Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 52, 1, pp. 47-60, Warsaw 2014 MULTI-OBJECTIVE CONCEPTUAL DESIGN OPTIMIZATION OF A DOMESTIC UNMANNED AIRSHIP Sasan Amani, Seid Hossein Pourtakdoust, Farshad Pazooki Department of Mechanical and Aerospace Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran; e-mail: s.amani@srbiau.ac.ir; pourtak@srbiau.ac.ir; pazooki fa@srbiau.ac.ir Autonomous airships have gained a high degree of importance over the last decades, both theoretically as well and practically. This is due to their long endurance capability needed for monitoring, observation and communication missions. In this paper, a Multi-Objective Optimization approach (MOO) is followed for conceptual design of an airship taking aerody- namic drag, static stability, performance as well as the production cost that is proportional to the helium mass and the hull surface area, into account. Optimal interaction of the afo- rementioned disciplinary objectives is desirable and focused through the MOO analysis. Standard airship configurations are categorized into three major components that include themain body (hull), stabilizers (elevators and rudders) and gondola.Naturally, component sizing and positioning play an important role in the overall static stability and performance characteristics of the airship. Themost important consequence ofMOOanalysis is that the resulting design not only meets the mission requirement, but will also be volumetrically optimal while having a desirable static and performance characteristics. The results of this paper are partly validated in the design and construction of a domestic unmanned airship indicating a good potential for the proposed approach. Key words: airship, multi-objective optimization, Pareto optimality Nomenclature x,y,z – body fixed coordinates X,Y,Z – aerodynamic forces L,M,N – aerodynamic moments α,β – angle of attack and side slip angle ρair,ρHe – air and helium density V0 – velocity magnitude of point O F,T – external forces and external moments vector ω – angular velocity in body coordinates m,M ′ – total and addedmass I,E – moment of inertia tensor and identity matrix rG – reference vector from center of gravity to any arbitrary point I0,I ′ 0 – moment and addedmoment of inertia about point O F0 – external forces acting through point O lf1, lf2 – x-distance from origin to the aerodynamic and geometric center of fins lf3 – y,z-distance from origin to the aerodynamic center of fins lgx, lgz – x- and z-distance from origin to the aerodynamic center of gondola CXu – derivative of CX (axial force coefficient) with respect to the translational velocity u 48 S. Amani et al. CYβ – derivative of CY (lateral force coefficient) with respect to sideslip angle CZα – derivative of CZ (normal force coefficient) with respect to angle of attack CLβ – derivative of CL (rolling moment coefficient) with respect to sideslip angle CMα – derivative of CM (pitching moment coefficient) with respect to angle of attack CNβ – derivative of CN (yawing moment coefficient) with respect to sideslip angle a1,a2 – parts of the semi-major axis of ellipse b – semi-minor axis of ellipse Sf,Sg,A – fin area, area of gondola and reference area, respectively D,CD – drag force exerted and drag coefficient on the airship body 1. Introduction Airships have gained significant attention for municipal and industrial applications since the 1980s. There are a few key reasons contributing to this fact. First, they exploit the natural buoyancy effect for descending and hovering purposes that greatly reduces expenditure of extra energy for these tasks. Second, they are able to operate at low tomedium altitudes required for many civilian and surveillancemissions such asmonitoring, observation and communication link purposes. In addition, these lighter than the air (LTA) vehicles need a lower resolution camera formonitoring purposes,making them superior to LowAltitudeOrbit (LEO) satellites forEarth observations. These facts make the airship an ideal candidate for surveillance, reconnaissance and communication relays. There are generally two approaches available for the design of hybrid airships. The first and more conventional approach is to rely on experienced designers to select the appropriate configurationandcomponentsusing theavailable