Professional Documents
Culture Documents
ScienceDirect
Weili Jiang a,*,1, Fengmin Du b,1, Klaus Drechsler a, Jinyang Zheng c,**
a
Chair of Carbon Composites, Technical University of Munich, 85748 Garching bei München, Germany
b
Laboratory for Physical Chemistry, ETH Zurich, 8093 Zurich, Switzerland
c
Institute of Chemical Machinery and Process Equipment, Zhejiang University, 310027 Hangzhou, China
Article history: Strict requirements for cost and reliability of compressed gaseous pressure vessels as on-
Received 24 October 2022 board hydrogen storage necessitates in-depth understanding on the mechanical responses
Received in revised form of the composite structure. General numerical methodology to assess this involves a two-
17 January 2023 step setup: composite layup generation and mechanical property calculation, which be-
Accepted 19 January 2023 comes highly time and resource-consuming, particularly if multiple designs are scanned
Available online 8 February 2023 for optimization. This study presents a multi-functional analytical tool, combining both
aforementioned steps in a single modeling framework. The model enables quick prediction
Keywords: of the strain/stress distribution in correlation with uncomplicated inputs describing the
High pressure gaseous hydrogen layup design, i.e. stacking sequence with desired winding angles. Additionally, the me-
storage chanical evaluation specially focuses on the dome region whose calculation has hardly
Type IV hydrogen tank been inspected analytically in the literature. The results are validated against a three-
Carbon composites dimensional finite element calculation implemented in ANSYS Workbench, showing quali-
Layup design tative and quantitative agreement. Overall, the study paves the way for composite tank
Analytical methodology
* Corresponding author.
** Corresponding author.
E-mail addresses: weili.jiang@tum.de (W. Jiang), jyzh@zju.edu.cn (J. Zheng).
1
These authors contributed equally to this work.
https://doi.org/10.1016/j.ijhydene.2023.01.223
0360-3199/© 2023 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
17566 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 8 ( 2 0 2 3 ) 1 7 5 6 5 e1 7 5 7 6
v2 c D11 vD11 r3 vc vεs vεq A11 D11 B211
D11 þ þ B11 þ B11 þ B12 G2 ¼ (27)
vs2 r3 vs r2 vs vs vs A11
B12 vD12 =vs D12 vr3 =vs þ D12 D22 Following the evaluation of Ms, the transverse force Q can
¼ c
r3 r23 be expressed as:
r3 B11 B21 vB11
þ A11 εs v v2 εq B11 pR
r2 r3 vs Q¼ G1 $εq R$G2 2 þ (28)
vs vs A11 2
r3 B12 B22 vB12 r3
þ A12 εq N*s (19)
r2 r3 vs r2 Analogously, Nq can be expressed by:
The last equation of the ODE system can be taken from the
v2 εq A12 pR
geometrical constraint [46]: Nq ¼ D1 εq R$D2 þ (29)
vs2 A11 2
vεq εq εs c Again, the auxiliary variables D1 and D2 as follows:
þ ¼ (20)
vs r3 r2
A11 ðA22 B22 =RÞ A12 ðA12 B12 =RÞ
Subject to appropriate boundary conditions, the three D1 ¼ (30)
A11
variables c, εs and εq can be calculated based on equations (17),
(19) and (20). Note that the so obtained ODE system represents
A11 B12 A12 B11
a complete analytical description of the mechanical proper- D2 ¼ (31)
A11
ties on the dome for a composite pressure vessel, which, to
our best knowledge, has never been addressed in the litera- Inputting the expressions of Nq and Q into equation (23), we
ture before. The solving procedure is further elaborated in finally obtain the following fourth-order ODE with respect to
Implementation. The solution of the ODE system can further εq:
be applied for evaluating the profiles of e.g. Ms, Q, Nq, etc. 2
v2 G1 D1 vG1 vεq v2 G2 v εq
εq þ 2 þ G1 R 2 þ D2
vs 2
R vs vs vs vs2
Force equilibrium on cylindrical body 2
vG2 v εq
3
v εq
4
A12 R v B11
In contrast to the dome region, on the cylindrical body, the 2R RG2 4 þ p 1 þ $ 2 ¼0 (32)
vs vs 3
vs 2A11 2 vs A11
curvatures can be expressed by equation (21):
Equation (32) therefore represents the mechanical re-
v2 εq εq sponses on the cylindrical body. Analogous to the ODE system
ks ¼ R$ ; kq ¼ (21)
vs2 R
in the previous section, by applying appropriate boundary
At the same time, the force balance is according to equa- conditions, εq can be obtained from the above ODE (equation
tion 22 and 23: (32)), which also allows the evaluation of other variables, such
as εs and Q. The coupling procedure of the ODEs on the cy-
R
Ns ¼ N*s ¼ p$ (22) lindrical body and on the dome is presented in
2
Implementation.
