Temporal Convolutional Network-based Approach for Forecasting
Fluctuations Differential Pressure in Reverse Osmosis Systems
The Son Phan
1
, Thanh-Ha Do
1
, Phuc Do
2
1
HaNoi University of Science, Hanoi, Vietnam
phantheson t65@hus.edu.vn
dothanhha@hus.edu.vn
2
University of Lorraine, Nancy, France
phuc.do@univ-lorraine.fr
ABSTRACT
Providing forecasts of pressure fluctuations and changes will
aid in selecting appropriate maintenance strategies to op-
timize efficiency and costs. This paper presents a deep-
learning-based model to forecast the degradation evolution of
membrane biological fouling in RO (Reverse Osmosis) sys-
tems. Although applying deep learning in forecasting still
faces many challenges, applying convolutional operations in
convolution 1D has yielded promising results for sequential
data, particularly time series data. Thus, in this paper, we
study and develop the 1D convolution operation-based Tem-
poral Convolutional Network (TCN) model to predict pres-
sure dynamics at both ends of the RO vessel. In addition,
since the deep learning technique has yet to be widely ex-
plored in this field, thus we also need to pre-process the data
collected from the Carlsbad Desalination Plant in California,
such as the proposed model can identify complex relation-
ships between timestamps and pressure features. The experi-
ment results were evaluated and compared with other existing
models, such as LSTM, CNN & LSTM, and GRU. The results
show that the TCN-based prediction model had the slightest
error in the test dataset.
1. INTRODUCTION
Water covers approximately 71 % of the Earth’s surface, and
more than 97 % of the Earth’s water is saltwater. Under popu-
lation growth, the need for clean water is highly critical. Vari-
ous methods have been developed to generate clean water for
industrial and domestic use to meet this demand, one such
method being desalination Nour et al., 2022 found. There-
Phan et al. This is an open-access article distributed under the terms of the
Creative Commons Attribution 3.0 United States License, which permits un-
restricted use, distribution, and reproduction in any medium, provided the
original author and source are credited.
fore, RO (Reverse Osmosis) technology-based desalination
plants have been put into operation. Nour et al., 2022 show
that the issue arises when these plants face system impair-
ment after operation due to membrane fouling. Depending
on the type of accumulated residue, membrane fouling can be
categorized into particle fouling, organic fouling, inorganic
fouling, and biofouling, with biofouling being considered the
most severe and challenging to solve.
One of the common causes of biofouling is algal blooms (van
Rooij, Scarf, and Do, 2021), (Villacorte et al., 2017) and
(Jiang, Li, and Ladewig, 2017). Organic compounds pro-
duced during algal blooms are the leading cause of biofouling
on membrane surfaces. These compounds create a slippery
layer on the membrane surface, increasing the salts and pres-
sure passing through the filter membrane. When these mem-
branes become fouled, they can reduce filtration efficiency
or, more seriously, damage the membrane system (Koutsakos
and Moxey, 2007) found. There are three ways to improve
membrane longevity: (a) membrane performance monitoring,
(b) membrane cleaning (clean-in-place (CIP) method) (Kout-
sakos and Moxey, 2007; van Rooij, Scarf, and Do, 2021), and
(c) membrane replacement (Koutsakos and Moxey, 2007; van
Rooij, Scarf, and Do, 2021).
Many works have been proposed to mitigate the impact of
biofouling, and some of them have yielded positive results.
In the direction of maintaining membrane biological foul-
ing, there are many causes. However, biological fouling is
challenging to resolve, and the algal bloom is the root cause
of membrane fouling (Nour et al., 2022). (Koutsakos and
Moxey, 2007) proposed methods for monitoring membrane
systems and maintenance options for the system. Mainte-
nance of this system includes cleaning the membranes with
chemical solutions, known as the CIP method. Replacing
and rearranging the elements is essential when the system be-
1
