Turbine blade implants measure at 32,000g.
By combining modern micro-machined silicon technology
with 19th Century type vitreous enamel, an Oxford University
research group has come up with a means of mounting
sensors on turbine blades which does not interfere with
True turbulent flow conditions inside a turbine operating in a realistic
manner can now be accurately monitored for the first time, unhindered
by possible interactions between sensors, their leads and the gases
Mounting a turbine rotor in a test rig.
The shaft mounted electronics associated with testing.
This will lead to the ability to model fluid flow under conditions of
extreme turbulence much more accurately and confidently than has
been possible up to now. And it also demonstrates a neater, more
effective way of building in pressure, temperature and other types
of sensors into test rigs, than is normally the case.
Fluid flow under extreme conditions of turbulence,-those to be
expected in turbines, hydraulic valves and most other situations
of real engineering interest- have so far proved impossible to
model fully on computers. The problems are caused not so much
by lack of computing power, as a mathematics which does not lend
itself to solution with conventional machines. One Oxford physicist
is even now engaged in developing a totally new kind of computer
to try to solve the problem but until he succeeds, proper mathematical
modelling and design of systems involving highly turbulent flow still has
to be based on painstaking experimental measurement.
Micro-machined silicon pressure sensors bonded into a turbine blade.
Nowhere is this more true than in the case of gas turbines, and for some
time now, Roger Ainsworth and his team from the university's Department
of Engineering Science have been making full scale measurements of
turbine gas flow behaviour in what used to be the old Oxford Power Station.
They have a test bed, which allows 1 .6m3 of air at 115 lb/in2 (8 bar) to be
exhausted through a single 2ft diameter rotor and stator in 0.2s (representing
a transient power output of 2MW). It would all be in vain were it not for the
effort the team has put into finding a way of building sensors and circuitry
into the turbine blades without distorting gas flow.
Both pressure and temperature can be measured, by making use of the
properties of vitreous enamels of the same types as used on domestic
cookers. In the case of temperature sensors, the blades are enamelled
before platinum resistance gauges and gold leads are painted on, using
four coats of 'Hanovia Liquid Bright' from Englehard. The blades are then
fired at 640°C, a process which results in measurement sensors and
leads being endowed with an overall resistance of about 50ohms.
But even more remarkable are the measurements with micro-machined
silicon pressure chips from Kulite. First, a coating of vitreous enamel is
laid down and fired to provide an insulating layer, followed by wires, in
a machined groove. Then follows a second coating and firing followed by
milling to expose the wires. Fifteen Kulite micro-machined silicon pressure
sensors are then stuck in the groove with epoxy and connected up to
provide a ring of sensors capable of withstanding 32,000g.
Personally developed by the president of Kulite Semiconductor products,
Anthony Kurtz, the chips use piezo-resistive strain gauge elements diffused
into a flexing diaphragm bonded onto glass. Most remarkable from Roger
Ainsworth's point of view has been Kulite's ability to turn round new chip
designs in no more than a week.
The project has been referred to by Geoff Bancroft, Kulite's British managing
director as the "Jewel in their Crown", both because of the importance they
are giving it and the jewel-like look of the sensors mounted on the blade.
The turbine test rig is also of more than passing interest. Invented in its
original form in the early 70s by Professor Jones to test stationary
cascades of turbine blades, the team have modified it to incorporate
a fully rotating turbine stage, supported by Rolls-Royce and SERC.
The driving reservoir of air is compressed by a piston in a cylindrical
chamber, 5.5m long and two feet in diameter, pressed on its outer side
by compressed air from cylinders. The air in the reservoir is then held
in place by a rotary gate valve of 30 segments, across the entire
cross-section of the cylinder. The turbine is initially run up to 6,500rev/min
by an air motor, whereupon a perspex cylinder is shattered by a detonator
to release the gate valve, which until then has been held open against
compressed air at 65 lb/in2 (4.5 bar).
The valve opens fully in 40ms and in the subsequent 200ms, the turbine
accelerates by a further 3,000 rev/mm. Data is grabbed from 8 of the
32 sensors as the turbine passes 8,400 rev/mm via slip rings, gathering
signals which have already been pre-amplified by electronic circuitry
built into the turbine shaft.
The bearing have to be preloaded during the run up, since they were
designed to run only under load, and the outward side of the turbine is
initially evacuated to 0.1 lb/in2 (7 mbar), a pressure which rises to
0 lb/mm2 (0.7 bar) by the end of the run.
The same embedding and painting techniques using vitreous enamel
could also be applied to the painting or implanting of strain gauge and
other sensors and the whole method is of general application to the
mounting of sensors on test rigs, whether to maintain unhindered gas
flow or merely to protect them.
The ideas can be further extended to the production of vitreous enamel
on steel-printed circuit boards. Such items have been manufactured
experimentally but are not in production because of micro-cracking
problems which appear in long-term use - particularly in the domestic
cooker environment for which they were originally intended.
From an article by Tom Shelley, published in the magazine 'Eureka'