Application of an Aircraft Design-To-Noise Simulation Process

Application of an Aircraft Design-To-Noise Simulation Process Lothar Bertsch∗, Wolfgang Heinze†, and Markus Lummer‡ German Aerospace Center (DLR) and Technical University of Braunschweig (TU-BS)

System noise has been integrated as an additional design objective within conceptual aircraft design. The DLR system noise prediction tool PANAM accounts for individual noise sources depending on their geometry and operating conditions. PANAM is integrated into the existing aircraft design framework PrADO from the Technical University of Braunschweig in order to realize a design-to-noise simulation process. In addition, a ray-tracing tool from DLR, SHADOW, is incorporated into the simulation framework in order to account for structural engine noise shielding. The overall simulation process is then applied to identify promising low-noise aircraft concepts. The presented application aims at fan noise reduction through shielding. For the selected reference aircraft, the fan is a major noise source during both landing and takeoff. It is demonstrated, that the aircraft designers influence on the environmental vehicle characteristics is significant at the conceptual design phase. Usually, a trade-off between extensive engine noise shielding and economical flight performance is inevitable. The new design-to-noise process is well suitable to assess all four measures of ICAOs balanced approach.

Nomenclature Tools & methods PANAM Overall aircraft noise prediction, DLR PrADO Aircraft design synthesis code, TU BS SHADOW Prediction tool for noise shielding effects, DLR VarCycle Engine design & performance code, DLR Vehicle categories (symbol) v-rx Modified reference (”◻”) v-0 Over-the-wing shielding (”△”) v-1 Empennage shielding (”◁, ▷, ▽”) v-2 Wing-fuselage shielding (”◇”) v-x Wing-tail shielding (”o”) Final vehicle design (symbol) V-r Ref. (”x”) V-rx Mod. ref. (”∎”) V-0 Over-the-wing shielding (”▲”) V-1 Empennage shielding (”◀”) V-2 Wing-fuselage shielding (”◆”) V-x Wing-tail shielding (”●”)

SPL SPL(A)

Sound pressure level, [dB] A-weighted SPL, [dBA]

Multiple criteria design evaluation Ki Scenario dependent overall environmental weighting, applicable to ξ Λi Scenario dependent overall economical weighting, applicable to ζ σ Overall scenario score ξ1,2 Environmental performance indicator (vehicle specific) ξ Overall environmental performance indicator (vehicle specific) ζ1,2,3 Economical performance indicator (vehicle specific) ζ Overall economical performance indicator (vehicle specific) Abbreviations VTP Vertical Tailplane HTP Horizontal Tailplane

Noise metrics EPNL Effective Perceived Noise Level, [EPNdB] ∗ DLR,

Institute of Aerodynamics and Flow Technology, Bunsenstr. 10, 37073 Goettingen, Germany, [email protected] Institute of Aircraft Design and Lightweight Structures, Hermann-Blenk-Str. 35, 38108 Braunschweig, Germany ‡ DLR, Institute of Aerodynamics and Flow Technology, Lilienthalplatz 7, 38108 Braunschweig, Germany † TU-BS,

1 of 17 American Institute of Aeronautics and Astronautics

I.

Introduction

During the conceptual design phase, each decision can have significant impact on the final aircraft design. At this phase, major design modifications are still feasible and will ultimately define the final vehicle layout. During the development of an aircraft, only a small percentage of the overall program costs are attributed to conceptual design but a major share of the costs and the final vehicle layout are determined. Moving further along the design evolution path as depicted in Fig. 1 will require final decisions that freeze certain parameters and subsystems, e.g. wing design. Later modification to this early and basic parameter setting

Figure 1. Overall project evolution (adapted from W. Heinze: ”Aircraft design lecture”, TU Braunschweig)

is not practicable or feasible because it will require to revisit earlier design phases. In the worst case, this will cause significant additional costs and can have negative implications on other disciplines and/or overall vehicle characteristics. Moreover, this could even require further modifications to other subsystems and/or to the initial vehicle layout. As a consequence, noise has to be accounted for as early as reasonable within the overall design process as depicted in Fig. 1. Within conceptual design, even major design modifications are still feasible, thus selected low-noise modifications can be fully exploited. Furthermore, relevant interdependencies can be identified and associated implications on other (sub-) systems can be tracked throughout this design phase. In the past, noise has usually not been accounted for in conceptual aircraft design because of the lack in appropriate noise prediction capabilities. Available noise simulation methods have either been too complex or just not appropriate for application toward a design-to-noise task. High-fidelity tools are too complex in input data request or computational demand, thus are not applicable for a full aircraft simulation. Consequently, fast tools for a quick simulation of the overall aircraft and its major subsystems are required. Available fast prediction tools from other research organizations or companies can be classified into two groups, i.e. best-practice and scientific methods, see Ref.1 . The best-practice tools are immediately ruled out for application toward any design-to-noise activity1 . These methods are based on a simplified noise source modeling, i.e. the entire aircraft is simplified by one overall noise source that represents both airframe and engine noise contribution. Measured noise level impact of the overall aircraft is translated back into noise emission of one single, overall noise source. Consequently, only existing technology and vehicles can be assessed with such an empirical methodology. Assessment of specific low-noise design modifications is prevented due to missing separation into individual noise components or at least into airframe and engine noise contribution. Only if individual noise sources are modeled by a parametrical model, geometry modifications to individual subsystems and components, e.g. high-lift system, can be simulated. Moreover, best-practice simulation results for individual approach flights 2 of 17 American Institute of Aeronautics and Astronautics

are questionable because realistic configurational scheduling along typical approach procedures cannot be accounted for. As a consequence, only scientific noise prediction methods are applicable toward the selected task of developing design-to-noise capabilities1 . A well-known and probably the most accepted scientific prediction tool is ANOPP2, 3 , a NASA development. Earlier attempts in recent years to apply ANOPP in the context of aircraft design had the focus more on the optimization process rather than on dedicated noise prediction or comprehensive vehicle design studies4 . Other activities using ANOPP5 had the focus on isolated design modifications and did not take all the inherent implications onto other disciplines into account, e.g. snowball effect on flight performance. Furthermore, none of these activities includes validation of their simulation results or feasibility checks of their implemented methods. Motivated by these early design-to-noise studies with ANOPP, DLR has launched its own design-to-noise research activity. The main goal behind this activity is to incorporate existing in-house knowledge and tools1 . Therefore, all relevant disciplines and interdependencies are identified and accounted for within the overall simulation process. Missing simulation capabilities had to be developed, i.e. including dedicated noise source models. In case of system noise prediction, multiple disciplines and their interactions have to be accounted for simultaneously, i.e. aerodynamics, weights, engine simulation, flight mechanics, and noise prediction. Comparison with experiments and high-fidelity computations has been an inevitable step in order to confirm the feasibility and reliability of the overall process. Ultimately, the DLR activities aim at a system level evaluation of the overall aircraft with all relevant subsystems along (a) departure procedures, (b) during design missions, and (c) along approach operation.

