IAHR, founded in 1935, is a worldwide independent member-based organisation of engineers and water specialists working in fields related to the hydro-environmental sciences and their practical application. Activities range from river and maritime hydraulics to water resources development and eco-hydraulics, through to ice engineering, hydroinformatics, and hydraulic machinery.
Log On
About IAHRDirectoryCommitteesMy IAHRNews & JournalseLibraryeShopEventsJoin IAHRWorld CongressDonate
spacer.gif
spacer.gif eLibrary
spacer.gif eLibrary
You are here : eLibrary : IAHR World Congress Proceedings : 36th Congress - The Hague (2015) FULL PAPERS : THEME 2- HYDRO-ENVIRONMENT : ESTIMATING COD LOADS IN COMBINED SEWER OVERFLOWS WITH MULTIVARIATE AND NEURAL NETWORK MODELS UNDER S...
ESTIMATING COD LOADS IN COMBINED SEWER OVERFLOWS WITH MULTIVARIATE AND NEURAL NETWORK MODELS UNDER SEMI-ARID RAINFALL CONDITIONS
Author : IGNACIO ANDRES-DOMENECH, M. ESTHER GOMEZ-MARTIN, JOSEP R. MEDINA & JUAN B. BARCO
Estimation off pollution load ds from combined sewer ov verflows (CSO O) is a major is ssue for mitiga ation of impac cts on the wate er environment.. According to o EU directive es, pollutant lo oads must be estimated in a frequency-m magnitude analysis to bette er assess their impact on the e water bodie es. This study focuses on e estimating the chemical oxy ygen demand (COD) load in CSO from on ne of the main sewer trunks in Valencia (S Spain) to asse ess impacts on n the waterfront. 42 events were recorded d, modelled and d analysed du uring the perio od 2008-2012 2 (quantity and quality data a). For each e event, anteced dent dry perio od (T), rainfall duration (D), peak rainfall inttensity (I), rain nfall volume (R), runoff volu ume (V) and C COD load (M) spilled into th he receiving wa ater body were e obtained. T is related to pollutant acc cumulation in the catchmen nt (build-up), R R and V to th he event magnittude and I to erosive proce esses (wash-o off). In this pa aper, two diffe erent models a are analysed to estimate M M. First, an ana alytical multiva ariate regress sive model is adjusted con nsidering relev vant explanatto ory variables. On the sam me basis, an arttificial pruned neural netwo ork (NN) was trained to es stimate M, dep pending on in nput variables s with a hidde en layer. Both m models highligh ht the same co ounterintuitive e result in the s studied case: M does not de epend on T. T The multivariatte model best fiit shows a quiite linear relationship betwe een R (or V) a and the COD loads. This strrong dependence between R and M is also o deduced fro om the NN mo odel, which eliminates the T T, D and I inp puts, and only considers R to estimate th he COD load (M M) with a 10% % relative mea an squared errror on test da ata. Semi-arid conditions of f the Valencia rainfall regim me lead to very large anteced dent dry perio ods. Accumula ated pollutants s in the catch hment have re eached their m maximum rate es and are not already influe enced by T. C Consequently, the higher ra ainfall or runofff volumes are e, the higher pollutant load ds because of th he huge amou unt of pollutantts accumulate ed in the syste em and mobilis sed during eac ch event.
File Size : 873,403 bytes
File Type : Adobe Acrobat Document
Chapter : IAHR World Congress Proceedings
Category : 36th Congress - The Hague (2015) FULL PAPERS
Article : THEME 2- HYDRO-ENVIRONMENT
Date Published : 18/04/2016
Download Now