The measurement principle is illustrated in the following drawing:
A Ronchi ruling with parallel black and transparent stripes of equal width is projected onto a plane surface S1, and the image of the stripes is focused onto a reference ruling. The stripes of the imaged and the reference ruling are parallel to each other, and superimpose to a Moiré interferogram. If the surface is then translated to location S2, the stripes on the surface move in the x-direction, and a projection ray indicated in the drawing shifts on the surface from P to P´. P´ in turn is now focused onto R´, thus producing an interference pattern with a different intensity.
If the observed surface is curved, the resulting fringes in the Moiré interferogram represent lines of equal elevation. This implies that large surface gradients lead to large spatial fringe frequencies in the interferogram. As an example, the next picture shows the Moiré pattern on the wing (lower side) of the HiReTT full span model in the European Transonic Windtunnel ETW. The distance in z-direction between two adjacent Moiré fringes is called the layer distance, which is a function of the optical set-up and usually has to be determined by a calibration measurement. By using image evaluation algorithms the z-coordinate of the surface can be determined for each pixel from the Moiré pattern. Furthermore, if the surface is deformed or shifted in z-direction (e.g. due to aerodynamic load), the change of the z-coordinates of the surface can be calculated by comparing both surfaces for the no-wind and the on-wind condition, yielding wing twist and bending.
In this case, the observation area of the Moiré system is about 0.60 x 0.45 m2 at an observation distance from 0.9 to 1.1 m from the top wall of the test section. The resolution is better than 0.05 mm for wing bending (z-direction) and better than 0.1° for wing twist.
Comparative measurements of the wing twist for different dynamic pressures q (caused by an increase in tunnel pressure) at a constant Reynolds number of 8.1 million (achieved at temperatures of 300 K and 214 K, respectively) are given in the following figure. Here, positive twist values represent a reduction in angle of attack. Reliable results have been gained between 30 and 90% span. Raising q by about 60% leads to growing twist levels amplified by the lift level cL.