exclusive orproprietarydatabases.Thesecond and recently favored approach is throughMOO techniques to attain the optimal configuration. The latter approach is followed in this paper, which produces a design that while meeting the mission requirements, optimizes the aerodynamic drag, static stability and performance indices as well as the airship geometry. The latter objective also affects the construction cost of the airship. In other words, theMOO approach allows the designer to select the best configuration which optimizes a series of previously mentioned objectives at the same time. According to the literature, someprevious researches considered the aerodynamic discipline in the optimal airship shape design (Lutz andWagner, 1998; Nejati andMatsuuchi, 2003;Wang and Shan, 2006). The effect of the structure and solar energy in the airship design was also investigated (Wang et al., 2009; An et al., 2007). Wang et al. (2009) used Simulated Annealing (SA) to minimize a composite single-objective function consisting of volumetric drag, structural stress, surface area and solar arrays. An et al. (2007) considered minimization of drag and structural weight using the MOO approach. They also determined forbidden design regions to be avoided. Wang et al. (2011) studied a winged airship design (high altitude stratospheric platform) and focused on verifying merits of lift to buoyancy ratio and its effects on the airship hull sizing and usage of solar energy. A thorough study of the previous works reveals that objectives such as static stability and performance characteristics as well as factors affecting the construction cost has so far not been considered. In this study, the MOO design approach is considered with new contributions to those previously utilized by researchers, An et al. (2007). First, the complete airship configuration including the tail and gondola are taken into account. Second, the static stability as well as performance (related to glide range) characteristics plus the production cost related to the hull surface and volume are considered as objectives. Considering the above objectives, an optimal configuration that simultaneously satisfies a number of objectives is determined using the results of several paretos. Finally, the proposed design configuration is implemented in the real world Multi-objective conceptual design optimization of a domestic unmanned airship 49 to verify the pragmatic aspects of the proposedMOO conceptual design approach. This paper is arranged as follows: Section 2dealswith the problem formulation and represen- tation of the airship six degree of freedom nonlinearmodel used to evaluate the gliding distance traveled by the airship. Stability derivatives are discussed in Section 3 as a reference for static stability and performance requirements. Optimal configuration design is the subject of Section 4 where discussions are made about the optimality definition of the airship configuration, the utilized objective functions as well as the design variables. MOO and genetic algorithm are the subjects of Sections 5 and 6, respectively. Section 7 presents simulations followed by discussion of the results. Conclusions are drawn in Section 8 followed by a description of the future work under consideration by the authors. 2. Problem formulation In this section, the governing dynamic equations for the nonlinear airship are presented. The assumptions that lead to these equations as well as the utilized coordinate system are discussed. The interested reader can refer to Mueller et al. (2004), for more information about the details of the nonlinear model development. It is important to note that the presented equations will be utilized to calculate the distance traveled by the airship. 