This paper describes the results of a study focusing on the possibility of identifying nonlinear helicopter models in the time domain. In recent years, identification techniques working mainly in the frequency domain were applied to estimate the parameters in helicopter models. Recently, the time domain identification method of the DLR was improved to allow the identification of fully nonlinear models. To investigate the applicability of this method to helicopters, a nonlinear derivative model with explicit equations for the individual blade flapping angles was formulated. In comparison to the linear derivative models nonlinear models are physically more realistic and are not restricted by small perturbation assumptions. Although a large number of unknowns had to be determined, a nonlinear model was successfully identified.

Equations of motion have been developed for the situation in which a substantial load, carried internally by an aircraft, is drawn along a ramp by an extraction parachute and is then dropped. The motion of the load is assumed to remain in the plane of symmetry. The perturbations to the aircraft motion are assumed to be small so that the equations can be linearised following the representation of the aerodynamic out-of-balance forces using conventional aerodynamic derivatives. The resulting system of six ordinary differential equations consists of the four normally associated with the longitudinally-perturbed motion of the aircraft (slightly modified by the reaction forces due to the load), together with two describing the motion of the load.

Numerical solutions are presented for a generic aircraft, the investigation examining the effects of a number of different parameters such as the ratio of the mass of the load to that of the aircraft, the length and angle of the ramp, the friction between the load and the ramp and the direction of the parachute extraction force. In addition the effects of employing various control strategies to limit the disturbed motion have also been computed.

It was found that the acceleration of the load, relative to the aircraft, was sensibly constant. In the absence of any resetting of the controls (elevators and throttle), the disturbances computed exceeded those for which the linearisation is justified. When the controls were reset, either as the load began to move or as it was jettisoned, to the trim values appropriate to the unloaded aircraft flying at the same speed, the disturbances were reduced, but remained large, the phugoid mode being dominant. The incorporation of various kinds of feedback from the disturbance variables to the elevator was also investigated and a successful control strategy was identified, that limited the perturbations and minimised the steady-state errors in airspeed and angle of climb.

Two new techniques for estimating aircraft stability and control derivatives (parameters) from flight data using feed forward neural networks are proposed. Both techniques use motion variables and control inputs as the input file, while aerodynamic coefficients are presented as the output file for training a neural network. For the purpose of parameter estimation, the trained neural network is presented with a suitably modified input file, and the corresponding predicted output file of aerodynamic coefficients is obtained. Suitable interpretation and manipulation of such input-output files yields the estimated values of the parameters. The methods are validated first on the simulated flight data and then on real flight data obtained by digitising analogue data from a published report. Results are presented to show how the accuracy of the estimates is affected by the topology of the network, the number of iterations and the intensity of the measurement noise in simulated flight data. One of the significant features of the proposed methods is that they do not require guessing of a reasonable set of starting values of the parameters as a popular parameter estimator like the maximum likelihood method does.

The aim of the work outlined in this paper is to compare three different variable cycle jet engine concepts for future SSTs. These engines are: the turbofan-turbojet, the mid-tandem fan engine and the double bypass engine. The comparison is carried out on the basis of unin-stalled and installed performance, handling and sizing issues.

This preliminary analysis compares SFC, size, variable geometry and cycle changes for each engine. The installed performance was estimated by calculating the air friction, the pre-entry and the afterbody drags, together with the wave drag due to the shock waves. A sizing calculation was carried out for the whole nacelle. The uninstalled and installed fuel bill, for two standard missions, is also estimated.

These preliminary results indicate that the turbofan-turbojet and the mid-tandem fan engines are quite similar in terms of general suitability. The mid-tandem fan appears to be an attractive proposition from the point of view of sizing, however, this comes with a small penalty in fuel consumption. The present double bypass engine was found to be the least attractive for the application, although the differences are small.

Measurements of drag have been made, in a shock tunnel, on a simple integrated vehicle-engine combination for hypersonic cruise with hydrogen scramjet propulsion. The test flow Mach number was 6·4, and the velocity was 2·45 kms-1. Zero drag, which is the necessary condition for cruise, was achieved as the equivalence ratio approached one. It was found that an analysis using established aerodynamic concepts was adequate for predicting drag in the case of no combustion. When combustion occurred results of direct connect experiments provided a qualitative guide to the measured levels of drag, and indicated that thrust nozzle combustion was taking place. An heuristic analysis is used to point to the important effect this may have on propulsive lift.

Simulated hailstones were made to impact on the rotating spinner and fan assembly of a Williams FJ44 engine. The mass distribution of ice behind the fan was determined by use of a suction tube technique. Suction was added to ensure that the tube did not affect the flow through the fan assembly. The strong air flow behind the fan meant that the ice caught by the tube melted and evaporated. This made it difficult to accurately determine the mass distribution of ice. As a result, the simulated hailstones were made from a water-salt solution so that the weight of salt residue could be measured after the water had evaporated, and hence the amount of ice caught was determined. A parametric study into the hail ingestion characteristics of the fan assembly was carried out. The parameters studied included the radial position of the impact point, the rotational speed of the fan and the position of the splitter between the core engine and bypass duct. The results showed that the impact position had a major effect on the overall ice distribution, which was determined by the combination of blade geometry at the impact point and the rotational speed. The splitter position was shown to have a significant effect on the amount of ice passing into the bypass duct.

A low Reynolds number turbulent boundary layer is subjected to a relatively strong rate of suction, which is applied through a short porous strip. Downstream of the strip, the effectiveness of a riblet surface is assessed by comparison,to that of a smooth wall. Measured mean velocity and longitudinal turbulence intensity profiles indicate that the riblets do not hinder the action the suction can have on the flow.