Matlab/Simulink

Matlab and Simulink are used regularly in both academic and industrial work.  The Mechanical Controls Laboratory exercises teach modeling in Simulink.  Each system's components are modeled piece by piece by running tests on them.  Component modeling yields a number of simple Simulink models that then must be assembled into a working simulation of the entire system.

For example, in the two-tank experiment in Cal Poly's controls lab, each tank is modeled as a first order system with a time constant and a steady state gain.  These tank models are then assembled into a combined system for both tanks.  Models of the control valve and the controller are added to comprise the entire tank level control loop.

Once all of a system's components are modeled and assembled into the overall control loop, a step input is imposed on the system, whose response is recorded.  This is done both with the actual, physical system and with the simulated system.  If the simulation is true to the actual system, its response to the step input should be close to the actual response.  If the responses differ, the simulation does not model the real system accurately.  In this case, students must troubleshoot the simulation to find out where the simulation diverges from the actual system.

In industrial work a simulation often is created to model a physical system that has been proposed.  The simulation, if done correctly, will allow an industry to see how a proposed system will behave even before it is prototyped.  Often an industry will want to modify an existing product to scale it up or down or make significant changes to it.  They might simulate the system to see the effect of the scaling or modifications.  I was involved in such an endeavor for a downhole tractor.