Date of Award
Campus Access Thesis
Roger A Dougal
In this thesis, two methods were developed for applying quantized discrete event methods to simulation of naturally-coupled systems. The first method, which did not meet the ultimate objective but did prove useful as a learning experience, applied a DEVS solution to the modified nodal analysis method. The second method applied a combination of quantized discrete event techniques with latency insertion techniques to produce a circuit solver.
In the first part of this thesis (chapter 2), the first method for applying quantized discrete event simulation (DEVS) methods to naturally coupled system models is defined. Pairing the DEVS formalism with the Modified Nodal Analysis technique allows for asynchronous discretization of events, instead of the typical uniform time discretization while automatically enforcing natural conservation laws, unlike signal flow methods that propagate the output of one model to the next without adding physical constraint equations. The mathematical equations for naturally coupled DEVS (NCDEVS) are presented along with example formulations for inductor and capacitor circuit elements. Results from application of the technique to an RLC circuit and to an LC filtered Half-Wave rectifier driving a Permanent Magnet DC Motor (PMDC) are presented and shown to be close to the well-known analytical solutions. A significant benefit of the NCDEVS formulation is that it allows bi-directional power flow, as shown in the case of the PMDC motor, therefore the solver doesn't need to be aware if the PMDC motor acts as a motor
or as a generator. Moreover using NCDEVS, the motor is modeled by one set of equations, regardless of the operating mode, in contrast to the signal flow method, where the solver needs to have two separate models, one for each case. In the second part of the thesis (chapter 3), we present the second method that solves electrical circuits by using the properties of Quantized Discrete Event Simulation (QDEVS) and Latency Insertion Method (LIM). The combination of these two methods
allows us to perform natural-coupled simulations using LIM while speeding up the simulation through asynchronous variable updating capabilities of DEVS. Through a proof-of-concept implementation in ACSL (Advanced Computer Simulation Language), we were able to calculate the time that each component`s events will occur thus defining the specifications for a more general future solver that can support independent time stepping for each component. The second part also presents a hybrid DEVS -LIM solution of a series RLC electrical circuit and the ACSL code that implements the combined DEVS-LIM scheme. Furthermore an analysis of the computational performance is provided as well as a comparison between the DEVS-LIM implementation and the traditional LIM in discrete time.
MAMAI, C.(2012). Applications of Quantized Discrete Event Simulation Methods. (Master's thesis). Retrieved from http://scholarcommons.sc.edu/etd/2193