The latest advances in sensor technology, data generation and computing have the potential to profoundly change areas of our economic and daily lives. The complete automation and control of entire systems such as cities (smart cities), power grids (smart grids) or industrial processes (Industry 4.0) is increasingly becoming a reality in the course of digital transformation.
The aim of the NNCR "Dependable Ubiquitous Automation" is to advance the methodological and technological bases for the large-scale implementation of such complex systems. By improving decision-making and control procedures and developing new algorithms and computer methods, the reliability and flexibility of intelligent systems can be improved. The new findings will be applied in the fields of energy management, mobility and advanced manufacturing. One of the NCCR’s key projects is to develop and implement a fully automated and decentralized energy management system at district or commune level. This will allow the economic potential and social impact of automated applications to be tested in real life.
K3 - Handwerkcity
The overall objective of this project is to show a system-related contribution of the Swiss gas industry to the implementation of the Swiss Energy Strategy 2050. Empa will provide a quantification of energetic flexibilities and the free capacities. Additionally, we are conducting an optimization for future adaptions of operation. We evaluate the system stability based on measurement data, and we will evaluate economical factors related to electrical self-sufficiency in the K3 building complex.
K3, a commercially used building complex, will serve as validation area. Its energy system consists of roof and façade PV systems, air-to-water heat pumps, water-to-water heat pumps, a CHP unit and several hot- and cold-water storage units.
A major barrier limiting the installation of photovoltaics are voltage quality problems they cause in the distribution grids. Instead of expensive grid reinforcements, these issues can be solved by using the photovoltaics’ inverters to optimize the feed-in of reactive power. Today parameters controlling reactive power are pre-set at the factory and do not consider the situation in the local grid.
By using grid measurements and a self-learning algorithm the proposed method can optimize parameters in-situ without the need for a network model. The parameters can be updated regularly, i.e. to reflect different seasonal power flows or the addition of new plants.
In this project the performance of a developed algorithm is tested on the ehub infrastructure.
Existing infrastructure in buildings, such as heat pumps or domestic hot water heaters is usually operated to purely satisfy the thermal needs of residents. However, thanks to thermal storage capacities available in buildings' thermal mass or water tanks, there exists energy flexibility in when to charge these storages. These flexibilities can be coordinated to satisfy other interests, e.g. from other players in the electric or thermal distribution grid. In a joint project with aliunid AG, controllers that optimally utilize these flexibilities are developed and implemented in the NEST demonstrator.
Data-driven, self-tuning controllers for large scale deployment
More than 90% of industrial controllers are still based on proportional-integral-derivative (PID) and rule-based (RB) control algorithms, as no other algorithms match the simplicity, clear functionality, and applicability of these two types. However, during the lifetime of a system, due to aging or exchanging of some components of the system, the overall process conditions typically change. This leads to sub-optimal control performance, with direct or indirect operational costs and calls for the manual re-tuning of the controllers. More sophisticated controllers exist, such as ones based on model predictive control (MPC), but they typically require complex physics-based models for proper functioning, which can be quite costly to obtain for certain systems.
In this project we aim to develop new data-driven self-adaptive control strategies and compare them to the classical and state-of-the-art adaptive strategies. We will particularly look at the scalability potentials of these algorithms. Our current efforts involve the development of data-driven self-learning controllers for heat pump systems, room temperature control, and smart charging of bidirectional EVs when coupled to buildings and grids.
Link to the project page
Coherent Energy Demonstrator Assessment (CEDA)
In Switzerland, six energy demonstrators serve as research platforms for different technologies, systems, and scaling. In their analysis, CEDA will standardize these demonstrators in order to show the impact of existing technologies on nationwide implementation in Switzerland. For this purpose, Switzerland-wide communication and collaboration between four SCCERs will be carried out with a total of six demonstrators. The harmonized data collection and modelling of state-of-the-art technologies makes it possible to plan their use in industry more effectively. Case studies will be carried out in order to obtain a direct benefit from the foundation that has been developed. Empa's ehub provides ideal conditions for other demonstrators that are still in the planning or construction phase.
Algorithmic Regulation & Control: A Novel Hybrid Data-driven Approach for Enhancing Building Performance along the Life Cycle (ARC)
In recent years, performance monitoring of buildings has become more common and the volume and resolution of data produced has increased significantly. This data opens up new possibilities for systematically improving building performance throughout the life cycle. Unlocking the potential for large-scale, data-driven performance improvement, however, necessitates the realization of effective feedback loops across different timescales (seconds to years) and spatial scales (building, city, canton, etc.). These feedback loops must be driven by a combination of rich, real-time data describing the state of the building stock and intelligent learning algorithms capable of effectively identifying and adapting feedback signals.
The aim of this project is to develop a transferrable methodology for algorithmic regulation and control of buildings and districts. A prototype ARC system will be implemented at NEST, the ehub facilities and the emerging dhub infrastructure. The project will result in the methodological elaboration and validation of the ARC approach, and a quantification of its potential to improve the energy performance of the Swiss building stock.
Renewable Management and Real-Time Control Platform (ReMaP)
The multidisciplinary demonstration project ReMaP will develop a flexible, software- and hardware-based, modular research platform for assessing potential energy system solutions for the neighborhood of the future. ReMaP will enable the testing, analysis and optimization of multi-component, multi-energy carrier systems on the distribution level, fostering the collaboration of multidisciplinary research teams from both academia and industry, and will furthermore provide a control and communication infrastructure for the joint operation of existing platforms and demonstrator sites. A large number of institutes at ETH Zürich, Empa, and PSI have committed to carrying out research projects using the platform through projects that feature inherent commonalities and that set ideal conditions for fostering further collaboration between these research groups.
Ecobilan Dynamique des Bâtiments (EcoDynBat)
The aim of the EcoDynBat project is to study the influence of temporal variability when calculating the environmental impacts of consumed electricity in buildings. This work will consider the temporal fluctuations of 1) national electricity generation, 2) electricity imports, 3) network losses and conversions, 4) decentralized electricity generation and 5) electricity demand within buildings.
Data Predictive Control
The building sector is responsible for more than one third of the global final energy consumption. Heating and cooling of buildings requires approximately half of this energy. Improving the operation of heating and cooling systems therefore has a significant impact on the mitigation of the climate change. Model Predictive Control (MPC) has been shown to be an energy efficient approach for building climate control. However, the costs to generate the required first-principle based models might be a bottleneck for widespread industrial application of MPC in the building domain.
This project aims at the replacement of first-principle based building models by data-driven models in the MPC framework (called Data Predictive Control (DPC)). Recent experiments with data-driven models based on adapted Random Forests revealed the high potential of DPC for energy efficient climate control in residential buildings. Further comparisons with state-of-the-art controllers like conventional MPC as well as the implementation of the DPC algorithm into an embedded system, are part of this project. With the implementation into an embedded system, we are able to test and evaluate a thermostat retrofit case for residential buildings.
Swiss Competence Center for Energy Research – Future Energy Efficient Buildings & Districts SCCER FEEB&D
The vision of the Swiss Competence Center for Energy Research on Future Energy Efficient Buildings & Districts (SCCER FEEB&D) is to develop solutions for the Swiss building stock, which will lead to a reduction of the environmental footprint of the sector by a factor of three by 2035 thanks to efficient, intelligent and interlinked buildings.
The SCCER FEEB&D is addressing this challenge in a combined effort of leading Swiss and international partners from academia, industry and the public sector.
Funding body: Innosuisse
Partners: Empa, ETH Zurich, EPFL, HSLU, Uni Geneve, FHNW
Contact: Matthias Sulzer