(PDF) Modelling of Coagulant Dosage in A Water Treatment Plant

10/7/2020 (PDF) Modelling of coagulant dosage in a water treatment plant
See all › See all › Recruit

Download citation Share researchers Join full-text
Download for free PDFLogin
49 Citations 15 References
coagulant dosage in a water treatment plant
Available) in Artificial Intelligence in Engineering 11(4):401-404 · October 1997 with 544 Reads

6/S0954-1810(97)00010-1 · Source: dx.doi.org
publication
Jules Thibault
Claude Gagnon Bernard P.A. Grandjean
40.28 · University of Ottawa
agulation-flocculation is a major step in the drinkable water treatment process allowing the removal of colloidal particles. The water treatment
s of the City of Sainte-Foy have been well instrumented and process variables such as temperature, pH, turbidity, conductivity of raw and
water along with actual coagulant dosage available have been measured and stored each 5 min for several years. Using such a data bank,
ective of this paper is to report on the development of a neural network predictor of coagulant dosage in order to facilitate process operation.
ward neural models have been built using a quasi-Newton method along with the early stopping approach to avoid overfitting. Annual and
al models have been built and their performances are discussed.
er the world's research
million members
+ million publications
k+ research projects
oin for free
nt
uploaded by Jules Thibault Author content Download full-text PDF

may be subject to copyright.
MODELLING OF COAGULANT DOSAGE IN A WATER TREATMENT PLANT

N. Valentin(1,2), T. Denoeux(2), F. Fotoohi(1)
(1) CITI - Suez Lyonnaise des Eaux, Technopolis, ZAC de Mercières, 14, rue du Fonds Pernant, 60471
Compiègne, France
Email: Nicolas.Valentin, Farrokh.Fotoohi @citi.suez-lyonnaise-eaux.fr
(2) Heudiasyc, UMR CNRS 6599, Université de Technologie de Compiègne, BP 20529, 60205
Compiègne, FRANCE
Email: Thierry.Denoeux@hds.utc.fr
Abstract: Artificial Neural Network (ANN) techniques are applied to the control of
coagulant dosing in a drinking water treatment plant. Coagulant dosing rate is non-linearly
correlated to raw water parameters such as turbidity, conductivity, pH, temperature, etc. An
important requirement of the application is robustness of the system against erroneous
sensor measurements or unusual water characteristics. The hybrid system developed
includes raw data validation and reconstruction based on a Kohonen self-organizing feature
map, and prediction of coagulant dosage using multilayer perceptrons. A key feature of the
system is its ability to take into account various sources of uncertainty, such as atypical
input data, measurement errors and limited information content of the training set.
Experimental results with real data are presented.
I INTRODUCTION
https://www.researchgate.net/publication/222981343_Modelling_of_coagulant_dosage_in_a_water_treatment_plant 1/11
See all ›
I. INTRODUCTION
See all ›
The water
49 Citations industry is facing increased pressure to produce higher quality treated water at a Download
15 References lower cost.
citation Share Download full-text PDF
The coagulation-flocculation process is a major step in the production of potable water, allowing the
removal of colloidal particles. The main difficulty is to determine the optimum coagulant dosage related
to the influent of raw water. Excessive coagulant overdosing leads to increased treatment costs and public
health concerns, while underdosing leads to a failure to meet the water quality targets and less efficient
operation of the water treatment plant. For the moment, both manual and automatic methods are available
to predict optimum coagulant dosage rate [1-2]. Manual methods mainly include jar testing. Jar testing
involves taking a raw water sample and applying different quantities of coagulant to each sample. After a
short period of time each sample is assessed for water quality and the dosage that produces the optimal
result is used as a set point. Operators change the dose and make a new jar test if the quality of treated
water changes. Disadvantages associated with jar testing are the necessity to perform manual intervention,
and the limitation to feedback control. Automatic coagulant control is ensured mainly by streaming
current detectors (SCD) [3], which measure the residual charge on colloidal colour and turbidity particles
in the water. As these particle have a negative charge and the coagulant ions have a positive charge, the
amount of coagulant added dictates the magnitude and sign of the net electrical charge. The system
controls this net charge at a set point which has been shown by jar testing to provide close to optimum
coagulation under a certain range of raw water conditions. Disadvantages associated with the SCD are its
operation cost and its lack of adaptation to all types of raw water quality.
This paper addresses the problem of on-line determination of optimal coagulant dosage from raw water
characteristics such as turbidity, pH, conductivity, etc. Some previous studies [4-5] have shown the
potential effectiveness of such an approach based on ANN's. The innovative aspect of this work resides in
the integration of various techniques in a global system including data validation and reconstruction,
modelling of coagulant dosage and analysis of uncertainties. Given the high variability of the inputs and
the low reliability of available sensors, an important requirement of the application is robustness of the
system, running without human supervision, against erroneous sensor measurements or unusual water
characteristics, due to accidental pollution for instance. Special attention has also to be paid to the
automatic design and training of such a system from learning data (including the phases of data validation,
input selection and model choice), which should allow the portability of the system at low cost from one
site to another.
This paper is organized as follows. The water treatment operation is first explained in Section II. The
methodology used to build ANN models for the prediction of coagulant dosage is then described in
Section III. Finally, experimental results are presented and discussed in Section IV.
II . WATER TREATMENT OPERATION
Water treatment involves physical, chemical and biological processes that transform raw water into
drinking water. However, contrary to most industrial processes, for which the quality of the input raw
material is under control, the quality of the given raw water source may fluctuate due to natural
perturbation or occasional pollution. The Viry-Chatillon water treatment plant, which was used as an
application site for this study, provides water to more than 300 000 inhabitants and has a nominal capacity
to process 120 000 m3 of water per day. Figure 1 presents a schematic overview of the various operations
necessary to treat the water, the available measurements, and the coagulant dosing point.
Raw wat er
Conductivity Residual Chlorine
Turbidity Chemical Dosage Sludge Turbidity
pH pH
Temperature Coagulant Drinking wat er Temperature
Dissolved Oxygen Residual Ozone
UV’s a bsorption Total Organic Carbon
GAC GAC
Clarifier Filter Ozone Filter Tank
Drinking
Water
Seine
GAC Filter : Soda
Granular activated Chlorine
See all › See all › carbon filter Chemi cal Dosage

