A benchmark two-tank heated system emulating typical fault conditions

Abstract

The need exists for a benchmark multi-domain practical system that could emulate typical process faults for the purposes of condition monitoring. A proposition was made to develop a heated two-tank system, due to its multi-domain nature i.e. thermal-fluid, mechanical, electrical as well as its non-linear nature. The focus of the study is to design and construct a two-tank heated system that is capable of emulating typical faults found in industrial processes. This is done for fault detection and identification (FDI) studies, allowing researchers to test their methods and algorithms on a practical system to validate simulation results. A conceptual design of the two-tank heated system is created by preforming preliminary design calculations in EESTM. The log mean temperature difference (LMTD) method is used to obtain the approximate specifications of the system that will be used during the detail design of the system. A detailed process flow diagram (PFD) of the two-tank system is developed concurrently with a Solidworks computer aided design (CAD) model of the system. A Steady state model of the two-tank heated system is developed in EESTM by making use of the Number of Transfer Units (NTU) method. The EESTM model is verified in FlownexTM using the FlownexTM steady state simulation. The transient behaviour when faults are introduced in the system is simulated in FlownexTM and validated by comparing the results to the experimental results. The experimental system is constructed for the purposes of model validation. The system is designed to allow fault injection by making use of dedicated control valves which are placed throughout the system. Leaks and blockages are emulated by opening and closing the control valves while sensor faults are implemented on software level. Sensor drift is emulated by introducing an offset to the value measured by the sensor. Each of the faults emulated in the system is tested at various intensities. The steady state EESTM and FlownexTM models showed to correlate. The percentage error when comparing the process tank temperatures in the models for each of the emulated faults did not exceed 2%. The experimental results showed to correlate with both the FlownexTM model and the EESTM model producing a percentage error of less than 1.5% when comparing the temperatures within the process tanks.