betaFeedback

WorksPhysics and Astronomy

Exploring the design space for nonlinear buckling of composite thin-walled lenticular tubes under pure bending

Abstract

This paper presents an automatic finite element simulation scheme accounting for high geometric nonlinearity and the difference between linear and nonlinear buckling of composite thin-walled lenticular tubes (CTLTs). Parameterizing of cross-section shapes and generation of design space for CTLTs with both circular and parabolic arcs were accomplished, and several key factors were identified, in particular the contrary effect of lumbus length and parabolic coefficient on the bending stiffness anisotropy. The first quantitative comparison of triangular rollable and collapsible (TRAC) booms and CTLTs is given in terms of bending performance in two directions, showing that the optimal CTLT carefully selected from the design space demonstrates a comparable or even better performance than the TRAC boom. This is of great importance from both academic and engineering perspectives. Our efforts enhance the understanding of nonlinear buckling and post-buckling behavior of CTLTs, and provide guidelines for future design of CTLTs with desirable performance.

Introduction

The composite thin-walled lenticular tube (CTLT), which is also known as the collapsible tube mast (CTM), is a closed tube with lenticular cross-section that can be commonly used for deploying large structures in space. The CTLT is flattened and coiled for storage before and during launch, and capable of deploying spontaneously for use once in orbit. This type of structure, since was first introduced by NASA a few decades ago [1], has been developed and employed as a basic component to deploy various large space structures, such as solar arrays and antennas in many space missions. The first CTM ever used in space was jointly developed by NASA and ESA for the ULYSSES mission [2]. The development of CTLT is also considered as a key technique of German Aerospace Center DLR’s solar sailing technology [3, 4, 5, 6], and in 2009 an agreement was made between ESA and DLR by which they started a three-step project aiming to develop, prove, and demonstrate that CTLT can serve as a safe and reliable component for long-lasting and deep space missions [7, 8]. NASA has also recently expressed an interest in small CTLTs as a candidate solar sail boom for low-cost deep space exploration and science missions [9, 10, 11, 12, 13, 14]. As an example of commercial application, Oxford Space Systems Ltd (OSS) in UK, is currently developing a 2.7-m-diameter wrapped-rib antenna where 48 CTLTs are employed as wrapping ribs to deploy the metal mesh reflector surface [15, 16, 17].

The CTLT may be subjected to various kinds of mechanical loads during different stages of its working process such as flattening, coiling and deploying, and as a slender thin-walled structure it could demonstrate a complex and nonlinear behavior. A large number of fundamental research has been carried out to study the flattening and wrapping process of CTLTs. Hu et al. [18] performed a combined experimental, numerical, and analytical investigation of the mechanical response of both compressive and tensile flattening deformations of deployable CTLTs. Chen et al. [19] carried out some experiments to test the large deformation behaviors of CTLTs in flattening and wrapping process and developed three-dimensional finite element models to predict the mechanical characteristics identified by experiments. Similar studies were also reported by Bai et al. [20, 21, 22] for determining tensile, compression and folding behaviors of CTLTs. The above-mentioned efforts have been focused on the deformation and the associated failure during flattening and folding, and localized buckling that occurs during flattening and folding was observed as the dominant failure mode of deployable CTLTs [23, 24]. However, little attention was paid to study the instability or buckling behavior of CTLTs. The only relevant work that we are aware of focused on the linear buckling and nonlinear post-buckling response of the CTLT subjected to uniaxial compression [25]. It was found that the critical buckling load estimated directly by eigenvalue analysis is far greater than the realistic experimental measurement. A closer critical buckling load was eventually obtained by introducing proper initial imperfections and performing a post-buckling analysis. However, the initial imperfections introduced in the post-buckling analysis were determined from the experimental observations and this may reduce the predictive significance of the model [26]. Therefore, reliable simulation techniques need to be developed to accurately predict the critical buckling load of CTLTs prior to conducting experiments [27].

A more accurate prediction of the critical buckling load can be obtained by first performing a geometrically nonlinear response analysis and then estimating the buckling load by a following eigenvalue analysis on the deformed configuration. This method is referred to as the nonlinear buckling analysis, and it has been well-developed and documented, and successfully applied in the field of mechanical engineering in the past few decades. A number of analytical solutions can be found for the nonlinear buckling predictions of simple structures such as beam, truss, and shell problems with ideal boundary conditions. Brendel et al. [28] formulated the eigenvalue problem with the information at deformed states on the nonlinear pre-buckling path and studied the effect of imperfections on the critical loads of several cylindrical shells under uniform pressure and wind load. Kounadis et al. [29] proposed a simplified approach for the nonlinear buckling analysis of frames subjected to either bifurcational or limit-point instability. Wu et al. [30, 31] derived the design derivatives of the nonlinear critical load and proposed an optimization procedure to maximize the load-carrying capacity of truss-structures. For complex systems, analytical solutions are rarely obtainable and numerical modelling is often required. Lindgaard et al. [32, 33, 34, 35] developed a combination of numerical approach and optimization procedure to maximize the critical nonlinear buckling load of laminated composite shell structures and provided validation with several benchmark problems. Nguyen et al. [36] formulated a finite element model using three-noded Timoshenko beam elements to analyze the nonlinear buckling load of an inflatable beam made of orthotropic technical textiles. Liang et al. [37, 38] proposed a reduced order model that capable of predicting the nonlinear buckling behavior of variable stiffness composite plates efficiently compared with typical full order finite element model.

Structural instability arises as one of the biggest concerns for structural design of thin-wall composite boom structures such as CTLTs in aerospace engineering, because these structures are usually thin in thickness and have long aspect ratio. TRAC (triangular rollable and collapsible) booms and CTLTs have been considered promising (competitive) candidates for solutions of various light-weight deployable structures [39, 40, 41]. Very recently a few studies have been carried out to understand the nonlinear buckling behavior of TRAC booms subjected to pure bending. Murphey et al. [42] analyzed the basic structural mechanics including deployment stiffness, buckling strength, and packaging constraints of TRAC booms using closed form analytical and finite element approaches. Leclerc et al. [43] performed a study of the nonlinear elastic buckling behavior of triangular rollable and collapsible (TRAC) booms under pure bending and reported a good agreement between numerical predictions and experimental measurements. Bessa et al. [44] constructed the design space diagram for the TRAC booms and proposed a data-driven computational framework combining Bayesian regression for optimizing the nonlinear critical buckling load of TRAC booms. Despite the available study on instability of TRAC booms and CTLTs, either numerical or experimental, reliable prediction of critical loads with emphasis on variation of configuration and initial stress field in the context of nonlinear buckling analysis remains elusive. A recent study on TRAC booms reveals the subtle difference between classical linear buckling analysis and nonlinear buckling, and the linear buckling may give biased estimation on critical loads [45, 46]. But this has never been carried out for CTLTs.