2.1. Airship model The airship configuration plus fins and gondola position are depicted in Fig. 1. According to the airship configuration shown in this figure, three fins are mounted in the aft section of the airship. These include two elevators (right and left) for longitudinal control and two rudders (top and bottom) for lateral-directional control purposes. Fig. 1. Airship configuration 2.2. Equations of motion The six degree of freedomdynamic equations for a rigidbody in the inertial framearewritten in accordance to the basic laws of translational and rotational motion (Mueller et al., 2004) mv̇=F Iω̇+ω× Iω=T (2.1) Themoment equations arewritten about the center of gravity (CG).For flight dynamic analysis, the airship is assumed to be a rigid body. The effects of the added mass and inertia are also taken into consideration. According to the geometry shown in Fig. 2, it is possible to transform 50 S. Amani et al. Equations (2.1) from CG to any other arbitrary point within the body. For airships, this point is the so called the center of buoyancy (CB), denoted by O. Fig. 2. Coordinate frames Since the inertial frame has no rotation, it is possible to write the relative velocities by vG =vO+vG/O vG =vO+ω×rG (2.2) Now it is possible to differentiate Eq. (2.2)2 and obtain the translational equation of motion in the body axis system m ( v̇O+ω×vO−rG× ω̇+ω× (ω×rG) ) =FO (2.3) A similar procedure leads to the rotational equation ofmotion (about point O) in the body axis system as well IOω̇+ω× (IOω)+mrG× (v̇O+ω×vO)=T (2.4) Combining equations (2.3) and (2.4), the coupled nonlinear equations will become [ mE −mr∗G mr∗G IO ][ v̇O ω̇ ] + [ m ( ω× v̇O+ω× (ω×rG) ) ω× (IOω)+mrG× (ω×vO) ] = [ FO T ] (2.5) where r∗G is the skew symmetric matrix made out of rG r ∗ G =    0 −z y z 0 −x −y x 0    rG =    x y z    (2.6) Finally, the six degree of freedomnonlinear equations ofmotion for a rigid bodymoving through a fluidmediumwith respect to the CB is given by [ mE+M′ −mr∗G mr∗G IO+ I ′ O ][ v̇O ω̇ ] + [ m ( ω× v̇O+ω× (ω×rG) ) ω× (IOω)+mrG× (ω×vO) ] = [ FO T ] (2.7) The aerodynamic and stability derivatives are a function of the airship geometry and flight conditions which are estimated using the Digital Datcom software released for public use (Dat- com+Pro version 3.0). It is important to note that the aforementioned motion equations are used to estimate the airship performance in terms of the distance traveled as one of the key objectives. The traveling distance in a finite time is among the objectives considered, which is directly related to the overall airship configuration sizing. Multi-objective conceptual design optimization of a domestic unmanned airship 51 3. Stability derivatives Stability derivatives play an important and critical role in the determination of static stability and performance characteristics of an airship. They show the sensitivity of aerodynamic force and moment coefficients with respect to flight, atmospheric and control parameters. In turn, the stability and control derivatives are functions of the configuration sizing parameters and the flight condition of the airship. For example, extremely large values of CXu will ruin the airship performance in attaining the desired translational velocity and, in return, the requirement of higher range or the distance to be travelled. However, a very small value of this derivative may not satisfy the desired static stability requirements. Naturally, the above statement necessitates the use of a trade-off study to reach a balanced value of the stability derivatives. This task is handled by the MOO procedure, a subject that will be addressed in Section 4. Table 1 summarizes some of the key longitudinal and lateral- directional stability derivatives for the airship. Table 1. Longitudinal and lateral-directional stability derivatives criteria u v w α β X < 0 ≈ 0 ≈ 0 ≈ 0 ≈ 0 ∗CXu < 0 Y ≈ 0 < 0 ≈ 0 ≈ 0 CYβ < 0 Z ≈ 0 ≈ 0 < 0 ∗CZα < 0 ≈ 0 L ≈ 0 ≈ 0 ≈ 0 ≈ 0 < 0 CLβ < 0 M ≈ 0 ≈ 0 ≈ 0 < 0 ≈ 0 ∗∗CMα < 0 N ≈ 0 ≈ 0 ≈ 0 ≈ 0 > 0 CNβ > 0 ∗ Thevalue of thesederivatives shouldbenegative for stabilitybecause it is assumedthat the aerodynamic coefficients have the same sign as the aerodynamic forces. ∗∗ The contribution of airship hull on this derivative is positive because the produced lift will produce a positive pitching moment. 