A MORE INTEGRATED DESIGN APPROACH FOR EMBEDDED …

A MORE INTEGRATED DESIGN APPROACH FOREMBEDDED MECHATRONIC SYSTEMS: APPLICATION TOELECTRICAL THRUST REVERSER ACTUATION SYSTEMS

Scott Delbecq1,2 , Marc Budinger2 , Jérôme Piaton1

1Safran Electronics & Defense , 2ICA, Université de Toulouse, CNRS

Keywords: Sizing, Multidisciplinary System Design Optimization, Design Space Exploration,Electromechanical Actuator, Thrust Reverser

Abstract

The design of embedded mechatronic systemslike electromechanical actuation systems in-volves heterogeneous knowledge due to the in-terference of several engineering specializationsand the multiple physical laws that govern theirbehaviour. This results in costly iterations dur-ing the design process and non-optimal solutions.Multidisciplinary System Design Optimizationtechniques provide theoretical foundations andcomputational tools for optimizing large andmultidisciplinary systems. The approach takenuses these techniques within a proposed designand sizing methodology. It allows a holistic siz-ing of mechatronic engineering systems with em-phasis placed on reusability and rapid decisionmaking. This methodology has been applied inan industrial context to size an electrical thrust re-verser actuation system. This more integrated de-sign approach has resulted in significant assess-ments and insights in early trade-off studies forthis complex aircraft subsystem.

1 Introduction

Environmental protection is a key driver for nextgeneration aircrafts as drastic goals of Flightpath2050 are asked by ACARE european council [1].In order to reduce drastically fuel burn, emis-sions and noise, new aircraft concepts have beenproposed. They investigate innovative propul-sion systems [2], propulsion-airframe integration

for boundary layer ingestion [3], structural al-loys, systems and equipment architectures. Oneresearch axis consists in electrifying propulsion,systems and equipment by betting on the advan-tages of electrical technologies in terms of inte-gration, maintenance and power efficiency. Morethan ever, this more electrical aircraft course ofaction is driving the research and technology ofthe aeronautical industry and academics. Theresearch axes that concerns non-propulsive sys-tems can be separated in two main axes. The firstaxis investigates the concept of bleedless aircraftswhich consists in replacing pneumatic systemsby electrical systems. The second axis investi-gates the concept of hydraulicless aircrafts whichconsists in replacing hydraulic systems by elec-trical systems. These electrical systems are moreeco-friendly than hydraulic systems due to the re-moval of irritating Skydrol fluid, more efficientand are also easier to integrate and maintain. Thehydraulicless aircraft concept has lead to the elec-trification of actuation systems for flight controls[4], landing gears [5] and thrust reversers.

The integration of these new technologiesincreases risks of unexpected and critical be-haviours of systems leading to costly re-designs.These new electrical systems have to prove thatthey meet the severe reliability and availabilityrequirements of commercial aircraft when oper-ating in a harsh vibratory and thermal environ-ment. Therefore, one of the challenges for devel-oping a more electrical aircraft is to master the

1

SCOTT DELBECQ, MARC BUDINGER , JÉRÔME PIATON

design and integration of the forthcoming mecha-tronic systems like actuation systems. Hence,emphasis must be placed on developing new de-sign approaches and methodologies to design andsize these new technologies that would ensure anoptimized and successful integration during fu-ture aircraft developments. In parallel, Multidis-ciplinary Design Optimization (MDO) tools havesignificantly increased in sophistication in the re-cent decades. Thus, they have lead to many ap-plications such as aircraft conceptual design (0Danalytic models), aircraft trajectory optimization(0D-1D lumped parameter models) or high fi-delity wing design (3D Finite Element Methodmodels).

This paper will focus on a developed de-signs and sizing methodology and its applica-tion to electrical thrust reverser actuation systems(ETRAS). For that, the paper is organized as fol-lows. Section 2 presents the methodology andits associated framework. Section 3 outlines thearchitecture, design drivers and sizing scenariosof the ETRAS. Section 4 describes the modellingtasks achieved for this case study. Section 5presents the design optimization/exploration for-mulation and the results obtained. Conclusionsand future work are presented in Section 6.

2 Sizing methodology and associated frame-work

The design of mechatronic systems involves sev-eral stakeholders which have different type of en-gineering specialization knowledge. The knowl-edge is available in different engineering teamsof the design office as shown in Fig. 1.