vQ Nq
þp¼0 (23)
vs R Methodology
Applying the ABD relationship in equation (2) to equation
(22) yields the following relationship: Implementation
1 pR v2 εq εq The differential equations were numerically solved within
εs ¼ A12 εq þ B11 R 2 þ B12 (24)
A11 2 vs R MATLAB® 2018b, applying built-in solvers for ODE systems
(bvp4c). All derivatives of variables (e.g., ABD, r3, G, D, refer to
Equation (21) describe the curvatures and εs as function of
the Theory section) were evaluated by numerical differentia-
εq. With these relationships, it is convenient to express
tion. Overall computational time of the model is around 10 s
equation (23) only with εq. In order to evaluate the transverse
without parallel computing.
force resultant Q, we first calculate the moment Ms:
In the following, we present the boundary and coupling
B11 pR v2 εq εq v2 εq D12 conditions of the model. It shall be emphasized at this point
Ms ¼ A12 εq þ B11 R 2 þ B12 þ B12 εq RD11 2 εq
A11 2 vs R vs R that the coupling of the two ODE (systems) which were pre-
v εq B11 pR
2
sented in the Theory section is of great significance, as the
¼ G1 $εq R$G2 2 þ
vs A11 2 junction part is well-known to induce “discontinuity stresses”
(25) [46], primarily due to the sudden change in curvature.
Where for better readability, we define two auxiliary vari- At the dome polar opening (s boundary), we consider that
ables G1 and G2 as: the displacement in the radial direction (w) and the rotation
(c) are zero due to the connection to the boss part, which was
A11 ðB12 D12 =RÞ B11 ðA12 B12 =RÞ assumed to be perfectly rigid. This gives the following
G1 ¼ (26)
A11 boundary conditions:
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 8 ( 2 0 2 3 ) 1 7 5 6 5 e1 7 5 7 6 17571
vðr3 εq Þ
εs ¼ 0 ðPolar openingÞ (36)
vs
Results and discussion
It shall be noted that the mechanical stresses at the polar
opening are not explicitly defined, but rather become the Stress resultant
calculation results, and are therefore dependent on the rest of
the model. The CLT requires inputs of the stress resultant (N), the values
At the middle of the cylindrical body (þs boundary), the of which in axial and circumferential direction are depicted in
symmetry condition indicates zero-gradient for the strains: Fig. 5 (solid lines). It shall be additionally noted that N is equal
to the integral of the stress in the corresponding direction (s or
vεs vεq
¼ 0; ¼ 0 ðCylinder middleÞ (37) q) with respect to the thickness direction.
vs vs
To understand the profiles shown in Fig. 5, the predictions
With the boundary conditions obtained, the next step is to
of the idealised membrane theory (N*s , N*q ) from equation (6)
find the coupling conditions of both ODE (systems) at the
are also illustrated in Fig. 5 as dotted lines. Evidently, the
junction. We note that the displacements (u, w) must be
membrane theory recovers the well-known Barlow's formulae
continuous at the junction. Therefore, following equation (35),
on the cylindrical body, where N*q has a doubled value
εq is continuous yet no direction relationship can be found for
its meridional counterpart εs (du/ds is not necessarily contin- compared to its axial counterpart, N*s .