Moreover, due to the unique cross-section shape of CTLT, it exhibits strong bending stiffness anisotropy [47, 48, 49, 50], and thus a rational design of CTLT is only achievable provided an accurate design space diagram is available, and a trade-off is made by compromising bending performance in different directions. Another benefit of generating and exploiting through a design space is the possibility of unveiling some key design parameters which might be missed otherwise. Last but not least, the obtained design space and exploration permits parameter tailoring for optimal designs, and more importantly, allows a possible quantitative comparison of the two counterparts, TRAC booms and CTLTs.

In this paper, we develop a finite element method based numerical scheme that predicts the nonlinear buckling and post-buckling behavior of CTLTs under pure bending. We first study the mechanical response of a particular CTLT bent along two perpendicular directions, and demonstrate the validity of the developed analysis methods. A good agreement was obtained between the predictions of critical buckling load extracted by the nonlinear buckling analysis and that identified from the post-buckling response. We then perform a systematic numerical study to construct the design space against nonlinear buckling for a variety of CTLTs with equal weight but different cross-section shapes. The cross-section shape is shown to be a convenient parameter for controlling the pre-buckling stiffness and the nonlinear critical buckling load of CTLTs. More specifically, we start by investigating the effect of the size of web and lumbus on the bending resistance performance of conventional CTLTs with circular arcs. We then introduce the concept of parabolic CTLTs by replacing the conventional circular arcs with parabolas, and show how it is possible, by just tuning the parabolic shape, to alter the range of attainable mechanical responses and leverage the trade-off between the bending resistance performance in the two directions. Finally, we made a comparison of the bending performance between conventional circular CTLTs, parabolic CTLTs and TRAC booms with equal weight. It is shown that the optimal CTLT carefully selected from the design space demonstrates a comparable or even better performance than TRAC boom loaded concurrently in two loading directions.

The paper is organized as follows. Section 2 describes the geometry design of the traditional CTLTs with circular arcs. Section 3 introduces the numerical analysis techniques that are used to investigate the nonlinear buckling and post-buckling response of CTLTs. Section 4 presents the results obtained from parametric studies, highlighting the effect of cross-section shape on the nonlinear mechanical response of CTLTs. Finally, concluding remarks are included in Section 5.

CTLT Geometry

Figure 1 shows a schematic of the conventional CTLT with circular arcs. This type of structure consists of joining two thin omega-shaped cylindrical shells of thickness tt. The cross-section of each omega-shaped shell consists of four circular arcs of radius RR subtending an angle 60∘60^\circ, two straight segments at the ends of width ww constituting the web, and one straight segment called lumbus of length LL at center [51]. The two shells are mirror-symmetric and the end flat parts are bonded together in the unstressed configuration. The geometry of the CTLT is then fully characterized by the longitudinal length ll, thin-shell thickness tt, and the cross-section parameters: web width ww, lumbus length LL, and circular arc radius RR. In order to make a fair comparison between the performance of CTLTs with different cross-section geometries and TRAC booms, the longitudinal length ll, thin-shell thickness tt, and the flattening length of the cross-section ss are set the same as those of TRAC booms reported in literature [43, 44], i.e., l=504l = 504 mm, t=0.071t = 0.071 mm and s=27.43s = 27.43 mm. Finally, the flattening length of the cross-section ss, which can be expressed as

Schematic of the CTLT with circular arcs

Figure 1. Schematic of the CTLT with circular arcs. (a) A CTLT in the deployed, i.e., unstressed configuration. (b) Cross-section of the CTLT with geometric parameters indicated. Web width ww and lumbus length LL are considered as the two independent variables, and circular arc radius RR is found from the flattening length equation.

s=2w+L+43πR,s = 2w + L + \frac{4}{3}\pi R,

is fixed constant, and therefore the mass of the structure remains unchanged. The web width ww and the lumbus length LL are selected as two independent variables that are used to adjust the cross-section geometry of circular CTLTs.

Finite element analyses

In this section, we will detail the finite element analyses that are used to investigate the nonlinear buckling and post-buckling behavior of CTLTs subjected to pure bending. First, we will introduce the finite element model with the definition of boundary conditions and material properties (see § 3.1). Then, we will focus on the analyses to predict the critical loads for the onset of buckling and the associated buckling modes (see § 3.2). Finally, we will discuss the post-buckling response analysis (see § 3.3).

Finite element model

The finite element model is constructed utilizing the commercial software Abaqus 2020 with python scripts allows for automated parametric studies. Four-node general-purpose shell elements with reduced integration (Abaqus element type S4R) were used and the accuracy of the mesh was ascertained through a mesh refinement study, resulting in a relative mesh density of around 50 elements along the cross-section profile.

The nodes forming the two end cross-sections are kinematically coupled to two reference points, in effect creating rigid cross-sections that match the end conditions of the experimental setup [43]. All six degrees of freedom of the reference point at end 1 are constrained, defining a clamped condition. At the other end, a pure moment load is applied to the reference point 2. The location of the two reference points is shown in Figure 1(a). In the case of a moment applied about xx-axis MXM_X, the translational degrees of freedom along YY and ZZ as well as the rotational degree of freedom around XX are left free, while the other three degrees of freedom are all fixed, i.e., u1=0u_1 = 0, u2=FREEu_2 = \text{FREE}, u3=FREEu_3 = \text{FREE}, ur1=FREEur_1 = \text{FREE}, ur2=0ur_2 = 0, ur3=0ur_3 = 0. In the case of a moment applied about yy-axis MYM_Y, the same boundary conditions are used with XX and YY inverted.

In accordance with previously reported data for TRAC booms [43, 44], each CTLT omega shell consists of four layers of unidirectional carbon fibers in an epoxy resin and the layers are arranged in the stacking sequence [0∘,90∘]S[0^\circ, 90^\circ]_S. The nominal orthotropic elastic material properties of each layer are set as E1=128.0E_1 = 128.0 GPa, E2=6.5E_2 = 6.5 GPa, ν12=0.35\nu_{12} = 0.35, G12=G13=G23=7.5G_{12} = G_{13} = G_{23} = 7.5 GPa. In addition, a composite shell section is defined to specify the material properties, thickness, and orientation angle of each layer.