4. Optimal configuration design In this section, the optimality criteria needed for theMOOare addressed. It is assumed that the airship consists of themain hull, tail mounted elevators, rudders and gondola. Clearly, the shape of the hull, the reference area and position of the stabilizers and gondola play an important role in the static stability and performance (distance traveled) of the airship. On the other hand, the configuration sizing and the heliummass determines the major cost of the construction. In order to have a better view of the optimization procedure, Fig. 3 is provided to illustrate the optimization loop and flowchart. In this paper, the following optimization objectives are considered. The aerodynamic drag coefficient is sought to be minimized. The hull surface area and, in turn, the helium mass are to be minimized in order to reduce the production cost. The glide distance traveled by the airship is to be optimized tomeet the customer stipulated requirement. Stability derivatives are to be optimized in order to make the airship resistant to initial disturbances, while keeping an acceptable performance (trade-off). Mathematically speaking, the above objectives need to bemodeled for optimization. 52 S. Amani et al. Fig. 3. Optimization flowchart For the aerodynamic drag, the drag force exerted on the airship is to beminimized (Mueller et al., 2004) D = 1 2 ρairV 2 0 ACD (4.1) For the airship hull surface area, the following objective is to beminimized (web-formulas) S =2π ( p √ 2a p 1b p+ b2p 3 + p √ 2a p 2b p+ b2p 3 ) (4.2) where pis a constant equal to 1.6075. The heliummass is obtained as follows mh = 2 3 ρHeπ(a1+a2)b 2 (4.3) In this equation, the glidedistance traveledwill be optimized according to the followingobjective function DT = √ x2final +y 2 final +z 2 final (4.4) where DT – distance traveled. Table 2.Boundaries of the design variables Configuration Boundaries Configuration Boundaries parameters parameters b [2.14,3.57]m lf3 [1,1.74]m a1 [5,7.14]m lgx [1.75,2.25]m a2 [120,12.85]m lgz [2.28,3.8]m lf1 [7,9]m Sf [9,27]m lf2 [7,78.10]m Sg [0.5,1.5]m The key stability derivatives and their desired signs as objectives to be met are tabulated in Table 1. The design variables are those defining the airship geometry which be led to their Multi-objective conceptual design optimization of a domestic unmanned airship 53 optimum set via a genetic algorithm. The design variables are illustrated in Fig. 1, while their boundaries are presented in Table 2. Considering the previous notations, deriving the airship geometrical parameters in amanner that optimizes our objectives (Eqs. (4.1)-(4.4)) simultaneously, there arises a need for theMOO that is further discussed in the next section. 5. Multi-objective optimization In many optimization problems, there exist several complementing objectives like minimizing the fuel and time. In such cases, it is possible to form an aggregate form of the objectives in one cost function, while allowing for weight coefficients to show the relative importance of each objective. This concept arises the meaning of single-objective optimization. However, in theMOOmethodology, objectives are usually conflicting andmust be optimized simultaneously without giving special attention to any objective. Consequently, a set of optimal solutions will be generated called thePareto front. Therefore, instead of one optimal solution (single-objective optimization), a set of optimal solutions can be usually generated (Censor, 1997). 6. Genetic algorithm Agenetic algorithm is a global search stochastic optimization strategy that resembles thenatural biological behavior and is based on the hypothesis that the nature gives more opportunity of survival to individuals who are more worthy. In other words, the main idea behind the genetic algorithm is to give more opportunities to better results (generations) in order to guide the solution toward its optimal set.Thegenetic algorithm is anelitist one.Regardless of its structure, this heuristic method is implemented via three operations denoted by reproduction, cross-over andmutation. Individuals are generated according to their objective values at the reproduction stage. The cross-over is a genetic operator to change the intrinsic value of one chromosome from one gene- ration to another. In other words, the cross-over reproduces the new individuals by combining random information in the mating pool. The mutation operator preserves the diversity of the algorithm in the search space. Because of the fact that inmost of the heuristicmethods, there is a possibility of getting caught in local minima, several schemes are proposed and used to escape these local minima. A genetic algorithm uses mutation for this purpose (Miller et al., 2010). 