Christian J. L. Hermes


Cláudio Melo

Date of publication





A first-principles model for simulating the dynamic behaviour of fan-and-damper controlled refrigerators is proposed in this thesis. The model has been used to simulate a typical Brazilian 440-litre top-mount household refrigerator, in which the compressor is on-off controlled driven by the freezer temperature, while a thermomechanic damper is used to set the fresh-food compartment temperature. Although the model development has been centred on a top-mount cabinet, it can be easily extended to any other cabinet configuration. The system model takes into account each one of the cycle components. The heat exchangers (condenser and evaporator) have been modelled according to a finite-volume techinique, using an explicit scheme for the time formulation. New air-side heat transfer correlations are also proposed for both finned-tube evaporators and wire-on-tube condensers, which are able to predict experimental data within –10% error bands. The compressor model has been divided into two sub-domains named compressor shell and compression process. While the first takes into account the mass and energy interactions taking place within the compressor shell, the second is based on the first law of Thermodynamics and the piston kinematics. Both of them have been calibrated using experimental data taken through a hot-gas calorimeter facility. A new model for the capillary tube has also been implemented, which is able to simulate non-adiabatic flows of HFC-134a and HC-600a as fast as adiabatic models. Such model was validated against more than one thousand experimetal data points. Results for mass flow rate predictions have shown that 85% of all data fall within –10% error bands. The cabinet model takes into account the dynamic thermal loads by considering the influence of the damper into the airflow blown to each compartment. Such model was developed based on measurements using a wind-tunnel facility. The component-level models were gathered together and integrated according to a predictor-corrector ODE solver, which has both order and stepsize controllers. Numerical predictions have been compared to experimental data showing a reasonable level of agreement for the whole range of operation, including startup and cycling transients: energy consumption have been predicted within –10% error bands, while air temperatures predictions have fallen within –1°C error bands.

Material for download

Access material


Learn more about one of the best post-graduate courses in mechanical engineering in Brazil

I want to know