Instability analysis

In a nonlinear finite element formulation, the response of a structure is calculated by iteratively solving the equation [52, 53, 54]:

KTδu=R(1)K_T \delta u = R \tag*{(1)}

where KTK_T is the current tangent stiffness matrix where the loads are applied, δu\delta u is the incremental displacement vector, and RR is the force residual for the iteration. For large deflection problems, the tangent stiffness matrix KTK_T consists of the sum of the initial stiffness matrix K0K_0, the displacement stiffness matrix KLK_L, and the stress stiffness matrix KσK_\sigma as described by:

KT=K0+KL+Kσ(2)K_T = K_0 + K_L + K_\sigma \tag*{(2)}

The structure is stable only if the tangent stiffness matrix KTK_T is positive definite; in other words, the structure becomes unstable and buckling is likely to occur when the tangent stiffness KTK_T is singular. Therefore, the onset of buckling can be predicted by looking for the loads for which the model tangent stiffness matrix becomes singular, and so that the problem

KTδu=0(3)K_T \delta u = 0 \tag*{(3)}

has non-trivial incremental displacement vector δu\delta u solutions.

Linear buckling analysis

In the linear buckling analysis, the pre-buckling displacement is assumed to be small thus the displacement stiffness matrix KL=0K_L = 0, and the tangent stiffness matrix is approximated using only the initial stiffness matrix K0K_0 and the stress stiffness matrix KσK_\sigma. By assuming the stress stiffness matrix to be linearly proportional to an incremental load QQ, Eq. (3) is reduced to an eigenvalue problem:

(K0+λiKσ)ϕi=0(4)(K_0 + \lambda_i K_\sigma)\phi_i = 0 \tag*{(4)}

where K0K_0 is the initial global stiffness matrix defined at the undeformed configuration, KσK_\sigma is the initial stress stiffness matrix caused by the incremental load QQ, λi\lambda_i are the eigenvalues, i.e., critical buckling load factors, and ϕi\phi_i are the eigenvectors, i.e., buckling modes. The eigenvalues and eigenvectors are ordered increasingly in magnitude, such that λ1Q\lambda_1 Q is the lowest linear critical buckling load and ϕ1\phi_1 is the corresponding linear buckling mode. This eigenvalue problem is readily solved using Lanczos method by an eigenvalue buckling analysis step (*Buckle) in Abaqus.

Nonlinear buckling analysis

In the nonlinear buckling analysis, the effect of the change in configuration in the pre-buckling stage is considered and the full tangent stiffness matrix from Eq. (2) must be utilized [55]. The nonlinear buckling analysis generally consists of two analysis steps: (i) a nonlinear static analysis step (*Static) and (ii) an eigenvalue buckling analysis step (*Buckle). In the first step, a preload (”dead” load), PP, is applied to the structure to attain a deformed state, during which the large-displacement formulation is used (NLGEOM=ON) to capture the geometric nonlinearity. Then the deformed state of the model at the end of the first step is identified as the base state for the second eigenvalue buckling step. Namely, the nonlinear critical loads and buckling modes are calculated by applying a seperate incremental load QQ to the deformed state of the structure with change of configurations and attained stress-field incorporated. The eigenvalue problem is a bit complicated as follows:

(K0+KL+λiKσ)ϕi=0(5)(K_0 + K_L + \lambda_i K_\sigma)\phi_i = 0 \tag*{(5)}

where K0K_0 is the initial global stiffness matrix defined at the undeformed configuration, KLK_L is the global displacement stiffness matrix determined at the deformed state by applying the preload PP, and KσK_\sigma is the initial stress stiffness matrix caused by the incremental load QQ.

While large deformation is included in the static analysis, the eigenvalue buckling theory relies on there being little geometric change due to the ”live” buckling load, λiQ\lambda_i Q. Then the final nonlinear critical loads predicted by the two-step nonlinear buckling analysis are given by P+λiQP + \lambda_i Q and ϕi\phi_i denote the corresponding nonlinear buckling modes. Notice that the amount of preload would also have an effect on the nonlinear buckling predictions of structures. To capture the nonlinear buckling behavior of the thin-walled composite structures which are known for their high sensitivity to geometric imperfections and having many buckling modes with closely spaced eigenvalues, it often helps to apply enough preload to deform the structure to just below the buckling load prior to performing the eigenvalue extraction [56]. On the other hand, the structure should not be preloaded above the buckling load; otherwise the adopted Lanczos algorithm in Abaqus will issue an error message and terminate the analysis.

In the nonlinear buckling analysis, with the aim of finding the appropriate preload which should be as large as possible while allowing the extraction of eigenvalues, the following tricks could be helpful: (i) applying a relatively large preload (greater than nonlinear buckling load) in the nonlinear static analysis step, and thus this step is expected to be conducted up to the pre-determined load or fail to converge due to buckling; and (ii) performing an eigenvalue analysis in the deformed state starting from the gradually decreasing last available increment until the eigenvalues can be properly extracted. The above procedure, illustrated as pseudocode in Algorithm 1, is implemented into a Python script in Abaqus to run the simulations automatically.

In real practice, the preload in the nonlinear static analysis is set equal to the linear critical buckling load obtained by conducting a linear buckling analysis, because the linear buckling analysis never underestimates the critical load of the CTLT as will be demonstrated in the following sections.

Algorithm 1 Pseudocode for the nonlinear buckling analysis

Step 1. Run a nonlinear static analysis with a preload being applied and write the model definition and deformed state at every increment to the files required for restart.

Step 2. Perform an eigenvalue analysis from the last available increment until the eigenvalues can be successfully extracted.

while True: do

Perform the eigenvalue analysis

if Eigenvalue analysis completes successfully then

break

end if

n=n−1n = n - 1 (nn is the maximum number of available increments obtained in Step 1)

end while

Post-buckling analysis

The nonlinear post-buckling response of the CTLT is investigated by introducing an imperfection in the form of the two most critical buckling modes, v1Mv_1^M and v2Mv_2^M, obtained from the nonlinear buckling analysis. Therefore, the mesh is perturbed by v1Mv_1^M and v2Mv_2^M scaled by a factor η\eta, such that

δvM=η(v1M+v2M)(6)\delta v^M = \eta(v_1^M + v_2^M) \tag*{(6)}

where η\eta is chosen as 5% of thickness of the CTLT, i.e., η=0.05t\eta= 0.05t.

Results and discussions

In this study numerical simulations are performed to explore the nonlinear buckling behavior space of CTLTs characterized by different cross-section geometries. First, the linear and nonlinear buckling and post-buckling behavior of a particular CTLT is investigated (see § 4.1). Then, the behavior space of circular CTLTs is explored, highlighting the effect of the size of web and lumbus on the pre-buckling stiffness and nonlinear critical buckling load of conventional CTLTs (see § 4.2). Next, the concept of parabolic CTLTs is introduced and the design space of which is also probed, demonstrating a wider tunability range of the mechanical response (see § 4.3). Finally, the bending resistance performance of both circular and parabolic CTLTs is compared to that of TRAC booms with equal weight, and a discussion is presented (see § 4.4).

