Project info

Buildings systems and railway traffic systems are examples of complex technical systems that consume significant energy amounts so as to enable their inner processes to evolve as required (required indoor comfort or travelling time). Thereby the requirements on dynamic functioning of these systems can be achieved by different system interactions, whereas some of them are more preferable than the others from the standpoints of energy consumption or other criteria like price of operation or equivalent pollution.

Drastic changes in energy grids are obvious to happen in few years time due to large penetration of dispersed renewable energy sources and new types of consumers like electric vehicles chargers. These changes will most of all be visible in very dynamic shortages and excesses of energy that will have to be balanced in real time. The lack of mature, efficient and accessible large-scale energy storage technologies will require to enforce coordination between dispersed production and consumption points in the grid. The coordination of such legally independent systems will have to be done on economically sound bases. Thus time-varying prices of energy exchange and of ancillary services on a very dynamic time-scale of hours or
less are expected to become a reality soon. A prerequisite for viability and success of such a scenario is to have an automated intelligent agent on the consumer side that decides how to balance local production, consumption and storage to yield most advantageous results for the consumer. This will unlock significant possibilities on the grid side.

Especially challenging nowadays becomes to internally manage large consumers like buildings or traffic systems in order to benefit from a triplet of grid exchange conditions, environmental conditions and internal states/requirements. The inherent complexity of such systems prohibits to consider all the interactions in a single control problem, such that distributed or hierarchical control approaches stand up-front.

In both buildings and railway traffic systems a clear hierarchical structure can be identified, consisting of energy consumption and power flows management levels.

The level of energy consumption in buildings is responsible to achieve proper users comfort in an energy-efficient and cost-optimal way. The level of balancing power flows between building production, storage and consumption points takes into account the current and near-future: availability of local renewable energy, required consumption, storages state and grid exchange conditions. Thus the actual price of energy consumption to achieve proper comfort heavily depends on the management strategy on the power flows balancing level, while that strategy itself depends on the presumed consumption profile. One may say that prices of energy exchange with the grid are transformed through the level of power flows balancing and with such transformed content affect the consumption level.

In railway traffic systems each train can be controlled to achieve least travel costs while maintaining the time-table and travellers comfort. Meanwhile, the cost of electricity can be quite different in time, depending whether it is sourced from another train in breaking or from the electricity grid. The higher-level railway traffic coordination system is here introduced for trains coordination with respect to external grid conditions, routes conditions, time-table requirements and trains current position on the routes. Again one may say that price of energy exchange with the grid is for an individual train transformed through a traffic coordination system level by cooperative action of all the trains in balancing power flows.

The global objective of the project 3CON is to research and develop optimal control techniques on significantly different large consumers – building systems and railway traffic systems – in conceptually identical way: by applying the principle of hierarchical decomposition of systems and cooperative optimal controls between the hierarchy levels. This will enable the proactive role of these largest consumers in energy grids of the future (smart grids), and will make them responsive and adaptive to varying energy exchange terms. The secondary global objective of this installation grant is to build an internationally competent research group in control of energy-intensive systems on the Faculty which is also recognized and accepted by relevant domestic companies.

Building and railway traffic systems have of course been in focus of software/hardware industries for decades since significant benefits by such large consumers can be achieved even in few percents reduction of energy consumption, and thus industrial tools are available. There is however a lack of research that considers mentioned grid prices transformations through levels of energy flows management towards consumption levels which consider the predicted consumption itself, availability of renewable sources and offered energy exchange terms with the electricity grid. The 3CON project intends to fill in this gap with a vision to fully exploit the smart grid possibilities and local renewables potential through design of cooperative hierarchical control levels. On the grid level such controlled buildings or neighbourhoods will represent responsive consumers ready to fully exploit dynamic energy pricing conditions. In this way vast possibilities in energy grid management through dynamic pricing will be opened – foremost for reducing losses between generation and consumption points and increasing grid reliability.