II.

Aircraft Design with Integrated Noise Prediction Capabilities

In the presented simulation process, noise is integrated as an additional objective within the vehicle design. This new design-to-noise simulation process as established at DLR is comprised of three major tools: PrADO6–8 , SHADOW9 , and PANAM10 . These tools are in-house developments of TU Braunschweig (PrADO) and DLR (SHADOW and PANAM). The tools and the overall simulation process are briefly described in the following. A.

PrADO

The aircraft design synthesis code PrADO6–8 is an in-house development of the Institute of Aircraft Design and Lightweight Structures, Technical University Braunschweig, Germany. PrADO can assess the feasibility of new aircraft concepts at the conceptual aircraft design stage. The PrADO framework is comprised of individual design modules, each dedicated to a certain task or discipline. All modules are embedded into a monolithic structure resulting in one large program code and therefore very fast computation times11 . Each of PrADO’s design modules offers a selection of methodologies to solve its designated task. The selected methodology determines the overall computational requirement, result accuracy, and most importantly the input data requirement. The top level aircraft requirements and a basic vehicle layout comprise the input for the design synthesis process. The overall design process is a sequential execution of individual PrADO design modules in a predefined order, see Fig. 2. This simulation sequence is repeated until predefined design parameters reach convergence. Selected design parameters can be for example aircraft mass, thrust requirement, design mission performance, and field length requirement. If the predefined parameters reach convergence, the simulation process is successfully ended and a final vehicle design is identified. This final design is then subject to a DOC evaluation and optionally to a PrADO internal noise prediction with PANAM embedded as a PrADO modulea . B.

SHADOW

The analysis of noise shielding effects is of growing importance with regard to future aircraft concepts, e. g. blended wing bodies or the DLR low-noise aircraft concept (LNA, see Fig. 9(f)). At DLR, the ray tracing a Without

consideration of shielding effects.


Checking of design constraints MD 30

DOC MD 29

Aircraft noise prediction (DLR PANAM) MD 48

yes Parameter convergence ?

no

Control & stability (HTP/VTP) MD 26,27

Take-off & landing performance MD 24,25

Mass, C.G., inertia MD 21,22 (structure analysis)

Load assumption MD 20

Flight simulations MD 16-19,34,38,40

Landing gear analysis/LCN MD 15

Propulsion design MD 6,33,14

Aerodynamics MD 12

Aircraft geometry MD 2-11

Requirements MD 1

Automatic design iteration

Automatic parameter variation/optimization

Figure 2. PrADO workflow: Aircraft design synthesis process

tool SHADOW9 has been developed in order to investigate different engine installations with respect to structural noise shielding. SHADOW is based on a high frequency approximation of the linearized Euler equations. Thereby, the pressure field is calculated by solving ordinary differential equations along lines in space, so-called rays. These rays originate in a point source, which approximates the center of the fan disc. So far, no mean flow effects are taken into account. The aircraft geometry is approximated by a triangulated surface, i.e. great flexibility in representation of complex geometries. Furthermore, a local 2nd-order polynomial geometry approximation is applied to take into account the surface curvature, thus enable proper calculation of the ray reflection. The pressure amplitude along each ray is calculated based on an energy conservation principle, which requires the calculation of the Jacobian of the ray-field. Evaluating the Jacobian with a difference approximation can lead to problems due to strong divergence of the rays after multiple reflections. Therefore, a differential equation for the Jacobian was derived and integrated along with the ray equations9 . In order to calculate the acoustic shielding at prescribed points in space, a shooting procedure using a Newton algorithm is applied. In the high frequency limit the diffractive part of the solution of the wave equation is lost. Thus, in ray-tracing the diffraction has to be taken into account by special approximations, e.g. the geometrical theory of diffraction12 . While this approach yields quite accurate results, its implementation for arbitrary geometries is complicated. Therefore, a more simple approach based on the Maggi-Rubinowicz formulation of the Kirchhoff diffraction theory has been implemented13, 14 . In this case, the diffracted field is calculated by a line integral along the shadow boundary on the surface of the diffracting body. In contrast to the ray field, the diffraction correction is frequency dependent. Since only the acoustic far field is calculated in the implemented form of the high frequency approximation, it is sufficient to consider acoustic point sources9 . Ultimately, the tool combines fast computation and at the same time simplified input requirements. As a consequence, the tool is well suited for application within conceptual aircraft design. The required input data for SHADOW, i.e. aircraft geometry and engine alignment, can directly be generated with PrADO. For a selected aircraft/engine design, frequency dependent noise attenuation factors are predicted with SHADOW and then fed back into PANAM for further processing. The attenuation factors are then accounted for in PANAM when simulating the fan noise contribution1 . Fig. 3 visualizes the significant impact of fan noise


shielding on the overall aircraft noise emission under approach flight conditions. Relevant emission angles are subject to large noise level reductions due to the shielding. Z Z X

Y

X

Y

SPL: 70 75 80 85 90

20

20

10

10

SPLZ [dB]

SPLZ [dB]

0

0

-10

-10

-20 20

SPL: 70 75 80 85 90

20

20 0 -10

10 SPL Y [dB]

0

10 ] [dB

-20 20

LX SP

-20

(a) reference vehicle without relevant shielding

0 -10

10 SPL Y [dB]

0

10 ] [dB

LX SP

-20

(b) vehicle with significant fan noise shielding

Figure 3. Impact of fan noise shielding on overall noise emission (under approach flight conditions)

C.

