ÐÓ°ÉÔ­´´

THE STEADY rise in the price of oil means that engineers must again
meet the technical challenges of placing people and machines in hostile
environments, such as the North Sea and the Arctic Circle. On these exposed
sites wind, waves and ice conspire to make loads on oil platforms, ships
and other offshore structures difficult to predict. Although we are familiar
with the damage these elements can cause, from the vicious storm that hit
southeast England in October 1987 to the sinking of the Titanic nearly 80
years ago, designers have trouble calculating the forces generated with
accuracy. Lasers are beginning to shed some light on the problems; at the
same time, they have started to unravel the complexities of other types
of fluid motion, in the heat exchangers of nuclear reactors and in the turbine
blades of aircraft engines.

The motion under a wave at sea, for instance, is not as simple as you
might think from looking at the regular undulations on the surface. It consists
of surges with large volumes of water orbiting in a plane at right angles
to the crest of a wave. The motion produces both a ‘drag’ force and an ‘inertial’
force. Drag increases four times for every doubling of the speed of the
water (the force rises as the square of the speed) and inertia is directly
proportional to the acceleration of the water. Vessels also suffer a ‘slamming’
force when a mass of water suddenly strikes them. In addition a steady current
produces a ‘wake’ in the lee of a structure that generates unsteady forces,
which are potentially destructive.

The flows are too complicated to model using numerical methods alone,
even for those researchers with a supercomputer and a thorough grounding
in computational fluid dynamics. (CFD is a way of trying to solve the set
of three-dimensional, time-dependent, nonlinear, partial differential equations,
known as the Navier Stokes equations, which govern the physical behaviour
of gases and liquids.) Engineers have to turn to physical models in wind
tunnels, water tanks and ice laboratories to supplement the knowledge gained
from numerical models.

The greatest difficulty is in measuring the velocity of the flow, that
is the speed and direction of the wind or waves, during dynamic, sometimes
violent, experiments. (ÐÓ°ÉÔ­´´s calculate the force of a flow from its
velocity.) A breaking wave, for example, is particularly difficult to analyse;
its haphazard motion includes random effects, such as foaming and swirling,
that are peculiar to individual breakers. An ideal experiment would measure
the velocity at all points under the wave and on the surface simultaneously.
Similarly, the transient effects of a gusting wind on the motion of a vessel
means that studies in wind tunnels should record the velocity of the wind
at as many points as possible around the structure simultaneously.

At Heriot-Watt we are using powerful lasers to do just that, and we
have obtained some remarkable results. We can already record the velocities
of waves and wind at every point in a longitudinal slice, or ‘sheet’, of
flow about 1 millimetre thick. This, for us, is two-dimensional measurement;
we have just started to extend our techniques to investigate flows of any
thickness, that is in three dimensions. We are using lasers to study the
forces generated by eddies, or vortices, which are among the most intricate
of fluid motions, and by the accretion and impact of ice in polar waters.
We have started to examine how offshore pipelines cope with mixtures of
gases, liquids and solids. The ratio of the components of these ‘multiphase’
flows varies enormously and unpredictably: sudden changes cause surges that
can damage sensitive equipment.

One of the principal advantages of using light to measure experimental
effects in a test rig is that it can do the job without influencing the
conditions being investigated. We use laser light in particular because
we can manipulate it with precision. We introduce particles, just a few
micrometres in diameter, into the flow as ‘seeds’, or ‘tracers’, to help
us to track the motion. In studies of airflows we add smoke, water or oil
droplets. In water, aluminium powder, pollen or latex spheres have been
used. More recently we came across an industrial coating powder, used to
make metals more resistant to corrosion, that is proving an ideal tracker;
it has a specific gravity near unity, which means it is neutrally buoyant
and will not rise or fall in water under its own steam. In some instances
natural seeds, such as dust particles in air and grit in water, are sufficient.

One way we measure fluid velocity is based on the doppler effect, the
change in the apparent frequency of a wave as a result of relative motion
between the observer and the source. In this technique, known as laser doppler
anemometry (LDA), we divide a laser beam using a ‘beam splitter’, a device
that may be no more than a simple mask with two holes in it. The resulting
two beams, which are about 1 millimetre in diameter, are ‘coherent’. This
means any changes in the shape of the light waves take place at the same
time in both beams. By making these beams cross, we exaggerate the effects
of the changes to produce ‘fringes’, or alternate light and dark bands.
You will recognise this ‘interference’ effect if you have stood near a harbour
wall watching reflected and incoming waves wash together to produce peaks
and troughs.

When a particle in the flow crosses the fringes, it reflects large amounts
of light when it is in the bright bands and small amounts when it is in
the dark bands. If we monitor the way the reflected light, or brightness,
changes and note the time period, or frequency, between the highest and
lowest intensities of brightness, we need only to know the distance between
bands, or the ‘fringe spacing’, to determine the velocity of the flow. A
typical fringe spacing, which is always regular and uniform, is often no
more than about one-tenth of a millimetre across. Knowing the wavelength
of the light and the angle of intersection of the two beams, scientists
can calculate the spacing from the laws of geometrical optics.