During system design the design drivers ofthe system are well known by the system archi-tect which enables him to manage the other en-gineering specializations (electronics, electricalmachine design, mechanical design, software...)design tasks. The other engineering special-ization experts are well aware of technologicallimits of their components and therefore assessrapidly their performances.

Experienced system architects are able tomake system design trade-offs without inter-

System Architect

ElectricalMachine Engineers

Power ElectronicsEngineers

Digital ElectronicsEngineers

Embedded SoftwareEngineers

System Engineers

Mechanical Design

Engineers

Fig. 1 Different engineering teams and knowl-edge flows

acting with other engineering specialization ex-perts. However, for unconventional systems or inthe absence of a multi-domain experienced sys-tem architects it is not possible. The proposedmethodology permits better consideration for en-gineering specialization knowledge during sys-tem sizing. A methodology consists in a process,methods and a framework. This section is orga-nized as so.

2.1 Process

A process can be defined as successive steps.The first step consists in the definition of require-ments, design drivers and sizing scenarios of thesystem. Once sizing scenarios are defined, mod-elling needs are identified and modelling activi-ties can be outlined. The second step is thus theimplementation of the models requested for siz-ing. The third steps assembles elementary mod-els into reusable groups that are manageable bya single individual. These groups correspondto component level sizing models. The fourthstep assembles these component sizing models toform a complete system sizing model. In the fifthstep, the system sizing model is analyzed and op-timized. The final and sixth step explores the de-sign space using the system sizing model. Thenext part summarizes these steps and outlines theassociated methods.

2

A more integrated design approach for embedded mechatronic systems: Application to Electrical ThrustReverser Actuation Systems

2.2 Methods

2.2.1 Step 1: Requirements, design drivers andsizing scenarios definition

The definition of requirements, design driversand sizing scenarios has to be achieved throughexchange of information between stakeholdersduring early development phase. The informa-tion consists in design knowledge represent needsand constraints of each of the stakeholders. Theseneeds and constraints are translated into require-ments and design drivers used to design and sizethe system. An efficient method to represent eachof this knowledge and their interaction is througha system breakdown structure. Nodes representcomponents, their design drivers and the sizingscenarios. Using different shapes of nodes helpdistinguish them.

2.2.2 Step 2: Elementary computational modelgeneration

In order to provide low computational cost mod-els, the elementary computational model usedare algebraic. Algebraic models can be gener-ated using different methods. Analytic modelscan be obtained by deriving equations represent-ing the laws of physics. Scaling laws are al-gebraic equations based on reference component[6]. Linear regression of component data sheetsprovide the possibility to obtain algebraic modelthat compute component geometrical parame-ters (length, mass) with respect to performances(torque, speed). Surrogate modelling can be usedto fit empirical data or simulation data. Surro-gate modelling techniques that generate algebraicmodels using Finite Element Method (FEM) sim-ulations are available [7]. Surrogate modellingcan also be used for representing lumped param-eter simulations.

2.2.3 Step 3: Elementary models assembly intoreusable component sizing models

The assembly of elementary algebraic modelsenables to use acausal modelling for increasingreusability of models. The combination of graph-based methods and symbolic computation meth-

ods provides techniques to implement an acausalenvironment. Such methods also permit to detectcomputation singularities. This step enables togenerate flexible inputs/outputs component siz-ing models composed of elementary algebraicmodels whilst checking their solvability.

2.2.4 Step 4: Component sizing models assem-bly into a system sizing model

The assembly of component sizing models toform a total system sizing model is intricate be-cause of interconnections between variables thatcome from the system layer. The method usedto represent the system model is the N2 diagram[8]. It is a diagram that represents in the shapeof a matrix the functional or physical interfacesbetween system elements. In addition, a hierar-chical decomposition method is used to deal withthe complexity of system containing a large num-ber of elements.

2.2.5 Step 5: Analysis and optimization usingthe system sizing model

Analysis and optimization using the system siz-ing model can be achieved using Multidisci-plinary System Design Optimization techniques.Gradient-based methods are used with symbolicdifferentiation for elementary models derivativesand Unified Derivatives Equations [9] for totalderivatives. This leads to low analysis and opti-mization times which is very interesting for rapiddecision making during the sizing process.

2.2.6 Step 6: Design space exploration usingthe system sizing model

Design space exploration using the system siz-ing model can be achieved using Multidisci-plinary System Design Optimization techniquesfor multidisciplinary analysis and Design of Ex-periments. Visualization methods like parallelcoordinates and scatter plots provide efficientfeatures for design and sizing purposes.

