.AixLib.Fluid.HeatPumps.HeatPump

Information

This generic grey-box heat pump model uses empirical data to model the refrigerant cycle. The modelling of system inertias and heat losses allow the simulation of transient states.

Resulting in the choosen model structure, several configurations are possible:

1. Compressor type: on/off or inverter controlled
2. Reversible heat pump / only heating
3. Source/Sink: Any combination of mediums is possible
4. Generik: Losses and inertias can be switched on or off.

Concept

Using a signal bus as a connector, this heat pump model can be easily combined with the new HeatPumpSystem or several control or safety blocks from AixLib.Controls.HeatPump. The relevant data is aggregated. In order to control both chillers and heat pumps, both flow and return temperature are aggregated. The mode signal chooses the type of the heat pump operation. As a result, this model can also be used as a chiller:

• mode = true: Heating
• mode = false: Chilling

To model both on/off and inverter controlled heat pumps, the compressor speed is normalizd to a relative value between 0 and 1.

Possible icing of the evaporator is modelled with an input value between 0 and 1.

The model structure is as follows. To understand each submodel, please have a look at the corresponding model information:

1. InnerCycle (Black Box): Here, the user can use between several input models or just easily create his own, modular black box model. Please look at the model description for more info.
2. Inertia: A n-order element is used to model system inertias (mass and thermal) of components inside the refrigerant cycle (compressor, pipes, expansion valve)
3. HeatExchanger: This new model also enable modelling of thermal interias and heat losses in a heat exchanger. Please look at the model description for more info.

Parametrization

To simplify the parametrization of the evaporator and condenser volumes and nominal mass flows there exists an option of automatic estimation based on the nominal usable heating power of the HeatPump. This function uses a linear correlation of these parameters, which was established from the linear regression of more than 20 data sets of water-to-water heat pumps from different manufacturers (e.g. Carrier, Trane, Lennox) ranging from about 25kW to 1MW nominal power. The linear regressions with coefficients of determination above 91% give a good approximation of these parameters. Nevertheless, estimates for machines outside the given range should be checked for plausibility during simulation.

Assumptions

Several assumptions where made in order to model the heat pump. For a detailed description see the corresponding model.

1. Performance data 2D: In order to model inverter controlled heat pumps, the compressor speed is scaled linearly
2. Performance data 2D: Reduced evaporator power as a result of icing. The icing factor is multiplied with the evaporator power.
3. Inertia: The default value of the n-th order element is set to 3. This follows comparisons with experimental data. Previous heat pump models are using n = 1 as a default. However, it was pointed out that a higher order element fits a real heat pump better in.
4. Scaling factor: A scaling facor is implemented for scaling of the heat pump power and capacity. The factor scales the parameters V, m_flow_nominal, C, GIns, GOut and dp_nominal. As a result, the heat pump can supply more heat with the COP staying nearly constant. However, one has to make sure that the supplied pressure difference or mass flow is also scaled with this factor, as the nominal values do not increase said mass flow.

Known Limitations

• The n-th order element has a big influence on computational time. Reducing the order or disabling it completly will decrease computational time.
• Reversing the mode: A normal 4-way-exchange valve suffers from heat losses and irreversibilities due to switching from one mode to another. Theses losses are not taken into account.

Contents

Revisions

• May 22, 2019 by Julian Matthes:
  Rebuild due to the introducion of the thermal machine partial model (see issue #715)
• November 26, 2018  by Fabian Wüllhorst:
  First implementation (see issue #577)