Download citation Share Download full-text PDF
Figure 1 - Simplified synopsis of the water treatment plant.
Raw water is extracted from the river Seine and pumped to the treatment works. The treatment consists in
coagulation-flocculation, settling, filtration, ozonation, filtration and final disinfection. The water is then
stored in a tank and ready to be transported through the water supply network. The coagulation-
flocculation step, which requires the addition of a chemical coagulant, is the critical process to remove
colloidal solids.
III. METHODOLOGY
The system developed was divided in three modules : single-parameter data validation, multi-parameter
data validation and reconstruction, and modelling of coagulant dosage. Figure 2 illustrates the structure of
the system.
Reconstruction
Validation
Water Model of Coagulant

Sensors
treatment coagulant dose

plant dosage
Figure 2 - Structure of the system for the prediction of the coagulant dosage.
3.1 Single-parameter data validation
As the system has to operate on-line with unreliable input data coming from a real process, the data
validation step is fundamental. The objective of this first step, referred to as single-parameter sensor data
validation, is to associate to each raw data item a confidence level. Confidence is measured by a real
number belonging to [0,1]. A zero confidence means "no confidence, the data is invalid" while a
confidence equal to one means "the data is perfectly valid". At this stage, the validation procedure is quite
simple and is based on a comparison of each signal and its derivative to a range of values typically
obtained in the absence of erroneous measurements. Raw data whose confidence level is less than a given
threshold are declared as invalid data. Although this simple approaches proves to be sufficient in most
cases, the detection of inconsistences in the data involving more than one parameter requires the use of
more sophisticated techniques such as Kohonen maps.
3.2 Multi-parameter data validation and reconstruction
We propose an approach based on the use of a Self-Organizing Map (SOM) [6] for multi-parameter data
validation and reconstruction of input data. The process in which the SOM is formed is an unsupervised
learning process. The SOM defines a mapping from the input data space ℜn (raw water quality
parameters) onto a regular two-dimensional array of nodes. A reference vector, or prototype, mi ∈ ℜn is
associated to every node i. Each input vector x ∈ ℜn is compared with the mi, and the best match defines
the winning prototype. The input is then mapped onto the corresponding location on the grid.
With this technique, the evolution of raw water quality can be visualized in two dimensions, and atypical
data or outliers can be detected by measuring the distance between each input vector and its closest
reference vector. For reconstruction, each missing value of a given input variable is estimated by the value
of the corresponding component in the winning prototype.
10 10
9.5 9.5
9 9
Dissolved Ox ygen (mg/l)
Dissolved Ox ygen (mg/l)
8.5 8.5
8 8
7.5 7.5
See all › See all ›

