UWE Bristol Engineering showcase 2015
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Charles Taylor<br />
BEng (Hons) Electrical/ Electronic <strong>Engineering</strong><br />
Project Supervisor<br />
Hassan Nouri<br />
Real Time Fault Analysis of AC Induction Motors<br />
Introduction<br />
Water companies can face large prosecutions and<br />
fines if the water or sewerage quality parameters<br />
goes outside of the consented limits. Any<br />
unforeseen breakdowns in mechanical or electrical<br />
equipment can result in critical items of plant<br />
being left un-operational for periods of time whilst<br />
a repair is carried out, leaving the plant exposed to<br />
potential failures<br />
This project set out to investigate, design and build<br />
an instrument that can monitor real time<br />
condition of plant and to detect potential failures<br />
at their inception.<br />
There are various methods of fault detection<br />
though they often tend to rely on mechanical<br />
instruments which are inherently expensive to fit<br />
and maintain. The use of motor current signature<br />
analysis to carry out real time fault diagnosis is<br />
one that has been widely researched and proven.<br />
Common Faults That Were Identified<br />
Below is a list of common faults that regularly<br />
occur in items of plant in the water treatment<br />
industry.<br />
Electrical Faults<br />
Electrical Over load, nine to ground fault, Line to<br />
line fault, two phase line to ground fault, three<br />
phase line to ground fault, unbalanced supply<br />
voltage, over voltage, under voltage and single<br />
phasing<br />
Pumps Failures<br />
Impellor blockage, Non return valve failure and<br />
Pressure head loss due to pipe burst or rupture<br />
Mechanical Faults<br />
Bearing failure, Motor Unbalanced load or rotor<br />
shaft misalignment<br />
Fault Signature Analysis<br />
In order to be able to identify various current fault<br />
signatures a number of real life experiments on<br />
various pieces of electrical equipment on Wessex<br />
Water’s Sewerage treatment works was set up.<br />
Additionally current signatures were collected on<br />
both healthy and unhealthy equipment and a<br />
comparison made between the two. The<br />
signatures were logged and collected using a<br />
Power Quality Analyser. The data for each test was<br />
then download into Matlab where a Fast Fourier<br />
Transform was carried out to display the current<br />
signature harmonics.<br />
Fuzzy Logic<br />
A hybrid fault identification algorithm was<br />
developed based on the fault identification<br />
method of fuzzy logic but with the added ability to<br />
self learn and back propagate as the Neural<br />
Network does.<br />
Hardware Design<br />
A design by Microchip for a three phase energy<br />
meter with integrated DSPIC was used to become<br />
the Pump Condition Monitor Tool. A PCB was built<br />
to provide three phase 415V voltage and current<br />
signatures .<br />
Software Design<br />
The PCM tool was designed to communicate its<br />
current status, power quality parameters and fault<br />
alarms via Modbus to a Programmable Logic<br />
Controller.. Additionally a Graphical User Interface<br />
was also built to allow the user a detailed insight<br />
into the condition of a pump as it is running.<br />
Project summary<br />
This project set out to investigate, design and<br />
build a system that can monitor real time<br />
condition of pant and to detect failures of<br />
plant at their inception.<br />
The instrument is intended to be used by<br />
utility company's so is required to be able to<br />
communicate via Modbus to a Programmable<br />
Logic Controller<br />
Project Objectives<br />
The objectives of this project were to<br />
investigate a method of carrying out real time<br />
fault analysis of AC induction motors, in<br />
particular the centrifugal pump widely used<br />
by utility company's.<br />
Once a method of fault identification had<br />
been established an instrument needed to be<br />
designed and built that would work on any<br />
centrifugal pump as a stand alone unit, 24<br />
hours a day and that would integrate and<br />
communicate with a Programmable Logic<br />
Controller to alert operational staff of the<br />
need to carry out potentially invasive<br />
maintenances at the faults inception.<br />
Project Conclusion<br />
The project proved to be a good success. A<br />
method of detecting and identifying all of the<br />
faults required was achieved. A instrument<br />
was built that was able to operate on three<br />
phase pumps and detect failures.<br />
The instrument was able to communicate via<br />
Modbus to a Programmable Logic Controller.