Hanby, V.I.; Dil, A.J.
August 1995
Applied Mathematical Modelling, vol 20-8, p. 608-613
Markov modeling offers the possibility of extracting long-term results from dynamic simulations with a significantly reduced execution time over that which would be necessary with an equivalent time-series simulation. The discretization of the problem variables which is necessary for Markov modeling introduces an inaccuracy into the simulation process which means that a balance must be struck between the accuracy required and the time reduction that is possible. The paper describes an approach to the quantification of the errors introduced by discretization of the simulation variables. Two error expressions were tested by running a representative building/HVAC system dynamic model in both time-series and Markov modes with varying degrees of discretization of the variables. A simplified error bound was found useful in identifying near-optimal discretization schemes: further refinement of this approach led to an expression that was found to give a good prediction of the error in Markov simulations of this type of system.
Source: www.sciencedirect.com