7 7
6.5 6.5
0 50 100 150 200 250 0 50 100 150 200 250
S a m p le S a m p le
Figure 3: Example of data validation (*: invalid data) and reconstruction (thin line: reconstructed
signal) on dissolved oxygen.
Figure 3 illustrates data validation and reconstruction for dissolved oxygen data. At this stage, the pre-
processing phase is terminated and the data is ready to be processed by the coagulant dosage model.
3.3 Modelling of coagulant dosage
1 1
For the modelling of coagulant dosage, a sigmoid
multilayer perceptron (MLP) was trained using the Σ f
Optimal Brain Damage (OBD) [7] learning and Turbidity
pruning algorithm. This approach is based on the Σ f
following general procedure. Conductivity
pH
Coagulant
First, a relatively large network (Figure 4) is trained ……. Dose
Tem perature
using the back-propagation algorithm. The network
is then examined to assess the relative importance Dissolved Ox yg en
of the weights, and the least important are deleted. UV ‘s absor ption
Σ f
The implementation of this technique consists of the

following steps (Figure 5): Σ f
Figure 4: Initial network
1. Choose a relatively large initial network architecture.

2. Train the network using a back-propagation algorithm applied to a sum of squares error function.
3. Compute the second derivatives Hii for each of weights and evaluate the saliencies.
4. Sort the weights by saliency and delete some of the low-saliency weights.
5. Go to step 2 and repeat until some overall stopping criterion is reached.
For practical use, a software system to predict the optimal coagulant dosing rate should not only provide
point estimates but also confidence intervals. Bootstrap sampling [8] was used to generate confidence
intervals for the system outputs [9]. As shown in Figure 6, 50 bootstrap sets of training data were created
from the original data training data by resampling with replacement. These bootstrap training sets were
used to train 50 bootstrap MLP models using the same architecture and training procedure described
previously. Lower and upper limit confidence bounds for any input were obtained by sorting these outputs
and selecting the 10% and 90% cumulative levels. The confidence interval provides upper and lower
limits on the prediction.
Start Data Upper

Bootstrap MLP1
Limit
Bootstrap MLP2 Calculate
OBD Back- confidence
Pruning Propagation
interval
Lower
Bootstrap MLP50 Limit
Yes Stop Learning No
and Pruning Stop
Raw water Coagulant Coagulant

quality Dose
Figure 5 : Learning and pruning algorithm
model (MLP)
Figure 6 : Bootstrap sampling to generate

confidence interval.
IV. RESULTS
The water treatment plant of Viry-Chatillon has been well instrumented for several years. Various process
variables (see Figure 1) such as conductivity, turbidity, pH, temperature, dissolved oxygen, UV absorption
of raw water and coagulant dosage are available. The raw database is made of 100000 measurements of
each variable during a period of 12 month (November 97 – November 98) sampled every 5 minutes.

Many sensor faults were detected and only 1600 measurements of each parameters were available for
training. Figure 7 shows the wide range of raw water conditions that can exist on the treatment plant.
C o n d u c t ivi ty ( µ S )
(N T U )
200 600
150 500
Tu rbid ity
100 400
50 300
0 200
500 1000 1500 500 1000 1500
T e m p e r a t u re (° C )
S a m ple S a m ple
8.5 30
8 20
pH
7.5 10
7 0
500 1000 1500 500 1000 1500
S a m ple S a m ple
O xygen (m g/l)
15 20
(d o/m )
15
10 10
U V
5 0
500 1000 1500 500 1000 1500
S a m ple S a m ple
Figure 7 : Raw Water Data
We used 70% of the data set to develop the model, find the best structure of the ANN and estimate the
prediction accuracy by bootstrap. The other part of the data set was used to validate the resulting model.
The prediction accuracy and confidence interval of the ANN are shown in Figure 8 on the validation set.
6 6
5.5 5.5
5 5
4.5 4.5
Predicted Co agulant Dose (ppm )
Coagulant Dos e (ppm)
4 4
3.5 3.5
3 3
2.5 2.5
2 2
1.5 1.5
1 1
0 20 40 60 80 100 120 14 0 160 18 0 200 1 1.5 2 2.5 3 3.5 4 4.5 5 5. 5 6
S a m p le Actual Coag ulant Dose (ppm)
Figure 8: Actual (thick line) versus predicted (thin line) coagulant dosage with ANN model on test
data and confidence interval (shaded region).
A linear regression model was also developed for comparison with the neural network model. Figure 9
shows the outputs of the linear model whose parameters were estimated with the same data as the ANN.
The prediction accuracy is clearly inferior to that of the ANN model.