3


2.3 Framework

Step 3 to 6 are supported by a in-house frame-work called BOA (Bind your models, Optimizeyour system, Accelerate your design) [10]. Tomatch the process this framework is decomposedin four environments: Block Generation, SizingProcedure, Design and Exploration. It was devel-oped using Python programming language. Thecomputational core relies on key scientific com-puting packages like Scipy, Numpy, Sympy, Net-workX, pyDOE and OpenMDAO [11]. The userinterface was implemented using PyQt5. Fig. 2illustrates the user interface of the sizing proce-dure environment.

Design variable block

Input block

Objective block

Causalblock

Fig. 2 Sizing Procedure environment

This framework has been used to achieve thesizing of aileron rotary [12] and linear [13] elec-tromechanical actuators and more recently a pri-mary flight control actuation system [14]. Forthe first time, it is used to achieve a conceptualsizing of an ETRAS. This example is describedin the following sections. The problem is non-exhaustive but includes the main challenges ofETRAS design and sizing.

3 Architecture, design drivers and sizing sce-narios

The ETRAS is supplied by the AC voltage net-work of the aircraft. It has to deploy and stow theengine nacelle transcowl(s) during landing. Thetranscowl is part of the thrust reverser system andcan have either two symmetrical transcowls (C-duct) or one same transcowl (O-duct). A O-ducttranscowl is considered in this case study.

3.1 Architecture

The ETRAS includes an electrical power chainand an electromechanical power chain. The elec-trical power chain main components are an Auto-transformer Rectifier Unit (ATRU), an inverter, abraking resistor and a housing. The electrome-chanical power chain main components are abrushless motor, flexshafts and linear actuators.This architecture is illustrated in Fig. 3.

Actuator

Actuator

Actuator

Actuator

Housing/Heat sink

ATRU Motor

BrakingResistor

InverterAC Network

Flexshaft

Flexshaft

BallscrewTB + Tube

BallscrewTB + Tube

Bevel gear

Flexshaft

BallscrewTB + Tube

BallscrewTB + Tube

Flexshaft

Bevel gear

Bevel gear

Bevel gear

COWL

Mechanical

Electrical

Thermal

Fig. 3 ETRAS power architecture

The ATRU includes an autotransformer, tworectifiers and two interphase inductors. An actu-ator includes a bevel gear, a thrust bearing, a ballscrew and its tube (housing). We focus here onthe sizing of components that have a significanteffect on the total mass of the system. Hence, thesizing of the system includes the autotransformer,the braking resistor, the housing which acts as aheat sink for the braking resistor and inverter, themotor, flexshafts and actuators.

3.2 Design drivers

The system design drivers considered for thisstudy are the requirements such as fail-safe sys-tem, aerodynamic loads and the transcowl mass.The Aborted Take Off (ATO) deploy and stowperformances are specified as a maximum timeto achieve full stroke. Hence, position/speed mis-sion profile for both sequences are considered asa system design driver.

Design drivers are given for each component.The main design driver of the autotransformer,

4


the braking resistor and the housing is their max-imum operating temperature. Since the ETRASapplication is very short (typically a few sec-onds), only heat capacities of components areused to estimate there maximum temperature.The braking resistor is also selected on its peakpower. The braking resistor resistance has alsoto be carefully adapted to DC bus voltage. Themain design drivers of brushless motors for dy-namic and transient applications are maximumtorque, maximum speed, maximum temperatureand inertia. For flexshafts their maximum/fatiguetorque and maximum speed are technologicallimits whereas rotational stiffness and inertia par-asitic characteristics. Bevel gears are selected ontheir reduction ratio and maximum/fatigue torquewhereas inertia is a parasitic characteristic thathas an effect on the sizing problem. Thrust bear-ings are selected using the maximum/fatigue ax-ial force. The main design drivers of ball screwsare their pitch, the maximum/fatigue axial forceand axial stiffness.

3.3 Sizing scenarios

The first sizing scenario considered is the ATOdeploy at cold temperatures (high friction in actu-ators and flexshafts). The torque and speed mis-sion profile depends on the deploy position/speedprofile, inertias, transcowl mass, aerodynamicloads and mechanical efficiency. It has an signif-icant effect on motor maximum torque and speedand the autotransformer peak power. The aero-dynamic load are position dependent. The de-ploy sequence begins with a phase where the mo-tor drives the transcowl and follows with a phasewhere the motor brakes the transcowl and thusgenerates energy.

