An article about how aircraft designs are tested for their CAA certificate in the UK, with regard to their ability to ditch in water.
SPLASHDOWN (from Engineering Magazine 2001).
The safety of aircraft that have to ditch at sea is one area where physical testing
is still ahead of computational analysis. Mike Farish finds out how it is done.
Anyone who travels with any regularity by air is familiar with the pre-flight safety
instructions that are intoned over the intercom and acted out by the cabin crew. They
invariably contain the information that: 'In case of a landing on water you will find
a lifejacket under your seat.'
That part of the announcement is usually greeted with as much indifference as the rest
of the routine. But some people have to treat the possibility of aircraft landing on
water with real attention - those who manufacture and operate them. In turn that means
that many of them beat a path to the Isle of Wight. For there, at Osborne only a mile
or so behind East Cowes on the island's north coast, is the UK's only fully commercial
location for testing the ditching and flotation characteristics of both fixed and rotary
The facility operated by GKN Westland Aerospace comprises four testing tanks housed
in low brick sheds. The two largest - No2 measuring 250x12x5.5ft and No3 measuring
650x12x5.5ft - are used for the full range of ditching and flotation tests for helicopters,
as well as ship testing in the case of the larger tank. The third, smaller tank, measuring
180x12x2.5ft and known as the Ditching Tank, is used solely for testing the ditching
performance of fixed wing aircraft. More precisely, as the dimensions of the tanks
indicate, they are used for testing scale models of ships, aeroplanes and helicopters
from which the performance of the full size equivalent can be extrapolated. No4 tank,
however, only 80ft in length, is used for testing full-size flotation components.
The site, which now employs some 23 people and also includes various static test rigs
for materials and components, is managed by David George, head of test facilities. He
says that the location has been in use for over 50 years.
It was originally set up in 1946 to test the design of flying boats made by the
now-defunct Saunders Roe company and the unbroken use of the site since then is attested
by the continuing presence of models of aircraft that have long since become familiar
sights in the sky or in some cases never quite made it to production. The Ditching Tank
shed, for instance, still houses the wind-tunnel test model of the ill-fated SR56, a
rocket-powered fighter whose career ended when a prototype exploded in a fireball at
Farnborough back in the 1950s. Hanging from the wall, however, is the unmistakable shape
of the scale model used for the ditching tests of Concorde. More prosaically the slightly
weatherbeaten model of a BAe 146 regional jet stands outside.
But over recent years helicopter testing has tended to predominate. Indeed, says George,
there have been extended periods when utilisation of tank No2 has been "100 per cent."
Machines that have been tested at the facility include the Boeing V22 'tilt rotor' and
its civilian derivative the Bell 609, the Sikorsky S70 and S92 and all of the machines
built by Westland Helicopters in Yeovil, including the EH101.
The day-to-day management of test programmes is the responsibility of Dave Eldridge,
who describes himself simply as a 'test engineer', but whose formal job title is the
more long-winded fixed wing and helicopter ditching group leader. The starting point
for any test programme, he says, is the receipt by the company of sufficient design
data from the manufacturer to enable the construction of a scale model accurately
representing the aircraft's external shape. The data is received in the Catia CAD format,
but from there the further design and manufacturing processes for the models are in
purely technical terms necessarily regressive.
GKN Westland converts the CAD data into paper drawings, a process that generally takes
about six weeks, and then passes that information on to a specialist model-making
company Plastech Products, also located on the Isle of Wight, which then needs about
another 18 weeks to build the scale models. The lengthy build schedule is the direct
result of the fact that the models are, as Plastech's general manager John Slater
confirms, built entirely by hand by the company's employees, whom he describes as
The process involves the production of templates in jelutong wood which are then used
as guides for the construction of the models themselves. Helicopter models are made
through the layering, coating and air-curing of two glassfibre cloths to produce
fuselage shapes only 10 thou in thickness internally reinforced with simple plywood
sections. Fixed wing aircraft are made through the shaping of lengths of balsa wood
to produce fuselages 10mm in thickness, again with internal plywood reinforcement.
Right now, by the way, the company is making a model of the forthcoming Dornier 728
regional passenger aircraft.
Model scales are variable. Helicopter ditching models are generally on a 1/10 scale,
though the models used purely for flotation testing can be much smaller at 1/24.