6 6
5.5 5.5
5 5
4.5 4.5
Predicted Co agulant Dose (ppm )

Coagulant Dos e (ppm)
4 4
3.5 3.5
3 3
2.5 2.5
2 2
1.5 1.5
1 1
0 20 40 60 80 100 120 14 0 160 18 0 200 1 1.5 2 2.5 3 3.5 4 4.5 5 5. 5 6
S a m p le Actual Coag ulant Dose (ppm)
Figure 9 : Actual (thick line) versus predicted (thin line) coagulant dosage with linear regression
model.
V. CONCLUSIONS
An integrated coagulant dosing system based on unsupervised and supervised neural network models, as
well as various statistical techniques, has been described. Experimental results using real data have
demonstrated the efficiency and soundness of this approach. Field testing is currently underway to fully
validate the system before its widespread dissemination to other sites. More robustness, if required, of the
system could be provided several techniques such as (1) input probability density estimation using, e.g.,
the EM algorithm with a Gaussian mixture approach (allowing to determine the validity domain of the
system), and (2) propagation of the imprecision of reconstructed inputs using sensitivity analysis and
interval arithmetic. The performance of the network is dependent on the quality and completeness of data
provided for system training. As such, continuous updating of training data would certainly improve the
performance of the system. This model, however, is only based on the previous behaviour of operators
and jar-test results. Further work is needed to develop a model taking into account the dynamics of the
system, and to predict treated water parameters (mainly turbidity) at the output of the clarification process.
REFERENCES
[1] C. Lind (1994). Coagulation Control and Optimization : Part One. Public Works for October, 56-57.
[2] C. Lind (1994). Coagulation Control and Optimization : Part Two. Public Works for November, 32-33.
[3] F. Bernazeau, P. Pierrone, J.P. Duguet (1992). Interest in using a streamline current detector for
automatic coagulant dose control. Water Supply 10 (4), 87-96.
[4] A. Mirsepassi, B. Cathers, H.B. Dharmappa (1995). Application of Artificial Neural Networks to the
Real Time Operation of Water Treatment Plants. In Proceedings International Joint Conference on
Neural Networks, Volume 1, pp. 516-521.
[5] J. Evans, C. Enoch, M. Johnson, P. Williams (1998). Intelligent based auto-coagulation control
applied to a water treatment works. In Proceedings of International Conference on Control, pp. 141-145.
[6] T. Kohonen (1995). Self-Organizing Maps. Heidelberg: Springer Verlag.
[7] Y. Le Cum, J.S. Denker, S.A. Solla (1990). Optimal Brain Damage. In D.S. Touretzky (Ed.),
Advances in Neural Information Processing Systems, Volume 2, pp. 598-605. San Mateo, CA: Morgan
Kaufman.
[8] B. Efron, R.J. Tibshirani (1993). An Introduction to the Bootsrap. New York: Chapmann & Hall.
[9] R. P. Lippmann, L. Kukolich, D. Shahian (1997). Predicting of Complications in Coronary Artery
Bypass Operations using Neural Networks. In Neural Information Processing System 7, 1055-1062.
(49) References (15)
s applications were performed in order to develop a neural model for the online estimation of optimal coagulant dosage from raw water
istics. Previous researches [22][23][24] [25] [26] show the eiciency of such approach using neural networks. ...
kinds of problems are addressed in the literature by a range of diferent approaches used to perform tasks such as classiication,
ation, anomaly detection, and prediction. Applying deep neural network techniques also into treatment water process [25, [39][40][41][42][43]
6] has
Seebeen
all › gaining momentum
See all › due to its state-ofthe-art performances and high-quality recommendations. In contrast to traditional
ndation models, deep15learning Download between
provides a beter understanding of user's demands, item's characteristics, and historical interactions citation Share Download full-text PDF
49 Citations References
al works [25, [39][40][41][42] have already shown the potential of these techniques for modeling the coagulation process. All these studies
o relate the coagulant dose to diferent Desalination and Water Treatment descriptors parameters of the quality of raw water, such as turbidity,
uctivity, etc., using a neural network. ...
Deep Learning Methods for WTP Control and Monitoring