The second sizing scenario considered is theATO deploy at hot temperatures (low friction inactuators and flexshafts). The high mechanicalefficiency leads to more braking torque for themotor, higher peak power ad more energy to dis-sipate for the braking resistor.

The third sizing scenario taken into account isthe stow at cold temperatures. Loads are smallerthat the ATO deploy but there is no braking phase

so the motor is always driving the transcowl.Hence, it does not determine peak torque orpower for motor and autotransformer but gener-ate important energy losses that they have to con-tain thanks to their intrinsic heat capacity.

The fourth and fifth sizing scenarios takeninto account are jamming failure modes. As thesystem has to be fail-safe, all components shallresist to such event. The loads taken are for ATOdeploy. The stress generated in mechanical com-ponents depends on inertias, stiffness, transcowlmass and motor torque/speed operating point atjamming.

Fig. 4 outlines the architecture considered forthe sizing problem, the different design driversand sizing scenarios as well as their interactionsand therefore the high complexity of the ETRASdesign.

4 Sizing models

The sizing models used enable to compute thequantity of interest of sizing scenarios. Estima-tion models are also used to determine to compo-nent parameters involve in sizing scenarios mod-els. Efforts are made to use algebraic models foroptimization by introducing 0D models (scalinglaws, analytic), surrogate models of 0D-1D (Dy-mola) and 3D models (FEM).

4.1 Sizing scenarios models

A surrogate model of lumped parameter modelsimulation is used to represent ATO deploy (coldand hot) and stow (cold). A 0D-1D lumped pa-rameter model is achieved using Dymola Soft-ware. It computes the maximum power, torque,speed and energy at transcowl level during mo-tor and generator (Deploy only) phases with re-spect to aerodynamic loads, total inertia of thesystem and a one degree of freedom trapezoidalspeed profile. The trapezoidal speed profile is im-plemented so that the only degree of freedom ismaximum speed at transcowl level. The modeluses the inverse simulation features of Dymolaas shown in Fig. 5.

A Latin Hypercube Sampling of 1000 sam-

5


ETRAS

Autotransformer Braking ResistorHousing MotorFlexshafts Actuators

Max Temperature

Max Temperature

Max Temperature Max TorqueMax SpeedMax TemperatureInertia

Max Torque

Max SpeedStiffness Inertia

Max Torque/Force

Max SpeedStiffness Inertia Reduction ratio

ATO Deploy (cold)

ATO Deploy (hot)

Stow (cold)Jamming sliderJamming actuator

Stow mission profile

Deploy mission profile

Aerodynamic loadsTranscowl mass

Fail-safe

Fig. 4 Non-exhaustive system breakdown graph of the ETRAS sizing problem with component designdrivers (oval), system design drivers (octogone) and sizing scenarios (hexagone)

Kinematics+

aerodynamic loads

Fig. 5 Stow mission profile model at transcowllevel in Dymola

ples is achieved on total system reflected inertiaand speed profile parameter. The simulations areexecuted using a co-simulation Functional Mock-up Unit. Post-analysis computes the quantity ofinterest (energy, peak power...) for each simula-tion and then a third-order polynomial responsesurface surrogate is used to represent them withrespect to total system inertia and speed profileparameter.

As the ETRAS operates during a very shorttime, components maximum temperature sizingscenario are computed with respect to the equiv-alent energy of losses and heat capacity.

Maximum speed of motor and flexshafts arechecked using analytic models.

Jamming sizing scenarios models computethe peak torque/force applied on mechanicalcomponents with respect to inertias, stiffness,aerodynamic loads, motor torque and speed.

They use an analytic model (Equation 1) that as-sumes that the equivalent kinetic energy of spin-ning inertias is transformed into potential elasticenergy of mechanical stiffness.

Fjam =√

MeqKeqV 2 (1)

Where Fjam is the jamming force, V the equiv-alent mass’s (Meq) speed and Keq the equivalentstiffness at the jamming point.

4.2 Estimation models

The autotransformer sizing is achieved usingscaling laws and an existing design. The auto-transformer is sized for constant magnetic satura-tion as it operates during a very short time. How-ever, the thermal behaviour has to be checked.Hence, the heat capacity is also estimated using ascaling law.