Fixed wing aircraft models are usually at a 1/11 scale meaning that the Dornier
model currently under construction will be 11ft in length, have a wingspan of 8.5ft
and weigh around 60lbs.
The helicopter ditching models in particular, however, are more complex than they
might at first appear. Not only are their rotor blades correctly aerodynamically
profiled, they also contain an intricate-looking mechanism to enable the blades to
be rotated at an appropriately scaled rate as they are released into the water.
Nevertheless it still comes as a slight shock to learn that the construction costs
for the models range from around £30,000 for smaller helicopters up to as much as
£45,000 for a fixed wing aircraft.
Once they are back at GKN Westland, moreover, the capital cost of the models is
ramped up sharply again as they are loaded with the range of sensors and monitoring
equipment required to record the forces they encounter on splashdown. The models are
designed with access panels in their sides to enable the equipment to be fitted and
Typically, says Eldridge, a model will be fitted with six data loggers, each capable
of recording four separate input channels and each costing £4,000. These will be
connected to six accelerometers costing £500 each, four strain gauges with a unit
cost of £1,000, a gyroscope costing about £1,000 with its associated electronics
and 12 pressure transducers. The latter units, adds Eldridge, now cost only £20
each after he found a much cheaper alternative to the surface mount devices formerly
used which cost around £500 per unit. But even with that cost-saving a quick
calculation shows that well over £60,000-worth of model and test equipment hits
the water on each test.
For helicopter ditching the models are fitted below a monorail that can run at a
speed of 30ft per sec several feet above the test tank. The moving carriage to which
the model is attached contains both a release mechanism and a motor to rotate the
rotor blades at a speed which is the real life equivalent multiplied by the root of
the model scale. Hence the rotors of a 1/10 model will be turning a little over three
times the full-size equivalent rate. The impact speed, however, will be correspondingly
reduced. Fixed wing aircraft models, in contrast, are launched into the water by a
gravity-powered catapult capable of propelling the models at 93 feet per second.
Those speeds alone make it evident that the models need to be fairly robust. According
to Eldridge, moreover, each testing programme will last about eight weeks and contain
around 180 ditchings. It will be followed by six weeks of intensive analysis
culminating in the delivery to the manufacturer of a hefty document containing test
data and analysis.
Nevertheless he stresses that test programmes can also be highly iterative. Data from
a ditching can be downloaded from a model, ported into a spreadsheet program and emailed
to the aircraft manufacturer's design team within 20 minutes of the model being retrieved
from the water. Hence the testing can be integrated into the overall development
programme for a new aircraft which can continue in parallel with it. One aspect of
aircraft design where this feedback is particularly useful, notes George, is calculating
the right strength for engine mountings to ensure they are not ripped off when an
aircraft hits the water.
Flotation testing, meanwhile, is absolutely crucial for helicopters, which in reallife,
as George points out, have a high centre of gravity and as a result are naturally
unstable when in water with their rotors stationary. The smaller scale models used
for flotation testing are accordingly fitted with replicas of the flotation bags
which inflate within five seconds of a real helicopter hitting the water. They can
be placed in the test tank by hand and their performance observed in various simulated
The actual legal requirements imposed on flotation performance are complicated but
at the most extreme require a ditched helicopter with all its flotation bags intact to
remain secure with wave heights of 30ft - what the World Meteorological Organisation
defines as 'Sea State 7'. Achieving the required performance is, moreover, particularly
difficult in the North Sea, the scene of intensive over water helicopter flying, because
the north-to-south funneling effect of the UK and European landmasses generates
unusually high-energy wave patterns and can also produce clashes between wave and
So despite the advances made in computer simulation and modelling elsewhere in the
aerospace industry the neat, and only apparently simple, models that enjoy a few
brief seconds of flight before splashdown or bob in the agitated waters of one of
Osborne's test tanks, have a crucial role to play in ensuring passenger safety.
Indeed there seems little prospect of their being replaced by digital equivalents
at any point in the foreseeable future. When one North American company tried to
simulate fixed wing aircraft splashdowns with a software system, says George, they
found the results to be some 25 per cent at variance with what model testing confirmed.
The next time you hear that pre-flight safety routine, therefore, maybe you will not
take it so much for granted.
This article was published online in Engineering Magazine in 2001, subsequently lost
during a system upgrade and is copied here until
reproduced again on the
Engineering Magazine web site.