under supervision by
Dirk Hartmann (Siemens Industry Software GmbH)
Felix Sievers (Siemens AG)
Friedrich Menhorn (TUM)
Introduction
Now more than ever, saving energy has become an urgent necessity more than a nifty possibility. For starters, saving energy would mean saving our planet: it is clear and evident to everyone that global warming and climate change are not myths, and they are indeed phenomena that could have tragic consequences for Earth’s habitability. And they will, if they are not stopped, or at least, slowed down in time. Here,nearly 20% of global energy waste is attributed to heating of buildings. Another important motivation can be found in the rise of energy prices: comparing February 2022 with February 2021 expenses in Germany, there is an increase of +129.5% for imported energy, +68.0% for energy produced internally and +22.5% for energy used by households. Optimized control of building heating systems could translate into a 30% reduction in energy consumption due to building operation, and a 2.2 billion tons reduction in global annual CO2 emission. Therefore, improving building efficiency is a crucial task.
To be able to control the heating system of a building in an efficient manner, it is necessary to build a thermal model, which is able to describe the heat exchange. This task is not trivial, since it requires a lot of information about the building which, in the current solutions, needs to be user-defined. The main status quo is EnergyPlus, an open-source program that can model energy consumption, including heating, in a whole building. EnergyPlus is a very well established simulator, but it has a pretty significant liability, namely the need for an input file that is far from intuitive to create for a generic user.
This project’s scope is to create a thermal model that can be fed to a smart heating system, which can efficiently regulate heating in buildings. We start with a simple input that any user would have such as a building plan. In this way, we can generate a model to reduce energy costs quickly, just using a building plan as input, which makes our program much easier to use for anyone.
Our product works in the following way: given the building plan, first it detects rooms and wall connections among them as bounding boxes through a Python based machine learning application. Using the data from the previous algorithm, the program is able to generate a connectivity graph, which fully captures the adjacency of all rooms in the building. Thanks to this connectivity graph generation, we are finally able to generate our thermal model, by creating a complex composition of RC circuits, and simulate it, through efficient Julia packages. This whole procedure results in a model that can be fed to a control system, which can then optimize parameters to produce an efficient, low-energy consuming heating system for the input building. Since our main proposition is to make this procedure accessible to any user, the backend product is paired with an intuitive user interface application. The model is coupled with a PID controller for regulating the heating. We call this application Energy++.
The easy usage for non-expert customers is such a crucial part of our product: thanks to Energy++, anyone can generate and control their own energy-saving heating systems directly in their home. Such widespread use could save a lot of energy and, at the same time, drastically reduce the emissions of carbon dioxide. Our hope is to make our contribution towards stopping climate change, making everyone’s life better today and guarantee a better future, for everyone.