Vous êtes sur la page 1sur 13

52nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference<BR> 19th 4 - 7 April 2011, Denver, Colorado

AIAA 2011-1968

52nd Structures, Structural Dynamics, and Materials Conference, April 47, 2011, Denver, CO

Inated-Wall Members and Guidelines for Cross Section Design


Gyula Greschik
TentGuild Engineering Co., Boulder, CO 80303

In order to improve the strength of inated members, the structuring of the member walls is considered. Put forth is the option of a cellular member wall which consists of compartments separated by partitions that comprise a hierarchical load bearing structure. It is the compartments of this wall, rather than the enclosure within (around the member centerline), that are pressurized when deployed. Straight inated-wall members with prismatic compartment geometries of stepwise rotational symmetry are examined. In particular, patterns of interlocking inner- and outer rib compartments are studied. Via the principles of maximum pressurized volume and equilibrium, some characteristics of the deployed shapes are studied, as well as the need for, and a procedure to ensure, cross section stability are established. The design procedure presented is developed semi-empirically: via a derivation that relies on an observation made during numerical studies. It is also shown that member strength is greatly improved over the capacity of pressurized members with smooth walls. Technological mass penalties (such as rigidization overheads) are not considered, nor are fabrication challenges addressed.

Nomenclature
E h l n R r t 0 1, 2 cr e Youngs modulus. Height above base line. Length in the cross section context. Rib number (number of inner or outer ribs). Radius; cross section radius; radial position. Local radius of curvature. Film thickness; member wall thickness. Reference to the web. Reference to the inner and outer ribs. Reference to limit (critical) state. Reference to equivalent monocoque tube. Unit (fundamental) cell central angle, = n . Rib arc central angle. Auxiliary variables to simplify expressions. Poissons ratio. Angle between web and cell border-line. Stress.

Subscripts

Symbols

Design

Engineer, Senior Member AIAA. Copyright c 2011 by Greschik. Published by the American Institute of Aeronautics and Astronautics, Inc. with permission.

1 of 13 American Institute of Astronautics, Aeronautics and Copyright 2011 by Greschik. Published by the American Institute of Aeronautics and Inc.,Astronautics with permission.

I. Introduction
ROAD ideas for, as well as detailed specics of, inatable (and inatable-rigidizable R/I) structures in space have attracted interest since the dawn of the space age. In the literature accumulated over the decades, low weight is repeatedly mentioned as a characteristic feature of space inatable technology. In fact, along with low stowage volume, low mass may be the most frequently claimed advantage of pressurized structural concepts. Already in his 1964 review of deployable (expandable) structures and applications, Forbes mentions lightness as the rst advantage when summarizing the characteristics of pressurized solutions a feature he doesnt even list for mechanically deployable (variable geometry) options [1][Figs. 2025]. The claim that the low weight and packaged volume of inatables ... has long been known opens Thomas and Frieses paper [2] on pressurized space antennas in 1980. Freeland, et. al., begin their 1998 inatable space technology review [3] by highlighting low weight (along with low cost, packaging efciency, and deployment reliability) for this technology. Darooka and Jensen in 2001, in the rst paragraph of their structures concept review paper [4], also claim that overall mass can be reduced with inatable rigidized solutions. Yet another example is the 2004 inatable deployment test report [5] by Campbell, et. al., in the Introduction of which the potential for weight minimization via (unique solutions of) inated structures is mentioned. A recent example is the statement by Cobb, et. al., who, when introducing their description of a space experiment [6], state that inatable structure concepts are a low mass alternative to conventional hardware. Comments like those just cited, while not pervasive (e.g., see Ref. [7] for a work where no mass advantage is mentioned), occur quite often. These statements are trivially true for applications to which alternative technologies are hardly adaptable. (Examples include bulky but topologically simple enclosures such as habitat modules [8, 9] as well as large near-perfect spherical objects for calibration [10], passive communication [11], etc.) However, for applications with meaningful technological tradeoffs (most subsystem-level components: booms, struts, trusses, antenna- and other device structures, etc.), this narrative can be misleading if specications, performance, and various overheads are not discussed. In particular, the impression of superior performance is given because, if little attention is paid to mass overheads and stiffness and strength, the last issues appear to be deemed insignicant. It is implied that inatable technology can generally match with lower mass the structural performance of alternative solutions. This message may raise unrealistic expectations for the technology if application in a structurally critical role is considered. (Concurrently, the spotlight is stolen from a truly pivotal advantage of inatable structures: the unparalleled exibility of stowage design coupled with the simplicity and robustness of deployment.) Contrary to what is suggested by some broad comments on mass advantages, structural performance (inversely, the mass needed to achieve specied strength, stiffness, precision) is more often a liability than an asset for inatedrigidized structures. This is easily seen for R/I columns (masts, booms, beams), the performance of which directly suffers from two causes. One, R/I materials are poor compared to those fabricated rigid in controlled conditions: stiffness moduli (on which column strength also depends) of even advanced fabrics, rigidized after deployment, are worse and less uniform than traditional materials. Two, to be inatable the column must be tubular: a shape less than ideal for performance [12]. (More efcient, skeletal, boom structures such as lattices or isogrids could only be used if stretched within [13] or wrapped on [14, 15] an inated tube, reducing the latter to non-structural overhead.) The present work addresses the strength deciency of inatable columns via the revision of the prevailing R/I member conguration. Namely, the traditionally smooth and uniform tube wall, Fig. 1, is replaced with a cellularcorrugated structure, Fig. 2. During deployment, the set of cavities within the wall, rather than the tube interior, is pressurized. The resulting set of interconnected wall segments and partitions together resist local buckling better than a standard wall of the same mass would. Thus member strength, which is the result of the interplay of local and global stability phenomena [16], can be signicantly increased. Attention in this paper is directed to a simple class of wall corrugation patterns. Namely, prismatic compartment geometries of interlocking inner- and outer rib compartments with stepwise rotational symmetry are examined. (Such a structure is exemplied in Fig. 2.) Design guidelines and procedures are established, and the achievable strength improvement is estimated.

