The surface accuracy of a rectangular membrane subjected to anisotropic tension in a reflector surface is investigated. The rectangular membrane is assumed to be located at an arbitrary point on the reflector surface with an arbitrary rotation angle alpha between the side of the membrane and the direction of the principal curvature of the reflector surface. The deflection and surface error of the membrane are calculated based on an analytical solution of the linear membrane equation. Three kinds of rectangular optimum membranes, whose surface errors are minimized with respect to parameters that specify the edge deflections, are considered and evaluated. These surface errors, D(1) (surface error of an equal-curvature-edge-optimum membrane), D(2opt) (surface error of an optimum-2 membrane), D(3) (surface error of an optimum-3 membrane), are also compared with D(copt) (surface error of a coincident-optimum-edge membrane) that is studied in other literature. In a certain range of parameters, it is found that the appropriate tension ratio reduces the surface error of each optimized membrane as compared to isotropic tension. The more slender the membrane is and the higher the ratio of tension (ratio of higher tension on the longer side to lower tension on the shorter side) is, the smaller the surface error becomes. The minimum surface error is obtained when the longer side is placed parallel to the direction of the larger principal curvature at the location and the longer side is loaded with higher tension. When alpha=0, D(2opt) is equal to D(3). As for a slender membrane whose longer side is loaded with higher tension, D(2opt) [equal to D(3)] is approximately equal to D(1) and D(copt) when alpha is equal to 0, and D(2opt) [approximately equal to D(copt)] is smaller than D(1) [approximately equal to D(3)] when alpha is not equal to 0. It is found that D(2opt) is the smallest among other membranes, and the surface error is about 0 - 15% less than D(copt). Each surface error decreases as the distance from the vertex increases.
View full abstract