@MASTERSTHESIS\{IMM2001-0733,
    author       = "O. Knop",
    title        = "Stokastiske modeller for varmedynamiske systemer",
    year         = "2001",
    school       = "Informatics and Mathematical Modelling, Technical University of Denmark, {DTU}",
    address      = "Richard Petersens Plads, Building 321, {DK-}2800 Kgs. Lyngby",
    type         = "",
    url          = "http://www2.compute.dtu.dk/pubdb/pubs/733-full.html",
    abstract     = "The present master thesis concerns stochastic dynamic modelling of heating systems. The thesis is organized into two parts. In the first part a mathematical model for a thermostatic valve is developed based on knowledge of the physical properties of the system. The key issue is that the hysteresis effects due to friction forces in the valve is compensated by an adaptive friction model. 

The presented valve model is a continuous-discrete state space model in terms of stochastic differential equations. Statistical methods and empirical data from controlled experiments are used to estimate the parameters in the model. The model performance is illustrated using independent empirical datasets. Furthermore, the model is extended to account for the heat dynamics of the valve thermostat. The parameters are estimated and the dynamical model is validated. 

In the second part of the project a micro combined heat and power (CHP) unit is studied. A simplified model for a heating system is suggested by exploiting the physical knowledge of the micro{-CHP} in conjunction with empirical energy demand data. 

The heating system is simulated when the micro{-CHP} is operated according to a commonly used heating strategy. This strategy is compared with a optimum operating solution found by dynamic programming. It is found  with a optimum operating solution found by dynamic programming. It is found that the loss of electricity could be reduced 50-75 \% if the {CHP} are controlled in an optimal way. As a result an optimum operation strategy for the micro{-CHP} is proposed based on prediction of the energy demand and dynamic programming."
}