Nonlinear buckling behavior of a particular circular CTLT

A beam develops compressive stresses on its inner surface when it is subjected to a pure bending moment. For the CTLTs, the bending moment about xx-axis produces in-plane compressive stresses on the inner lumbus, while the bending moment about yy-axis produces in-plane compressive stresses on the inner web. The compressive stress is the driving force for most buckling phenomena. Moreover, the lumbus and web, as being parts of the CTLT, are thin-walled structures and have a low bending stiffness. Therefore, the in-plane compressive stress may lead to out-of-plane buckling within the lumbus or web when the critical value is reached.

To demonstrate the typical response characteristics of the CTLTs under pure bending, without loss of generality, in Figure 2 we present the analysis results of a particular CTLT characterized by w=5.5w = 5.5 mm and L=3.5L = 3.5 mm. Figure 2a and 2b show the mechanical response of the CTLT when the bending moment is applied about xx-axis. We start by determining the critical loads for buckling of this CTLT, only the lowest critical load is of interest here. The critical load is determined from different perspectives. First, we estimate the critical load by solving the eigenvalue problem through performing a liner buckling analysis or a nonlinear buckling analysis as discussed previously. The results are presented by two dashed lines perpendicular to the moment axis. In Figure 2a the red dashed line indicates the critical load obtained from the linear buckling analysis, which we refer to as linear critical load (LCL), and the blue dashed line indicates that obtained from the nonlinear buckling analysis, which we refer to as nonlinear critical load (NLCL). Moreover, the critical buckling load can also be identified from the moment-angle curve by performing a post-buckling analysis. For the post-buckling analysis, the imperfections afore-mentioned should be introduced. It can be seen from Figure 2a that there exist two distinct regimes in the moment-angle curve (as shown in blue solid line), i.e., a pre-buckling regime and a post-buckling regime. The transition point on the moment-angle curve from pre-buckling regime to post-buckling regime can be identified as the critical load for buckling. From these results shown in Figure 2a, we conclude that (i) linear buckling analysis would overestimate the overall stability of the structure in this case, i.e., LCT>NLCTLCT > NLCT, and (ii) the nonlinear critical load (NLCLNLCL) estimated by the nonlinear buckling analysis agrees very well with that identified from the post-buckling response. Furthermore, the difference of linear and nonlinear buckling analysis is also manifested in terms of buckling modes. As shown in Figure 2b, linear buckling analysis predicts a larger number of winkles within the lumbus than nonlinear buckling analysis in this case. It should be pointed out the difference between linear and nonlinear buckling highlighted herein is firstly pointed out by [45]in the analysis of TRAC boom. The difference is elucidated once again for CTLT bent about xx−axis.

Nonlinear buckling and post-buckling response of a particular CTLT

Figure 2. Nonlinear buckling and post-buckling response of a particular CTLT characterized by w=5.5w = 5.5 mm and L=3.5L = 3.5 mm when subjected to a pure bending moment about (a-b) xx−axis and (c-d) yy−axis.

In sharp contrast, a different buckling behavior is observed when the pure bending moment is loaded about yy−axis. It can be seen in Figure 2c that for this case the critical loads predicted by linear and nonlinear buckling analysis are basically the same, i.e., LCL=NLCLLCL = NLCL, and both are in good agreement with that identified from the post-buckling moment-angle curve. This difference is attributed to the different boundary conditions of the web and lumbus in compression. More specifically, when the bending moment is applied about xx−axis the lumbus undergoing compressive stresses is constrained on both sides by surrounding structures; when the bending moment is applied about yy−axis, only one side of the web in compression is constrained while the inner edge remains free, which reduces the sensitivity of buckling load to geometric nonlinearity. On the other hand, the post-buckling response under the two loading conditions exhibits similar features, as they both contain two distinct stages divided by NLCLNLCL, namely, a pre-buckling stage starting from initiation to NLCLNLCL, and a post-buckling stage starting from NLCLNLCL and ending with buckling collapse. We also show, in Figure 2a and Figure 2c, that the pre-buckling regime predicted by the post-buckling analysis is observed coincided with the curve obtained by performing a nonlinear static analysis without introducing any imperfections, which exhibits approximately a linear behavior. Then the pre-buckling stiffness of CTLTs can be calculated as

K=dMdθ(7)K = \frac{dM}{d\theta} \tag*{(7)}

where the derivative can be directly obtained through linear regression of the data in pre-buckling regime.

Construction of design space for circular CTLTs

Having identified typical mechanical behavior of CTLTs under pure bending, we then move on to construct the design space, aiming to probe the evolution of pre-buckling stiffness KK and nonlinear critical buckling load NLCLNLCL. The design space is constructed by varying the two independent cross-section geometric parameters, which are the web width ww and lumbus length LL as shown in Figure 3a. More specifically, ww is varied from 1 to 10 mm and LL is varied from 0 to 7 mm, and a number of diverse cross-section shapes are determined as shown in Figure 3b. Note that L=0L = 0 indicates a CTLT design without lumbus. Figure 3c and 3d present the evolution of the pre-buckling stiffness KxK_x and the nonlinear critical load (NLCL)x(NLCL)_x about xx-axis as a function of the web width ww and the lumbus length LL. It is shown that the pre-buckling stiffness KxK_x strongly depends on the web width ww but slightly on the lumbus length LL, while the nonlinear critical load (NLCL)x(NLCL)_x depends both highly on the web width ww and the lumbus length LL. Specifically, KxK_x increases as ww decreases but tends to remain unchanged as LL varies, while (NLCL)x(NLCL)_x increases as ww and/or LL decreases. Furthermore, the evolution of the pre-buckling stiffness KyK_y and nonlinear critical load (NLCL)y(NLCL)_y about yy-axis is presented in Figure 3e and 3f. It is shown that the pre-buckling stiffness KyK_y depends both highly on the web width ww and the lumbus length LL, while the nonlinear critical load (NLCL)y(NLCL)_y depends solely on the web width ww. Specifically, KyK_y increases with the increase of ww or LL, while (NLCL)y(NLCL)_y increases with the decrease of ww. In conclusion, there exists a common region of the design space that maximize the pre-buckling stiffness about xx-axis and nonlinear critical buckling load about both xx and yy-axes at small web width, although a decrease in the pre-buckling stiffness about yy-axis is expected. Given the fact that the minimum pre-buckling stiffness about yy-axis (∼20\sim20 Nm/rad) is yet much greater than the maximum pre-buckling stiffness about xx-axis (∼6\sim6 Nm/rad), the smallest value of 1 mm is determined to be the optimal size for the web.