The braking resistor electrical resistance andthermal resistance are obtained using manufac-turer analytic model. The housing heat capacity,which acts as a heat sink for the braking resis-tor and the inverter, is obtained with an analyticmodel.

The motor electromagnetic model is obtainedusing surrogate modelling of FEM simulations(Fig. 5). It considers the magnetic saturationof the core material. It computes the electro-magnetic torque with respect to three high levelvariables, the stator diameter, the yoke thickness

6


and current density in the windings. Models thatcompute Joules and iron losses are also imple-mented. Details of this model can be found in[7].

Fig. 6 Brushless motor electromagnetic FEMmodel using FEMM

Flexshafts diameter, stiffness (Equation 2),inertia and mass are estimated using scaling lawswith respect to peak torque.

K f s = G−1lre f

(Tf s

Tf sre f

)43 L−1

f s (2)

Where K f s is the flexshaft stiffness, Glre f andTf sre f respectively the linear torsional deflectionand peak torque of a reference flexshaft, Tf s theflexshaft peak torque and L f s the length of theflexshaft.

The bevel gear is sized using a scaling lawthat assumes the tooth contact stress to be themain effect on gear mass.

Ball screw parameters are estimated usingscaling laws with respect to peak axial load.However, buckling load is check using a scalinglaw as well.

Thrust bearing scaling law assumes Hertzcontact stress as the main design driver.

4.3 System sizing model

The system sizing model assembles all the sizingscenarios and parameter estimation models. Themission profile analysis evaluates the ATO de-ploy (cold and hot) and Stow (cold) with respecttheir respective speed profile parameter and totalsystem inertia. It analyzes the more constrain-ing variables in the different operating quad-rants for the components sizing such as maxi-mum speed, maximum load and energy. These

sizing inputs are then distributed to the differentcomponent sizing bricks. The system level de-sign variables are two speed profile parameters,spur gears reduction ratio, motor stator diameter,yoke thickness and current density. The systemlevel outputs of the sizing model are autotrans-former, braking resistor, housing and motor tem-peratures. Motor and flexshaft maximum speedshave also to be evaluated.

The sizing problem includes two multidisci-plinary couplings. First, the total system inertia isrequired to compute mission profile analysis butis evaluated using the inertia of components esti-mated with respect to the outputs of mission pro-file analysis. The second multidisciplinary cou-pling comes from the jamming scenarios. The in-ertias and stiffness of components are required toachieve jamming load analysis but are evaluatedusing components sizing models that require thevalue of this jamming load. Hence, these multi-disciplinary couplings have to be solved to pro-vide a consistent design solution.

A Multidisciplinary System Design Opti-mization monolythic formulation is used to solvethem. It consists in adding a normalized de-sign variable and an inequality constraint for eachcoupling and the system total mass as an objec-tive. For example, the first coupling can be de-scribed by:

Fnom = f (Js,α)

Js = g(Fnom)(3)

Where Fnom is the nominal force from the missionprofile analysis, α the ATO deploy speed profileparameter and Js the system total inertia.

Is transformed into:

Fnom = kos ·h(α)Js = g(Fnom)

Fnom ≥ f (Js,α)

(4)

Where h = f (Js = 0,α).This formulation has a lighter computational

cost compared to other monolythic formulationslike Individual Discipline Feasible (IDF) or AllAt Once (AAO) [15]. Nevertheless, the use ofthe inequality is only possible because decreasing

7


the nominal force Fnom decreases the total systemmass and that is the objective.

5 Design optimization and exploration

5.1 Formulation

The representation of the system sizing modeland optimization problem is given in Fig. 7 usingthe Extended Design Structure Matrix diagram[16].

The sizing optimization implemented mini-mizes the total mass of the ETRAS Mtot with re-spect to deploy and stow speed profile parameterαd , αs, motor diameter Dmot , yoke thickness ey,current density Jcur and consistency variables formission profile load analysis kosF and jammingscenario kos jam . The constraints come from tech-nological limits of components like maximumspeed Ω∗ and maximum temperature Θ∗. Twoconsistency constraints introduced previously areused to solve the two multidisciplinary couplings.The optimization problem formulation is the fol-lowing:

minimize Mtot

with respect to αd,αs,Nred,Dmot ,

ey,Jcur,kosF ,kos jam

subject to Ωmot −Ωmotmax ≤ 0Ω f s −Ω f smax ≤ 0Fdyn −Fnom ≤ 0Fjams −Fnommech ≤ 0Fjambs −Fnommech ≤ 0Θat −Θatmax ≤ 0Θhous −Θhousmax ≤ 0Θbr −Θbrmax ≤ 0Θmot −Θmotmax ≤ 0

(5)

Performing this optimization provides thepossibility to assess rapidly integration parame-ters such as mass and dimensions of the systemin order to make system integration trade-offs.

