Skip to content

Custom COP model

This document describes how COP of arbitrary heat pump models can be modelled and integrated, based on the model developed by Tekniker by deploying the DOE-2 building energy analysis program, to yield a statistical model by leveraging measurements of select heat pump manufacturers under predefined operational conditions.

COP model training

The training of the Tekniker COP model produces a JSON file that contains the COP model parameters for a specific heat pump. This file will subsequently be used by the HeatPumpStrategy model in the Grid Singularity Exchange, in order to more accurately approximate the select heat pump COP model. To train the COP model, performance data for the selected heat pump is required, pursuant to the characteristics described in the following section.

Heat Pump performance data characteristics

Heat production data

A table of heat production (Q) in kW needs to be provided for the full load operation of the heat pump. The heat production is represented as a function of air temperature and condenser temperature. The COP model will produce valid data only for the provided temperature range. Consequently, a wide range and a high resolution of input data will result in more realistic output values.

Electric power consumption data

For the same data points as for the heat production data, the corresponding electrical power consumption (P) in kW under full load needs to be provided.

Partial-load data

The partial-load data represents the full range of load to be simulated by the model. A combination of Q and P for different loads of the heat pump needs to be provided in table form. There must be a minimum of three distinct partial-load data points for the same temperature combination.

How to train COP model for the heat pump strategy

  1. Install the software requirements with pip install -r requirements.txt
  2. Create an Excel file in the input_data folder with 3 tabs each containing tables listed in the former section. Example files can be found in the input_data folder.
  3. Run python cop_model_fitter.py that will read all input data in excel files, fit the models and save the model parameters into JSON files.

How to integrate the COP model into the gsy-e

  1. Move the newly generated JSON file to src/gsy_e/models/strategy/energy_parameters/heatpump/cop_models/model_data
  2. Add an entry in the COPModelType enum ** that is named after the heat pump model.
  3. Add entry in the MODEL_TYPE_FILENAME_MAPPING** mapping where the value should be the JSON file name that contains the model parameters.

** = can be found in the /strategy/energy_parameters/heatpump/cop_models/cop_models.py module

Now, the pre-trained COP model can be selected in the HeatPump strategies by providing the newly added enum to the cop_model_type input parameter.