Traction force planning on the rail route is from the standpoint of the proposed research in a situation similar to building energy management, if not somewhat worse due to very conservative approach in railways stemming from safety requirements. The methods for optimal traction force calculation are developed [Franke et al., Liu et al., Vašak et al. 2009]. It was also shown on a conceptual level that the latter approach is extendable on the level of railway traffic system [Vašak et al. 2010] based on the availability of energy cost characterisation for different travelling times, which enables optimal resolution of conflicts with the fixed time-table. Recent work pinpoints the lack of train interaction concepts on the level of neighboring dispatching areas and two-level hierarchical control structure is proposed. The 3CON project undertakes to perform a research on cooperative hierarchical control structure that will optimize the performance of multi-train system operation by taking into account time-varying energy costs stemming from cooperative trains behaviour on single energy links. Similar benefits are expected on the grid management level as in the case of buildings. Taking into account that buildings and traffic systems are the two consumer groups with highest energy consumption share in the overall consumption, vast possibilities on the smart grid side will be opened.

Specific objectives

The specific objectives required to reach the global objective and the secondary global objective are posed in 3CON as follows:

  • SO1. Consolidate, research and develop optimal control methods on the energy consumption level of the building and train-on-the-route system, taking into account the required cooperation abilities with the power flow optimisation level;
  • SO2. Consolidate, research and develop optimal control methods on the power flow optimisation level of the building and railway traffic system, taking into account the required cooperation abilities with the energy consumption level;
  • SO3. Research and develop coordination methods between the energy consumption and power flow optimisation levels for the considered building and railway traffic system;
  • SO4. Verify the developed procedures on building case study of the FER skyscraper building and on railway traffic system case study assessed in cooperation with the railway operator in Croatia;
  • SO5. Derive exploitation strategy for project results in cooperation with the real sector in Croatia and disseminate the project results on the world level.

Since control comes on top of the condensed knowledge of system relevant dynamics and constraints, control of large-scale systems, such as the one proposed in 3CON, regularly opens interdisciplinary and multidisciplinary issues. Among interdisciplinary ones one may notice that realistic electricity prices  fluctuations and data communication constraints in the planned applications may be the issues raised. The multidisciplinary issues could be the ones of binding control and mechanical engineering in the buildings case or control and traffic engineering in the railway traffic system case.

Control-based maximising benefits from the complex system operation will for sure open new research issues in the mentioned interdisciplinary and multidisciplinary fields and the research performed here will set a solid platform to successfully address them.

Methodology overview

In principle, two possible methodological approaches are foreseen to interconnect the consumption and power flow hierarchy levels in both building and railway application: (i) parametric representation of optimal solution and cost on one level, that is then integrated in optimisation on the other level; or (ii) iterative sequential optimisation procedures on both levels where solution on one level is used as an input on the other, for which game theory will be used to provide convergence and optimality guarantees.

For optimal control of linear, piecewise affine and linear hybrid systems parametric representation of the optimal control solution and of the corresponding value function is developed during 2000s, whereas the 3CON collaborators were either directly contributing (Mato Baotić) or closely linked to that research through its applications (Mario Vašak, Nedjeljko Perić). The idea pursued in 3CON will be to parameterise the value function on the power flows optimisation level, related to economical gain for the prosumer, with respect to the consumption amount and timing. This parameterisation would actually represent the transformation of grid-relevant energy prices to the consumption-relevant energy prices. It will intrinsically incorporate also the local energy production (e.g. renewables), energy storages state in the system (e.g. batteries, kinetic and gravitational potential energy of trains), as well as dynamic grid conditions for energy exchange. The consumption amount and timing will be then optimised on the consumption level by directly taking into account the parameterised economical gain for the prosumer.


If the nature of energy storages on the power flow optimisation level will lead to optimisation problems whose value function cannot be explicitly parameterised, two possibilities will be explored: (i) approximate descriptions through piecewise linear models to enable parameterisation and proceed as explained above and (ii) sensitivity analysis of the power flow optimisation problem with respect to optimized consumption which will enable local parameterisation of the value function on the power flow optimisation level, and then re-optimisation on the consumption level with this information included. The approach (ii) will lead to an iterative computation scheme for which convergence and optimality properties will be explored, presumably within the domain of games theory and distributed control.