uous in equation (34)). Instead, we consider the continuity of At the junction, the membrane theory predicts a smooth
transverse force (Q) and moment (Ms). For the solver, the transition of N*s (note that r2 / R at the junction), yet a
following conditions were directly applied: considerable jump of N*q depending on the value r1. Even
negative values of Nq may occur locally if the dome is of rather
Q ¼ Qþ ; εq; ¼ εq;þ ðCoupling at JunctionÞ (38)
flat nature (r2/r1 > 2) [50], which is exactly the case as shown in
here, the index represents the dome region immediately Fig. 5. Overall, the sudden change of N*q from cylinder to dome
adjacent to the junction, while index þ the cylindrical body. region induces the well-known effects of “discontinuity
To meet this requirement of Ms-continuity, the value of εq stresses”, which leads to significant deviations of the mem-
at the junction was iterated until the condition as described by brane theory from the reality due to the neglecting of realistic
equation (39) is met. shear/bending effects.
Fig. 4 e Liner geometry and dimension applied in the present study. Winding angle a is defined as the angle between roving
and meridional direction.
17572 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 8 ( 2 0 2 3 ) 1 7 5 6 5 e1 7 5 7 6
(S3)(Supplementary Material) before being implemented in derivative of ABD-entries plays a significant role on the me-
the look-up-table. chanical response. The strongly varying ABD values could
Meshing of the model was performed with higher order 3D have induced numerical inaccuracies and explain the differ-
SOLID187 elements, which is well suited for modeling of ences observed in Fig. 7.
irregular meshes, especially in the case of helical layers with
varying thickness on the dome [51]. Overall, the model con- Application to a different layup design
sists of 584 944 elements, with a element size of 3 mm. The
model was constrained by fixation on only one side, whereas For a further confirmation of the model accuracy, we inves-
the other side was set as free. The simulation time was tigate a significantly different composite layup with both
approximately 8 h with six-core parallel computing. Fig. 7 analytical and FE simulation approaches. The layer stacking
compares the stress distribution of various layers from the sequence and estimated stress are demonstrated in Fig. 8.
developed analytical methodology with that from the FE Analogous to Fig. 7, the stress profiles were averaged within
simulations. For a better visualization, the stress profiles were each layer group with the same winding angle for visualiza-
averaged within each layer group with the same winding tion purposes.
angle. This scenario was selected to emphasize the mechanical
Overall, the calculated stresses using the presented model response near the transition from dome to cylinder. In the
shows excellent qualitative and good quantitative alignment layup design, the composite layers over the junction was
with the 3D FEM results. The occurrence of maximum stress intentionally designed to be relatively weak, applying only
within all layers was accurately predicted (inner hoop layers, four layers of 54 helical layers. Thus, it is not surprising that
blue curves in Fig. 7), which gives indication on the maximum the overall stresses become the most critical near the junction
pressure-bearing capacity of the current design. Within the part, particularly in the case of “responsible” layers, i.e. the
dome region, the findings that the helical layers are respon- inner/outer hoop and 54 helical layers. The peak for the he-
sible for less stresses with increasing winding angle, are lical layers as well as the local maximum for the hoop layers
confirmed by the FE results. are very well captured by both modeling approaches, which
Nonetheless, some quantitative discrepancies between the also show excellent quantitative agreement here. Analogously
FE and our analytical model can still be observed, for instance to the previous design, this high “discontinuity stresses”
regarding the peak positions of stresses in individual layers on observed near the junction is dampened on the cylindrical
the dome. The discrepancies could be explained by the dif- body, reaching a stable and uniform distribution within a
ference in smoothing algorithm of our layup generation tool distance of around 100 mm. Here, the predicted trend of the
compared to the commercial pre-treatment module in ANSYS® analytical model agrees reasonably well with the FE results.
ACP, which further leads to small discrepancies of individual On the dome, both modeling approaches yielded nearly
layer thicknesses. At the same time, it shall be noted that this identical qualitative trend of stress profile on all helical layers.
region involves a rapid change of layer structure (with respect For instance, at the axial coordinate of around 30 mm
to the axial coordinate), which results in strongly varying ABD (30 mm from the junction on the dome), a peak may be
properties of the composite laminate. As it was demonstrated observed for 15 -helical layer, as well as two troughs for 30
in the governing ODE system for the dome mechanics, the first and 54 layers.