The design space for nonlinear buckling of CTLTs with circular arcs

Figure 3. The design space for nonlinear buckling of CTLTs with circular arcs. (a) Schematic illustration of the cross-section geometry of the circular CTLT. (b) A variety of cross-section shapes of circular CTLTs characterized by web width w∈[1,10]w \in[1, 10] mm and lumbus length L∈[0,7]L \in[0, 7] mm. Heat map illustrating the (c) pre-buckling stiffness KxK_x and (d) nonlinear critical load (NLCL)x(NLCL)_x as a function of the web width ww and lumbus length LL when the moment is applied about xx-axis. Heat map illustrating the (e) pre-buckling stiffness KyK_y (f) and nonlinear critical load (NLCL)y(NLCL)_y as a function of the web width ww and lumbus length LL when the moment is applied about yy-axis.

Next, we take a closer look at the effect of the lumbus length LL on the nonlinear buckling behavior of CTLTs by considering a group of samples with ww fixed to be 1 mm and LL varies from 0 to 7 mm. Figure 4a describes a trade-off between the bending resistance performance about xx-axis and that about yy-axis. Specifically, an increase in the lumbus length LL leads to a better bending resistance performance about xx-axis but in the meantime leads to a worse bending resistance performance about yy-axis. The effect of the lumbus length is also reflected in terms of the nonlinear buckling modes as shown in Figure 4b. Firstly, for the cases the bending is applied about xx-axis, two distinct nonlinear buckling patterns were observed according to whether the CTLT has lumbus or not. For the CTLT without lumbus, i.e., L=0L = 0, the buckling mode is a diamond wave pattern, as often found in the buckling of thin-walled cylindrical shells [57, 58, 59]. In contrast, for the CTLT with lumbus, i.e., L>0L > 0, the buckling mode is recognized as ”wrinkles” in the lumbus, and the number of the wrinkles increases as the lumbus length LL increases from 1 to 7 mm. Secondly, for the cases bending is applied about yy-axis, the buckling mode is a wave pattern occurs in the compressed web, and the number of waves also increases as the lumbus length increases from 0 to 7 mm. Finally, considering the trade-off between the bending resistance performance about xx- and yy-axes, a value of 3 mm, which is in the middle of the range for the lumbus length LL, is selected as the optimal one. Note that for the particular cross-sectional shape of CTLT of interest here, the lumbus length has profound effect on the parameter selection and sectional shape design. This effect, however, is omitted in current literature, and it is elucidated for the first time with the aid of constructed design space diagram.

Effect of the lumbus length $L$ on the post-buckling response and nonlinear buckling modes of CTLTs

Figure 4. Effect of the lumbus length LL on (a) the post-buckling response and (b) nonlinear buckling modes of CTLTs with w=1w = 1 mm when subjected to a pure bending moment about xx–axis (left) and yy–axis (right).

Design space for parabolic CTLTs

So far, we have explored the nonlinear buckling behavior space of conventional CTLTs with circular arcs, and reported the effect of web and lumbus sizes on the bending performance of circular CTLTs. However, the constant curvature in circular arcs may limit the potential use of conventional CTLTs. A new design of CTLT with variable curvature arcs is achieved by replacing the circular arcs with parabolic arcs, which is referred to as the parabolic CTLT [47]. The cross-section geometry of the parabolic CTLT is illustrated in Figure 5a. In this section, we follow our previously developed numerical strategy for parameterizing sectional shape and generating design space for circular CTLTs and extend them to parabolic CTLTs. As shown in Figure 5a, each parabolic arc segment is defined by a parabola function:

Design space for nonlinear buckling of parabolic CTLTs with schematic geometry, cross-section shapes, and heat maps of pre-buckling stiffness and nonlinear critical load

Figure 5. The design space for nonlinear buckling of parabolic CTLTs. (a) Schematic illustration of the cross-section geometry of the parabolic CTLT. (b) A variety of cross-section shapes of parabolic CTLTs characterized by the parabolic coefficient a∈[10−2,100]a \in[10^{-2}, 10^{0}] and the lumbus length L∈[0,7]L \in[0, 7] mm. Heat map illustrating the (c) pre-buckling stiffness KxK_x and (d) nonlinear critical load (NLCL)x(NLCL)_x as a function of the parabolic coefficient aa and the lumbus length LL when the moment is applied about xx-axis. Heat map illustrating the (e) pre-buckling stiffness KyK_y and (f) nonlinear critical load (NLCL)y(NLCL)_y as a function of the parabolic coefficient aa and the lumbus length LL when the moment is applied about yy-axis.

y=ax20≤x≤x0(8)y = ax^{2} \qquad0 \le x \le x_{0} \tag*{(8)}

where aa is the coefficient of the parabola. The arc length of a single parabolic segment, LPL_{P}, is calculated using the following formula:

LP=∫0x01+4a2x2 dx(9)L_{P} = \int_{0}^{x_{0}} \sqrt{1 + 4a^{2}x^{2}}\,\mathrm{d}x \tag*{(9)}

and then the total flattening length of the cross-section of the parabolic CTLT is expressed as s=2w+L+4LPs = 2w + L + 4L_{P}. Recall that ss is set to 27.43 mm as the same as its circular counterpart, and in this case the web width ww is fixed to 1 mm. The parabolic CTLT thus has two parameters that can be varied independently, i.e., the lumbus length LL and the coefficient of parabola function aa. Figure 5b presents various cross-section geometries obtained by varying the lumbus length LL from 0 to 7 mm and varying the coefficient aa from 10−210^{-2} to 10010^{0}. The coefficient aa determines how wide or narrow the parabola is; the greater the coefficient aa, the narrower the parabola. Therefore, as shown in Figure 5b, the cross-section becomes narrower in xx-axis and wider in yy-axis as the coefficient aa increases for a given value of lumbus length LL. In theory, this will cause a decrease in the stiffness about xx-axis and in the meantime an increase in the stiffness about yy-axis [47, 48]. Heat maps in Figure 5c and 5d present the effect of the coefficient aa and the lumbus length LL on the bending performance of parabolic CTLTs about xx-axis. The pre-buckling stiffness KxK_x and the nonlinear critical load (NLCL)x(NLCL)_x both depend significantly on the coefficient aa and the lumbus length LL. Specifically, a greater coefficient aa and/or a smaller lumbus length LL leads to an increase in both the pre-buckling stiffness KxK_x and the nonlinear critical load (NLCL)x(NLCL)_x. Figure 5e shows the evolution of the pre-buckling stiffness about yy-axis KyK_y as a function of the parabolic coefficient aa and the lumbus length LL. In contrast with the performance about xx-axis, a greater coefficient aa and/or a smaller lumbus length LL leads to a decrease in the pre-buckling stiffness about yy-axis KyK_y. In addition, as shown in Figure 5f the parabolic coefficient aa has greater effect on the nonlinear critical load (NLCL)y(NLCL)_y than the lumbus length LL, and indicatively, as aa increases from 10−210^{-2} to 10010^{0}, (NLCL)y(NLCL)_y increases first, reaches a peak at a≃10−1a \simeq10^{-1}, and then decreases.