PANAM

The Parametric Aircraft Noise Analysis Module (PANAM10 ) accounts for individual noise sources depending on their geometry and operating conditions. Major noise sources are accounted for by individual noise source models10 . These models are semi-empiric but physics-based. Their parametric setup allows for a modification of both geometry and operating conditions. These modifications are directly accounted for, thus define the overall noise emission of each noise source. According to the operating condition and the configurational setting of the aircraft, noise impact on the ground can be predicted along simulated flight procedures. The implemented models for engine and airframe noise sources come from the literature and/or are DLR in-house developments. Ultimately, all required input data for the implemented noise source models can automatically be generated within the design processb . The airframe noise source models are derived from a data base comprised of measurements or simulations, i.e. high fidelity computation, wind tunnel, and flyover experiments, see Refs.15, 16 . Obviously, derivation of models and approximations from a fixed data base sets up inherent limitations to the applicability. General application of a noise source model is not feasible. Each noise source model can only be applied according to the underlying and case-specific data, i.e. general noise generating mechanisms and physics need to be consistent with the data. Reasonable and reliable results are achieved, if principal design features are kept constant or if empirical constants are adapted. According to Ref.16 , the prediction accuracy for the selected airframe noise source models is within approximately 1 dB(A). Result uncertainty increases if existing noise source models are applied toward vehicle concepts far off the design space specified by the underlying empirical data base. b Note:

A few detailed engine design parameters, e.g. rotor-stator-spacing, cannot be generated within the presented simulation process but have to be provided as input.


Reliability and validity of engine noise predictions is limited to the underlying database because extrapolation is not recommended17 . Jet noise modeling with Stone18 does not impose significant restrictions because the relevant parameter domain for the presented work is covered, i.e. preferred engine sizes and thrust requirements are mapped. The selected fan noise model by Heidmann19 is based on a somewhat antique database hence application should be limited to older engine types with low bypass ratios. In order to apply Heidmann’s approach to modern turbofan engines, adaption and modification of the inherent empirical constants becomes necessary. Recent studies from NASA20 show that such a modified approach can be applied for quite a wide range of fan pressure ratios, tip speeds, and bypass ratios. It is demonstrated, that predicted results are usually within a 4 dB margin with respect to the measurements20 for bypass ratios up to 13.3, i.e. a satisfying agreement to the experimental data. PANAM’s implemented fan noise model features similar modifications of Heidmann’s original constants. Therefore, it is assumed that the implemented engine noise source models can be applied to modern turbofan engines up to high bypass ratios around 15. A model to account for noise absorption due to acoustic liners has been implemented. The method as developed by Moreau, Guérin, and Busse21 further improves overall result accuracy especially under take-off conditions. Returning PANAM results back to the simulation framework completes the overall process, i.e. introducing noise as a new design constraint. Arbitrary noise results can be selected as overall design criteria, e.g. noise isocontour areas along approach and departure. D.

Overall simulation process

The overall aircraft design-to-noise simulation process, see Ref.1 , is briefly described in the following. The tools PrADO, SHADOW, and PANAM are arranged into the simulation process as depicted in Fig. 4. Each tool has specific tasks and generates the required input data for the subsequent tools. 1. PrADO is applied to define the final aircraft and engine design. After successful integration and convergence of the aircraft and engine design modules, a feasible vehicle design is identified and the generated results are provided to the subsequent tools. In addition to the vehicle design synthesis, PrADO simulates the operation of the final aircraft design during (a) take-off, (b) design mission, and (c) landing, thus generates additional input data for the overall system noise prediction. 2. If required, SHADOW is applied to evaluate the impact of noise shielding effects on the overall ground noise impact for each vehicle design. The PrADO vehicle layout and engine locations are provided as input for SHADOW. 3. PANAM is applied to evaluate the vehicle’s acoustic performance as perceived on the ground. PANAM inputs the vehicle and engine geometry, the engine performance map, and the flight procedures from PrADO. SHADOW’s predicted shielding factors are accounted for in PANAM’s simulation of the fan noise emission. As input for the process, the user has to provide a basic vehicle concept and the Top Level Aircraft Requirements (TLAR). With this input, the aircraft design synthesis and engine performance simulation can be initiated. The basic vehicle as specified by the user is evaluated under the given TLAR. Simultaneously, the engine performance is simulated and engine design parameters can be adapted to the requirements, if required. If a selected vehicle design fulfills the given TLAR and passes the design modules, a feasible aircraft concept is found and the iterative design cycle is successfully completed. The resulting aircraft is then passed on to the module for noise shielding, if required. Based on a three dimensional model of the vehicle and the precise engine location, noise shielding effects can be accounted for. After a successful SHADOW run, the vehicle flight simulation is initiated. The vehicle is simulated along a predefined flight procedure for take-off, design mission (if required), and during approach. According to the specific vehicle design and the engine performance, a final flight track with required engine operation and defined configurational setting is generated. After successful computation of the flight tracks, all necessary input data for an overall noise simulation have been generated and PANAM can be initiated. Ultimately, PANAM predicts the ground noise impact based on each vehicle’s specific data, i.e. overall design, engine data, shielding factors, and


individual flight tracks. Simulation results of all tools are then returned back to process for the final and overall assessment of the vehicle’s environmental and economical performance. For the final vehicle ranking,

TLAR, vehicle concepts

environmental & economical vehicle performance

aircraft design synthesis

structural noise shielding

engine performance simulation

overall aircraft noise prediction

tool / framework:

flight simulation

PrADO, TU BS

SHADOW, DLR

PANAM, DLR

Figure 4. Aircraft Design-to-noise simulation process

arbitrary design objectives can be selected within the simulation process. In the context of this work, certain acoustical and environmental performance parameters have been selected according to engineering judgment and experience. The overall process as depicted in Fig. 4 enables to run a complete design-to-noise analysis, thus enables to evaluate the impact of selected technologies and design modifications on overall vehicle performance.

