Bento Silva de Mattos

Artificial neural networks to predict aerodynamic coefficients of transport airplanes

Purpose Multidisciplinary design frameworks elaborated for aeronautical applications require considerable computational power that grows enormously with the utilization of higher fidelity tools to model aeronautical disciplines like aerodynamics, loads, flight dynamics, performance, structural analysis and others. Surrogate models are a good alternative to address properly and elegantly this issue. With regard to this issue, the purpose of this paper is the design and application of an artificial neural network to predict aerodynamic coefficients of transport airplanes. The neural network must be fed with calculations from computational fluid dynamic codes. The artificial neural network system that was then developed can predict lift and drag coefficients for wing-fuselage configurations with high accuracy. The input parameters for the neural network are the wing planform, airfoil geometry and flight condition. An aerodynamic database consisting of approximately 100,000 cases calculated with a full-potential code with computation of viscous effects was used for the neural network training, which is carried out with the back-propagation algorithm, the scaled gradient algorithm and the Nguyen–Wridow weight initialization. Networks with different numbers of neurons were evaluated to minimize the regression error. The neural network featuring the lowest regression error is able to reduce the computation time of the aerodynamic coefficients 4,000 times when compared with the computing time required by the full potential code. Regarding the drag coefficient, the average error of the neural network is of five drag counts only. The computation of the gradients of the neural network outputs in a scalable manner is possible by an adaptation of back-propagation algorithm. This enabled its use in an adjoint method, elaborated by the authors and used for an airplane optimization task. The results from that optimization were compared with similar tasks performed by calling the full potential code in another optimization application. The resulting geometry obtained with the aerodynamic coefficient predicted by the neural network is practically the same of that designed directly by the call of the full potential code. Design/methodology/approach The aerodynamic database required for the neural network training was generated with a full-potential multiblock-structured code. The training process used the back-propagation algorithm, the scaled-conjugate gradient algorithm and the Nguyen–Wridow weight initialization. Networks with different numbers of neurons were evaluated to minimize the regression error. Findings A suitable and efficient methodology to model aerodynamic coefficients based on artificial neural networks was obtained. This work also suggests appropriate sizes of artificial neural networks for this specific application. We demonstrated that these metamodels for airplane optimization tasks can be used without loss of fidelity and with great accuracy, as their local minima might be relatively close to the minima of the original design space defined by the call of computational fluid dynamics codes. Research limitations/implications The present work demonstrated the ability of a metamodel with artificial neural networks to capture the physics of transonic and subsonic flow over a wing-fuselage combination. The formulation that was used was the full potential equation. However, the present methodology can be extended to model more complex formulations such as the Euler and Navier–Stokes ones. Practical implications Optimum networks reduced the computation time for aerodynamic coefficient calculations by 4,000 times when compared with the full-potential code. The average absolute errors obtained were of 0.004 and 0.0005 for lift and drag coefficient prediction, respectively. Airplane configurations can be evaluated more quickly. Social implications If multidisciplinary optimization tasks for airplane design become more efficient, this means that more efficient airplanes (for instance less polluting airplanes) can be designed. This leads to a more sustainable aviation. Originality/value This research started in 2005 with a master thesis. It was steadily improved with more efficient artificial neural networks able to handle more complex airplane geometries. There is a single work using similar techniques found in a conference paper published in 2007. However, that paper focused on the application, i.e. providing very few details of the methodology to model aerodynamic coefficients.

Wing-Fuselage Drag Prediction Using Artificial Neural Networks

The conceptual design of an aircraft involves several multidisciplinary analisys in order to reach the optimum configuration according to market and certification requirements. An application is being developed in the Technological Institute of Aeronautics aiming to group such analisys in a single interface. Previously, aerodynamic drag was calculated through semi-analytical methods and empirical data interpolations, demanding strong computational resources, which became critical in optimizations processes. In this work, an artifical neural network was developed to replace the old aerodynamic module. This neural network was trained with a database containing more than one hundred thousand configurations, which were previously analysed with the BLWF 28. The metamodel gave acurrate results with low computational costs, thus attaining the required performance for multidisciplinary optimizations.

Conceptual Optimal Design of Airliners with Noise Constraints

Noise generated by airplane during the takeoff and landing flight phases is a matter of increasing concern for both aviation and aeronautical industries. Air traffic is steadily increasing and airports are operating close to their capacity limit. Airlines are under high pressure from communities surrounding airports to operate quieter airplanes and/or change their operating procedures. For this reason, quieter airplanes will be welcome by the airlines and comply to the expected traffic growth. From an airplane manufacturer point of view, the reduction of noise generated by airplane to acceptable levels is a very challenging task. It is to be expected that such drastic noise reductions will not be achieved by merely working on mitigating noise sources on the airplane in isolated form. Instead, the interactions of noise sources as well as shielding effects have to be taken into account. Airplane noise becomes a configuration issue and thus has to be considered in the conceptual design phase. Quieter airplanes could be charged with some penalties like performance degradation and higher fuel consumption, in the latter case generating more pollutants. In order to carry out the design for a quieter airplane considering all this aspects, it is useful to incorporate a noise assessment methodology into a multi-disciplinary design and optimization framework. This approach is justified because an airplane should not be designed considering just a few requirements. Field performance, stability and control, operating costs, manufacturing costs, passenger comfort, embedded technology, all this must be simultaneously considered for designing an airliner that airlines need. A Parametric Airliner Noise Prediction Architecture (PANPA) has been developed at Instituto Tecnologico de Aeronautica (ITA), which is able to predict noise levels generated by an airliner along arbitrary flight trajectories. The related noise levels are estimated for an observer positioned on ground, as required by certification authorities. The module takes into account major airframe and engine noise sources, as well as diverse effects on sound propagation. A multi-disciplinary integrated conceptual airplane design framework, designated AIDMIM, has been in development for some time at ITA. This framework features a modular structure written in MATLAB® language, which allows for manageable incorporation of additional disciplines and analysis methods to the overall design process. PANPA was integrated into AIDMIN and some design tasks with and without noise constraints were carried out.