7. Presentation of results The simulation results are presented in this section. Subsection 7.1 includes all of the necessary information for the simulation of the previously described airship dynamics. Parameters of the MOOare presented in Subsection 7.2. In Subsection 7.3, drag based optimization that considers dragminimization and the relevant results are presented. Subsection 7.4 considers an analysis to show the effects of static stability derivatives on the performance characteristics of the airship. Subsection 7.5 considers the gliding distance traveled as a requirement for the optimization. Subsection 7.6 presents the results of the proposed optimal design approach for the airship configuration and its real world implementation. Section 8 draws appropriate conclusions and lists possible future related areas of study being considered by the authors. 7.1. Airship dynamic parameters Thesolver for nonlinear airshipdynamic simulation is basedonode3 (ShampineandReichelt, 1997)with integration time stepof one second.The simulation is performed for 100 second,where the initial conditions are given in Table 3. 54 S. Amani et al. Table 3. Initial conditions of the airship Initial Values Initial Values conditions conditions Body u0 0.9908m/s Body p0 0rad/s v0 −0.0042m/s q0 0rad/s w0 −0.0042m/s r0 0rad/s Inertial x0 97.17m Inertial φ0 0rad y0 0.2075m θ0 0rad z0 2500m ψ0 0rad 7.2. Multi-objective optimization parameters TheMulti-Objective Genetic Algorithm (MOGA) parameters are presented in Table 4. Table 4.Multi-objective genetic algorithm (MOGA) parameters MOGA parameters Values Population size 15 ·number of variables= 105 Number of generations 200 ·number of variables= 1400 Tournament size 2 Cross over fraction 0.8 Migration fraction 0.2 7.3. Drag based designs Undoubtedly, drag reduction is one of themost important objectives pursued inmanyairship preliminary designs. In this paper, the drag is considered to be an objective function (Eq. (4.1)) and itsminimization is sought.According toFig. 4,minimizing thedragcoefficient (and therefore the drag itself!) can lead to a larger hull surface area. Fig. 4. Hull drag coefficient versus hull surface area Accordingly, though minimization of the aerodynamic drag seems to be a good choice, its implementation requires a higher production cost due to the resulting larger hull surface area. Therefore, the authors have proposed a circular region within the Pareto, in which the design point should be selected for equally weighing the two objectives. Similar regions will also be introduced for the other Paretos as well, all of which would eventually be utilized to select the final best design point. Multi-objective conceptual design optimization of a domestic unmanned airship 55 7.4. Stability based designs Asdiscussed in Section 3, static stability derivatives are in competitionwith the performance indices.This section presents the results of a sensitivity analysis to show this competitive nature. In this regard, the effect of hull configuration, position and size of the stabilizers and gondola on two key stability derivatives CZα and CYβ are investigated, and the pertaining results are shown in Fig. 5. Fig. 5. CZα vs. CYβ for different configurations According to Fig. 5, there are several configurations that lead to unstable airship behavior in terms of CZα and CYβ derivatives. In order to demonstrate their effect, two design points (1 and 2) are chosen to represent the airship dynamics with the initial conditions specified in Table 3. The corresponding six degree of freedom nonlinear airship response is presented in Fig. 6. Assuming the initial trim condition, Fig. 6 indicates that design point 1 leads to a more stable behavior that is desirable from flying quality point of view. Table 5 is presented to show related configuration sizing parameters for these two points of the Pareto front. According to this table, albeit using point 1 as the design choice brings about several advantages (like more stable behavior, less helium mass and ...), but it has a drawback due to additional expenses required for larger fin areas. Table 5.Comparison of configuration parameters for design points 1 and 2 from Fig. 4 Airship parameter Design point 1 Design point 2 Hull length to diameter ratio 3.87 2.19 Hull surface area [m2] 203 299 Helium mass [kg] 26.53 58 Fin area [m2] 24 10.8 lf1 [m] 8.88 7.51 lf2 [m] 9.80 8.29 lf3 [m] 1.11 1.74 lgx [m] 2.20 1.86 lgz [m] 2.43 3.80 These considerations compact the Pareto to a useful encircled design area shown in Fig. 5. This circular region shows the best design points that satisfy the overall requirements. 