III. A.

Validation

Aircraft and engine design

Each tool has been validated with available data, measurements, or theoretical findings. PrADO results for existing aircraft can directly be compared with available data from literature or manufacturers6, 7 . Deviation between this data and PrADO simulations can be in the order of a few percent for global parameters of conventional transport aircraft1 , e.g. overall system mass or global flight performance. Overall, the agreement is well suitable for the intended task. The conceptual aircraft design with PrADO yields simulation results that are in accordance with available data1 . Inherent deviations due to approximations and simplifications in PrADO’s computations are systematic and affect all vehicles in equal measure. Furthermore, the context of the presented work lies on level differences rather than precise absolute values. Consequently, the performance ranking of individual vehicles should not be significantly affected due to uncertainties in the PrADO conceptual design phase. Increasing uncertainties have been experienced if the engine performance is only simulated with simplified models. Therefore, in order to minimize uncertainties in the overall process validation, an external and high-fidelity engine performance deck is applied at this point. This engine data has been generated with VarCycle, a high-fidelity tool from DLR23 . B.

Engine installation effects

SHADOW simulation results have been compared to experimental and high-fidelity data (here: Boundary element methods), see Refs.9, 24 . Overall agreement of SHADOW for a full scale aircraft are quite promising9 .


The outcome of this comparison and the fact, that the main focus lies on a comparative design study instead of a precise and detailed level prediction, support the feasibility and reliability of a SHADOW application within the presented design-to-noise simulation process. C.

System noise prediction

PANAM’s implemented noise source models have individually been compared to dedicated component measurements. In order to assess the overall system prediction capabilities, prediction results are directly compared to flyover noise measurements. In general, such a comparison with flyover data can be understood as an assessment of the feasibility of the overall process. The simulation is based on input data from the Level time history: departure rec003

SPL(A) [dBA]

90

2000

departure 7

Y [m]

1000

12

4

90

panam sum [dBA] exp sum [dBA]

10

80

Level time history: departure rec003

100

SPL(A) [dBA]

100

6 3 1

70

panam sum [dBA] exp sum [dBA] 80

8

2

5

70

11 0

10 9

-1000

-2000

6

8

3

5000

X [m]

10000

60

50 32400

2

5

0

60

1

32450

32500

15000

(a) observer locations

32550

Time [s]

Microphone position

50 32400

32600 0m microphones (2012_AzB)

32450

32500

(b) along flight ground track

32550

Time [s]

Microphone position


(c) sideways

Figure 5. Departure SPL time-histories at selected observer locations

presented simulation processc . Dedicated comparison of simulation and measurement for an Airbus A319 indicates feasible overall aircraft noise prediction capabilities. Prediction results and noise related effects are in good agreement to available experimental data and theoretical knowledge. Exemplary, the time-level-history for one approach and one departure flight are presented. The measurement locations for this campaign are depicted in Figs. 5(a) and 6(a). Results along the flight ground track, i.e. Figs. 5(b) and 6(b), and sideways, i.e. Figs. 5(c) and 6(c), are depicted. Furthermore, initial comparison of prediction results versus other aircraft types is available. PANAM predictions are compared to experimental data for DLR’s former flying testbed ATTAS (Refs.25 ), the new DLR Airbus ATRA (not published yet), a Boeing B737 (Ref.22 ), and a Boeing B747 (not published yet). Initial Level time history: approach rec004

100


2000

approach

Y [m]

1000

25 22

19

21

18

16 0

-1000

-2000 -30000

24

20

17

15

15 80

18 24

60


80

14

70

17

23 20

60

13

50 33500 X [m]

21

70

90

14

23

-20000

13

SPL(A) [dBA]

SPL(A) [dBA]

90

Level time history: approach rec004

100

-10000

(a) observer locations

0

33600

Microphone position

33700

Time [s]

33800

50 33500

0m microphones (2012_AzB)

(b) along flight ground track

33600

Microphone position

33700

Time [s]


(c) sideways

Figure 6. Approach SPL time-histories at selected observer locations

results according to the implemented noise source models are quite promising for all four vehicles. Yet, alternative high lift and landing gear design principles (e.g. B747) require modifications to the empirical factors. If these empirical factors are modified, the results show good agreement with the measurements15, 16 . A more accurate simulation of these other aircraft types requires detailed engine design with higher fidelity c except

the engine performance computation as described above, e.g. Ref.23


tools, e.g. Varcycle23 , not being available at the time of finalization of this work. Therefore, the simulation of ATRA, B737, and B747 are based on simplified engine models, thus should be seen more as an initial feasibility check than a final validation. Obviously, the dominating influence of engine operating conditions on noise emission is crucial for accurate modeling and a feasible noise prediction. It is demonstrated, that result accuracy can be significantly improved if more detailed engine data is available, e.g. high-fidelity thermodynamic engine modeling22 . A direct tool-to-tool comparison with other existing scientific noise prediction tools has been initiated and is currently under progress. As presented in Ref.26 , an initial comparison with the University of Manchester’s FLIGHT code confirms the feasibility and result accuracy of PANAM. In the long run, these tool-to-tool comparisons will contribute to establish defined validation standards and realistic accuracy margins for scientific noise prediction tools. Ultimately, the predictions and the tool-to-tool comparison confirm the reliability and feasibility of the simulation process as depicted in Fig. 4.

IV.