We also investigated in detail the effect of the parabolic coefficient aa on the post-buckling response of parabolic CTLTs by considering a group of samples characterized by L=3L = 3 mm and a∈[10−2,100]a \in[10^{-2}, 10^{0}]. An possible compromise is made once again between the bending resistance performance about xx-axis and that about yy-axis. It can be seen from Figure 6 that a greater value of parabolic coefficient aa leads to a better bending performance about xx-axis but a worse bending performance about yy-axis. In addition, CTLTs with circular arcs (black dash lines in Figure 6) show a moderate performance in both cases. These results indicate that a further tunability of the bending performance of CTLTs can be achieved by altering the parabolic coefficient aa, and it could be useful in real-world engineering applications when anisotropic stiffness property, e.g., a higher stiffness in one direction and a lower stiffness in the other direction, is desirable.

Effect of the parabolic coefficient $a$ on the post-buckling response of parabolic CTLTs

Figure 6. Effect of the parabolic coefficient aa on the post-buckling response of parabolic CTLTs characterized by L=3L = 3 mm when subjected to a pure bending moment about (a) xx-axis and (b) yy-axis.

Comparison with TRAC booms

Having explored the design space of circular and parabolic CTLTs, in this subsection, we will perform a comparative study on the bending resistance performance of circular and parabolic CTLTs to that of TRAC booms of equal weight.

With the design space in hand, the variations of bending performance of both circular and parabolic CTLTs are extracted readily from the heat maps in Figure 3 and Figure 5. The same physical quantity is taken from the nonlinear buckling of TRAC booms reported very recently in [44]. A comparison of variation of the nonlinear critical loads between TRAC booms, circular CTLTs and parabolic CTLTs is given in Table 1. In summary, circular CTLTs has a narrower range of variation (NLCL)x(NLCL)_x but a significantly wider range of variation of (NLCL)y(NLCL)_y than TRAC booms for the case considered herein. By introducing parabolas, the maximum value of (NLCL)x(NLCL)_x of parabolic CTLTs is beyond that of TRAC booms. The maximum value of (NLCL)y(NLCL)_y of parabolic CTLTs is slightly lower than that of circular CTLTs, but still much greater than that of TRAC booms.

(NLCL)x(NLCL)_x(NLCL)y(NLCL)_y
TRAC booms [44]0 − 0.77 Nm0 − 0.31 Nm
Circular CTLTs0 − 0.6 Nm0 − 2.0 Nm
Parabolic CTLTs0 − 0.82 Nm0 − 1.5 Nm

Table 1. Comparison of variation of nonlinear critical buckling moment (NLCLNLCL) between TRAC booms, circular CTLTs and parabolic CTLTs

In addition, as reported in [44], in the design space of TRAC booms there is a common region that maximizes the nonlinear critical buckling load for both loading directions at large flange angles and small web height, and this gives rise to an optimal TRAC boom design characterized by a web height of 2 mm and a flange angle of 300∘300^\circ. In Figure 7 the moment-angle curves of the optimal TRAC boom were reproduced (blue lines) and used as the reference to evaluate the performance of CTLTs. Notice that the results of TRAC boom that reproduced by our approach are in excellent agreement with that reported in [44], which validate the accuracy of the proposed model. A circular CTLT characterized by w=1w = 1 mm and L=3L = 3 mm and a parabolic CTLT characterized by a=100a = 10^0 and L=3L = 3 mm are chosen as the optimal candidates from the design space for comparison. As shown in Figure 7, the circular CTLT demonstrates a worse performance than the others in xx-axis but better in yy-axis, while the parabolic CTLT shows a comparable performance in both xx and yy-axes to the TRAC boom. These results show that the optimal CTLT with tailored parameters derived from the design space demonstrates a comparable or even better performance than the TRAC boom in either the two loading directions.

Comparison of bending resistance performance around the x-axis and y-axis

Figure 7. Comparison of bending resistance performance around (a) xx− axis and (b) yy−axis between the optimal TRAC boom (blue), circular CTLT (black), and parabolic CTLT (red).

Conclusion

TRAC booms and CTLTs have been proposed as two promising candidates for solutions for deploying various light-weight space structures. Since the thin-walled tubes or booms are thin in thickness and have long aspect ratio, structural instability arises as one of the biggest concerns for structural designs. Recent study on nonlinear buckling analysis of TRAC booms reveals that classical linear buckling may biasedly estimate the critical load of TRAC booms under pure bending and necessitates the classification of nonlinear and linear buckling. However, the recurrence of this difference needs to be justified in CTLTs. Moreover, a design space is desperately needed, with which a thorough exploration and tailoring of design parameters is admissible.

In the present paper, we carried out a comprehensive computational investigation aiming at exploring the design space for nonlinear buckling behavior of CTLTs with both circular and parabolic arcs. Our results reveal the effect of cross-section geometric parameters such as the size of web, lumbus, and the parabolic coefficient, on the bending resistance performance of CTLTs. More specifically, we found that there is a common region at small web width that maximizes the nonlinear critical buckling load of CTLTs concurrently about both xx and yy−axes under pure bending, whereas there is a trade-off between the two loading directions in terms of pre-buckling stiffness, namely, an increase in the web width and/or a decrease in the parabolic coefficient would result in a decrease in the pre-buckling stiffness about xx-axis but an increase in the pre-buckling stiffness about yy-axis. More importantly, the contrary effects of the lumbus length on bending performance of CTLTs in two directions are also unveiled. This profound effect of the lumbus, nevertheless, has been ignored in the literature to our best knowledge [51].

Moreover, we made a quantitative comparison on the bending performance of TRAC booms and CTLTs under the same weight. The optimal circular CTLT demonstrates a better bending resistance performance than the optimal TRAC boom in yy- axis but a worse performance in xx-axis, while the optimal parabolic CTLT demonstrates a comparable or better performance in both loading directions than TRAC boom.

We believe our efforts provide guidelines for engineers and scientists attempting to design CTLTs with desirable bending resistance performance. We also note that the nonlinear buckling simulation scheme proposed in this work is readily extended for analyzing and optimizing the bending performance of other kinds of thin-walled composite deployable structures [44, 60, 61, 62].The paradigm of parameterizing cross-section shape and then constructing a design space in an automatic way can also be developed for optimal design of other thin-walled slender structures.

Acknowledgement

This research is supported by National Natural Science Foundation of China (grant 11972277). Q. J. also acknowledges the support from Shanghai Rising-Star Program (19QB1404000).

Data availability

The Abaqus scripts used for the numerical analyses that support the findings of this study are available from the corresponding author upon reasonable request.