One of the most onerous practical problems is in setting up the experiment.
The laser, which fires the light into the flow, and the photo detector,
which focuses and records the illuminations, must be aligned precisely.
The seed particles reflect light in all directions but we usually install
the laser and detector on opposite sides of the flow, an arrangement known
as forward scatter, or on the same side, which is known as back scatter.
These are the easiest ways to observe the illuminated flow. We fix the beam
splitter and the lenses, which guide the laser light and collect the reflections,
on rigid frames to make them stable. We mask the detector to cut out the
uniform light from the two beams but so that it still records the changing
levels of brightness per second, or frequency, of the particles. The trouble
comes when we move these components to scan the flow; trying to keep them
aligned is a painstaking task. We must move the equipment because it can
cover only a small region of the flow at any one time, say one cubic millimetre.

There is no way round this difficulty. In small-scale tests, where the
volume of flow being investigated is only a few cubic millimetres, we tend
to move the flow rig and keep the instruments fixed. On larger scale ones,
such as in wave tanks or wind tunnels containing several cubic metres of
fluid and typical of the sort of work we are doing at Heriot-Watt, frequent
and frustrating realignment of the equipment is normal practice. There is
another problem. The detector is often a metre or more from the volume of
flow it is measuring and so any small flexing of our components causes misalignment
and a loss of signal. LDA has its advantages: it generates a continuously
varying signal, that is it gives us an exact measurement of velocity in
real time. But there are drawbacks: the experiment is difficult to set up;
it measures velocity in only one direction when, in three-dimensional flow,
velocity has three components; and it covers a small volume of the flow.

To overcome the limitations of LDA, scientists developed particle image
velocimetry (PIV), which uses a second important feature of the laser previously
neglected in the measurement of fluid flows. The feature is a laser’s ability
to deliver large amounts of light energy that can illuminate bigger areas
of flow in air or water. LDA illuminates an area of less than 1 square millimetre,
which gives us a ‘point’ measurement; PIV illuminates an area of many square
centimetres or even square metres. In principle, we simply flash light onto
particles in the fluid and photograph the illuminations, or speckling: the
result is a series of images at different instants that record the paths
of individual particles. From these, we can calculate the velocity of the
flow at various points and the forces the fluid is generating at them. To
do this we use a combination of spherical and cylindrical lenses to guide
the laser, usually from below a wave tank or wind tunnel, into the shape
of a sheet with a constant thickness of about 1 millimetre. The drawbacks
of PIV are that it provides an historical record rather than a real-time
one and that it measures velocities at a point in only two directions. We
are developing techniques to allow PIV to produce real-time records of three-dimensional
flow.

For the moment, the seed particles crossing the sheet look like a moving
pin cushion. During the experiment we use a camera, installed to one side
of the flow, to record the illuminated particles in the sheet of light.
By chopping the illuminating laser beam into pulses of light to create a
stroboscopic effect, we can record on film two or more images of the same
region of flow at different instants. We chose the length of the pulse according
to the speed of the flow. Typically, the period is one of nanoseconds or
microseconds.

We chop the polarised light of a laser beam into separate flashes in
one of two ways. When we want very short pulses to freeze high-speed flows,
greater than about 1 metre per second, we use a Pockels cell, which passes
light polarised only in one particular plane, determined by the magnitude
of a voltage applied across the cell. In this way we can use rapid electronic
pulses to open and close the ‘shutter’ of the Pockels cell very quickly
indeed. For slower flows we use a mechanical chopper. This consists of a
rotating disc with a slit cut along a radius to produce a string of pulses
from a continuous laser beam.

At Heriot-Watt we use a pulsed ruby laser, which incorporates a Pockels
cell. In wind tunnels, it can measure flow speeds of many hundreds of metres
per second although we tend to use it to study gales blowing at tens of
metres per second. The laser delivers 10 joules of light energy spread over
one, two or four pulses, depending on the number of stroboscopic images
we wish to record. The pulses are 25 nanoseconds long and between 1 and
800 microseconds apart. The power of the pulse can be as great as 168 megawatts
– easily enough to penetrate sheet steel, shatter glass and ionise air.
We allow only experienced specialists to use the device.

Our group also has another high-powered laser: this 18-watt argon laser
uses a rotating disc to chop the beam into pulses. Its blue-green light
passes through water with only a small reduction in intensity. This makes
the laser suitable for hydraulic studies in which there is plenty of light
in the fluid and where researchers are investigating flow velocities measured
in only centimetres per second. Red light, on the other hand, is quickly
absorbed in water so we use the ruby laser in aerodynamic studies, in which
its higher output of power compensates for the lack of dispersed light in
the air and the greater flow speeds.