2 of 13 American Institute of Aeronautics and Astronautics

cross section tm pressurized: tube interior R Rm

cross section tM pressurized: wall chambers

load bearing structure: solid smooth cylindrical wall


Figure 1: Customary inated member.

load bearing structure: "tubelet ribs" of structured wall


Figure 2: Inating the member wall.

II. Novelty and Signicance

HE idea to deploy a surface structure (shell, plate, wall, etc.) via the pressurization of its structured wall is not new. Air mattresses and -beds are perhaps the most trivial examples, but there are also several pieces of inatable architecture with the walls, rather than the interior space, pressurized. Most common of the last are inatable sport installations and tents such as those shown in Fig. 3.

Figure 3: Inatable tennis court and tent. (Sources: www.surebeatswork.com, www.tradevv.com.) Space concept examples include tension trusses or drop lines internal to inated envelopes such as the airmat [1] or the ISIS idea [13], as well as actual compartmentalization [17]. A recent example of the last is the Inatable Reentry Vehicle Experiment (IRVE [18]) launched on August 20, 2009, with a funnel-shaped ballute body of (at some locations structurally, elsewhere structurally and pneumatically separated) ring compartments. Airborne applications where an envelope is pressurized through its internal structure can also be found. Blimp appendages with interior tension-reinforcement are one common example. Less common but also well known are pressure-deployable airfoils such as inatable wings [19] with compartments within. However, simple structural members such as struts or beams with their walls, rather than the interior space, pressurized are unknown to the writer. While there exists a structural concept with a longitudinal-cellular a tubular members wall, this solution is achieved via poltrusion, not ination. In fact, this innovation (which is referred to as an articial stem, technische Panzenhalme in German) by the Institute for Textile Technology and Process Engineering Denkendorf (ITV Institut fr Textil- und Verfahrenstechnik der Deutschen Institute fr Textil- und Faserforschung Denkendorf) has been put forth [20, 21] as a biomimetic (biologically inspired) achievement, Fig. 4. According to the inated-wall paradigm examined in the present work, a wall structure somewhat similar to the articial stem is achieved via the pressurization of an otherwise collapsible fabric or membrane (composite) shroud.
3 of 13 American Institute of Aeronautics and Astronautics

(a)

(b)

(c)

Figure 4: Cross section of scouring rush and articial stems by the Inst. for Textile Techn. and Process Eng., Germany. (Images from Ref. [21] and the German Fabric Research Foundation web site www.textilforschung.de.) Thus a pivotal advantage this concept offers over the articial stem is deployability: the key to large space applications. The other structural advantages of wall-ination over the articial stem higher stiffness, precision, larger dimensions, and fundamentally different fabrication technologies set the two concepts even further apart. In fact, the two are entirely different except that both improve member strength by increasing the level of structural hierarchy downward, by topological renement. (By better trading various modes of failure against the amount of material used, advanced structural hierarchies can generally increase performance [12, 22].) However, the concept of such a hierarchical enrichment for an inatable tube wall hasnt been explored before. Accordingly, the design for such a member calls for the consideration of new trades, and for new means of optimization. These issues are explored in the remainder of this paper.

III. Basic Considerations for Cross Section Design


tube wall can be made inatable with many kinds of internal structuring. Attention here is restricted to prismatic solutions (where the member cross section is uniform). Some geometries are shown in Fig. 5. The simplest topology is the basic stacked pattern shown in Fig. 5 (a) which echoes the air mattress idea called a dual-wall (a) (b) structure by Bair [17]. (If the chambers are circular, this cross section reduces to a ring of circles the member becomes a tube of tubes.) In the slightly more complex conguration of Fig. 5 (b) simplysimply the chambers are wedge shaped and are placed in an alternatingstacked interinterlocking pattern. The circular continuity of the zig-zagging parconfig. locked titions in this arrangement achieves a healthier structural integration between the rings of bulging exterior walls: those facing toward the member centerline and outward. This should increase performance (c) (d) and robustness (e.g., stiffen the counter-rotational vibration modes of the inner and outer walls). If, in the latter design, the pressures in the inward- and outward-facing compartments differ, the partition curved web walls curve, Fig. 5 (c), increasing their buckling strength. The cost of webs scaffolds this performance improvement is the need to maintain (at least durp1 ing rigidization, if applicable) different pressures in the two sets of p1 > p2 compartments. Geometrically, even the hierarchy could be further increased with more delicate partition patterns, cf. Fig. 5 (d). In reference to the skeletal microstructures called bone tissue scaffolding Figure 5: Some cross section options. in anatomy, the name web scaffolds could be used for this last option. This paper focuses on the simple interlocking compartment pattern of Fig. 5 (b) as this conguration is deemed to best balance the potential to improve tube strength against design complexity. The practical consequences of