Nonetheless, some quantitative discrepancy up to around
20% may still be observed in Fig. 8 within both dome and
cylinder regions. the stress distributions manifest a quanti-
tative discrepancy. As noted in the previous section, differ-
ences in the layup generation algorithm could be one
explanation for this discrepancy. Since FE simulations are
typically also subject to certain error rate, we consider the
observed quantitative differences between the two modeling
approaches acceptable. It shall be emphasized that this ac-
curacy of the presented methodology was achieved while
reducing the computational cost by a factor of > 10 000 times
compared to 3D FE simulation (~10 s serial vs. ~8 h parallel on
six cores).
Conclusion
serves the purpose to investigate intra-laminar failure of [13] Mair GW, Thomas S, Schalau B, Wang B. Safety criteria for
composite laminates as well as the initiation and evolution of the transport of hydrogen in permanently mounted
delamination phenomenon between the layers [32]. composite pressure vessels. Int J Hydrogen Energy
2021;46(23):12577e93.
[14] Hua TQ, Ahluwalia RK, Peng J-K, Kromer M, Lasher S,
McKenney K, Law K, Sinha J. Technical assessment of
Declaration of competing interest compressed hydrogen storage tank systems for automotive
applications. Int J Hydrogen Energy 2011;36(4):3037e49.
The authors declare that they have no known competing [15] Cohen D, Mantell SC, Zhao L. The effect of fiber volume
financial interests or personal relationships that could have fraction on filament wound composite pressure vessel
appeared to influence the work reported in this paper. strength. Compos B Eng 2001;32(5):413e29.
[16] Park JH, Hwang JH, Lee CS, Hwang W. Stacking sequence
design of composite laminates for maximum strength using
genetic algorithms. Compos Struct 2001;52(2):217e31.
Appendix A. Supplementary data [17] Nguyen BN, Roh HS, Merkel DR, Simmons KL. A predictive
modeling tool for damage analysis and design of hydrogen
Supplementary data to this article can be found online at storage composite pressure vessels. Int J Hydrogen Energy
https://doi.org/10.1016/j.ijhydene.2023.01.223. 2021;46(39):20573e85.
[18] Ramirez JPB, Halm D, Grandidier J-C, Villalonga S, Nony F. Bar
type iv high pressure hydrogen storage vessel
references burstesimulation and experimental validation. Int J
Hydrogen Energy 2015;40(38):13183e92. 700.
[19] Wang L, Zheng C, Luo H, Wei S, Wei Z. Continuum damage
modeling and progressive failure analysis of carbon fiber/
[1] Ajanovic A, Haas R. Prospects and impediments for hydrogen epoxy composite pressure vessel. Compos Struct
and fuel cell vehicles in the transport sector. Int J Hydrogen 2015;134:475e82.
Energy 2021;46(16):10049e58. [20] Zhang Q, Xu H, Jia X, Zu L, Cheng S, Wang H. Design of a 70
[2] Sharma P, Bera T, Semwal K, Badhe RM, Sharma A, MPa type IV hydrogen storage vessel using accurate
Ramakumar S, Neogi S. Theoretical analysis of design of modeling techniques for dome thickness prediction. Compos
filament wound type 3 composite cylinder for the storage of Struct 2020;236:111915.
compressed hydrogen gas. Int J Hydrogen Energy [21] Wang R, Jiao W, Liu W, Yang F. Dome thickness prediction of
2020;45(46):25386e97. composite pressure vessels by a cubic spline function and
[3] Sinigaglia T, Lewiski F, Martins MES, Siluk JCM. Production, finite element analysis. Polym Polym Compos
storage, fuel stations of hydrogen and its utilization in 2011;19(2e3):227e34.
automotive applications-a review. Int J Hydrogen Energy [22] Sofi T, Neunkirchen S, Schledjewski R. Path calculation,
2017;42(39):24597e611. technology and opportunities in dry fiber winding: a review.
[4] Felderhoff M, Weidenthaler C, von Helmolt R, Eberle U. Adv Manuf Polym Compos Sci 2018;4(3):57e72.