Once we have photographs of successive instants in a flow, we determine
the velocity by first calculating the distances that particles have travelled
during the time that separates the images. We can either analyse the emulsion
of the negative microscopically or, for less accurate but quicker results,
we can examine the pattern of light produced when a low-power laser beam
passes through the negative. In the first method, a camera records the enlarged
image in digital form as an array of 256 by 256 picture elements, or pixels,
each with a specific ‘grey level’, or brightness. A computer searches the
data to locate the same particles at different instants by identifying pixels
of the same brightness, or ‘spots with partners’. Knowing the time between
the illuminating flashes, researchers derive the velocity of the flow.

The second method of analysis is handy for making rapid measurements
of the average rate of flow using a low-powered laser. Passing laser light
through the image spots of the developed negative divides the beam, just
as a beam splitter does in LDA. Recombining, or focusing, the beams has
the same effect as crossing them does in LDA; it produces bright and dark
bands, or fringes. The spacing of these fringes depends on the gap between
the spots, or the distance a particle travelled while a camera recorded
its path. The distance can be calculated from classical optics; in our experiments,
however, we calibrate our photographs using a negative on which we already
know the spacing of the particles.

As we increase the number of flashes of the laser during the experiment
in the wind tunnel or wave tank (so increasing the number of images for
a particle), the sharpness and quality of the fringes improve. We can enhance
the contrast of the fringes if we use a positive transparency, by making
a contact print from the negative, to analyse the results. The transparency
only allows laser light to pass through useful areas of the negative, that
is through images of the particle. This is because these images are transparent
as they are the most exposed areas of the film, while the rest of the photograph
is opaque.

In principle, this fringe analysis should enable us to gather information
about the turbulence of a flow from the quality of the fringes. In practice,
we have found that the presence of random reflections of light from the
grain of the film emulsion, known as film speckle, and the varying density
and spacing of the particles introduced into the flow make this an unreliable
approach. For the moment, direct measurement of the spacing of the particles
is the most satisfactory way of assessing turbulence.

Since we have to scan the negative whichever of the two methods we use,
we have developed and built our own automatic scanning systems linked to
a computer. The negative (or positive transparency) of the flow is placed
in a holder that moves to and fro across a light beam: the computer analyses
the result. We started with a personal computer, and then earlier this year
we began to use a parallel-processing supercomputer instead. We can now
do in seconds what used to take minutes on the PC: at last we can obtain
results as quickly as we can measure flows.

The future for PIV looks bright; the excitement for us lies in the unexpected
directions and developments that will follow. We have just been investigating
the flow in heat exchangers in nuclear reactors for Britain’s Central Electricity
Generating Board. Later this year, we expect to be working with other specialists
in our field from West Germany on improvements in the control of flame jets
in combustion chambers where coal slurry is turned into coal gas. Early
next decade, we could be providing more accurate details of the forces encountered
in supersonic flows.

Ian Grant was recently appointed to a Personal Chair in Offshore Engineering
at Heriot-Watt University, Edinburgh. Over the past decade he has developed
techniques for using lasers to determine the effects of flow on offshore
platforms and vessels.

* * *

ICE BREAKS AND SPECKLES IN THE FROZEN LIMELIGHT

ICE IS a problem on offshore platforms and vessels, especially for oil
and fishing companies working in areas within the Arctic Circle, such as
the Beaufort Sea. Designers need to know how the cross section of a platform
or the shape of a ship’s hull will influence the forces generated by ice
hitting the structure, and whether accretions of frozen ice spray will cause
the structure to topple over.

At Heriot-Watt, we built a scale model of an Arctic platform to study
the accretion of ice. We put a wind tunnel into a mobile food transporter,
capable of maintaining a temperature of around -20 Degree C, and placed
a water droplet generator, rather like a garden sprayer, at the air intake
of the tunnel. The water droplets were carried down the wind tunnel, cooling
as they travelled. By the time they reached the model they had ‘supercooled’,
their temperature was below freezing point but they had not frozen. The
droplets turned to ice only when they hit the model.

We have used particle image velocimetry (PIV) to measure the speeds
of the water droplets. This enabled us to determine the behaviour of the
whole flow round the platform in an instant.

To study the impact of ice packs, we used a technique known as laser
speckling, a forerunner of PIV. We built a channel, or trough, of marine
ply, 4.5 metres long by 1.2 metres wide by 0.5 metres deep. We filled the
channel with water and placed it in the transporter. Above the channel,
we laid a small railway and successively fitted models of different platforms
to a carriage on the track; the models were fixed so that they were submerged
in the water.

We left the apparatus to freeze before we towed the models through the
ice. Strain gauges fixed to the models recorded preliminary measurements
of the forces generated by the impact of the ice on the structure. Under
laser illumination, the ice produces bright speckles, or pin-prick highlights.
These reflections create changing patterns of light and dark bands, or interference
effects, that will help us to make more accurate recordings of the freezing
forces.