4 of 13 American Institute of Aeronautics and Astronautics

complexity (e.g., technical challenges of fabrication, pressurization, and rigidization if applicable) are beyond the scope of this work. A. Simply Interlocking Chambers

The tube wall architecture herein considered consists of interlocking wedge-shaped inner- and outer rib unit cell with line lengths: outer rib compartments, Fig. 6. The partition walls separating these chambers are herein called l2 webs, and the chamber wall sections exterior to l0 l1 the inated wall, bulging from the internal pressure, the ribs. The latter are called inner or outer possible definitions of inner rib depending on whether they face the tube interior cell of rotational web or exterior. Equal pressure is assumed in all comsymmetry: partments, rendering the webs straight. The unit cell of general symmetry which characterizes the entire cross section geometry conchambers sists of the adjacent halves of an inner and an (rib comouter rib with the web between, delimited by two partments) cross section radii. Denote the angle between these radii as , as shown in Fig. 6 (upper right Figure 6: Nomenclature and cells of symmetry. corner). (However, in the strict context of [stepwise] rotational symmetry the fundamental cell is twice the above: it includes a full inner and outer rib and its denition is not unique, cf. the lower right part of Fig. 6.) If the rib number (the number of inner or outer ribs) is n then

/n

(1)

The rib number n (equivalently, the central wedge angle ) along with the lengths of the inner and outer half-ribs and of the web, l1 , l2 , and l0 as shown in Fig. 6, fully dene any symmetric hardware design. (Fabrication procedures can be developed from the geometric denition these four parameters provide.) B. The Shape of the Cell of Symmetry

Given a particular set of hardware geometry parameters (n, l1 , l2 , and l0 as described in Section A), the shape of the unit cell in the context of n-step rotational symmetry depends on the location of the ribs-web assembly within the principal wedge. As alluded to in Figs. 7 (a) and (c) with gray arrows, this location can be interpreted as how (to what extent) the walls slide inward, toward the member centerline, or outward, away from the latter. (The geometry of Fig. 7 (b) is shown in Figs. 7 (a) and (c) with dot lines.) Geometric variables are also shown in order to avoid clutter, only in one of the three sub-gures even if generally applicable. As indicated in Fig. 7 (a), each rib wall contour length li (with i being 1 for the inner, 2 for the outer rib) is divided into a free, bulging, part ai and a contact section wi which presses against the similar section of the adjacent cell. li = ai + wi i = 1, 2 (2)

(a) Rw1 w 1 a1, h1' (b)

h1 h2 Rw2 1 1 r1

h2' w2

a2 e l2 f 2 l0e 1 r2 2

R1 l1 R2 (c)

l0

l0f

Figure 7: Unit cell position and variables.

Obviously, wi = 0 means no contact, cf. Figs. 7 (b) and (c). The heights of the web endpoints and of the rib contact line endpoints hi and hi over the wedge border-lines e and f are also shown. The distances from the cross section center to the ends of the rib contact regions are denoted with

5 of 13 American Institute of Aeronautics and Astronautics

Rw1 and Rw2 . As indicated in Fig. 7 (b), the inner and outer web endpoints are at R0 and R1 distances from the cross section center, and the web subtends acute angles 0 and 1 at these locations with e and f . The freely bulging rib sections ai are circular with central angles i and radii ri , Fig. 7 (c). These arcs subtend i angles with e and f . 1. Wedge Solution Approach From the assumption of membrane mechanics two straightforward approaches arise for the solution of the wedge geometry. One is based on the variational concept of maximum pressurized volume, the other is according to the Newtonian convention of formally establishing equilibrium. However, neither approach leads to convenient design formulas because of the complexities involved. Symbolically coupling the location and orientation of the web line in the wedge, by itself already leads to formulas with little aesthetic appeal. These are further complicated by the conditional appearance at each web end of the contact region w, Fig. 7 (a). As the equations of rib geometry with and without contact are both complicated, this ambiguity constitutes a serious hurdle if general equations are sought. Finally, note that even if the geometric parameter expressions were derived, the subsequent formulation of either the maximum pressurized volume, or the conditions of equilibrium, would add yet another layer of complexity. Manageable symbolic relations have been successfully developed only for two special details. One, a basic and intuitive rule of thumb for rib shapes has been formally proven and, two, a geometric condition for maximum volume has been established for a degenerate case. These are described below in Sections III.B.2 and III.B.3. Once the search for symbolic solutions was exhausted in the present work, wedge solution, volume calculation, and equilibrium control have been performed numerically. Step-wise equations have been derived and programmed, the ones solved earlier providing input to the next, as listed in the following. In the context of a specied wedge angle , the ribs-web assembly moves in the unit cell (wedge) with one degree of freedom (dof) kinematics; its congurations can be dened with a single variable. This variable has been herein chosen as 2 (the angle between the web and f at the outer web end). From this then 1 is obtained via

= 2 +

(3)

The web endpoint heights hi , radii Ri , and web projection lengths l0 f and l0e (Fig. 7 (c)) on e and f are subsequently h1 l0 f R1 = = = l0 sin 2 l0 cos 2 h1 / sin (4) (5) (6) h2 l0e R2 = l0 sin 1 = l0 cos 1 = h2 / sin (7) (8) (9)

The remaining parameters are then calculated, after the introduction of the i and wi auxiliary variables, as follows:

1 w1
w1 a1 R w1 h1 1 r1 1

= ( /2 ) tan = (l1 1 R1 )/(1 1) = 0 w1 if w1 < 0 otherwise

(10) (11) (12) (13) (14) (15) (16) (17) (18)

2 w2
w1 a2 R w2 h2 2 r2 2

= ( /2 + ) tan = (l2 2 R2 )/(1 + 2) = 0 w2 if w2 < 0 otherwise

(19) (20) (21) (22) (23) (24) (25) (26) (27)

= l1 w1 = R1 w1 = Rw1 sin = (a1 sin 1 ) / h1 = a1 / 1 = /2 1

= l2 w2 = R2 + w2 = Rw2 sin = (a2 sin 2 ) / h2 = a2 / 2 = /2 2 +

of which Eqs. 16 and 25 are iteratively solved. Finally, the extreme member wall radii are obtained as R min R max = R2 l0e w1 cos r1 (1 cos 1 ) = R1 + l0 f + w2 cos + r2 (1 cos 2 )
6 of 13 American Institute of Aeronautics and Astronautics

(28) (29)

Expressions for the inated volume have been subsequently derived, and the conguration was solved via maximizing this metric with numerical optimization. Primary control for the results was provided via equilibrium conditions. Moreover, for an additional level of control, the solution has been independently programmed and executed both with MicroSoft Excel and a custom C program. 2. Rib Shapes The condition of maximum volume in the planar context of an (a) 2 cross section wedge leads to two simple rules for the rib shapes: r2 h r1 The freely bulging section of the rib (the section free of conl0 a1=l1 tact with the adjacent cell) has a circular arc contour. a2=l2 1 If there is an active contact region w > 0 (part of the rib is pressed against the other in the adjacent cell), then the cir(b) w1 cular arc contour of the freely bulging rib section osculates a1 r2 2 the line of contact at their shared endpoint. (Cf. Figs. 7 (a) and 8 (b), the latter showing the special case of a degenerate r1 a2=l2 1 =/2 l0 geometry.) While these rules of thumb postulate elementary conditions of Figure 8: The degenerate case of = 0. membrane mechanics, their explicit acknowledgment (and, in case of the second rule, formal proof) in the present work formed the initial premises on which the geometric derivations have been developed. 3. Special Case: Parallel Wedge Boundaries If the wedge angle diminishes ( 0) the rib number approaches innity (n ) then the compartment pattern becomes innitesimally ne. A cell of balanced proportions for this case degenerates into one with parallel wedge borders, Fig. 8, with 1 = 2 = and the web endpoint radial positions Ri become immaterial (they increase beyond all bounds with respect to the cell dimensions). Within this special scenario, the equilibrium of the simple case with no contact regions shown in Fig. 8 (a) can be formalized with the condition l0 cos 2 cos = l1 cos 2 cos 1 + l2 2 2 (30)

which is still implicit as it also involves i in addition to the section lengths li . Casting Eq. 30 in the sole context of the li lengths is possible only via complicated transcendent relations, limiting the design utility of this relation. Moreover, the onset of a contact region, Fig. 8 (b), already destroys the simplicity even of the implicit form Eq. 30, further reducing utility. Apparently, even for the degenerate case of parallel wedge boundaries, a symbolic solution has severe limitations. This illustrates the need for numeric approach. 4. Shape Solutions for Given Rib Numbers In the process of exploring characteristic cell geometries, cell shapes have been determined for various rib numbers n and wall section contour length li . The results of this exercise are illustrated via examples in Fig. 9.

(a) n = 18 l1/0/2 = 40/40/40 mm tM /Rmax = 41 %

(b) n = 18 l1/0/2 = 36.45/40/54.31 mm tM /Rmax = 41 %

(c) n = 36 l1/0/2 = 30/30/30 mm tM /Rmax = 31 %

(d) n = 60 l1/0/2 = 20/30/25 mm tM /Rmax = 14 %

Figure 9: Cell shapes (shown at a different scale each) for some rib number and contour section lengths, as indicated. The cells are arranged according to the rib numbers n. Examples Figs. 9 (a) and (b) correspond to the same n = 18 (i.e., = 10o ) they differ in the contour section lengths only. The latter are uniform in Fig. 9 (a), l1 = l0 = l2 = 40 mm,
7 of 13 American Institute of Aeronautics and Astronautics

but not so in Fig. 9 (b) where l1 = 36.45, l0 = 40, and l2 = 54.31 mm. The former case results in an active contact region on the inner rib but no contact outside. The non-uniform values shown are set to approach the contact limit state on both sides: the wedge borders osculate the rib arcs with innitesimal contact lengths. Uniform lengths are used in Fig. 9 (c) with n = 36, twice the number in the preceding examples. Accordingly, the maximum-volume solution features a contact region on the inner rib shorter, than before, with respect to the other lengths. There is no contact on the outer rib. The last conguration, Fig. 9 (d), features the highest rib number, n = 36, and a web contour length greater than the ribs. Consequently, the ribs are not in contact with those of adjacent cells. Cell designs with concurrently active inner and outer rib contacts can also be achieved with webs sufciently short in comparison to the rib contours. Such congurations, however, have little practical relevance. Also indicated in each gure legend is the ratio of the gross wall thickness, tM = Rmax Rmin (cf. Fig. 2), to the maximum radius Rmax . The values range from 41 to 14%, highlighting that the wall depths occupy signicant portions of the cross section radii. A seemingly attractive feature of congurations with uniform section lengths l1 = l0 = l2 is that the ribs and l l the web can smoothly collapse, Fig. 10. Such creaseand fold-free attening may be convenient during fabrication and storage. However, this paper stops short of recommending this or any other conguration, for two reasons. First, attention is herein limited to conceptual exploration. Acknowledging the challenges of fabricaFigure 10: Flattening of collapsed wall if l1 = l0 = l2 . tion, stowage, and deployment, care is exercised to imply no judgment on these issues. The second reason why attenable walls are not presented as desirable is that, in the framework of the general design recipe discussed in Section IV.A, they dont guarantee cross section stability.