References

  1. [1]B. Rennie, New closed tubular extendible boom (1967).
  2. [2]M. Aguirre, The collapsible tube mast (ctm), in: Second European Space Mechanisms and Tribology Symposium, Vol. 231, 1985, pp. 75–81.
  3. [3]M. Leipold, H. Runge, C. Sickinger, Large sar membrane antennas with lightweight deployable booms, in: 28th ESA Antenna Workshop on Space Antenna Systems and Technologies, ESA/ESTEC, European Space and Technology Research Centre Noordwijk, The Netherlands, 2005, p. 8.
  4. [4]J. Block, M. Straubel, M. Wiedemann, Ultralight deployable booms for solar sails and other large gossamer structures in space, Acta astronautica 68 (7-8) (2011) 984–992.
  5. [5]M. Hillebrandt, S. Meyer, M. Zander, M. Straubel, C. Hühne, The boom design of the de-orbit sail satellite, in: European Conference on Spacecraft Structures, Materials and Mechanical Testing, 2014.
  6. [6]W. K. Belvin, M. Straubel, W. K. Wilkie, M. Zander, J. M. Fernandez, M. Hillebrandt, Advanced deployable structural systems for small satellites (2016).
  7. [7]U. Geppert, B. Biering, F. Lura, J. Block, M. Straubel, R. Reinhard, The 3-step dlr–esa gossamer road to solar sailing, Advances in Space Research 48 (11) (2011) 1695–1701.DOI
  8. [8]P. Seefeldt, P. Spietz, T. Sproewitz, J. T. Grundmann, M. Hillebrandt, C. Hobbie, M. Ruffer, M. Straubel, N. Tóth, M. Zander, Gossamer-1: Mission concept and technology for a controlled deployment of gossamer spacecraft, Advances in Space Research 59 (1) (2017) 434–456.DOI
  9. [9]J. M. Fernandez, Advanced deployable shell-based composite booms for small satellite structural applications including solar sails (2017).
  10. [10]J. M. Fernandez, G. Rose, O. R. Stohlman, C. J. Younger, G. D. Dean, J. E. Warren, J. H. Kang, R. G. Bryant, K. W. Wilkie, An advanced composites-based solar sail system for interplanetary small satellite missions, in: 2018 AIAA Spacecraft Structures Conference, 2018, p. 1437.DOI
  11. [11]J. A. Firth, M. R. Pankow, Advanced dual-pull mechanism for deployable spacecraft booms, Journal of Spacecraft and Rockets 56 (2) (2019) 569–576.
  12. [12]J. A. Firth, M. R. Pankow, Minimal unpowered strain-energy deployment mechanism for rollable spacecraft booms: Ground test, Journal of Spacecraft and Rockets 57 (2) (2020) 346–353.
  13. [13]Y. Long, O. Rique Garaizar, J. M. Fernandez, A. Bergan, W. Yu, Multiscale simulation of deployable composite structures, in: AIAA SciTech 2021 Forum, 2021, p. 0199.DOI
  14. [14]O. R. Stohlman, M. E. Zander, J. M. Fernandez, Characterization and modeling of large collapsible tubular mast booms, in: AIAA Scitech 2021 Forum, 2021, p. 0903.DOI
  15. [15]J.-C. Angevain, A. Ihle, G. Rodrigues, J. Santiago-Prowald, Large deployable spaceborne reflector antennas in europe: Progress status and perspectives, in: 2019 13th European Conference on Antennas and Propagation (EuCAP), IEEE, 2019, pp. 1–5.
  16. [16]O. Yoshiro, J. Reveles, V. Fraux, D.-J. Ashley, Deployable wrapped rib assembly, uS Patent App. 16/348,390 (Nov. 28 2019).
  17. [17]A. d. S. Curiel, P. Whittaker, R. Bird, A. Haslehurst, P. Nejadi, I. Victoria, A. Cawthorne, C. Underwood, M. Sweeting, Synthetic aperture radar on a nanosatellite-is it possible?, in: Proceedings of the 12th IAA Symposium on Small Satellites for Earth Observation, International Academy of Astronautics (IAA), 2019.
  18. [18]Y. Hu, W. Chen, J. Gao, J. Hu, G. Fang, F. Peng, A study of flattening process of deploy- able composite thin-walled lenticular tubes under compression and tension, Composite Structures 168 (2017) 164–177.DOI
  19. [19]W. Chen, G. Fang, Y. Hu, An experimental and numerical study of flattening and wrapping process of deployable composite thin-walled lenticular tubes, Thin-Walled Structures 111 (2017) 38–47.DOI
  20. [20]J. Bai, J. Xiong, J. Gao, X. Yi, Analytical solutions for predicting in-plane strain and interlaminar shear stress of ultra-thin-walled lenticular collapsible composite tube in fold deformation, Composite Structures 97 (2013) 64–75.DOI
  21. [21]J. Bai, R. Shenoi, J. Xiong, Thermal analysis of thin-walled deployable composite boom in simulated space environment, Composite Structures 173 (2017) 210–218.DOI
  22. [22]J.-B. Bai, D. Chen, J.-J. Xiong, R. A. Shenoi, Folding analysis for thin-walled deployable composite boom, Acta Astronautica 159 (2019) 622–636.DOI
  23. [23]C. Sickinger, L. Herbeck, E. Breitbach, Structural engineering on deployable cfrp booms for a solar propelled sailcraft, Acta Astronautica 58 (4) (2006) 185–196.
  24. [24]J. Bai, J. Xiong, Temperature effect on buckling properties of ultra-thin-walled lenticular collapsible composite tube subjected to axial compression, Chinese Journal of Aeronautics 27 (5) (2014) 1312–1317.DOI
  25. [25]Y. Hu, W. Chen, R. Li, G. Fang, Mechanical characteristics of deployable composite thin-walled lenticular tubes, Composite Structures 153 (2016) 601–613.DOI
  26. [26]A. Shirkavand, F. Taheri-Behrooz, M. Omidi, Orientation and size effect of a rectangle cutout on the buckling of composite cylinders, Aerospace Science and Technology 87 (2019) 488–497.DOI
  27. [27]B. Devarajan, R. K. Kapania, Thermal buckling of curvilinearly stiffened laminated com- posite plates with cutouts using isogeometric analysis, Composite Structures 238 (2020) 111881.DOI
  28. [28]B. Brendel, E. Ramm, Linear and nonlinear stability analysis of cylindrical shells, Computers & structures 12 (4) (1980) 549–558.DOI
  29. [29]A. N. Kounadis, An efficient simplified approach for the nonlinear buckling analysisof frames, AIAA journal 23 (8) (1985) 1254–1259.DOI
  30. [30]C. Wu, J. Arora, Design sensitivity analysis and optimization of nonlinear structuralre- sponse using incremental procedure, AIAA journal 25 (8) (1987) 1118–1125.DOI
