Stress Imaging (aka Thermoelastic Stress Analysis) relies on a coupling between volumetric deformation and a reversible change in temperature, called the thermoelastic effect. This means, during elastic deformation, a solid will increase in temperature when in compression and decrease in temperature when in tension:
For solids deformed under adiabatic conditions the relationship is:
As ΔT is several orders of magnitude smaller than T, the relationship between Δσ and ΔT is, for all practical purposes, linear, which means:
Therefore, all you need to see stress, is an imaging system which can measure the stress signal (thermoelastic response) ΔT. Unfortunately, this stress signal is miniscule and typically in the range of only a few mK (millikelvin or thousandths of a degree Celsius). For example, 1 MPa (megapascal) of stress induces a temperature change of approximately 3 mK in aluminium alloy and approximately 1 mK in steel and titanium. This means that it is impossible to directly measure the stress on an object with traditional off-the-shelf cameras, as no device is sensitive enough to resolve these small temperature variations.
Additionally, as temperature is subject to the laws of heat diffusion, any stress-induced temperature gradients will quickly dissipate and be unmeasurable.
Our innovative approach is to use small and rugged uncooled LWIR (Long Wave Infrared) camera cores, combined with sophisticated signal processing along, process know-how and custom-built timing electronics to achieve a ground-breaking thermal sensitivity of less than 1 mK. The key requirements to make this work are:
This allows us to solve the problem of the fleeting nature of the stress signal, as applying a dynamic load allows the stress signal to persist and gives us the ability to measure it.
Our image processing requires access to the load signal. A load signal input connection is provided on our cameras and the in-built, custom electronics ensures precise time synchronisation between these data streams is achieved. If a load signal is not available, an accelerometer, laser displacement or strain gauge signal can be used.
This ensures our signal processing works correctly (and doesn’t give false results). For small displacements this is not a problem, and for larger displacements, the camera can be physically coupled to the moving object (to ensure a static view is achieved). This is one of the significant practical advantages of our systems over traditional cooled systems which are simply too big and heavy to physically attach to a structure.
As the core of our cameras is a LWIR (Long Wave Infrared) sensor, we require a high emissivity (non-reflective) surface to ensure we measure the stress on the object and not reflected background temperature signals. Aircraft primers will suffice for this, but reflective surfaces (i.e. high gloss coatings or bare metal) will require a thin coating of matt paint to be applied.
With these conditions in place, our Stress Imaging system performs a semi real time operation on the load signal and infrared video stream that extracts the stress signal while diminishing all other signals.
This is a semi real-time process which provides a progressively improving stress image every 10-20 seconds. After several minutes the camera has typically achieved a thermal sensitivity sufficient to generate a high-fidelity stress image akin to a Finite Element model simulation (depending on the load amplitude).
What makes our Stress Imaging system unique is the type of thermal imaging sensor at its heart. Rather than taking the well-established path of developing a system utilising a cooled, photonic sensor, which are the most sensitive sensors available, 1MILLIKELVIN did the opposite. We prioritised ease-of-use, ruggedness, and simplicity as our starting point, which required the use of uncooled, microbolometer technology.
This approach came with a serious technical challenge though, as the lower sensitivity and higher latency of these sensors make Stress Imaging more difficult. We pushed ahead undeterred, with the hope that the significant benefits in usability and deploy-ability would outweigh any loss in sensitivity. What eventuated was a rather surprising result. Due to the distinctive nature of the noise in an uncooled bolometer sensor it starts out noisier but as processing of the video stream continues it quickly catches up and ultimately outperforms a photonic sensor.
The result is a Stress Imaging camera that is easier to use, more rugged, more affordable, more sensitive, and better suited for real-world use.