IV. Symmetry and Stability


HE results discussed so far derive from the assumption of symmetry: the kinematics considered were strictly conned to specic wedge geometries directly dened by the rib number. However, symmetry for a pressurized conguration shouldnt be assumed a priori, even if the hardware itself is symmetric (cf. asymmetric pumpkin balloon congurations [23, 24]). If the maximum inated volume in the global conguration space is outside the subspace of symmetry, the cross section will assume that asymmetric shape whenever the opportunity arises. For example, the cross section may ovalize (collapse) unidirectionally as alluded to in Fig. 11, or it may take some other noncircular shape, if pressurized volume is gained with the transition. Quantitative insight into such phenomena could be gained Figure 11: The degenerate case of = 0. only with a model that captures the cross section in its entirety. Moreover, for an even more faithful representation of practical reality, the spatial interaction between member walls and end constraints would also be desirable to model. However, there exists an approach to identify robust cross sections without the analysis of complex interactions: cross sections immune to pressurization instabilities can be designed relying on the analysis framework of a single wedge, discussed above, as opposed to more complex models. This design procedure, used in the present work, is described next.

A.

Stability and Unconstrained Cell Shape

The procedure herein given for the design of stable cross sections is based on one basic observation. Namely, a cross section is bound to be stable if all of its cells take the shape that maximizes their own individual volumes. (Clearly, if this condition is satised, then no wall deformation can increase total pressurized volume.) The premise just stated is a sufcient condition: if satised, it guarantees stability. However, it is not necessarily a necessary condition also: it says nothing regarding the existence of stable cross sections for which not all cell volumes

8 of 13 American Institute of Aeronautics and Astronautics

are individually optimized. This simply sufcient condition, nevertheless, ts the needs of elementary design which merely aims at achieving stability, without requiring that no other stable conditions exist. The key to the optimization of a cell unconstrained by the wedge angle is that the latter be included in the control variables, rather than treated as a xed constant. Thus the pressurized volume calculated from the geometry dened by Eqs. 3 through 29 must be maximized in terms of both 2 and . The results of this optimization, performed numerically, for the sets of contour section lengths in Fig. 9 are shown in Fig. 12.

(a) n = l1/0/2 = 40/40/40 mm

(b) n = 18 l1/0/2 = 36.45/40/54.31 mm

(c) n = l1/0/2 = 30/30/30 mm

(d) n = 32.96 l1/0/2 = 20/30/25 mm

Figure 12: Unconstrained shapes (shown at a different scale each) for the cells in Fig. 9. (An innite rib number, n = , means that the maximum-volume conguration has a zero wedge angle, = 0o .) Note a few characteristics of the new, unconstrained, shapes. First, the conguration in Fig. 12 (b) doesnt differ from that in Fig. 9 (b) quantitative details, not presented here, reveal identity to numerical precision. Apparently, the design in which the contour lengths were tuned to achieve osculation between border lines and rib arcs was the absolute minimum energy conguration, despite that it had been developed for a pre-determined wedge angle. Second, the unconstrained states for the cells with uniform contour lengths, Figs. 9 (a) and (c), possess the degenerate geometry of parallel wedge borders (cf. Section III.B.3). Furthermore, the two results differ only in scale: their shapes are identical. The third noteworthy observation pertains the fourth, generic, design, Fig. 9 (d). The unconstrained shape of this cell is a wedge, Fig. 12 (d), with a central angle 5.46o different from the initial ini = 180o/60 3o value. The new rib number is thus n = 180o/ free =32.96, not an integer, revealing that a cross section (a full circle) couldnt be assembled by repetitively applying this unit. If a member were fabricated with a cross section of cells in Figs. 9 (a), (c), or (d) despite the nonconform unconstrained shapes involved, the result may not be stable. Under pressure, each cell would try to better approach its unconstrained shape of absolute maximum volume, undermining symmetry. On the other hand, a section design in which each cell assumes its unconstrained conguration, such as Fig. 9 (b), is unconditionally stable. B. An Empirical Observation

A last, very signicant detail of the freely optimized cell shapes shown in Fig. 12 is that, for each, the inner and outer rib arcs subtend the same angles with the wedge borders. Quantitatively,

1 1 1 1
In fact, the equality

= 2 5.31o = 2 0.00 = 2 5.31o


o

in Fig. 12 (a) in Fig. 12 (b) in Fig. 12 (c) in Fig. 12 (d)

(31) (32) (33) (34)

= 2 19.96o

1 = 2

(35)

characterized each and every unconstrained cell shape examined during research, regardless of contour lengths, wedge angles and cell shapes, and whether the unconstrained shape agreed with the constrained one or not. This observation, therefore, has been empirically accepted as a necessary condition for cell stability. Accordingly, the decision to add relation Eq. 35 to the design equations formally proven has been made. The set of equations so extended enabled the derivation of a practical design procedure without sophisticated cell shape optimization, as described next.

