1.39 g COD/g NO3 -N. The model formula was rewritten according to the above derivation
as Equation (5):
COD = 1.4721 + 0.0177ORP. (5)
The unit for COD is mg/L and that for ORP is mV. The relationship between COD
and ORP was established, which can be applied to carbon source dosing control systems.
The ORP online monitoring sensors can be used to determine the COD requirement as well
as the carbon dosage.
Processes 2022, 10, 2607 10 of 11

4. Conclusions
The relationships between ORP values and various water quality variables in wastew-
ater under different reaction environments were studied, and correlations between low
nitrate concentrations and ORP under anoxic conditions were investigated. The conclusions
are as follows:
(1) Under anoxic conditions, the ORP value exhibited a plateau period (during which
the denitrification rate decreased significantly) and then decreased significantly at
the end of the denitrification reaction. Under anoxic conditions, ORP was positively
correlated with DO, nitrate, and COD, and negatively correlated with phosphate,
NH4 -N, and pH;
(2) PCA showed strong collinearity among different water quality variables under anoxic
conditions and that three PCs can be used to replace all the original variables. The
change in the ORP value was the result of the combined influences of multiple factors.
However, changes in the COD and ORP values were not significant;
(3) The ORP value correlated with NH4 -N, nitrate, and DO, and the R2 value was 0.7195.
In theory, the ORP online sensors can assist in verifying the accuracy of the nitrate
online sensors or serve as its substitute;
(4) ORP sensors used for water sample measurements require a period of stability to
approach the true value;
(5) Under the conditions of a low nitrate concentration (3 mg/L), a direct relationship
between COD and ORP was obtained according to a series of transformations. The
results can be used for process control in carbon source dosing for nitrogen removal
at WWTPs.
Overall, the mechanism of ORP response to water quality variation was investigated
which may provide insights for online monitoring of anoxic process using ORP sensors.
The correlation models can serve as tools for COD prediction at anoxic environment and
carbon dosing control for denitrification in future applications.

Author Contributions: Conceptualization, X.W; methodology, Y.W. and X.W.; formal analysis, Y.W.
and H.P.; investigation, D.S., resources, Y.W. and X.W.; writing—original draft preparation, Y.W.;
writing—review and editing, X.W. and N.C.; project administration, Y.W.; supervision, H.R. and Z.M.;
validation, X.B.; funding acquisition, X.W. All authors have read and agreed to the published version
of the manuscript.
Funding: This work was funded by the National Natural Science Foundation of China (grant number
51908303).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data will be made available on request.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Wang, Y. The Mechanism and Processes Study on Denitrifying Phosphorus Removal. Ph.D. Thesis, Harbin Institute of Technology,
Harbin, China, June 2004.
2. Wen, D. Research progress and related applications of REDOX potential. Light Ind. Sci. Technol. 2017, 7, 101–103.
3. Peddie, C.C.; Mavinic, D.S.; Jenkins, C.J. Use of ORP for monitoring and control of aerobic sludge digestion. J. Environ. Eng. 1990,
116, 461–471. [CrossRef]
4. Koch, F.; Oldham, W. Oxidation-reduction potential—A tool for monitoring, control and optimization of biological nutrient
removal systems. Water Sci. Technol. 1985, 17, 259–281. [CrossRef]
5. Suslow, T.V. Oxidation-Reduction Potential (ORP) for Water Disinfection Monitoring, Control, and Documentation. University of
California, Berkeley, CA, USA, 2004. Available online: https://escholarship.org/uc/item/1730p498 (accessed on 1 November
2022).
6. Yu, R.F.; Chen, H.W.; Cheng, W.P.; Shen, Y.C. Dynamic control of disinfection for wastewater reuse applying ORP/pH monitoring
and artificial neural networks. Resour. Conserv. Recycl. 2008, 52, 1015–1021. [CrossRef]
Processes 2022, 10, 2607 11 of 11