Conceptual Optimal Design of Environmentally Friendly Airliners: AReview of Available Methodologies and their Integration into a ConsistentFramework for Everyday Use

Airplane design involves complex system integration and must comply with a set of requirements, which are set up by certification authorities, customers, manufacturing, and that coming from market studies. From the traditional perspective of an airline, an interesting airplane is one that is capable of generating the highest revenue with minimum cost-a maximum profit airplane. In later years, however, the airline industry is swiftly broadening its consideration of what constitutes a nice-to-buy airplane. Not only economics but also environmental considerations are taking part in fleet-planning considerations-a trend spurred by environmental-aware passengers. In a move to comply with this trend, airplane conceptual design has incorporated methodologies for preliminary assessment of airplane noise and emissions. As more approaches become available to address these issues, the choice between the most suitable methodologies becomes tougher. The present work analyzes some methodologies in order to select a subset of them. The objective is the evaluation of such methodologies and their integration into a framework to airliner conceptual design. In order to test the design methodologies and the optimization techniques, the authors selected two test airplane categories: first, a long range, transcontinental jet; and a mid-size regional jet. Design tasks based on optimization with noise footprint, direct operational cost, and emission profile or a combination of them as objectives are presented and analyzed.

Artificial Neural Networks Applied to Airplane Design

Multi-disciplinary frameworks for airplane optimal design require a lot of computational power, which grows enormeously if higher fidelity tools are used to model aeronautical disciplines like aerodynamics, loads, flight dynamics, performance, and structural analysis. In order to address properly and elegantly this issue, surrogate models are employed. In this highlight, the main goal of the present work is the design and application of an artificial neural network to predict aerodynamic coefficients of wing-body configurations of transport airplanes in order to replace a computational fluid dynamic code for future optimizations. The artificial neural network system that was developed is able to predict lift and drag coefficients for wing-fuselage configurations of transport airplanes. The input parameters for the neural network are the wing planform, airfoil geometry, and flight condition. An aerodynamic database consisting of 100,000 cases evaluated with the BLWF V2.81 full-potential code is used for the neural network training. The neural network training is carried out with the back-propagation algorithm, the scaled gradient algorithm, and the Nguyen-Wridow weight initialization. Networks with different numbers of neurons are evaluated in order to minimize the regression error. The optimum networks reduce the computation time of the aerodynamic coefficients in 4000 times when compared with BLWF V2.81, and with an average error of only five counts for the drag coefficient. We present an adaptation of the back-propagation algorithm that allows the computation of the gradients of the neural network outputs in a scalable manner, and then we use it for an airplane optimization task. The results are then compared with similar tasks that were performed by calling the full potential code. An adjoint-method is employed to use ANN in constraint functions as well. The resulting geometry obtained with the ANN methodology is compared to one designed directly with BLWF. The optimal geometry is practically the same, and the drag coefficients predicted by each method differ in four drag counts.

Considerations about Wingtip Devices Design of Transport Airplane

From the start of aviation, designers have been searching for methods and technologies to reduce the required fuel burn of commercial aircraft. Wingtip devices in special winglets offer a way of significant drag reduction for transport airplanes. Winglets alongside with tip tanks, raked wingtips as well as aligned fans belong to this class of wingtip devices. Although winglets are targeted to induced drag reduction, their effects can also be extended to wave and zero-lift drag. Induced drag alone is responsible for 30-40% of the overall drag of a transport airplane at long-range cruise condition and for considerably downgrading the climb performance of fixed-wing aircraft. Better investigation in this field employing CFD tools and extensive wind-tunnel testing has allowed the rise of efficient winglet designs in recent times. Several of newly designed aircraft configurations embody winglets and many of the older ones are being retrofitted. However, there are discussions concerning the best cost/benefit ratio of reducing induced drag of a transport plane with wingtip devices. Another big issue is the associated penalties to the configuration caused by winglets when compared to a simple wingtip extension. Two winglet optimization studies were carried-out for a 70-seat airliner. In the first simulation, wing and winglet were optimized for cruise and field performance. In the second study, optimization was carried out for wings without winglet. The Pareto front for the configuration with winglets lies above the one obtained from the optimization for wing-body alone, justifying the adoption of winglets from the aerodynamic point of view.

The Suited Airliner for an Existing Airline Network

Successful application of multi-disciplinary design and optimization (MDO) methodology in the aeronautical industry only started in the early 2000s. This was due to the extreme complexity and the low automation on all levels of aircraft design. Thanks to the development of high-performance computer systems, the real benefit of multi-disciplinary design and optimization frameworks could now finally be accounted to aircraft design. Nevertheless, even nowadays, usual MDO approaches still just consider the airplane under the manufacturers point of view, minimizing production cost and maximizing performance. The airline requirements are taken into account in the MDO methodology by minimizing the direct operating costs in general form only. A different approach for aircraft optimal conceptual design considering airline needs and employing an advanced engine model is described and analyzed in this paper. In this context, a methodology for the design of an airliner better suitable for an existing scheduled network was developed and two test cases were run. The MATLAB ® suite was employed to code and house the present methodology. Another additional feature in the present work is the incorporation of wing and engine position in the configuration, number of engines, and tail configuration as design variables. The remained design variables are related to the wing and horizontal tail geometry. In addition, a genetic algorithm was developed, tested and validated to perform the optimization task. 50th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference 17th 4 7 May 2009, Palm Springs, California AIAA 2009-2206 Copyright