9 of 13 American Institute of Aeronautics and Astronautics

C.

Semi-Empirical Design

In experimental research and design procedures that rely on it, it is common to tailor formulation to match observations. Coefcients and functional forms are imported in frameworks established with rigorous theory, if reality is approximated with acceptable accuracy. While this practice is unusual in conceptual-theoretical research, it has been nevertheless followed in the present work for convenience. As the observation relied upon, Eq. 35, is numerically established with computational accuracy, the precision of the developed model (and of the consequent results) will not suffer. Further, the validity of Eq. 35 as a condition for wall cell volume maxima has in fact been veried by the results developed from it. The need for a theoretical proof for Eq. 35 is herein deemed aesthetic and left for later work. The solution of wall cell geometry for contour lengths li and a wedge angle , as a function of 2 , has been outlined in Eqs. 3 through 29. With the following procedure, the input to this solution can be determined to ensure cross section stability (maximum volume for each cell). Namely, a recipe to obtain l0 and l2 (web- and outer rib contour lengths) for a given inner rib contour l1 and arc-to-wedge border angle 1 = 2 is given, within the context of a wedge angle . The step-wise equations, applicable to congurations with no inter-cell contact (w1 = w2 = 0), are as follows:

1 r1 h1 2

= 1 = l1 /1 = r1 sin 1 = sin(1 +2 )/sin 1 cos(2 ) = 1/2 tan1 (sin (2 ) /)

(36) (37) (38) (39) (40)

2 l0 h2 r2 l2

= 1 + 2 = h1 / sin 2 = l0 sin 1 = h2 / sin 2 = r2 2

(41) (42) (43) (44) (45)

These expressions have been derived from simple geometric relations combined with the force-balance equilibrium conditions of the web endpoints in the associated border-line directions e and f (cf. Fig. 7), and from the equality Eq. 35. Their output includes l0 and l2 which complete the input to Eqs. 3 through 29, of which only the ones not yet covered by Eqs. 36 through 45 need to be solved in order to complete the cell design. The cell for an example design is shown in Fig. 13, for wedge angle = 180o/n = 4o and cross section wedge contour lengths l1 = 12.70 mm (half inch), l0 = 15.09 mm, and l2 = 14.88 mm for the inner rib, web, and the outer rib. The rib contour arc-to-wedge border angles are 1 = 2 = 10o . Accordingly, the rib walls meet with 21 = 20o groove angles on both the inner and outer wall Figure 13: Stable cell, n = 45, surfaces. The inner cross section radius is Rmin = 125.6 mm, the outer one is l1/0/2 = 12.70/15.09/14.88 mm. Rmax = 153.5 mm. Thus the outer diameter is Douter = 307 mm, about a foot, and the total wall structure thickness is R = 24.9 mm, a little less than an inch. A part of the full cross section with several cells is shown as a contour in Fig. 14, and in the member context in 15. These images are snapshots of a nite element (FE) model of the design.

Figure 14: Cross section geometry detail.

Figure 15: Mesh detail perspective view.

10 of 13 American Institute of Aeronautics and Astronautics

V. Strength Estimate
Y virtue of their higher local curvatures, the ribbed member wall herein considered is generally able to withstand without local buckling higher compressive stresses than a weight-equivalent tube with smooth wall would. The assessment of this strength improvement, however, is not a trivial exercise, not even if the wall material is linear elastic and isotropic (with Youngs modulus E and Poissons ratio ) as assumed in the following. In fact, there are two kinds of local buckling for the partitioned wall: one in which the structured wall, as a whole, buckles, the other when individual wall segments (partitions or ribs) loose stability. Ultimate member strength depends on how local stability phenomena of these two kinds interact with the global context with global buckling for sufciently long members, or with constraints by member support otherwise. The interaction of the global and two local stability effects, in the context of the cellular wall structure, can only be captured with third order (geometrically correct) full three-dimensional numerical methodology such as capable nonlinear FE analysis. The delicate structure of the wall, the shell formulation needed, and the delity required render such an effort computationally expensive. This analysis has been attempted multiple times: a model, with imperfection seeds, has been incrementally compressed to probe the collapse load and mechanism. However, the analyses were unable to overcome numerical instabilities early in the solution. The problem is deemed to have arisen from web buckling which occurs in intricate patterns early (web atness invites buckling much earlier than the ribs on which wall strength effectively depends). Inated-wall member performance, however, can be estimated via well known design relations, even if in an imperfect manner. This is carried out here with the following simplifying assumptions: 1. Web contribution to member strength is ignored. 2. Web contribution to member mass is fully accounted for. 3. Member strength is the integral over all rib material cross sections of the minimum critical stress the lower one of the critical stresses assessed for the inner and outer rib walls. 4. Rib wall critical stress can be estimated via the relations presented for compressed circular-cylindrical columns in the 1968 Peterson report [26]. The rst assumption is, obviously, conservative. The second and third ones are, simply, realistic: they are neither conservative, nor non-conservative. The fourth, last, one is also conservative because it prescribes expressions developed for full, unconstrained, circular cross sections (full tubes) to a more constrained problem: to the wall stability of cylindrical sections with supported edges (by the webs and the adjacent ribs). Therefore, all in all, the assumptions spelled out are conservative. True performance should be better than the assessment obtained. Hardware imperfections are implicitly accounted for in the calculations by the statistical data embedded in the Peterson equations. While the imperfection patterns and magnitudes in an inated-wall member are likely different from those in simple tubes (the subject of the Peterson report), this discrepancy is accepted herein as inevitable.

