Optimization of a three-bed adsorption chiller by genetic algorithms and neural networks

Adsorption cycles have a distinct advantage over other systems in the ability to use low grade heat, especially waste heat of near ambient temperature. A Tri-bed twin-evaporator adsorption chiller constitute an innovative design in cooling production which allows more efficient conversion and management of low grade sources of thermal energy due to more effective way of utilization adsorptive abilities of the beds during a single work. Although it is the most effective way in chilled water production the complexity of the Tri-bed twin-evaporator adsorption chiller operation is still not sufficiently recognized and the improvement in cooling capacity (CC) of the cooler is still a challenging task. The paper introduces artificial intelligence approach for the optimization study of a Tri-bed twin-evaporator adsorption chiller using low-temperature heat from cogeneration. Genetic algorithms (GA) and artificial neural networks (ANN) are used to develop the model which allows estimating the behaviour of the adsorption heat pump. Cooling capacity (CC) as one of the main energy efficiency factor in cooling production is examined during the study for different operating sceneries. The presented non-iterative approach gives quick and accurate results as an answer to the input data sets. The CC of the chiller, evaluated using the developed model, is in good agreement with the experimental data. Maximum relative error between measured and calculated data is lower than +/- 10%. The developed model permits to study the influence of operating parameters on the cooling capacity of the chiller. For the considered range of input parameters the highest cooling capacity which can be obtained by the heat pump is equal 93.3 kW. The method constitutes an alternative, easy-to-apply and useful, complementary technique, comparing to the other techniques of data handling, including the complex of numerical and analytical methods as well as high costs of empirical experiments. The model can be applied for optimizations purposes and can constitute a sub model or a separate module in engineering calculations, capable to predict the CC of the Tri-bed twin-evaporator adsorption cooler, integrated into multigenerative systems.