Full-text available
mrini · El Khadir Lakhal
w abstract
orward controller is preferred to control the coagulant dosage in the water treatment process [4].The quality of the water is measured at the
e of the water treatment plant and the desired amount of alum must be added to take corrective action. In the previous literature studies,
n model of coagulant dosage was developed using neural network [5] . Worldwide, several million tons per year of metal salts (Al-and Fe-
agulants) are used for these purposes, and the world market for coagulants and flocculants in 2013 was estimated to be on the order of $5.4
There are many studies concerning the effect of coagulant dosage and mixing conditions such as pH and temperature on the flocculation of
ated suspensions using polymeric coagulants [7] ; coagulation of synthetic water by plant seeds [8] and coagulation of low turbidity water
ntonite [9]. ...
wn in Fig.8, the maximum removal efficiency occurred with pH=6. The obtained results are fully agreed with the results obtained from Ahmed
0 [11] where the optimum turbidity removal efficiency was at pH (5) (6). ...
e manner the effect of pH on turbidity removal for medium (120±10 NTU), high (240±10 NTU) and very high (480±10 NTU) are illustrated in
9-11) respectively, and it is clear from the figures that the optimum removal efficiency also done with pH between (5) (6). The obtained results
y agreement with the optimum pH range for NOM removal with alum coagulation which was generally between 5.5 and 6.5, depending on
ality factors (Crittenden et al. 2012) [12].Also According to a study done by Pernitsky [13], the optimum pH conditions for alum coagulation
do between pH of 6-7. ...
the Optimum Alum Dosages Used in Several Drinking

Paper Full-text available
w abstract
raw water treatment, the determination of coagulant dosage is one of the most considerations among various works conducted in unit
s [13]. Antecedently, the coagulant quantities are generally determined by the empiric Jar test technique that induces problems of excess or
nt reagent, particularly during the period of fast variation in water quality [14, 15] . These process of determination lead to increasing iron
nd residual iron content of drinking water, which creates an unpleasant taste, odor, color, and higher opaqueness [16]. ...
alum in surface water treatment