Application

The presented process has been applied to optimize a reference aircraft with respect to its environmental performance. The reference vehicle is similar to an A319, see Fig. 9(a). The reference has been selected due to the high market penetration of such single-aisle, medium-range transport aircraft27, 28 . Initial application of the simulation process to the reference vehicle indicates a relevant contribution of fan noise during both take-off and approach operation. Therefore, as an initial application, the application goal is to minimize this noise source. A very efficient way to reduce fan noise is through structural shielding of this rather localized source. The implemented DLR tool SHADOW, as presented earlier, allows to evaluate these shielding effects. Consequently, the first application of the design-to-noise process has the main focus on improved fan noise shielding. At the same time, the engine design and the aircraft TLAR (Top Level Aircraft Requirements) are kept constant. Modifications are limited to (1) wing & empennage (shape and location) and (2) engine installation detailsd . Different vehicle layouts seem promising with respect to engine noise shielding and are selected for further investigation. In order to achieve relevant aircraft noise reduction, more or less radical solutions have been proposed by various researchers. In the context of the presented work, the focus lies on more realistic, medium term solutions with respect to aircraft design. Futuristic vehicle concepts such as blended wing bodies are not in the scope of this work. The selected solution space is based on (or at least inspired by) earlier low-noise aircraft designs from various research institutions and universities, e.g. see Refs.29–31 . Fig. 7 shows the different concepts with increasing priority on noise reduction from left to right. Concepts on the right side of the figure focus on maximum noise reduction accepting possible negative implications on the vehicle economics. Vehicles more on the left side of the figure represent a compromise between noise reduction and vehicle performance. Each one of these 4 vehicle categories is furthermore subject to dedicated parameter variations, i.e. mainly a variation of wing shape, empennage layout, and engine location. Details of the parameter variation are summarized in Table 1. These parameter variations result in a total of almost 500 vehicle variants (marked with a lowercase v ) with each variant subject to the entire simulation process. A.

Evaluation metric

In order to rank-order the available vehicle variants (v -...) and to identify promising concepts, an evaluation metric is required. The existence of a feasible and reliable metric that combines both environmental and economical factors is not known to the authors. Due to their insignificant magnitude, existing noise regulations and noise fees are no incentive to develop or even investigate low-noise technology. Therefore, a multicriterial analysis has been selected. For a comparative analysis of different technologies and vehicles, such an approach seems appropriate. Yet, selection of the relevant parameters is critical. For the presented study, parameters are selected according to the authors engineering knowledge and experience in aircraft design and noise simulation. Of course, the new simulation process allows for an arbitrary parameter selection. d Note:

The exterior fuselage shape and the cabin layout are kept constant during the process.


over-the-wing

empennage

fuselage / wing integration

wing / tail integration

v-0

v-1

v-2

v-x

economics

design priority

environment

Figure 7. Solution space with vehicle categories

vehicle category v-rx, ”◻” v-0, ”△” v-0, ”△” v-1, ”▽” v-1, ”▽” v-1-1, ”▷” v-1-1, ”▷” v-1-2, ”◁” v-1-2, ”◁” v-2, ”◇” v-2, ”◇” v-x, ”o”

parameter wing aspect ratio wing area wing aspect ratio wing area wing aspect ratio wing location wing aspect ratio wing area wing aspect ratio wing location VTP location along wing area HTP location along wing area HTP sweep angle area ratio (main vs. HTP sweep angle VTP area wing aspect ratio wing area wing aspect ratio wing location VTP location along wing area

HTP fuselage

side VTP)

HTP

range 7.0 - 11.0 100.0 - 130.0 m2 7.0 - 11.0 100.0 - 130.0 m2 8.0 - 10.0 30 - 40 % of fuselage length 7.0 - 11.0 100.0 - 130.0 m2 8.0 - 10.0 30 - 40 % of fuselage length 10 - 80 % of HTP chord length 105.0 - 125.0 m2 1.5 - 3.5 m from fuselage end 110.0 - 130.0 m2 -30 to -50 ° 20-8, 22-7, 24-6, 26-5, 28-4, 30-3 -30 to -50 ° 22.0 - 36.0 m2 7.0 - 11.0 100.0 - 130.0 m2 8.0 - 10.0 30 - 40 % of fuselage length 10 - 80 % of HTP chord length 120.0 - 140.0 m2

Table 1. Selected parameter variation according to vehicle category


Selected parameters for environmental and economical vehicle performance are evaluated and compared to the parameters of the reference aircraft. These normalized parameters are dimensionless performance indicators and reveal advantages and disadvantages with respect to the reference performance. The selected environmental performance parameters are aircraft noise induced awakenings along approach and departure operatione . The economical parameters are block fuel, operational empty weight (OEW), and balanced field length (BFL). Block fuel is selected as an indicator for the direct operating costs of the vehicle. The operational empty weight is selected as an indicator for the overall vehicle price. Furthermore, the field length requirement is selected in order to reward short take-off and landing operation (STOL). Such low-noise STOL vehicle could be operated at underutilized airports with shorter runways enabling new transportation business models32 . In order to improve a variant’s overall vehicle performance, the selected individual parameters have to be minimized. In order to get one overall indicator for environment and one for the economy, the selected and normalizedf parameters have to be weighted and translated into an overall score. The selected environmental parameters, i.e. normalized number of aircraft noise induced awakenings along approach and departure, are weighted equally. The overall environmental performance ξ can then be expressed as the sum of 0.5 times the normalized approach awakenings ξ1 and 0.5 times the normalized departure awakenings ξ2 . For the economical performance parameter ζ, the normalized block fuel ζ1 is weighted by 0.7, normalized OEW ζ2 by 0.2, and normalized BFL ζ3 by 0.1. The overall economical performance score ζ can then be expressed as the sum of these weighted ratios ζ1 to ζ3 . ξ

=

ζ

=

1 1 ⋅ ξ1 + ⋅ ξ2 2 2 7 2 1 ⋅ ζ1 + ⋅ ζ2 + ⋅ ζ3 10 10 10

(1)

These two overall performance indicators ξ and ζ are vehicle specific and remain constant, thus are independent of prevailing boundary conditions, e.g. fuel price. Obviously, the selection of the weighting factors is somewhat arbitrary. Similar as for the performance parameter selection, the weighting factors represent the opinion and experience of the authors, thus can be arguable. B.

