At the division of Applied Mechanics and Energy Conversion (TME) of the Department of Mechanical Engineering, a research group is working on Thermal System Simulation (The SySi) to optimize performance of thermal systems in buildings and clusters of buildings through integration of control and design. Research topics include among others: development of an automated toolchain for model predictive control (MPC), integrated optimal design and control of GEOTABS buildings, packaged geothermal system solutions, integration of renewable energy sources (RES) in building energy systems, active demand response and flexibility offered by building thermal systems, the role of thermal energy storage in optimal design and control of building energy systems on building and district level, innovative energy conversion technologies, … The SySi group, led by Lieve Helsen, has gained significant expertise and international recognition in the field of innovative integration of control and design for performance optimisation of thermal systems. Today The SySi group consists of 1 professor, 3 post-doctoral researchers and 6 PhD students. The research has very close links and is integrated within EnergyVille, a research collaboration between KU Leuven, VITO, Imec and University Hasselt on sustainable energy and intelligent energy systems. The mission of The SySi research group within the Mechanical Engineering Department is to optimize performance of thermal systems through integration of control and design. Their experience in Model Predictive Control (MPC) of energy systems in buildings is now being valorized. The aim is to start up a spin-off company within two years. The path towards this spin-off company needs among others some improvements in the non-linear optimization solver. For this work we are looking for a post-doctoral researcher with expertise in non-linear optimization solvers, either for 1 year full-time or for a longer period (maximum 2 years) part-time (minimum 50%).
Thermal systems in buildings are responsible for 15 % of the primary energy use and Model Predictive Control (MPC) is a promising technology for achieving cost and energy savings. TACO - Toolchain for Automated Control and Optimization - is designed to significantly reduce the engineering time and expertise required for MPC and other optimization-based technologies for applications in the control and design of thermal systems in buildings with possible extensions, e.g. to building districts. TACO is implemented based on the open-source JModelica platform, which is a compiler for the Modelica modelling language. TACO uses this compiler to extract model equations from a hierarchical object-oriented Modelica model. Using these equations, the non-linear MPC code is constructed automatically, using the CasADi algorithmic differentiation framework. The resulting NLP is solved using the interior-point solver IPOPT. A Modelica model library that contains models for the building and thermal system components, IDEAS, is developed by the same research group (in collaboration with others). TACO is currently further developed with the goal to commercialize the toolchain.
Technical job description
This application-oriented position at the Thermal Systems Simulation (The SySi) research group of KU Leuven aims to improve the performance of the toolchain with respect to computation time, optimality, robustness and automated diagnosis in case of optimization problems, with a focus on the NLP solver side of the platform.
To improve the performance of IPOPT, it can be modified, changed to adifferent class of solvers, or a new solver that is tailored to the specificproblem structure of TACO will be developed. In addition to these convergence aspects,an approach for better solving problems including local minima and integerdecision variables will be developed, and the algorithms can be parallelized toreduce computation time.
We are looking for a highly motivated, enthusiastic, dynamic, mobile and communicative researcher with preferably a PhD degree (or equivalent experience) in Engineering or a related field. The primary focus of this position is on numerical non-linear optimization solvers. The applicant should have a strong background in non-linear optimization and in broader mathematical theory and should have experience with implementing efficient computer codes, preferably using C or C++.
Experience with Modelica, building energy simulation, a background in mechanical engineering, algorithmic differentiation and CasADi, integer programming, parallelized code development are additional benefits.
The candidate is expected to:
Be scientifically rigorous, initiative taking, results orientation, loyal.
Be proficient in English (spoken and written) to allow effective communication, knowledge of Dutch is an asset
Be able to work independently, accurately and methodically
Be a team player, using the broad network offered
Publishing & IP
This position is funded within the scope of an innovation project. For intellectual property reasons there may be constraints to what can be published publicly.
This post-doc position will be fully funded through a University Innovation Project. The research is to be performedwithin The SySi research group in close collaboration with two other post-doc researchers, and in the framework of EnergyVille (http://www.energyville.be/en),a research collaboration on sustainable energy between KU Leuven, VITO, Imec and University Hasselt.The location for this position will be Leuven (Department of MechanicalEngineering).
The successful candidate will receive:
Multiple benefits (health insurance, accessto university infrastructure and sports facilities, etc.)
A very competitive salary, including socialsecurity
The opportunity to participate in researchcollaborations and international conferences
The city of Leuven,just 20 km east of Brussels offers a stimulating, youngand multicultural working environment
A start date is to be agreed upon, preferably April 1, 2019.
This position is at the post-doc level, but having a PhD degree is not strictly required if the required skills are clearly present. This job can be taken either full-time for 1 year or part-time (down to 50%) for a longer period (maximum 2 years).