A.

Performance for the Example Design

To relate the performance of the design shown in Figs. 13 through 15 to that of a comparable monocoque tube, rst, specify some additional design details for the former and dene an equivalent design for the latter. Let the rib and web wall thicknesses in Fig. 13 be t1 t0 = t2 = 0.305 mm (12 mil) = 0.152 mm (6 mil) (46) (47)

Next, dene an equivalent monocoque tube with a radius that places the tube wall where the inated-wall unit cell center of gravity is, and a wall thickness that results in the same total material volume as for the inated wall member. In particular: Re te = = 139.7 mm 1.098 mm (48) (49)

Then assume the same, immaterial, density and Youngs modulus E for both tubes, and evaluate the critical load by simply applying the Peterson equations [26] for the monocoque tube, and using the four assumptions outlined in Section V for the inated-wall member.
11 of 13 American Institute of Aeronautics and Astronautics

The calculations reveal that this particular design achieves a performance about four times that of an equivalent traditional member:

cr,in f latedwall
B. Assessment of Performance Trend

4.01 cr,monocoque

(50)

With the use of some rule of thumb assumptions, the performance improvement expected from the considered member cross sections can be assessed. For this derivation, described here, use the classic [25, 26] tube wall stability limit E t (51) cr = 2 r 3(1 ) with cr the local-critical compressive stress, and apply this to the bulging rib segments as if the latter were (integral parts of) full tubes. In the monocoque context of a traditional inatable tube, the r in Eq. 51 corresponds to the tube radius Re r = Re (52)

in which the subscript e indicates that a tube equivalent to the inated-wall member is considered. Further, note that the r in Eq. 51 will be the (local) rib bulge radius. Assuming that the rib contours are semi-circles one can consequently approximate the value of r geometrically with the rule of thumb r Re / n (53)

in which n is the rib number, and Re is the equivalent tube radius. Further, assume that the member wall materials are identical and their total masses are similar accordingly, summarily take the wall thickness in the considered inated-wall members outer skin to be a third of that in the traditional tube t1 t2 te / 3 (54)

From the above relations it follows that the strength of the wall-inated member relates to that of a comparable traditional one with the equivalent radius Re according to cr n (55) cr, ribs cr, e cr, e 3 which is approximately unity if n = 9 and increases with n thereafter. An order of magnitude improvement is expected for n = 72, where the central angle from the member axis of each wall chamber is = 5o .

VI. Concluding Remarks

HE pressurization of a cellular member wall, rather the member interior, as a means of deployment and to improve pressurized member performance has been considered and investigated theoretically and numerically. Geometric equations for the design of stable cross sections composed of interlocking wedge-shaped chambers have been derived, and the achievable performance improvement has been shown to be substantial. However, as the direct numerical modeling of the collapse mechanism has not been successful, the independent verication of the performance estimates presented is still yet to be completed. Due to the conservative nature of the assumptions used, however, the strength numbers developed are deemed to be lower bound performance metrics. A practical investigation of the studied inated-wall concept is also left for future work. Fabrication, packaging, and deployment issues have not been investigated herein.

VII. Acknowledgments
HE initial core idea of boosting inatable member strength via a structured-pressurized member wall was rst proposed by the writer in 2005 in a non-public study commissioned by LGarde, Inc. All further work (illustrations, derivations, computer programs, numerical studies) was subsequently performed with no corporate or government support. The writer thanks LGarde for their permission to make the initial idea public.