Results

Fig. 8 shows the predicted performance indicators ξ and ζ for all vehicle variants under considerationg . Environmental performance indicators ξ are aligned along the x-axis in Fig. 8 and economical performance indicators ζ along the y-axis. Low values for both parameters indicate an improvement compared to the reference vehicle. Obviously, the reference vehicle (”x”) is located at ξ = 1 and ζ = 1. Each vehicle category is marked with different colored signs in Fig. 8, e.g. ”◻” depicts variants of the reference vehicle design. As shown in Fig. 8, certain patterns and sensitivities with respect to the selected design modification and/or vehicle layout can be identified. The differences in environmental and economical performance among the vehicle categories can be tracked back to the underlying physics. For example, the influence of noise shielding with respect to the individual engine location can directly be identified in the results. Within each vehicle category, i.e. v-r to v-x, vehicle specific correlations between the modified design parameters and the overall noise generation can be identified. For example, environmental performance levels ξ can directly be correlated to the underlying variant’s wing shape and aspect ration. The impact of the wing shape on both flight performance, i.e. flight speeds and configuration, but also on the noise generation due to the modified geometries can be identified in the results. Ultimately, one most promising design variant out of each vehicle category v-r to v-x can then be selected. These final 6 vehicle variants are then subject to a subsequent and ”manual” feasibility check with respect to flight stability. In order to guarantee sufficient flight performance, some final vehicle variants did require further modifications to the empennage or the location of the center of gravity (CG). This final postprocess results in a feasible and practical aircraft design out of each vehicle category. Final tailoring of the CG location and the empennage design can result in a shift of the final e Evaluation of aircraft noise induced awakenings is a weighting of max. SPL distribution. This weighted distribution can be transformed into one scalar value based on the population density, i.e. the number of awakened people. f Normalization: the parameters are divided by the corresponding value for the reference vehicle. g Note: The filled symbols depict final vehicle designs which will be introduced below.


Vehicle performance indicators: Environmental (ξ) vs. economical performance (ζ) v-1-1 v-1-2 v-2

v-r v-0 v-1

1.1

v-x

ζ

1.05

X

1

0.95 0.8

0.85

0.9

ξ

0.95

1

1.05

Figure 8. Performance indicators for all evaluated vehicle variants

(a) Reference (V-R, ”x”)

(b) Optimized reference (V-Rx, ”∎”)

(c) Over-the-wing (V-0, ”▲”)

(d) Empennage (V-1, ”◀”)

(e) Wing/fuselage (V-2, ”◆”)

(f) LNA concept (V-X, ”●”)

Figure 9. Reference vehicle and final design out of each category


vehicle V-rx (”∎”) V-0 (”▲”) V-1 (”◀”) V-2 (”◆”) V-x (”●”)

environment, xref,i dep. aw. [-] app. aw. [-] 8.19 14.00 7.79 13.76 7.80 12.33 7.59 11.02 7.66 11.62

block fuel [kg] 9753.7 10692.4 9929.5 10540.8 10396.9

economy, zref,k oew [kg] bal. field length [m] 39731.6 1922.7 39789.5 1962.1 37905.3 1909.3 38005.2 1734.2 39165.8 1827.0

ξ

ζ

0.99 0.96 0.91 0.85 0.88

0.97 1.04 0.97 1.01 1.01

Table 2. Vehicle performance parameters for the final vehicles

1.1

dominating economical impact

all "green" scenario

neutral

overall ranking σ

1.05

worse

reference

1

better

0.95 vehicle: V-R V-Rx V-0 V-1 V-2 V-X

0.9

0.85 0

1

2

3

4

5

6

7

8

9

10

scenario Figure 10. Scenario dependent weighting of final vehicles

vehicle with respect to the unmodified variants out of its category. As depicted in Fig. 8, the final designs are shifted downward and to the left due to the modifications. This confirms the importance of this final design check and the adaption; both environmental and economical performance can be improved for each selected variant. A similar shift would be experienced for any other variant out of the same category after the final design adaption. Each of these final aircraft designs is labeled with a capital letter, i.e. V-r to V-x, and a filled symbol, i.e. ”∎” to ”●”. Figs. 9(b) to 9(f) show the modified reference aircraft and the final aircraft design out of each category. According to the selected economical / environmental parameters and their weighting, each aircraft design has a specific value for ξ and ζ as depicted in Fig. 8. If environmental factors are of dominating importance (”green” scenario), low ξ values are obviously of much higher priority than the economical performance, which would be indicated by low ζ values. Consequently, the overall ranking of a vehicle depends on the selected scenario and its prevailing boundary conditions. In order to account for this fact, an additional weighting can be introduced: the weighting of environmental versus economical performance, i.e. ξ versus ζ. For a specified scenario ”i”, i.e. a fixed weighting of environmental (Ki ) and economical performance (Λi ), 13 of 17 American Institute of Aeronautics and Astronautics

an overall score σi can be assigned to each vehicle. σi

= Ki ⋅ ξ + Λi ⋅ ζ 1 1 7 2 1 = Ki ⋅ ( ⋅ ξ1 + ⋅ ξ2 ) + Λi ⋅ ( ⋅ ζ1 + ⋅ ζ2 + ⋅ ζ3 ) 2 2 10 10 10

(2)

For σi values less than 1.0, the corresponding vehicle has an improved overall performance compared to the reference vehicle. Obviously, the reference vehicle has a score σi of 1.0 in all scenarios. Fig. 10 shows the scenario dependent overall ranking σi of the 6 final vehicles. The scenario dependent weighting is modified along the x-axis, i.e. from 100% dominance of the economical performance (Λ0 = 1, K0 = 0) on the left to 100% dominance of the environmental performance (Λ10 = 0, K10 = 1) on the right. In the middle, environmental and economical performance are weighted equally (Λ5 = K5 = 0.5). It is obvious, that the vehicle ranking depends on the scenario and its underlying boundary conditions. For example, the design V-2 is the best choice if noise is a dominating factor, see Fig. 10. Of course, the overall vehicle ranking is depends on the selected design objectives and the weightings. A different selection of these factors would change the outcome of this study, see Ref.1 . Exemplary, the simulation results for this vehicle are presented. Noise directivities for the V-2 and the reference aircraft V-r are compared for a representative operating condition along an approach and a departure flight procedure. Obviously, due to the engine installation onboard of the V-2, significant fan noise shielding can be achieved. Along the approach procedure, fan noise is reduced from a level comparable to

(a) V-r approach

(b) V-2 approach

(c) V-r departure

(d) V-2 departure

Figure 11. Noise directivities under typical operating conditions (aircraft symmetry plane)

airframe noise levels down to a negligible level. Fan noise contribution, as experienced along the selected departure situation, could be significantly reduced from a dominating level to a level below the noise contribution by the jet, see Fig. 11. Obviously, jet noise represents a ”noise barrier” for the V-2. For further overall noise reduction, a jet noise reduction becomes essential.


