doi:10.1088/1755-1315/779/1/012110

Details

Publication Date

Tue Jun 01 2021

Journal Name

Iop Conference Series: Earth And Environmental Science

Volume

779

DOI

10.1088/1755-1315/779/1/012110

Choose Citation Style

Statistics

View publication

2

Statistics

(1)

Performance Evaluation of Al-Karkh Water Treatment Plant Using Model-driven and Data-Driven Models

Operational performance evaluation

water treatment plant

multiple linear regression

multivariate multiple regression

artificial neural network model

model-driven and data-driven modeling types.

Basim Hussein

Awatif Soaded

Basim H.

...Show More Authors

Abstract<p>There is a great operational risk to control the day-to-day management in water treatment plants, so water companies are looking for solutions to predict how the treatment processes may be improved due to the increased pressure to remain competitive. This study focused on the mathematical modeling of water treatment processes with the primary motivation to provide tools that can be used to predict the performance of the treatment to enable better control of uncertainty and risk. This research included choosing the most important variables affecting quality standards using the correlation test. According to this test, it was found that the important parameters of raw water: Total Hardness, Calcium, Magnesium, Total Solids, Nitrite, Nitrates, Ammonia, and Silica are to be used to construct the specific model, while pH, Fluoride, Aluminium, Nitrite, Nitrate, Ammonia, Silica, and Orthophosphate of the treated water were eliminated from the analysis. For modeling the coagulation and flocculation process temperature, Alkalinity and pH of raw water were the depended variables of the model. As for the modeling process turbidity of the treated water was used as the output variable. In general, the linear models including model-driven type, (Multivariate multiple regression, MMR and Multiple linear regression, MLR) have slightly higher prediction efficiencies than the, data-driven type (artificial neural network, ANNM). The coefficients of determination (R<sup>2</sup>) reached 66 to 85% for the MMR and MLR models and 65 to 81% for the ANN models.</p>

View Publication

1 2 3 4 ... 3197 3198 3199 3200