Title

Session

Authors

Abstract

Mechanical ventilation systems are becoming more and more necessary in increasingly airtight buildings and they can contribute to high energy efficiency when used in combination with heat recovery systems. The prerequisite for this is that the energy consumption of the fans is low. This can be reached by (i) reducing the required hydraulic power and (ii) making the conversion as efficient as possible. One approach to decrease the hydraulic power (i) is the decentralized placement of fans in a central duct network (so called semi-central systems) to avoid throttling losses. To increase the efficiency (ii), the interaction of components must be balanced. Especially with fluctuating loads and complex topologies, countless variants have to be compared and evaluated for this purpose. In order to still master the combinatorial explosion, algorithm-based design methods from the field of discrete mathematics are a promising approach.
Therefore, in this contribution we present an approach to design semi-central ventilation systems for non-residential buildings using discrete optimization methods. We first present the methodology to select the placement of the decentralized fans based on an assessment of the hydraulic power and throttling losses. Then we present the techno-economic model of the components and the system. Based on this, the design task – the selection, interconnection and operation of the fans – is formulated as a mixed-integer nonlinear optimization problem. Here, the life-cycle costs are minimized considering different volume flow demands due to the occupancy in the rooms. For the solution, exact optimization methods from the field of discrete optimization are used, which have the capability to ensure the global optimality of the solution. This is applied in a case study, in which we optimize the ventilation system of a building of the Technical University of Darmstadt. Finally, the results are discussed and the potential of our approach for cost and energy savings is demonstrated.