ull-text available
ement de la turbidité augmente progressivement et atteint une valeur maximale à la concentration critique de floculation, puis décroit ensuite,
upposer une restabilisation du système (Dihang, 2007 ;García-Fayos et al., 2010). Les résultats obtenus sont comparables à ceux trouvés
érentes études (Gagnon et al., 1997 ; Waissman-Vilanova, 2000 ;Baxter et al., 2002 ;Lamrini et al., 2005). ...
Universitaire
ull-text available
arfi
w abstract
s applications of ANN have been reported in the drinking water treatment industry: prediction of the drinking water consumption (Koffi et al.
hl and Brémond 2000), leak detection and calibration (Vítkovský et al. 2000), predicting membrane fouling (Delgrange-Vincent at al.
etty and Chellam 2003), the adequate chlorine dosage for potable water disinfection (Rodriguez and Serodes 1996) and the prediction of
ater quality parameters (Rak 2013;Kabsch-Korbutowicz and Kutylowska 2011). Specifically, many researchers focused their studies on the
n of the coagulant dosing: Bazer-Bachi et al. (1990), Baba et al. (1990), Gagnon et al. (1997 ), Van Leeuwen et al. (1999, Valentin (2000),
al. (2004), Lamrini et al. (2005), Wu and Lo (2008), Heddam et al. (2012) and recently Kumar et al. (2013) who developed a model to control
ulant dosing pump. ...
of the optimal dose of coagulant for various potable water treatment processes through artificial neural network
ull-text available
HYDROINFORM
d Hassen Baouab · Semia Cherif
w abstract
rch studies have shown that the simultaneous application of the techniques of combined analysis of the main components and the ANN
better and more accurate results than the situation of considering them separately (Oliveira-Esquerre et al., 2002), so this study will use the
thod Baxter et al., 1999) by using the multilayer perceptron (MLP) structure for the ANN for anticipating the turbidity and the color of the
ater at the Rossdale Water Treatment plant in Alberta, Canada. Gagnon et al. (1997) used the method of inverse models in ANNs, a similar
or anticipating the necessary amount of alum for the Sainte-Foy Waste Water Treatment Plant in Quebec, Canada (Joo et al., 2000).
has also been performed for the Chungju Wastewater Treatment Plant in Korea. ...
See all
oagulant ›
forecastingSee all ›
by modeling jar test experiments using ANNs Download citation Share Download full-text PDF
ull-text available
ghiri · Amin Daghighi · Sina Moharramzadeh
w abstract
orward controller is preferred to control the coagulant dosage in the water treatment process [4].The quality of the water is measured at the
e of the water treatment plant and the desired amount of alum must be added to take corrective action. In the previous literature studies,
n model of coagulant dosage was developed using neural network [5] . Worldwide, several million tons per year of metal salts (Al-and Fe-
agulants) are used for these purposes, and the world market for coagulants and flocculants in 2013 was estimated to be on the order of $5.4
There are many studies concerning the effect of coagulant dosage and mixing conditions such as pH and temperature on the flocculation of
ated suspensions using polymeric coagulants [7] ; coagulation of synthetic water by plant seeds [8] and coagulation of low turbidity water
ntonite [9]. ...
the Optimum Alum Dosages Used in Several Drinking Water Treatment Plants in WASIT Governorate (IRAQ) and Investigation the
on Alum Optimum Dosages
Paper Full-text available
· Sara Ali Zaalan
w abstract
rch studies have shown that the simultaneous application of the technique of combined analysis of the main components and the ANN
better and more accurate results than the situation of considerably them separately (Oliveira-Esquerre et al., 2002), so this study will use the
thod Baxter et al., 1999) by using the Multi-Layer Perceptron (MLP) structure for the ANN and its modeling for anticipating the turbidity and
of the refined water at the Rosedale Refinery in Alberta, Canada. Gagnon (Gagnon et al., 1997) using the method of 30 inverse models in
tworks, a similar method, for anticipating the necessary amount of Alum for the Ste-Foy Refinery in Quebec, Canada (Joo et al., 2000) A
has also been performed for the Chungju Refinery in Korea. Van Leeuwen prepared a neural network model based on the jar test procedure
n the collected surface waters in southern Australia (Van Leeuwen et al., 1999). ...
oagulant Forecasting with Modeling the Jar Test Experiments Using ANN
ull-text available
ri · Amin Daghighi · Sina Moharramzadeh
w abstract
odels to Predict Disinfection By-Products in Drinking Water: An Updated Review
adiq · Manuel Rodriguez · Haroon Rashid Mian
w abstract
dynamic system for chemical coagulation control

Paper
nha · Eric Morris · Chuhong Fei · Sol Jin
Show more
Recommendations

Project Download citation Share Download full-text PDF
Effect of Noise on Gas Membrane Characterization

Haoyu Wu · Jules Thibault · Boguslaw Kruczek
Time lag method; Gas membrane characterization; Noise; Resolution; Precision; Finite difference; Nonlinear regression
View project
Project
Optimisation
Jules Thibault
View project
Project
Modelling of Mass transfer in Mixed Matrix Membranes (MMMs)

Arian Ebneyamini · F. Handan Tezel · Jules Thibault
View project
Project
Undergraduate Chemical Engineering Thesis

Peter Gaudet · Jules Thibault · F. Handan Tezel
View project

Article Download citation Share Download full-text PDF
Characterization and coagulation–flocculation performance of a composite coagulant: poly-ferric-alum...

November 2015 · Desalination and water treatment
Wei Chen · Huaili Zheng · Jun Zhai · [...] · Yongjun Sun
Composite coagulant has received widespread attention and research for its excellent coagulation performance in recent years. In this study, a new
composite coagulant poly-ferric-aluminum-silicate-sulfate was synthesized using water glass, FeSO4·7H2O, Al2(SO4)3·18H2O, and NaClO3, their
structures and morphologies were characterized and compared by X-ray diffraction, infrared spectra, and scanning ... [Show full abstract]
Read more
Article Full-text available
Applications of some polyelectrolytes for the treatment of surface waters