For example such sizing code can assess theeffect of changes of requirements like deploytime on the total system mass. In addition, the

effect of changes of technologies like motor iner-tia can be evaluated rapidly.

5.2 Results

The study of the effect sizing scenarios on theglobal design is used to illustrate both the utiliza-tion of the methodology and the sizing problem.The sizing scenarios such as jamming and manyother have to be included in the design in orderto avoid non-compliant design solutions. Fig.8 outlines optimization results in terms of massbreakdown of the ETRAS when considering andnot considering the slider jamming and ball screwjamming.

Jamming scenarios not included Jamming scenarios included0.0

0.2

0.4

0.6

0.8

1.0

Mas

s [p.

u]

MotorActuatorsFlexshaftsHousingAutotransformer

Fig. 8 Mass breakdown optimization resultswithout (left) and with (right) jamming scenarios

Results show that these jamming scenarioshave a significant impact on the mass of the actu-ators. As they are the main source of mass, it di-rectly impacts the system total mass. Therefore,emphasis must be placed on implementing siz-ing codes that enable the consideration of a largenumber of sizing scenarios, design variables anddesign constraints.

5.3 Discussions

The proposed more integrated design approachuses a in-house framework. This frameworkprovides the possibility to capitalize the knowl-edge through sizing models libraries and sizingprojects libraries. This way sizing scenarios canbe reused for other projects. This is a tremendousfeature as generations of experts have started to

8


α(0)d , α

(0)s , N

(0)red, D

(0)mot, e

(0)y , J

(0)cur, k

t,(0)osF , k

t,(0)osjam

Mcowl

Optimizer αd, αs, ktosF

Nred, Dmot, ey, Jcur Nred, ktosjam

Nred Nred

Missionprofile

analysisEmot Egen, Pmax Egen, Emot Egen, Emot, Fnom, Vmax Fnom, Vmax Fnom Faero, Vmax

g1 : ΘatAutotransformer

sizingMat

g2 : Θbr

BrakingResistor

sizing

g3 : ΘhousHousing

sizingMhous

g4 : Θmot,ΩmotMotorsizing

Jmot Jmot, Tem Mmot

Mechanicalcomponents

loads

Tfs,Ωfs Fact Fact

g5 : ΩfsFlexshaft

sizingJfs Jfs,Kfs Mfs

Actuatorsizing

Jact Jact,Kact Mact

gc1 : FdynDynamic load

analysis

gc2 : Fjamslider

Jamminganalysis

f : MtotTotalmass

Fig. 7 XDSM diagram for the ETRAS sizing optimization problem

retire and leave organizations with their preciousexpertise. The sizing scenarios of the ETRAScase study presented in this paper come from themain design drivers of the system. Nevertheless,the sizing of the ETRAS includes also scenar-ios that have not been taken into account such asgeometric integration or different voltage/powersupply configurations. However, this confirmsthat design sizing methodologies are mandatoryto design such complex systems in a holistic andtightly coupled manner.

6 Conclusion and future work

The more electrical aircraft course of action haslead to the integration of new electrical technolo-gies like electromechanical actuation systems.The introduction of such technologies requires todeveloped design and sizing methodologies in or-der to achieve successful developments. Such de-sign and sizing methodology and its associatedin-house framework have been outlined. It en-ables to integrate different engineering special-ization knowledge to achieve a holistic sizing ofthe system. This permits to capitalize expertiseknowledge throughout the design process thatcan be reused easily for forthcoming projects. In

addition, the framework includes rapid optimiza-tion and exploration capabilities as emphasis isplaced on rapid decision-making during early de-sign phases. Then, this methodology was appliedto the sizing of an ETRAS where a large num-ber of disciplines and scenarios, various types ofmodels (0D,1D,3D) and multiple couplings wereinvolved as typical MDO applications.

This paper has shown that it is possible to in-tegrate and capitalize different engineering spe-cialization expertise in one same design and siz-ing environment. The ETRAS had not yet beentreated in scientific literature. This paper is anintroduction to the complexity of such system asonly the main sizing scenarios were considered.However, this offers interesting perspectives forfuture work. The mission profile has a significanteffect on the total system sizing. Hence, futurework will include the optimization of both mis-sion profile and motor control strategies.