7. Yu, R.F.; Cheng, W.P. Determination of chlorine demand in water and wastewater chlorination by oxidation-reduction potential.
Water Sci Technol. Water Supply 2003, 3, 313–320. [CrossRef]
8. Yu, R.F. Feed-forward dose control of wastewater chlorination using on-line pH and ORP titration. Chemosphere 2004, 56, 973–980.
[CrossRef]
9. Olsson, G.; Carlsson, B.; Comas, J.; Copp, J.; Gernaey, K.V.; Ingildsen, P.; Jeppsson, U.; Kim, C.; Rieger, L.; Rodriguez-Roda, I.; et al.
Instrumentation, control and automation in wastewater—From London 1973 to Narbonne 2013. Water Sci. Technol. 2014, 69,
1373–1385. [CrossRef]
10. Aguado, D.; Ribes, J.; Montoya, T.; Ferrer, J.; Seco, A. A methodology for sequencing batch reactor identification with artificial
neural networks: A case study. Comput. Chem. Eng. 2009, 33, 465–472. [CrossRef]
11. Zanetti, L.; Frison, N.; Nota, E.; Tomizioli, M.; Bolzonella, D.; Fatone, F. Progress in real-time control applied to biological nitrogen
removal from wastewater. A short-review. Desalination 2012, 286, 1–7. [CrossRef]
12. Kaelin, D.; Rieger, L.; Eugster, J.; Rottermann, K.; Bänninger, C.; Siegrist, H. Potential of in-situ sensors with ion-selective
electrodes for aeration control at wastewater treatment plants. Water Sci. Technol. 2008, 58, 629–637. [CrossRef]
13. Ohmura, K.; Thürlimann, C.M.; Kipf, M.; Carbajal, J.P.; Villez, K. Characterizing long-term wear and tear of ion-selective pH
sensors. Water Sci. Technol. 2019, 80, 541–550. [CrossRef]
14. Wang, H.; Fa, K. REDOX potential as a control parameter of biological phosphorus and nitrogen removal process. China Water
Wastewater 1998, 3, 285–290.
15. Su, G. Study on Characteristics of REDOX Potential in Organic Wastewater Treatment Engineering. Doctoral Dissertation,
Zhengzhou University, Zhengzhou, China, 2009.
16. Azis, K.; Ntougias, S.; Melidis, P. NH4 + -N versus pH and ORP versus NO3 − -N sensors during online monitoring of an
intermittently aerated and fed membrane bioreactor. Environ. Sci. Poll. Res. 2020, 28, 33837–33843. [CrossRef]
17. Khanal, S.K.; Huang, J.C. ORP-based oxygenation for sulfide control in anaerobic treatment of high-sulfate wastewater. Water Res.
2003, 37, 2053–2062. [CrossRef]
18. Li, S.; Mu, J.; Du, Y.; Wu, Z. Study and application of real-time control strategy based on DO and ORP in nitritation–denitrification
SBR start-up. Environ. Technol. 2019, 42, 114–125. [CrossRef]
19. Alattabi, A.W.; Harris, C.; Alkhaddar, R.; Alzeyadi, A.; Abdulredha, M. Online monitoring of a sequencing batch reactor treating
domestic wastewater. Procedia Eng. 2017, 196, 800–807. [CrossRef]
20. Fox, S.; Clifford, E. Detecting the end of nitrification in small and decentralized wastewater treatment systems using low-resource
real-time control methods. J. Environ. Eng. 2018, 144, 04018069. [CrossRef]
21. Weißbach, M.; Drewes, J.E.; Koch, K. Application of the oxidation reduction potential (ORP) for process control and monitoring
nitrite in a Coupled Aerobic-anoxic Nitrous Decomposition Operation (CANDO). Chem. Eng. J. 2018, 343, 484–491. [CrossRef]
22. Schneider, M.Y. Monitoring and Quantifying the Treatment Performance of On-Site Wastewater Treatment Plants; ETH Zurich: Zurich,
Switzerland, 2020.
23. APHA; AWWA; WEF. Standard Methods for the Examination of Water and Wastewater, 22nd ed.; American Public Health Association:
Washington, DC, USA, 2012.
24. Faraway, J.J. Practical Regression and ANOVA Using R; University of Bath: Bath, UK, 2002; Volume 168.
25. Wang, X.; Ratnaweera, H.; Holm, J.A.; Olsbu, V. Statistical monitoring and dynamic simulation of a wastewater treatment plant:
A combined approach to achieve model predictive control. J. Environ. Manag. 2017, 193, 1–7. [CrossRef]
26. Zou, L. The principle of least square method and its simple application. Sci. Technol. Inform. 2010, 23, 282–283.
27. Wareham, D.G.; Hall, K.J.; Mavinic, D.S. Real-time control of wastewater treatment systems using ORP. Water Sci. Technol. 1993,
28, 273–282. [CrossRef]
28. Takigawa, A. Intercross real-time control strategy in alternating activated sludge process for short-cut biological nitrogen removal
treating domestic wastewater. J. Environ. Sci. 2008, 20, 957–963.
29. Vega, P.; Salazar, E.M.D.; Jaramillo, M.A.; Cros, J. New contributions to the orp & do time profile characterization to improve
biological nutrient removal. Bioresour. Technol. 2012, 114, 160–167.
30. Chen, K.C.; Chen, C.Y.; Peng, J.W.; Houng, J.Y. Real-time control of an immobilized-cell reactor for wastewater treatment using
ORP. Water Res. 2002, 36, 230–238. [CrossRef]
31. Wouters-Wasiak, K.; Heduit, A.; Audic, J.M.; Lefevre, F. Real-time control of nitrogen removal at full-scale using oxidation
reduction potential. Water Sci. Technol. 1994, 30, 207. [CrossRef]