Hydrogen storage: the remaining scientific and technological [23] Hosseini SA, Scha € kel M, Baran I, Janssen H, van Drongelen M,
challenges. Phys Chem Chem Phys 2007;9(21):2643e53. Akkerman R. A new global kinematic-optical-thermal
[5] Mori D, Hirose K. Recent challenges of hydrogen storage process model for laser-assisted tape winding with an
technologies for fuel cell vehicles. Int J Hydrogen Energy application to helical-wound pressure vessel. Mater Des
2009;34(10):4569e74. 2020;193:108854.
[6] Singh R, Singh M, Gautam S. Hydrogen economy, energy, and [24] Zeng W, Hu W, Liu H, Tian H. Finite element analysis of glass
liquid organic carriers for its mobility. Mater Today Proc fiber winding molding of hdpe pressure vessel. In: Journal of
2021;46:5420e7. physics: conference series. vol. 1965. IOP Publishing; 2021,
[7] Zhou W, Wang J, Pan Z-b, Liu J, Ma L-h, Zhou J-y, Su Y-f. 012050.
Review on optimization design, failure analysis and non- [25] Quanjin M, Rejab M, Idris M, Kumar NM, Merzuki M. Robotic
destructive testing of composite hydrogen storage vessel. Int filament winding technique (rfwt) in industrial application: a
J Hydrogen Energy 2008;47(91):38862e83. review of state of the art and future perspectives. Int Res J
[8] Zheng J, Liu X, Xu P, Liu P, Zhao Y, Yang J. Development of Eng Technol 2018;5(12):1668e76.
high pressure gaseous hydrogen storage technologies. Int J [26] Liu PF, Xing LJ, Zheng JY. Failure analysis of carbon fiber/
Hydrogen Energy 2012;37(1):1048e57. epoxy composite cylindrical laminates using explicit finite
[9] Hassan IA, Ramadan HS, Saleh MA, Hissel D. Hydrogen element method. Compos B Eng 2014;56:54e61.
storage technologies for stationary and mobile applications: [27] Liu PF, Chu JK, Hou SJ, Xu P, Zheng JY. Numerical simulation
review, analysis and perspectives. Renew Sustain Energy Rev and optimal design for composite high-pressure hydrogen
2021;149:111311. storage vessel: a review. Renew Sustain Energy Rev
[10] Abohamzeh E, Salehi F, Sheikholeslami M, Abbassi R, Khan F. 2012;16(4):1817e27.
Review of hydrogen safety during storage, transmission, and [28] Rafiee R, Torabi MA, Maleki S. Investigating structural failure
applications processes. J Loss Prev Process Ind of a filament-wound composite tube subjected to internal
2021;72:104569. pressure: experimental and theoretical evaluation. Polym
[11] Nebe M, Soriano A, Braun C, Middendorf P, Walther F. Test 2018;67:322e30.
Analysis on the mechanical response of composite pressure [29] Rafiee R, Torabi MA. Stochastic prediction of burst pressure
vessels during internal pressure loading: FE modeling and in composite pressure vessels. Compos Struct
experimental correlation. Compos B Eng 2021;212:108550. 2018;185:573e83.
[12] Satyapal S, Petrovic J, Read C, Thomas G, Ordaz G. The US [30] Lin S, Yang L, Xu H, Jia X, Yang X, Zu L. Progressive damage
Department of Energy's national hydrogen storage project: analysis for multiscale modelling of composite pressure
progress towards meeting hydrogen-powered vehicle vessels based on puck failure criterion. Compos Struct
requirements. Catal Today 2007;120(3e4):246e56. 2021;255:113046.
17576 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 8 ( 2 0 2 3 ) 1 7 5 6 5 e1 7 5 7 6
[31] Garnich MR, Akula VMK. Review of degradation models for [43] Koussios S. Filament winding: a unified approach. Ph.D.
progressive failure analysis of fiber reinforced polymer thesis, Delft University of Technology; 2002.
composites. Appl Mech Rev 2009;62(1):010801. [44] Zu L, Koussios S, Beukers A. Design of filamentewound
[32] Zhang M, Lv H, Kang H, Zhou W, Zhang C. A literature review domes based on continuum theory and non-geodesic roving
of failure prediction and analysis methods for composite trajectories. Compos Appl Sci Manuf 2010;41(9):1312e20.