12 of 13 American Institute of Aeronautics and Astronautics

References
F. W., Expandable Structures For Space Applications, Technical Report XCAFAPL, Air Force Aero Propulsion Lab, WrightPatterson AFB, Ohio, July 30 1964, Accession Number: AD0607541. 2 Thomas, M. and Friese, G., Pressurized antennas for space radars, AIAA Sensor Systems for the 80s Conference, December 1980, AIAA 1980-1928. 3 Freeland, R. E., Bilyeu, G., Veal, G. R., and Mikulas, M. M., Inatable deployable space structures technology summary, 49th International Astronautical Congress, International Astronautical Federation, 3-5, Rue Mario-Nikis, 75015 Paris, France, Melbourne, Australia, September 28 October 2 1998, IAF-98-I.5.01. 4 Darooka, D. K. and Jensen, D. W., Advanced space structure concepts and their development, The 42nd Structures, Structural Dynamics, and Materials Conference and Co-Located Conferences and Forums, AIAA, Seattle, WA, April 1619 2001. AIAA 2001-1257 5 Campbell, J., Smith, S., Main, J. A., and Kearns, J., Staged Microgravity Deployment of a Pressurizing Scale-Model Spacecraft, Journal of Spacecraft and Rockets, Vol. 41, No. 4, July-August 2004, pp. 534542. 6 Cobb, R., Black, J., and Swenson, E., Design and Flight Qualication of the Rigidizable Inatable Get-Away-Special Experiment, Journal of Spacecraft and Rockets, Vol. 47, No. 4, 2010, pp. 659. 7 Bernasconi, M. C. and Reibaldi, G. G., Inatable, Space-Rigidized Structures Overview of Applications and their Technology Impact, Acta Astronautica, Vol. 14, January 1986, pp. 455465. 8 Cadogan, D., Stein, J., and Grahne, M., Inatable Composite Habitat Structures for Lunar and Mars Exploration, Acta Astronautica, Vol. 44, No. 712, AprilJune 1999, pp. 399406. 9 de la Fuente, H., Raboin, J. L., Spexarth, G. R., and Valle, G. D., TransHab: NASAs Large-Scale Inatable Spacecraft, The 41st Structures, Structural Dynamics, and Materials Conference and Co-Located Conferences and Forums, AIAA, Atlanta, GA, April 36 2000. AIAA 2002-1822. 10 Guidanean, K. and Veal, G., An Inatable Rigidizable Calibration Optical Sphere, The 44th Structures, Structural Dynamics, and Materials Conference and Co-Located Conferences and Forums, AIAA, Norfolk, VA, April 710 2003, AIAA-2003-1899. 11 C., E. D., Out From Behind the Eight-Ball: A History of Project Echo, Vol. 16 of AAS History Series, American Astronautical Society, AAS Publication Ofce, P.O. Box 28130, Dan Diego, CA 92198, 1st ed., 1995. 12 Mikulas, M. M., Structural efciency of long lightly loaded truss and isogrid columns for space applications, Tech. Rep. Technical Memorandum TM-78687, NASA Langley Research Center, July 1978. 13 Natori, M. C., Higuchi, K., Sekine, K., and Okazaki, K., Adaptivity Demonstration of Inatable Rigidized Integrated Structures (IRIS), Acta Astronautica, Vol. 37, October 1995, pp. 5967. 14 Rottmayer, E., Compression Tests of Wire-Film Cylinders, Ref. [27], pp. 519535, pp. 519535. 15 Greschik, G., Lichodziejewski, L., and Veal, G., A Counter-Intuitive Condition for the Wrap Reinforcement of Aluminum-Rigidized Tubes, The 43rd Structures, Structural Dynamics, and Materials Conference and Co-Located Conferences and Forums, AIAA, Denver, CO, April 2225 2002. AIAA-2002-1262. 16 Greschik, G., Global Imperfection-Based Column Stability Analysis, 48th Structures, Structural Dynamics, and Materials Conference and Co-Located Conferences and Forums, AIAA, Sheraton Waikiki, Oahu, 2255 Kalakaua Avenue, Honolulu, HI 96815, April 2326 2007. AIAA 2007-2225. 17 Bair, H. Q. and Fischer, W. H., Dual Wall Inatable Structures For Space Oriented Applications, Ref. [27], pp. 785802. 18 Hughes, S. J., Dillman, R. A., Starr, B. R., Stephan, R. A., Lindell, M. C., Player, C. J., and McNeil Cheatwood, D. F., Inatable Reentry Vehicle Experiment (IRVE) Design Overview, 18th AIAA Aerodynamic Decelerator Systems Technology Conference and Seminar , ADS Technology Seminar, AIAA, AIAA, Munich, Germany, May 2326 2005, AIAA 2005-1636. 19 Gal-Rom, Y. C. and Raveh, D. E., Analytical Failure Criteria of an Inated Wing, AIAA-2010-2637. 20 Ghomeshi, R., Milwich, M., and Planck, H., Entwicklung von biomimetisch optimierten, pultrudierten Faserverbundprolen mit hoher dynamischer Belastbarkeit und Schwingungsdmpfung, Forschungsvorhaben AiF 15141/1, Institut fr Textil- und Verfahrenstechnik der Deutschen Institute fr Textil- und Faserforschung (Institute for Textile Technology and Process Engineering Denkendorf), Krschtalstrae 26, d-73770 Denkendorf, Germany, October 2009. 21 Milwich, M. a., Fiber Composite Material with Four Models: Technical Plant Stem, Leaet by ITV Institut fr Textil- und Verfahrenstechnik der Deutschen Institute fr Textil- und Faserforschung Denkendorf (Institute for Textile Technology and Process Engineering Denkendorf), Krschtalstrae 26, 73770 Denkendorf, Germany, 2009, Downloaded on March 5, 2011, from www.itv-denkendorf.de. 22 Murphey, T. and Hinkle, J., Some Performance Trends in Hierarchical Truss Structures, Ref. [?], AIAA-2003-1903. 23 Calladine, C. R., Stability of the Endeavour Balloon, Buckling of Structures: Theory and Experiment, edited by et. al.. Elishakoff, I., Elsevier Science Publishers, 1988, pp. 133149, The Josef Singer anniversary volume. 24 Baginski, F. and Brakke, K., Estimating the Deployment Pressure in Pumpkin Balloons, AIAA-2010-2669. 25 Young, W. C., Roarks Formulas for stress and strain, McGraw-Hill, Inc., sixth ed., 1989. 26 Peterson, J. P., Seide, P., and Weingarten, V. I., Buckling of Thin-Walled Circular Cylinders, NASA Space Vehicle Design Criteria (Structures) NASA SP-8007, NASA Langley Research Center, Langley Station, Hampton, VA 23681-2199, August 1 1968, Revision of September, 1965, version. 27 Second Aerospace Expandable Structures Conference, Technical Report AFAPL-TR-65-108, Air Force Aero Propulsion Laboratory, Wright-Patterson Air Force Base, Ohio, February 1966, Proceedings of the 2nd Aerospace Expandable Structures Conference, sponsored by Air Force Aero Propulsion Laboratory in cooperation with Archer Daniels Midland Company, held May 2527, 1965, in Minneapolis, MN.
1 Forbes,

13 of 13 American Institute of Aeronautics and Astronautics