Steam condener perfomance evaliation by using neural network

Abstract EN

This work applied Artificial Neural Network (ANN) for performance evaluation of steam condensers which are widely used in power plants and refineries.

Two condensers were experimentally investigated.

Experimental data was obtained by use unit steam power from G.U.N.T Company and industrial condenser operates in Dura refinery.

The commonly used Back Propagation (BP) algorithm was used to train and test network.

Input of neural network include inlet water temperature, water flow rate, steam temperature and enthalpy difference.

The exit water temperature represented output of the neural network.

The maximum deviation between the predicted results and experimental data was less than 1 %.

It is recommended the (ANN) can be used to predicate the performance of thermal system in engineering applications, such as modeling condenser for heat transfer analysis.

Afterwards, ANN resulted used to find thermal parameters (convection heat transfer coefficient of water side and steam flow rate) based on software program built by Matlab language.

Comparing the resulted from modeling with experimental data reveals a good agreement (-3 % to-13 %).