December 1999
Carmen Zaharia · Carmen Teodosiu · Matei Macoveanu · G.Chiţanu
The paper presents the application of some inorganic coagulant, i.e. ferric sulphate and polymeric flocculants, such as: anionic polyelectrolyte,
PONILIT GT-2 and cationic polyelectrolyte, PRODEFLOC CRC 301 for water treatment. Synthetic waters with different total concentrations of
colloids were prepared and the coagulation-flocculation process followed by sedimentation and filtration was ... [Show full abstract]
View full-text
Article Full-text available
Performance of bentonite clay as a coagulation aid on water quality

January 2019 · Desalination and water treatment
Ahari M'hamed · Hassan Ddahim · Rachid Ramadane
The objective of this work is to observe the effectiveness of bentonite as a coagulation adjuvant on water treatment quality at Ait Baha station. In
our case, bentonite is used as a source of turbidity and is not an adsorbent. Jar tests were performed on raw water samples of low turbidity
according to the approach applied in the laboratory. The performance of the bentonite was tested in mixtures: ... [Show full abstract]
View full-text
Article
Response surface methodological approach to optimize the coagulation–flocculation process in drinkin...

July 2011 · Chemical Engineering Research and Design
Thuy Khanh Trinh · Lim Seok Kang
Performing jar tests often requires carrying out a time consuming iteration procedure to find out the right amount of chemical for coagulation–
flocculation process in water treatment plants. Applying the response surface method (RSM) in jar tests as an alternative to the conventional
methods was investigated in this study. The purpose is finding out the optimum combination of coagulant dose and ... [Show full abstract]
Read more
Discover more
Company Support Business solutions
About us Help Center Advertising

News Recruiting
Careers
© 2008-2020 ResearchGate GmbH. All rights reserved. Terms · Privacy · Copyright · Imprint

(PDF) Modelling of Coagulant Dosage in A Water Treatment Plant

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(PDF) Modelling of Coagulant Dosage in A Water Treatment Plant

Uploaded by

Copyright:

Available Formats

10/7/2020 (PDF) Modelling of coagulant dosage in a water treatment plant

See all › See all › Recruit

coagulant dosage in a water treatment plant

Available) in Artificial Intelligence in Engineering 11(4):401-404 · October 1997 with 544 Reads

er the world's research

oin for free

uploaded by Jules Thibault Author content Download full-text PDF

MODELLING OF COAGULANT DOSAGE IN A WATER TREATMENT PLANT

II . WATER TREATMENT OPERATION

See all › See all › carbon filter Chemi cal Dosage

Water Model of Coagulant

treatment coagulant dose

3.1 Single-parameter data validation

3.2 Multi-parameter data validation and reconstruction

Dissolved Ox ygen (mg/l)

See all › See all ›

3.3 Modelling of coagulant dosage

Optimal Brain Damage (OBD) [7] learning and Turbidity

pruning algorithm. This approach is based on the Σ f

following general procedure. Conductivity

The implementation of this technique consists of the

Figure 4: Initial network

1. Choose a relatively large initial network architecture.

Start Data Upper

Raw water Coagulant Coagulant

Figure 6 : Bootstrap sampling to generate

See all › See all ›

Figure 7 : Raw Water Data

See all › See all ›

Predicted Co agulant Dose (ppm )

(49) References (15)

Deep Learning Methods for WTP Control and Monitoring

mrini · El Khadir Lakhal

the Optimum Alum Dosages Used in Several Drinking

alum in surface water treatment

ghiri · Amin Daghighi · Sina Moharramzadeh

· Sara Ali Zaalan

ri · Amin Daghighi · Sina Moharramzadeh

odels to Predict Disinfection By-Products in Drinking Water: An Updated Review

adiq · Manuel Rodriguez · Haroon Rashid Mian

dynamic system for chemical coagulation control

nha · Eric Morris · Chuhong Fei · Sol Jin

See all › See all ›

Effect of Noise on Gas Membrane Characterization

Modelling of Mass transfer in Mixed Matrix Membranes (MMMs)

Undergraduate Chemical Engineering Thesis

See all › See all ›

Characterization and coagulation–flocculation performance of a composite coagulant: poly-ferric-alum...

Article Full-text available

Applications of some polyelectrolytes for the treatment of surface waters

Article Full-text available

Performance of bentonite clay as a coagulation aid on water quality

Response surface methodological approach to optimize the coagulation–flocculation process in drinkin...

Company Support Business solutions

About us Help Center Advertising

You might also like