56 S. Amani et al. Fig. 6. Six degree of freedom system response for the results shown in Fig. 4 7.5. Distance traveled based designs Figure 7 illustrates theminimumandmaximumdistances traveled by the airship for various optimized configurations in three dimensional space during 100 seconds of simulation. As shown in this figure, thedesigner can select an optimumconfiguration according to glide ranges covered. For example, design point 1 leads the airship to travel to a totally different location as compared to design point 2. Therefore, the designer can select design point 1 to travel farther. Table 6 presents the difference in configuration sizing parameters of the two design points for the Pareto front shown in Fig. 6. Table 6.Comparison of configuration parameters for design points 1 and 2 from Fig. 7 Airship parameter Design point 1 Design point 2 Hull length to diameter ratio 3.97 3.54 Hull surface area [m2] 186.6 166.81 Helium mass [kg] 23.09 20.45 Fin area [m2] 24 21.36 lf1 [m] 8.49 7.20 lf2 [m] 9.38 7.94 lf3 [m] 1.05 1.04 lgx [m] 2.11 1.78 lgz [m] 2.30 2.29 Multi-objective conceptual design optimization of a domestic unmanned airship 57 Anotable consequence of the presented table and its related Fig. 6 is that for the airship to glide farther, a larger heliummass is required.This is due to the fact that according toNewton’s second law of motion, increasing the mass will decrease the negative acceleration produced by the drag force and which, in turn, causes the airship to move farther in the inertial space. Again given the above mentioned facts, one can choose a spherical optimum design region shown in Fig. 7. This region satisfies various objectives (like acceptable helium mass) while meeting the required gliding range covered in the inertial space. Fig. 7. Final gliding position of the airship in inertial coordinates A similar approach is taken to show the rotational movement of the airship in terms of the terminal Euler angles. Pertaining results are depicted in Fig. 8. Table 7 demonstrates the configuration parameters corresponding to the design points 1 and 2 from Fig. 8. Fig. 8. Final attitude orientation for various configurations Table 7.Comparison of configuration parameters for design points 1 and 2 from Fig. 8 Airship parameter Design point 1 Design point 2 Hull length to diameter ratio 3.86 2.17 Hull surface area [m2] 260.53 285.92 Helium mass [kg] 38.58 54.71 Fin area [m2] 10.73 15.18 lf1 [m] 9.03 7.18 lf2 [m] 9.96 7.23 lf3 [m] 1.25 1.71 lgx [m] 2.24 1.78 lgz [m] 2.75 3.74 58 S. Amani et al. Similar to the previous analysis, a spherical optimal design area can be considered that satisfies the needs for reasonable rotational airship maneuver while observing other objectives such as the hull surface area and the heliummass. Taking all of the above into account, an airship configuration is proposed that lies within the realms of the spherical as well as circular design regions depicted in the Pareto plots. This configuration will be optimal form all aspects considered, namely aerodynamic drag, stability requirement (derivative value or sign), helium mass and hull surface area (production cost) and the gliding distance traveled . It means that the designer can be assured of the selected configuration to show the overall reasonable behavior. The proposed configuration is shown in Fig. 9, next to the configuration which is not Pareto optimal. Fig. 9. Two configurations. Design point 1 (not optimal), design point 2 (optimal) 7.6. Implementation of the described design procedure Figure 10 shows a schematic of the constructed airship called NAMA. This airship was built for practical research onLTAbuoyancy at 2500meter altitude andwas towithstandwind speeds of up to 13 meters per second. Fig. 10. NAMA, first prototype of the constructed domestic airship The system is a non-rigid dirigible whose body shape is achieved through preserving a dif- ferential pressure (0.05 to 0.2 atmospheres) with the ship outside environment. Obviously, this pressure differential is a function of system