C.W. Letchford

Doctor of Philosophy, Oxford University, Trinity 1987

Pneumatic Averaging and its Application in Wind Engineering


Wind tunnel model tests are a major source of loading information for the design of building structures. Of particular interest are loads on intermediate size areas on the building surface, e.g., cladding panels and areas supported by structural elements such as purlins.

Pneumatic averaging is a flexible technique for obtaining such area-averaged wind loads. It is achieved by linking, with tubing, several surface pressure tappings via a manifold to a single pressure transducer. The two requirements for a useful pneumatic averaging system are;

  1. the accurate transmission through the tubing and the instantaneous averaging by the manifold of the fluctuating surface pressures, and
  2. the adequate representation of the spatial variations in the fluctuating surface pressure field by a discrete pressure tapping distribution

The theory of Gumley for predicting the dynamic response of the manifold tubing networks is reviewed and its suitability for the design of a full-scale pressure measurement assessed. The analysis is extended to examine the effect of the interaction of the diaphragm flexibility and reference-pressure tubing.

A rigorous experimental investigation of the Gumley analysis for tubing networks used in wind tunnel model studies has shown the theory to be an effective predictor of the dynamic response as long as all the geometric parameters required for the analysis are accurately known. The application of the analysis for predicting the dynamic response of full-scale tubing networks was experimentally verified.

The effect of approximating the surface pressure field by discrete pressures at a grid of tappings is considered and a mathematical description of the overestimation of the variance and spectrum of the fluctuating pneumatically averaged pressures obtained. A method is presented for estimating the actual error involved without resorting to direct force measurements. A simplified version of this method, based on a simple homogeneous and isotropic model of the surface pressure field, has been shown to give reasonable estimates of the discrete approximation errors when compared with actual experimental results. The investigation confirmed that the discrete approximation error can be neglected for a reasonable choice of tapping spacing.

(no thesis available)