Journals Information
Environment and Ecology Research Vol. 11(3), pp. 493 - 504
DOI: 10.13189/eer.2023.110309
Reprint (PDF) (1220Kb)
Forecasting Wastewater Influent Parameters Using ARIMA and Holt-Winters Models (A Case Study)
A. Chaoui 1,*, W. Elkhoumsi 1, M. Laaouan 2, R. Bourziza 1, K. Sebari 1
1 Hassan II Institute for Agronomy and Veterinary Medicine, Morocco
2 International Institute for Water and Sanitation, Morocco
ABSTRACT
Forecast models are essential tools used to make predictions about future events based on past data. These models use a variety of methods and techniques, such as extrapolation, simulation modeling, and judgmental approaches, among others, to analyze time-series data and make forecasts. In the context of wastewater treatment plants (WWTPs), predicting the influent quality parameters is crucial to improving their operation and design. This allows plant operators to adjust their processes to optimize treatment efficiency, reduce costs, and minimize environmental impact. Therefore, forecasting plays a vital role in ensuring that WWTPs operate effectively and efficiently, and it is an essential component of modern wastewater treatment management. In this paper, Zaio's WWTP (Morocco) BOD5 variable will be used as a case example. This will lead to other furthered studies for much better plant optimization and operation. The current paper starts with analyzing the plants' operation data, checking for stationarity and applying the ARIMA as well as the Holt-Winters model in order to provide more or less accurate predictions that will provide support to WWTPs management. Models will be compared using error terms such as RMSE, MAPE and R-squared. Other statistical tools will also be used as ADF t-statistic test and Box-Ljung test. After defining and proving the statistical significance of models, the RMSE and the MAPE provided a conflicting decision. For this, the selection of the best model was based on the coefficient of determination R-squared. This latter has a value of 0.527 for the ARIMA model and 0.517 for the Holt-Winters model. This concludes that within the ARIMA model, the total variance of the dependent variable is explained by the independent variable with a higher percentage compared to the Holt-Winters model.
KEYWORDS
Wastewater, WWTPs, BOD5, Forecasting, Error Terms, ARIMA, Holt-Winters
Cite This Paper in IEEE or APA Citation Styles
(a). IEEE Format:
[1] A. Chaoui , W. Elkhoumsi , M. Laaouan , R. Bourziza , K. Sebari , "Forecasting Wastewater Influent Parameters Using ARIMA and Holt-Winters Models (A Case Study)," Environment and Ecology Research, Vol. 11, No. 3, pp. 493 - 504, 2023. DOI: 10.13189/eer.2023.110309.
(b). APA Format:
A. Chaoui , W. Elkhoumsi , M. Laaouan , R. Bourziza , K. Sebari (2023). Forecasting Wastewater Influent Parameters Using ARIMA and Holt-Winters Models (A Case Study). Environment and Ecology Research, 11(3), 493 - 504. DOI: 10.13189/eer.2023.110309.