crunching
The Surrey Satellite facility is very data intensive and the STFC RAL vibration testers are equipped with 500 data acquisition channels so there is a vast amount of information involved in performing these tests. Making sense of this much data is always a challenge as it can be difficult to analyse. According to Dr Richard Ahlfield, founder
of Monolith AI, there is an increase in the use of machine learning or AI to help with the analysis. This is particularly pertinent to the space industry as it is much more iterative than industries such as automotive. The example Ahlfield used was that a
typical car manufacturer uses 1,000 simulation iterations, occupying 20 test days over a six-month period at a cost of $1 million. In Aerospace engineering, this can grow to 100,000 simulations over 100 test days occupying 36 months at a cost of $100 million. Recognising that AI itself isn’t
straightforward and has had a mixed reception amongst test engineers, Monolith
Copper sheet has been laid into the floor of the chamber as well as copper wall and ceil- ing panels to create a shield through which electromagnetic waves cannot pass. Inside this quiet zone, satellite manufac-
turers will be able to accurately measure the noise that satellite antennas produce, ensur- ing a high quality signal is returned to Earth for TVs, weather forecasts and science oper- ations.
Shock measurement \\\ Part of the large STFC RAL premises is home
to a 16m3 vibration test facility with two Data Physics LE-5000-VH shakers providing 222kN sine force on a 3” stroke. The two tables com- prise a 230 tonne vertical shaker and a hori- zontal system with a 2.1m x 2.1m slip plate. This is the scale of the environmental test
facilities required for the industry but RAL is not alone in having such equipment. Surrey Satellite Technology in Guildford produces satellite systems of different sizes, all of which have to go through shock and vibration test- ing for pre-launch qualification.
All environmental tests for the space industry generate vast amounts of data, something Monolith believes can be analysed more easily with AI
AI has taken the step of simplifying and streamlining the use of AI to make it more accessible to a wider community. Simulation provides ways of interpolating
data to understand system behaviour outside observed conditions. For simple physical systems, that can be achieved with an equation, statistical modelling or higher level data modelling. For more complex systems, some aspect of AI or machine learning is required.
According to David Costello, head of me- chanical and propulsion at the company, the space industry is different from other sectors such as automotive or electronics, where components are made in volume. In such cases, industry standard 1/2 sine impulse shock tests are the norm and replacement of failed components is possible. For the space industry, volumes are low, shock levels are typically higher and across a wider frequency range and component replacement is often not possible. “Shock events can be caused by jettison of
the rocket fairing, staging, satellite separa- tion, release of deployable items etc. The de- signer needs to ensure that optics, moving parts, ceramic components and relays are protected from such events,” he says. Shock testing is done in two stages, first at
the satellite level to characterise the way the shock transmits to sub-systems and compo- nents. The second stage is the lower level qualification testing. The characterisation stage is important in determining whether there are any alter-
The AI software takes experimental data
from complex system tests and presents the user with a parametric dashboard. The AI makes predictions on performance when parameters are changed. Some predictions will have higher levels of confidence than others and the user can design physical experiments to verify output as required. This can reduce testing requirements by as much as 70-80 per cent, according to Ahlfield.
ations needed to design characteristics such as component selection, positioning or mounting. This is difficult to model and so physical testing is necessary. Once the design assesment is done, py-
rotechnic shock tests are carried out at spacecraft level using a spacecraft struc- tural model. A representative shock (eg sep- aration) is performed and the transmission of the shock is measured throughout the structure. This enables unit level tests to then be carried out to qualify each one indi- vidually. According to Dr Brian Le Page, materials
and testing functional manager at Surrey Satellite, the tests generate massive amounts of data with which response envelopes can be created. Shock response within the spacecraft can be affected by, for example, structural damping or the use of mass dummies. “A results envelope is the best way of visu- alising the data and giving some confidence in the results. This enables a test to be designed for component qualification outside the spacecraft,” he says. C&VT
2021 /// Climatic & Vibration Testing \\\ 13
Data
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40