References

[1] Kallas S, Geoghegan-Quinn M, Darecki M,Edelstenne C, Enders T, Fernandez E and Hart-man, P. Flightpath 2050 europe’s vision for avi-ation. Report of the High Level Group on Avia-

9


tion Research, European Commission, Brussels,Belgium, 2011.

[2] Charier G and Gallet F. System of com-pact contra-rotating propellers. U.S. Patent No.9,057,326., 2015.

[3] Gray J, Mader C, Kenway G and Martins J.Modeling boundary layer ingestion using a cou-pled aeropropulsive analysis. Journal of Air-craft, Vol. 55, No. 3, pp 1191-1199, 2017.

[4] Van Den Bossche D . Model Based Design andExperimental Validation of Electro-mechanicActuator Systems for a Nose Landing Gear.Proc International Congress of the Aeronauti-cal Sciences, Hamburg, Germany, 2006.

[5] Doberstein D and Thielecke F. Model BasedDesign and Experimental Validation of Electro-mechanic Actuator Systems for a Nose LandingGear. Proc International Congress of the Aero-nautical Sciences, Brisbane, Australia, 2012.

[6] Budinger M, Liscouët J, Hospital F and MaréJ-C. Estimation models for the preliminary de-sign of electromechanical actuators. Institutionof Mechanical Engineers, Part G: Journal ofAerospace Engineering, 226(3), 243-259, 2012.

[7] Sanchez F, Budinger M and Hazyuk I. Dimen-sional analysis and surrogate models for thethermal modeling of Multiphysics systems. Ap-plied Thermal Engineering, 10, 758-771, 2017.

[8] Kapurch S J. NASA systems engineering hand-book. Diane Publishing, 2010.

[9] Martins J R and Hwang J T. Review and unifi-cation of methods for computing derivatives ofmultidisciplinary computational models. AIAAjournal, 51(11), 2582-2599, 2013.

[10] Delbecq S, Budinger M, Hazyuk I, Sanchez Fand Piaton J. A framework for sizing embeddedmechatronic systems during preliminary design.Proc IFAC-PapersOnLine, 50(1):4354 – 4359,Toulouse, France, 2017.

[11] Gray J, Moore K and Naylor B. OpenM-DAO: An open source framework for mul-tidisciplinary analysis and optimization. 13thAIAA/ISSMO Multidisciplinary Analysis Opti-mization Conference, Fort Worth, Texas, USA,2010.

[12] Sanchez F and Delbecq S. Surrogate model-ing technique for the conceptual and prelimi-nary design of embedded actuation systems and

components. Proc International Congress of theAeronautical Sciences, Daejeon, Korea, 2016.

[13] Delbecq S, Tajan F, Budinger M, Maré J.-C andSanchez F. A framework for the conceptual andpreliminary design of embedded mechatronicsystems. Proc International Workshop on air-craft System Technologies, Hamburg, Germany,2017.

[14] Delbecq S, Budinger M, Piaton J, Daguse B.Optimization of primary flight control actuationsystem using parametric sizing models of actu-ators, power electronics and structural analysis.Recent Advances in Aerospace Actuation Sys-tems and Components, Toulouse, France, 2018.

[15] Martins J R and Lambe A. MultidisciplinaryDesign Optimization: A Survey of Architec-tures. AIAA journal, 51(9), 2049-2075, 2013.

[16] Lambe A and Martins J R. Extensions to the de-sign structure matrix for the description of mul-tidisciplinary design, analysis, and optimizationprocesses. Structural and Multidisciplinary Op-timization, 46(2), 273-284, 2012.

Contact Author Email Address

mailto: [email protected]: [email protected]

Copyright Statement

The authors confirm that they, and/or their companyor organization, hold copyright on all of the origi-nal material included in this paper. The authors alsoconfirm that they have obtained permission, from thecopyright holder of any third party material includedin this paper, to publish it as part of their paper. Theauthors confirm that they give permission, or have ob-tained permission from the copyright holder of thispaper, for the publication and distribution of this pa-per as part of the ICAS proceedings or as individualoff-prints from the proceedings.

10

A MORE INTEGRATED DESIGN APPROACH FOR EMBEDDED …

Documents

Transcript of A MORE INTEGRATED DESIGN APPROACH FOR EMBEDDED …