endurance, flight duration, ambient temperature and flight altitude. In addition, in order towithstand the high pressure on the vehicle body and skin, advancedmaterials (Vectran) were usedwith an ultimate stress capability of up to 3 gigapascal. Vectran is extremely endurable and tolerant in harsh environmental conditions such as UV and, at the same time, is lightweight (about 80 gram per meter squared) which makes it a quite appropriate for the airship body skin. Multi-objective conceptual design optimization of a domestic unmanned airship 59 Composite materials forming the inside layer of the body are produced out of thermo pla- stic polyurethane (TPU) which also has a good tolerance to UV and, additionally, exhibits low permeability against helium leakage. To overcome internal pressure on the body and stitch loca- tions, and also in order to have clamping positions for payload hanging, several circumferential belts were utilized. These belts are spaced out 1 meter apart through the entire length of the body, and each weighs around 1800 kilograms. 8. Concluding remarks Multi objective optimization for the configuration design of a domestic airship is performed considering static stability and flight performance merits. The role of other objectives such as the aerodynamic drag, the airship hull surface as well as the helium mass is also investigated. The latter two objectives directly affect the production cost of the airship.The proposed scheme can produce configurations with different levels of stability. Application of MOO enables the designer to select more efficient designs through consideration of multiple criteria without the need for additional trade off studies. Implementation of the proposedmethod has been utilized for the design and construction of a prototype domestic airship. Despite the fact that the presented paper has covered a large and useful class of objective functions for design optimization, other criteria such as the aerodynamic lift and the airship material could also be considered as complementary objectives. In addition, implementation of active controls through vectored thrusters and their optimal placement for station keeping and trajectorymanagement are of high importance. The above issues aswell as the flight test results of the first prototype are being utilized by the authors for the future work, results of which will help the improved second prototyping. References 1. AnW., LiW.,Wang H., 2007.Multi objective optimization design of envelop shape of a certain airship with deviation considered, Journal of Northwestern Polytechnical University, 25, 1, 56-60 2. Censor Y., 1977, Pareto optimality in multiobjective problems, Applied Mathematics and Opti- mization, 4, 41-59 3. Datcom+Pro version 3.0, http://www.holycows.net/datcom/ 4. Lutz T., Wagner S., 1998, Drag reduction and shape optimization of airship bodies, Journal of Aircraft, 35, 3, 345-351 5. Miller F.P., Vandame A.F., McBrewster J., 2010, Genetic Algorithm, VDM Verlag Dr. Mueller E.K., ISBN: 6130211791, 9786130211790 6. Mueller J.B.,PaluszekM.A., ZhaoY., 2004,Development of an aerodynamicmodel andcon- trol lawdesign for a high altitude airship,AIAA3rd “UnmannedUnlimited” Technical Conference, Workshop and Exhibit 20-23 September 2004, Chicago, Illinois 7. Nejati V., Matsuuchi K., 2003, Aerodynamic design and genetic algorithms for optimization of airship bodies, JSME International Journal, Series B: Fluids and Thermal Engineering, 46, 4, 610-617, doi:10.1299/jsmeb.46.610 8. Shampine L.F., Reichelt M.W., 1997. TheMATLAB ODE Suite, SIAM Journal on Scientific Computing, 18, 1, 1-22 9. Wang H., Song B., Zhong X., 2011, Configuration design and sizing optimization of a winged airship, International Conference on Network Computing and Information Security, 2, 41-45 60 S. Amani et al. 10. Wang Q.-B., Chen J.-A., Fu G.-Y., Duan D.-P., 2009. An approach for shape optimiza- tion of stratosphere airships based on multi disciplinary design optimization, Journal of Zhejiang University, Science A, 10, 11, 1609-1616 11. Wang X.L., Shan X.X., 2006, Shape optimization of stratosphere airship, Journal of Aircraft, 43, 1, 283-287 12. www.web-formulas.com/Math-Formulas/Geometry Surface of Ellipsoid.aspx Manuscript received February 4, 2013; accepted for print May 17, 2013