An OEW reduction (-1.6%) is predicted and can be attributed to the centered engine installation. In case of one-engine-out, significantly reduced tail surfaces are required to counterbalance the torque, thus decrease the overall vehicle weight. This OEW reduction results in a reduction in balanced field length (-8.4%) which is again advantageous for the environmental performance. Yet, the V-2 design is optimized for maximum noise shielding. Therefore, a large wing with low aspect ratio is selected. This causes reduced aerodynamic performance and increases the wetted area / parasitic drag of the wing. This leads to 3.6% additional block fuel requirement, which has been defined as the most important economical design objective. Finally, the overall economical performance indicator ζ is slightly increased compared to the reference vehicle, i.e. equal to a reduced economical flight performance. Obviously, the significant noise reduction potential of the V-2 comes with negative implications on the economical flight performance.

V.

Summary and Future Work

A new design-to-noise simulation process as established at DLR is presented. The process can be applied to investigate different aircraft designs along a specified departure procedure, design mission, and landing procedure. Environmental and economical parameters are predicted, that can be used to rank-order different vehicles in a comparative design study. Prediction results for an existing reference vehicle are compared to measured experimental data. This comparison indicates feasible overall noise prediction capabilities, thus is applicable to all measures as specified in ICAO’s Balanced Approach33 . This report presents the initial application of this process. A conventional medium-range transport aircraft, i.e. similar to an A319, is subject to modifications of the vehicle layout. The main goal is to reduce the dominating fan noise contribution. A multicriterial approach is installed in order to evaluate and directly compare various vehicle concepts. Selected design objectives and weighting functions are based on engineering judgment and represent the authors opinion (best-practice). A comparative analysis of both environmental and economical performance under various boundary conditions is established. The identification and initial evaluation of promising concepts out of the available solution space concludes the assigned task. It is demonstrated, that the aircraft designer’s influence on the environmental vehicle performance is significant at the conceptual design phase. Early modifications to the basic vehicle layout result in extensive reduction of ground noise levels. Initial PrADO evaluations indicate, that this can be achieved for small decreases in economical vehicle performance, e.g. V-2 design. Maximum engine noise shielding is achieved for this vehicle with fuselage/wing shielding of over-the-fuselage mounted engines. Compared to the reference vehicle, noise level reductions in excess of 10 EPNdB are predicted along both approach and departure procedure. Reduction in maximum SPL(A) is smaller along the departure because SPL does not account for tonal corrections and broadband jet noise is dominating. In conclusion, promising and feasible concepts are identified with the new simulation process. These concepts can be recommended for subsequent investigations of increased fidelity levels. It should be noted, that only fan noise shielding has been considered within the presented work. For a configuration like the V-2 it can be expected, that jet noise levels are decreasing as well. For the selected engine installation, jet noise could be shielded all along the fuselage. But, the necessary prediction methods are not yet available at DLR for implementation into the process. Furthermore, recent findings and technology toward high-lift noise reduction will be incorporated in the near future. Accounting for new leading edge devices, e.g. as documented in an accompanying publication34 , indicates significant airframe noise reduction up to 8 dBA. A combination of shielding effects with such an advanced airframe design promises maximum ground noise reduction.


Acknowledgments This contribution is mainly based on a 2013 doctoral thesis, Noise Prediction within Conceptual Aircraft Design 1 , and has been updated with most recent results and developments. For a PDF or a printed copy of the thesis, please contact the main author. The authors would like to thank the German Research Foundation (DFG - Deutsche Forschungsgemeinschaft) for supporting parts of this work in the framework of the collaborative research center SFB 880 (fundamental research on active high-lift systems for future transport aircraft, https://sfb880.tu-braunschweig.de).

References 1 L. Bertsch: Noise Prediction within Conceptual Aircraft Design, DLR Forschungsbericht, ISRN DLR-FB–2013-20, 2013 2 R.E. Gillian: Aircraft Noise Prediction Program Users Manual, NASA Langley Research Center, 1982 3 C. Burley, L. Lopez: ANOPP2: Progress Update, NASA Spring Acoustics Technical Working Group, presentation, April 21-22 2011, Cleveland, OH 4 E.D. Olson, D.N. Mavris: Aircraft Conceptual Design and Risk Analysis Using Physics-Based Noise Prediction, AIAA 2006-2619, 12th AIAA/CEAS Aeroacoustics Conference, 8 - 19 May 2006, Cambridge, Massachusetts, USA 5 N. Antoine, I. Kroo, K.Willcox, and G. Barter: A Framework for Aircraft Conceptual Design and Environmental Performance Studies, AIAA-2004-4314, 10th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, August 2004, Albany, New York, USA 6 W. Heinze: Ein Beitrag zur quantitativen Analyse der technischen und wirtschaftlichen Auslegungsgrenzen verschiedener Flugzeugkonzepte fuer den Transport groer Nutzlasten, ZLR-Forschungsbericht 94-01, ISBN 3-928628-14-3 7 W. Heinze, C.M. Oesterheld, P. Horst: Multidisziplinaeres Flugzeugentwurfsverfahren PrADO - Programmentwurf und Anwendung im Rahmen von Flugzeugkonzeptstudien, DLRK 2001-194, Deutscher Luft- und Raumfahrtkongress 2001, September 2001, Bonn, Germany 8 W. Werner-Westphal, W. Heinze, P. Horst: Multidisciplinary Integrated Preliminary Design Applied to Unconventional Aircraft Configurations, Journal of Aircraft, 45 (2), pages 581-589, DOI: 10.2514/1.32138, March - April 2008 9 M. Lummer: Maggi-Rubinowicz Diffraction Correction for Ray-Tracing Calculations of Engine Noise Shielding, 14th AIAA/CEAS Aeroacoustics Conference 2008, paper AIAA2008-3050 10 L. Bertsch, S. Gu´ erin, G. Looye, M. Pott-Pollenske: The Parametric Aircraft Noise Analysis Module - status overview and recent applications, AIAA-2011-2855, 17th AIAA/CEAS Aeroacoustics Conference, 5-8 June 2011, Portland, Oregon, USA 11 C. Liersch, M. Hepperle: A Unified Approach for Multidisciplinary Preliminary Aircraft Design, accepted for presentation at the CEAS 2009 European Air and Space Conference, 26 - 29 Oct 2009 12 J.B. Keller: Geometrical Theory of Diffraction, Journal of the Optical Society of America, vol. 52, no. 2, pages 116-130, Feb 1962 13 K. Miyamoto, E. Wolf: Generalization of the MaggiRubinowicz Theory of the Boundary Diffraction Wave - Part I, Journal of the Optical Society of America, vol. 52, no. 6, pages 615-625, June 1962 14 K. Miyamoto, E. Wolf: Generalization of the MaggiRubinowicz Theory of the Boundary Diffraction Wave - Part II, Journal of the Optical Society of America, vol. 52, no. 6, pages 626-637, June 1962