high-pressure hydrogen storage tanks. Int J Hydrogen Energy [45] Jois KC, Welsh M, Gries T, Sackmann J. Numerical analysis of
2019;44(47):25777e99. filament wound cylindrical composite pressure vessels
[33] Hajmohammad MH, Tabatabaeian A, Ghasemi AR, Taheri- accounting for variable dome contour. J Comp Sci
Behrooz F. A novel detailed analytical approach for 2021;5(2):56.
determining the optimal design of frp pressure vessels [46] Calladine CR. Theory of shell structures. Cambridge
subjected to hydrostatic loading: analytical model with university press; 1989.
experimental validation. Compos B Eng 2020;183:107732. [47] Jones RM. Mechanics of composite materials. CRC press;
[34] Kalali A, Hadidi-Moud S. A semi-analytical approach to 2018.
elastic-plastic stress analysis of fgm pressure vessels. J Solid [48] Mannigel M. Einfluss von Schubspannungen auf das
Mech 2013;5(1):63e73. Faserbruchgeschehen in kohlenstofffaserversta € rkten
[35] Akhtar MK, Qureshi WA, Jamshed R, Raza M. Analytical Kunststoffen (CFK): influence of shear stresses on the fibre
modeling and numerical analysis of the effect of mosaic failure behaviour in carbon fibre reinforced plastics (CFRP).
patterns on composite pressure vessels with dome. J 2007. Mainz.
Reinforc Plast Compos 2021;40(17e18):621e31. [49] Hinton M, Soden P, Kaddour A-S. The world-wide failure
[36] Ramos I, Ho Park Y, Ulibarri-Sanchez J. Analytical and exercise: its origin, concept and content. In: Failure criteria in
numerical studies of a thick anisotropic multilayered fiber- fibre-reinforced-polymer composites; 2004. p. 2e28. Ch. 1.1.
reinforced composite pressure vessel. J Pressure Vessel [50] Wells A. On the solution of beam-on-elastic-foundation
Technol 2019;141(1):011203. problems by means of a mechanical analogue. Proc Inst
[37] Altenbach H, Altenbach JW, Kissing W. Mechanics of Mech Eng 1950;163(1):307e10.
composite structural elements. Springer; 2018. [51] Madenci E, Guven I. The finite element method and
[38] Camilleri D, Ellul B, Muscat M. Design-by-analysis methods for applications in engineering using ANSYS®. Springer; 2015.
asymmetric or unbalanced cylindrical composite pressure [52] Son D-S, Chang S-H. Evaluation of modeling techniques for a
vessels. In: Pressure vessels and piping conference. vol. 46001. type III hydrogen pressure vessel (70 mpa) made of an
American Society of Mechanical Engineers; 2014, V003T03A011. aluminum liner and a thick carbon/epoxy composite
[39] Nebe M, Asijee T, Braun C, van Campen J, Walther F. for fuel cell vehicles. Int J Hydrogen Energy
Experimental and analytical analysis on the stacking 2012;37(3):2353e69.
sequence of composite pressure vessels. Compos Struct [53] Kathavate V, Pawar D, Bagal N, Adkine A. Progressive failure
2020;247:112429. analysis of fiber reinforced polymer matrix composites.
[40] Zu L, Xu H, Wang H, Zhang B, Zi B. Design and analysis of Mater Today Proc 2020;22:1524e34.
filament-wound composite pressure vessels based on non- [54] Leh D, Saffre P, Francescato P, Arrieux R, Villalonga S. A
geodesic winding. Compos Struct 2019;207:41e52. progressive failure analysis of a 700-bar type IV hydrogen
[41] van Hien D, Thanh TN, Lam VT, Van TTT. Design of planar composite pressure vessel. Int J Hydrogen Energy
wound composite vessel based on preventing slippage 2015;40(38):13206e14.
tendency of fibers. Compos Struct 2020;254:112854. [55] Suryan A, Kim HD, Setoguchi T. Three dimensional
[42] Gramoll K, Onoda J, Namiki F. Dome thickness of filament numerical computations on the fast filling of a hydrogen
wound pressure vessels. Trans Jpn Soc Aeronaut Space Sci tank under different conditions. Int J Hydrogen Energy
1990;33(100):66e79. 2012;37(9):7600e11.