Massey Documents by Type

Permanent URI for this communityhttps://mro.massey.ac.nz/handle/10179/294

Browse

Filters

1992 - 19962

Settings

Subject: search.filters.subject.Refrigerating machinery

Search Results

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    A dynamic modelling methodology for the simulation of industrial refrigeration systems : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Biotechnology and Bioprocess Engineering at Massey University
    (Massey University, 1992) Lovatt, Simon James
    A dynamic modelling methodology has been developed for the computer simulation of industrial refrigeration systems. A computer program, RefSim, has been developed which embodies the new methodology. RefSim contains a total of 33 separate models - 11 derived from existing models, six which are substantially enhanced and 16 which are new. In general, these models were derived from thermal considerations and ignored the effect of hydrodynamic processes in the refrigeration circuit. Models can be dynamically linked together as specified by the input data in order to simulate a complete plant. The program includes a set of simulation utilities which reduce the amount of work required to develop models. The object-oriented features of inheritance, encapsulation and polymorphism are used extensively. Substantial model development was carried out to achieve accurate predictions of the heat release profile during chilling and freezing of food product as product cooling makes the greatest contribution to both mean and peak heat loads in many industrial refrigeration plants. The new ordinary differential equation (ODE) model was tested against finite difference (FD) calculations for a range of product shapes and Biot numbers. The ODE model predicted to within ±10% of the FD calculation during almost all of the cooling process under the test conditions. The ODE model required several orders of magnitude less computation than FD while being capable of extension to shapes that could not be handled by FD. To test the new ODE model against experimental data, a differential air temperature method to measure the cooling food product heat load profile was developed. Both the FD and ODE methods predicted the heat load profile of freezing meat cartons to within the experimental margin for error (±10%). The ODE model also predicted the heat load profile of freezing lamb carcasses to a similar level of accuracy. Three refrigeration plants (a laboratory water chiller, a 18500 lamb per day meat processing plant, and a 6000 lamb/1000 beef per day meat processing plant) were surveyed to obtain data for testing the whole simulation environment. RefSim was found to predict the measured data satisfactorily in most cases. The results were superior to those from a commercial refrigeration simulation environment and comparable to an enhanced version of that environment which included the new ODE product heat load model. Differences between the measured values and those predicted by Refsim were probably more attributable to uncertainties in the simulation input data than to model deficiencies. RefSim was found to be a flexible environment which was general enough to simulate both simple and complex refrigeration systems. Unusual components could be simulated by combining existing models rather than implementing custom models. Nevertheless, the simulation results have indicated a number of areas for further model improvement. The effects of air mixing and the thermal buffering of structural materials were shown to be modelled poorly for some refrigerated rooms. There is some scope for improving the chilling stage of the ODE product heat load model.
  • Thumbnail Image
    Item
    Evaluation of dynamic models for refrigeration system components : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Process and Environmental Technology at Massey University
    (Massey University, 1996) Estrada Flores, Silvia
    There is a paucity of proven models for predicting the energy transients of walls used in low temperature applications and pressure vessels commonly used in refrigeration plants. The aim of this work was to investigate how the accuracy of feasible models for these situations was affected by changing model complexity. Wall models were developed by assuming each wall layer could be represented by one of five possible thermal behaviours: null, resistance only, capacity only, alternating resistance and capacity (lumped) or fully distributed resistance and capacity. A range of feasible models for each of four common low temperature wall types was investigated by comparison of simulated behaviour to predictions by a finite element model (which was itself validated by comparison to experimental data for wall systems). Several model evaluation measures are presented to aid engineering judgement in selecting appropriate wall models for particular applications. Only resistance needs be considered for accurate prediction of mean heat flux entering a room. It is recommended that metal layers be represented by capacity only models, thin insulation layers by resistance only models, thicker insulation layers by lumped or fully distributed models, and concrete layers by lumped or fully distributed models. The recommended number of zones within a lumped or distributed model for a layer rises as the amplitude of the expected repeating temperature cycle for that layer increases. Four models of different complexity were derived to represent a typical industrial intercooler (pressure vessel). These models were tested by comparison of predictions to the measured time-temperature response of two pilot plant calorimeters containing R134a, when subjected to changing heat inputs. The measured response rate was most strongly influenced by sensible heat storage in the calorimeter shells and liquid refrigerant. Little difference in predictions by the four models was obtained in spite of the less complex models neglecting many known physical phenomena. A model considering only the thermal capacity in the shell and liquid refrigerant predicted rates of temperature change within 10% of predictions by all other models, and also close to the experimental data. An industrial case study suggested that the conclusions from the calorimeter study may be valid over much wider ranges. Suggestions are made on ways to improve the simplest model accuracy, and to gain greater benefit from the more complex models.