Model-based compensation of sensor failure in industrial gas turbine

Panov, Vili and Maleki, Sepehr (2017) Model-based compensation of sensor failure in industrial gas turbine. In: 1st Global Power and Propulsion Forum 2017, 16 - 18 Jan 2017, Zurich, Switzerland.

Documents
GPPF_2017_paper_129.pdf

Request a copy
[img] PDF
GPPF_2017_paper_129.pdf - Whole Document
Restricted to Repository staff only

282kB
Item Type:Conference or Workshop contribution (Paper)
Item Status:Live Archive

Abstract

This study investigates application of analytical sensor
redundancy to improve reliability of gas turbine control
system. Analytical redundancy, which uses a reference
engine model to provide redundant estimates of a measured
engine variables, has been utilized as a basis for the proposed
sensor fault detection and accommodation method.
Model-based compensation of measurement fault is
usually resolved by introducing virtual engine sensors, which
are obtained via accurate engine modelling. In this paper, a
real-time dynamic gas turbine engine model is used in order
to generate the redundant virtual measurements. The engine
model accuracy directly determines the validity of the modelbased
approach for sensor fault diagnosis, and hence a model
with auto-tuning capability is deployed as a reference for the
gas turbine.
The proposed fault detection technique examines the
residuals between the redundant channels. Once the
discrepancy between the virtual and the sensor measurement
exceeds the prescribed tolerance levels, the sensor fault
diagnosis determines the state of the switching logic in the
dual lane control configuration. The deployed logic is also
used for reconfiguration of the auto-tuning process. When a
sensor fault occurs, the estimation process is affected, and
hence the tuning process must be adjusted to account for this
deficiency.
Single and multiple sensor failures are simulated during
the gas turbine transient manoeuvre to assess capability of
the proposed model-based detection and accommodation
method. Hard (large in-range) and soft sensor failures (small
in-range or drift) are injected during the numerical simulation
and results are presented.

Keywords:Gas Turbines, Sensor fault detection, Model based approach, Analytical redundancy, Control Systems
Subjects:H Engineering > H660 Control Systems
Divisions:College of Science > School of Engineering
ID Code:25919
Deposited On:02 Feb 2017 21:05

Repository Staff Only: item control page