  31. [31]C. Wu, J. Arora, Design sensitivity analysis of non-linear buckling load, Computational Mechanics 3 (2) (1988) 129–140.DOI
  32. [32]E. Lindgaard, E. Lund, Nonlinear buckling optimization of composite structures, Com- puter methods in applied mechanics and engineering 199 (37-40) (2010) 2319–2330.DOI
  33. [33]E. Lindgaard, E. Lund, Optimization formulations for the maximum nonlinear buckling load of composite structures, Structural and Multidisciplinary Optimization 43 (5) (2011) 631–646.
  34. [34]E. Lindgaard, E. Lund, A unified approach to nonlinear buckling optimization of compos- ite structures, Computers & Structures 89 (3-4) (2011) 357–370.DOI
  35. [35]E. Lindgaard, E. Lund, K. Rasmussen, Nonlinear buckling optimization of composite structures considering “worst” shape imperfections, International Journal of Solids and Structures 47 (22-23) (2010) 3186–3202.DOI
  36. [36]T.-T. Nguyen, S. Ronel, M. Massenzio, E. Jacquelin, K. L. Apedo, H. Phan-Dinh, Numerical buckling analysis of an inflatable beam made of orthotropic technical textiles, Thin-Walled Structures 72 (2013) 61–75.DOI
  37. [37]K. Liang, Q. Sun, Y. Zhang, Nonlinear buckling analysis of variable stiffness composite plates based on the reduced order model, Composite Structures 206 (2018) 681–692.DOI
  38. [38]K. Liang, P. Hao, B. Wang, Q. Sun, A novel reduced-order modeling method for nonlinear buckling analysis and optimization of geometrically imperfect cylinders, International Journal for Numerical Methods in Engineering 122 (6) (2021) 1456–1475.DOI
  39. [39]D. A. Spencer, L. Johnson, A. C. Long, Solar sailing technology challenges, Aerospace Science and Technology 93 (2019) 105276.DOI
  40. [40]J. A. Firth, et al., Deployment mechanisms for rollable spacecraft booms. (2019).
  41. [41]G. Sullivan, J. R. Blandino, T. Hayes, Boom deployment mechanism for cubesats, in: AIAA Scitech 2020 Forum, 2020, p. 1672.DOI
  42. [42]T. W. Murphey, D. Turse, L. Adams, Trac boom structural mechanics, in: 4th AIAA Spacecraft Structures Conference, 2017, p. 0171.DOI
  43. [43]C. Leclerc, L. L. Wilson, M. A. Bessa, S. Pellegrino, Characterization of ultra-thin composite triangular rollable and collapsible booms, in: 4th AIAA Spacecraft Structures Conference, 2017, p. 0172.DOI
  44. [44]M. Bessa, S. Pellegrino, Design of ultra-thin shell structures in the stochastic post-buckling range using bayesian machine learning and optimization, International Journal of Solids and Structures 139 (2018) 174–188.DOI
  45. [45]C. Leclerc, S. Pellegrino, Nonlinear elastic buckling of ultra-thin coilable booms, International Journal of Solids and Structures 203 (2020) 46–56.DOI
  46. [46]K. Cox, K. A. Medina, Scalability of triangular rollable and collapsible booms, in: AIAA Scitech 2019 Forum, 2019, p. 2026.DOI
  47. [47]A. Lee, J. M. Fernandez, Mechanics of bistable two-shelled composite booms, in: 2018 AIAA Spacecraft Structures Conference, 2018, p. 0938.DOI
  48. [48]A. J. Lee, J. M. Fernandez, Inducing bistability in collapsible tubular mast booms with thin-ply composite shells, Composite Structures 225 (2019) 111166.DOI
  49. [49]J. E. Salazar, J. M. Fernandez, Experimental characterization of the dimensional stability of deployable composite booms during stowage, in: AIAA SciTech 2021 Forum, 2021, p. 0195.DOI
  50. [50]W. K. Wilkie, J. M. Fernandez, O. R. Stohlman, N. R. Schneider, G. D. Dean, J. H. Kang, J. E. Warren, S. M. Cook, P. L. Brown, T. C. Denkins10, et al., An overview of the nasa advanced composite solar sail (acs3) technology demonstration project.
  51. [51]F. Royer, S. Pellegrino, Ultralight ladder-type coilable space structures, in: 2018 AIAA Spacecraft Structures Conference, 2018, p. 1200.DOI
  52. [52]K.-J. Bathe, E. N. Dvorkin, On the automatic solution of nonlinear finite element equations, Computers & Structures 17 (5-6) (1983) 871–879.
  53. [53]P. Wriggers, Nonlinear finite element methods, Springer Science & Business Media, 2008.
  54. [54]N.-H. Kim, Introduction to nonlinear finite element analysis, Springer Science & Business Media, 2014.
  55. [55]N. L. Pedersen, P. Pedersen, Buckling load optimization for 2d continuum models, with alternative formulation for buckling load estimation, Structural and Multidisciplinary Optimization 58 (5) (2018) 2163–2172.DOI
  56. [56]L. Zheng, Wrinkling of dielectric elastomer membranes, California Institute of Technology, 2009.DOI
  57. [57]O. K. Fajuyitan, A. J. Sadowski, M. A. Wadee, J. M. Rotter, Nonlinear behaviour of short elastic cylindrical shells under global bending, Thin-Walled Structures 124 (2018) 574–587.DOI
  58. [58]K. K. Yadav, S. Gerasimidis, Instability of thin steel cylindrical shells under bending, Thin-Walled Structures 137 (2019) 151–166.DOI
  59. [59]T.-U. Lee, X. Yang, J. Ma, Y. Chen, J. M. Gattas, Elastic buckling shape control of thin-walled cylinder using pre-embedded curved-crease origami patterns, International Journal of Mechanical Sciences 151 (2019) 322–330.DOI
  60. [60]H. Yang, L. Liu, H. Guo, F. Lu, Y. Liu, Wrapping dynamic analysis and optimization of deployable composite triangular rollable and collapsible booms, Structural and Multidisciplinary Optimization 59 (4) (2019) 1371–1383.DOI
  61. [61]H. Yang, H. Guo, Y. Wang, J. Feng, D. Tian, Analytical solution of the peak bending moment of an m boom for membrane deployable structures, International Journal of Solids and Structures 206 (2020) 236–246.DOI
  62. [62]H. Yang, H. Guo, R. Liu, S. Wang, Y. Liu, Coiling and deploying dynamic optimization of a c-cross section thin-walled composite deployable boom, Structural and Multidisciplinary Optimization (2019) 1–8.DOI

Paper details

vlnc:W_kh6hmssf
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents