For a long time there has been growing awareness in the international aviation community of the considerable danger which an encounter with the atmospheric phenomenon of windshear can bring to an aircraft in flight. Considerable statistical evidence is now available which points to the fact that aircraft windshear encounters in initial climb, final approach, or landing are extremely hazardous, often leading to fatal accidents. The paper presents details of a preliminary design for an airborne windshear detection system suitable for use in general aviation aircraft. First, an elementary explanation of windshear, and its most dangerous form, the microburst, is given, together with a short account of the hazards that such atmospheric phenomena can present to aircraft in flight, particularly at takeoff and landing. Then a novel windshear detection algorithm is described and associated simulation results are presented. The algorithm is based upon observer theory and uses only a restricted number of measurements. The system is shown to provide very good estimates of the horizontal and vertical components of some windshear encounters. These estimates of the windshear components are then used to provide the pilot with a warning of the presence of windshear together with an indication of its severity. Digital simulation has been used to show the effectiveness of the proposed design.

In 1995, 19 western built jet aircraft were totally destroyed in air crashes which killed 383 passengers and 39 crew members. Although this rate of loss has been steady for the past 10 years, there were three major losses in 1996 — the famous Valujet and TWA aircraft in the United States and the world's worst midair collision in history New Delhi, India, where a Saudia Boeing 747 with 312 persons aboard collided with a Kazak aircraft carrying 37 passengers and crew. All on board were killed.

Performance data are presented for eight annular diffuser families, a family being defined by the inner and outer cone angles and the inlet hub-tip ratio. These results include profiles of velocity and total pressure across the inlet and outlet lanes as well as overall pressure recovery coefficients. Diffuser outer cone angles range from 10° to 60° and inner cone angles from 59° to -19°. Swirl angles were varied from 0° up to 44°. Each diffuser family was tested over a range of lengths and swirl angles resulting in over ninety different tests. A procedure for extending the non-swirling definition of effectiveness to swirling flows is presented and used to evaluate a limiting swirl angle for separation for free-vortex-type inlet swirl. Comparisons are made to several measured values of flow separation. Some experimental diffusers exhibited a violently unsteady hub stall which apparently consisted of rotating reverse flow regions.

Measurements have been made in the exit flows of the fan, diffuser, settling chamber and contraction of a low speed blower windtunnel driven by a centrifugal fan. A velocity deficit and streamwise vortex were observed at the diffuser exit when the fan was operating at less than about 0·9 of maximum efficiency but not when operating above. Unacceptably large unsteadiness (-0·7% rms) occurred at low speeds when the fan speed was increased from rest, but not when the fan speed was decreased from full speed. This unsteadiness was at the rotor frequency, and appears to have been caused by laminar boundary layer separation on one or more adjacent aerofoil blades, coupled with slight asymmetry in the rotor. Unsteadiness at the rotor frequency and of comparable magnitude has also been seen in another, larger fan, but was not eliminated in the same way. Attaining high quality flow is likely to require careful attention to the characteristics of individual fans.

A new method has been used to solve the transonic flow equation for flow over a wavy wall. The results thus obtained are compared with those obtained from windtunnel experiments. A new fluctuating flow phenomenon has been observed both in the experiment and in numerical calculations which may enhance flow mixing. The numerical results also indicate that the flow becomes locally sonic for a transonic parameter value greater than 0·5, while experimental data indicate a value of 0·6363 for the transonic parameter. The classical value of transonic parameter has a value of one for the flow to become locally sonic.