EnergyPLus multi-objective and calibration optimization

edit

The OpenStudio Analysis Framework OSAF was created for this very purpose.

There are tutorials on how to setup and run analyses here

The PAT gui can help with setting up the problem formation here

You can setup the OSAF to run on local computing hardware using docker / k8s or leverage cloud providers here

Hello, I have used the JEPlus EA's implementation in DesignBuilder to attempt to multi-objective and calibration optimisation. As mentioned by Ammar (or as link provided), you can do it directly in JEPlus or same implementation within DesignBuilder.

In DesignBuilder we can use the Optimisation (JEA based) and Python scripting functionality to create an multi-objective optimisation problem, where you can aim to minimise CVRMSE and NMBE values and use any additional E+ outputs or custom outputs or statistical metrics as constraints.

You will define your typical range of variables such as loads, equipment efficiencies, their schedules or operational parameters using pre-defined or custom variables. The optimisation engine minimised the objectives (error metrics) and python scripting is used to calculate the error between measured results in measured data and simulation results for each iteration in each generation and reported back. Visualisation of such results would be in parallel coordinates.

You can use Idf import to get your EnergyPlus model's Geometry and construction into DesignBuilder.

To setup rest of the analysis you need to write a Python script like in the attached images.