Ankara üNİversitesi fen biLİmleri enstiTÜSÜ

2005, 154 sayfa 
 
Anahtar Kelimeler: Dinamik Benzetim, Dinamik Modelleme, Parametre Belirleme. 

---

  ii 
 
ABSTRACT 
 
Ph. D. Thesis 
 
A NEW DYNAMIC SIMULATION SOFTWARE FOR MONITORING  
WATER QUALITY IN RIVER STREAMS 
 
A. Mehmet YÜCEER 
 
Ankara University 
Graduate School of Natural and Applied Sciences 
Department of Chemical Engineering 
 
Supervisor: Prof. Dr. Ridvan BERBER 
 
As the river pollution affects the human life and agricultural development directly by distorting 
ecological balance, monitoring the river streams for pollution becomes one of the important 
areas of research today. In order to be able to assess the consequences of possible point 
discharges from existing nearby installations or future industrial investments throughout the 
river basin, one needs to have a reliable water quality model, which very much depends on some 
parameters. These parameters playing a vital role in validation of the model are still mostly 
determined by exhaustive trial-and-error procedures. In addition, currently most of the avaliable 
softwares for river water quality modeling do not have parameter estimation capabilities.   
 
Yeşilırmak river was modeled as a series of CSTRs and kinetic parameters that present inside 
the model were estimated for dynamically modeling of rivers. Since parameter estimation 
requires the solution of a dynamic optimization problem, a practical approach was used which is 
based on control-vector parameterization and also which does not seek for the sensitivity 
function information. Various kind of techniques as an optimization method such as Gauss-
Newton, Levenberg-Marquardt and SQP (Sequential Quadratic Programming) were used and 
tested. As a next step AKAB (river simulation) software, which is user interactive and has 
graphical user interface, was coded and developed that enables users to determine the variation 
of components with time which are responsible for the pollution in river along the river length 
while optimizing the model parameters without the aid of a dynamic simulation expert. 
Thereby, as the relevant information both about the river and the discharge were provided to the 
software, the variations of eleven state variables could graphically be monitored and 
represented, namely organic nitrogen, ammonium nitrogen, nitrite nitrogen, nitrate nitrogen, 
organic and dissolved phosphor, BOD
5
, dissolved oxygen, coliform, chlorine and algae. 
 
A series of experimental studies were performed in order to test the dynamic simulation 
software and parameter estimation strategy developed in Yeşilırmak River around the region of 
Durucasu at the city of Amasya where intensive pollution exists. Initially, dynamic field data 
obtained from two sampling stations, 500 m apart from each other, were used for identifying the 
model parameters. When the assumption that twenty CSTRs present for each five hundred 
meters river length was made, the difference between the experimental and predicted values 
reduced to a minimum. Measurements were taken along the 7 and 36.5 km river length at 
various intervals also including points right before and after an industrial baker’s yeast plant. 
The simulation result obtained from AKAB software and that obtained from QUAL2E software, 
which is the most used and recognized one all over the world for modeling of water quality, 
were compared with the experimental measured values. The predictions from AKAB indicated 
better agreement with the experimental data, compared to QUAL2E.