Co-simulation Between Trnsys and Simulink Based on Type155

Abstract

The interface between Trnsys and Matlab based on Type155 of Trnsys’ standard library is used to construct a co-simulation between Trnsys and Simulink. The high flexibility of this interface is demonstrated, which includes its ability to provide a communication in strong and loose coupling schemes. A simplified use case including a compact thermal energy storage is considered to discuss accuracy and computational demands for various settings of the co-simulation. Loose coupling with constant and linear extrapolation of the input variables is presented, as well as an application of the strong coupling scheme to estimate the inaccuracies of the loose coupling co-simulation.

Appendix: Detailed Settings and Results

The essential solver parameters used in this study are given in Table 1. In all cases discussed here, the Trnsys solver is the successive one (modified Euler) with a relaxation factor of 1. Table 1 also gives a comparison of the performance of the different co-simulation setups in terms of accuracy and computational costs.

Table 1. The parameter settings for the different simulations as well as performance indicators for the (co-)simulations for accuracy and computational demands are given. “Ref.” denotes the reference simulation, “Monolithic” a Trnsys simulation without co-simulation, “default” the default parameter setting ($t_\mathrm{step}$ = 1 s and tol. = $10^{-6}$), “lin. extrapol.” the loose coupling with linear extrapolation of the input variables, “relaxed” the relaxed solver parameters (to match the accuracy of the BCVTB/FMI based co-simulation). $t_\mathrm{step}$ denotes the Trnsys time step; “call.Mode” abbreviates “callingMode”, which is the parameter of the Type155 governing the commuication pattern; and “tol.” gives the relative tolerance. $\varDelta T_b$, $\varDelta T_s$, $\varDelta T_\mathrm{s,out}$ and $\varDelta T_\mathrm{s,in}$ refer to the respective maximum deviation of these variables in units of Kelvin, and “comp.” denotes the computational demand of the respective simulation. The simulations were executed on an Intel Xeon E5 6C/12T 2.2 GHz with 96 GB RAM.