Prediction of the behavior of C-4130 low alloy steel in HCl acid using artificial neural networks

Other Title(s)

التنبؤ في سلوك الصلب السبائكي في حامض الهيدروكلوريك C-4130

Dissertant

Abd al-Imam, Ula Habib

Thesis advisor

Hammadi, Nawal J.

Comitee Members

Muhammad, Haydar Muadh
Hassan, Abd al-Karim F.
Hammud, Ali S.

University

University of Basrah

Faculty

Engineering College

Department

Department of Material Engineering

University Country

Iraq

Degree

Master

Degree Date

2012

English Abstract

Many engineering industrial equipmentsneed acid cleaning for washing mill scale or removing complex scale adherent to the metallic surfaces as one of the steps of the maintenance program.

But, severe corrosion problem may be arising if they improperly controlled. The aim of the present work is to study the effect of different environmental factors on the corrosion behavior of bare (non scale – containing surface) and scale – containing alloy steel samples using corrosion tests and pickling rate tests respectively.

A pharmaceutical compound as " Ampicillin + Cloxacillin " (ACX) were evaluated as corrosion inhibitor for alloy steel in acidic medium. The proposed artificial neural network (ANN) models using the experimental data obtained from corrosion test and pickling rate tests involved : 1) Prediction of corrosion rate and inhibition efficiency in the corrosive environment in the absence and presence of organic inhibitor (ACX) using bare metal sample.

The input variables were : the concentration of corrosive medium, concentration of ACX inhibitor, and immersion durations; while the output variables were the corrosion rate and inhibition efficiency. 2) Prediction of pickling rate in the corrosive environment in the absence and presence of ACX inhibitor using the scale - containing metallic surface of alloy steel samples.

The input variables are the concentration of corrosive medium, concentration of ACX inhibitor, and immersion durations; while the output variables were the scale density removed and dissolved iron content in acid medium. Back propagation networks were used for the proposed ANN models of the present work to find the targets of each model.

The prediction results of the two models of corrosion testand pickling rate showed good agreement with the obtained experimental values.

The correlation coefficients for the first and second models were found to be (R = 0.99936, R = 0.99542) respectively.

The mean square error for both models were found to be (MSE = 3.670 × 10-4, MSE = 1.302 × 10-3) respectively. It was concluded that ANN model can predict the corrosion parameters for any set of conditions.

The predicted results showed a decrease of corrosion rate and increase of inhibition efficiency with the increase of inhibitor concentration.

The results of pickling rate showed reliable prediction of the optimum concentration of inhibitor at the maximum acid concentration required for complete removal of scale.

Main Subjects

Engineering & Technology Sciences (Multidisciplinary)

Topics

• Mechanical properties
• Neural networks(Computer science)
• Steel alloys

No. of Pages

104

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-317179