15 W. Dobrzynski, K.S. Rossignol, and D. Almoneit: Handbuch zum modularen Quellmodell fuer das Umstroemungsgeraeusch von Verkehrsflugzeugen, Version 4.0, Internal Report IB 124-2011/? (t.b.d.), Institute of Aerodynamics and Flow Technology, German Aerospace Center 16 W. Dobrzynski, M. Herr, M. Pott-Pollenske: Airframe Noise Source Modeling - Application and Limitations, Presentation, Campus Research Airport Symposium 2011: Active High Lift and Impact on Air Transportation, DLR Braunschweig, 11-12 April, 2011, http://www.aircraft-noise.info (accessed 23 July 2011) 17 S. Gu´ erin, A. Moreau, C. Weckmller: Engine Noise Source Modeling - Limitations and Perspectives, Presentation, Campus Research Airport Symposium 2011: Active High Lift and Impact on Air Transportation, DLR Braunschweig, 11-12 April, 2011, http://www.aircraft-noise.info (accessed 23 July 2011) 18 J.R. Stone, D.E. Groesbeck, C.L. Zola: Conventional profile coaxial jet noise prediction, AIAA Journal, 21(1) (1983), pp. 336-342 19 M.F. Heidmann: Interim prediction method for fan and compressor source noise, NASA Technical Report TMX-71763, 1979 20 E. Envia, D.L. Tweedt, R.P. Woodward, D.M. Elliott, E.B. Fite, C.E. Hughes, G.G. Podboy, and D.L. Sutliff:An Assessment of Current Fan Noise Prediction Capability, AIAA2008-2991, 14th AIAA/CEAS Aeroacoustics Conference (29th AIAA Aeroacoustics Conference), 5 - 7 May 2008, Vancouver, British Columbia Canada 21 A. Moreau, S. Gu´ erin, and S. Busse: A method based on the ray structure of acoustic modes for predicting the liner performance in annular ducts with flow, NAG/DAGA International Conference on Acoustics, Rotterdam, 23 - 26 March 2009 22 L. Bertsch, G. Looye, E. Anton, S. Schwanke: Flyover Noise Measurements of a Spiraling Noise Abatement Approach Procedure, Journal of Aircraft, 48 (2), pages 436-448, DOI: 10.2514/1.C001005, March - April 2011 23 F. Deidewig: Ermittlung der Schadstoffemissionen im Unter- und Ueberschallflug, German Aerospace Center, DLR Forschungsbericht 98-10, Collogne, Germany, 1998 24 K.-S. Rossignol, M. Lummer, J. Delfs: Validation of DLR’s sound shielding prediction tool using a novel sound source, AIAA-2009-3329, 15th AIAA/CEAS Aeroacoustics Conference, June 2009, Miami, FL (USA) 25 L. Bertsch, G. Looye, T. Otten, M. Lummer: Integration and application of a tool chain for environmental analysis of aircraft flight trajectories, AIAA-2009-6954, 9th AIAA Aviation Technology, Integration, and Operations Conference (ATIO), September 2009, Hilton Head Island, South Carolina, USA 26 A. Filippone, L. Bertsch, M. Pott-Pollenske: Validation Strategies for Comprehensive Aircraft Noise Prediction Methods, AIAA 2012-5411, 12th AIAA Aviation Technology, Inte-


gration, and Operations (ATIO) Conference, 17 - 19 September 2012, Indianapolis, Indiana, USA 27 Airbus Company: Airbus Market Forecast 2011-2030, online PDF version (accessed 09 January 2012) 28 Boeing Company: Boeing Current Market Outlook 2011-2030, online PDF version (accessed 12 January 2012) 29 M. Hepperle: Environmental Friendly Transport Aircraft, Notes on Numerical Fluid Mechanics and Multidisciplinary Design, Vol. 87, pages 26-33, Springer Verlag, 13th STAB/DGLR Symposium, 12 - 14 Nov 2002, Muenchen, Germany, ISBN 3-540-20258-7 30 M. Lummer, M. Hepperle, J. Delfs: Towards a Tool for the Noise Assessment of Aircraft Configurations, Aeroacoustics of New Aircraft and Engine Configurations, 8th ASCCEAS Workshop, Budapest, Hungary, 11 - 12 Nov 2004, ISBN 963-420-842-8

31 W. Dobrzynski: Konfigurationen 2020 Abschlussbericht - Konfigurative Lrmkriterien, final project report, ref.-nr.: K2020-DLR-000-017-P4-Final Report, Braunschweig, Germany, 16 July 2007 32 R. Henke, T. Lammering, E. Anton: Impact of an Innovative Quiet Regional Aircraft on the Air Transportation System, Journal of Aircraft, 47 (3), pages 875-886, DOI: 10.2514/1.45785, May - June 2010 33 International Civil Aviation Organization (ICAO), Air Transport Bureau (ATB): Concept of a balanced approach principle to aircraft noise management, appendix C of Assembly Resolution A35-5, 2007, http://www.icao.int/env/noise.htm (accessed 21 April 2011) 34 M. Pott-Pollenske, J. Wild, L. Bertsch: Aerodynamic and Acoustic Design of Silent Leading Edge Devices, AIAA 2014-tbd, 20th AIAA/CEAS Aeroacoustics Conference, 16-20 June 2014, Atlanta, Georgia, USA
