Proceedings of the 9th International Conference on System Simulation in Buildings - SSB2014 issue:accepted pages:1-24
International Conference on System Simulation in Buildings - SSB2014 edition:9 location:Liège, Belgium date:10-12 December 2014
The energy use saving potential by improving the operation of building control is estimated to range from 10 to 30% of total HVAC energy use, in addition to savings from major retrofits performed in the same buildings. Often, inefficient operation is caused by control flaws, equipment flaws or changes in the building use. Continuous Commissioning (CC) is the set of actions taken to optimize the operational efficiency of the HVAC system.
Recent advances in the framework of system identification can help to realize and automate part of this CC process, by using low order predictive models as controller models. However, a crucial aspect for model based CC is the selection of a suitable simplified model.
To demonstrate this point, one specific CC task has been chosen: Optimized heating start-up (OS). For this task, a case-specific comparison is made between several model based algorithms: a first order bi-linear, second order non-linear and high-order black box model, using a detailed multi-zone white box model as an emulator.
The prediction accuracy, simulation performance and control performance in this CC-task are investigated for each of these low order models, taking into account the required information, the model development time and the simulation time. This comparison sheds light on the (dis)advantages of each simplified model structure for the chosen CC-application.