Fatigue Performance Evaluations of Vehicle Toroidal ...akademikpersonel.kocaeli.edu.tr/ykisioglu/sci/ykisioglu04.04.2017... · experimental and finite element analysis (FEA) ...

Fuat Kartal1Engineering and Architecture Faculties,

Mechanical Engineering Department,

Kastamonu University,

Kuzeykent/Kastamonu 37100, Turkey

e-mail: [email protected]

Yasin KisiogluBiomechanics,

Faculty of Technology,

Biomedical Engineering,

Kocaeli University,

Kocaeli 41380, Turkey,

e-mail: [email protected]

Fatigue PerformanceEvaluations of VehicleToroidal Liquefied PetroleumGas Fuel TanksIn this study, fatigue performances of the vehicle toroidal liquefied petroleum gas (LPG)fuel tanks were examined to estimate the fatigue life and its failure locations using bothexperimental and finite element analysis (FEA) methods. The experimental investigationsperformed as accelerated fatigue tests were carried out using a hydraulics test unit inwhich the tanks were internally pressurized by hydraulic oil. The LPG tanks were sub-jected to repeated cyclic pressure load varying from zero to service pressure (SP) of thetank. The computerized FEA modeling of these tanks were developed in three-dimensional (3D) form using nonuniform geometrical parameters and nonlinear materialproperties. These models were also subjected to zero-based high cycle fatigue pressureload considering the stress life approach. The FEA modeling process was also simulatedin nonhomogeneous material conditions. Therefore, the fatigue life performance andfailure location of the toroidal LPG fuel tanks were predicted using the computer-aidedsimulations and compared with the experimental results. [DOI: 10.1115/1.4035976]

Keywords: vehicle LPG tanks, LPG fuel tanks, fatigue life, fatigue failure location, stresslife fatigue

1 Introduction

Liquefied petroleum gas is commonly used as an alternativefuel for internal combustion engines of vehicles in Turkey andEurope. The LPG is stored and transported based on the rules ofthe Turkish Standard Institute (TS) and Economic Commissionfor Europe Regulation (ECE-R). The LPG fuel tanks also knownas LPG tanks approved by these regulations are commerciallyfilled and used in the vehicle industry. About 75,000 of thesetanks including cylindrical and torispherical (toroidal) shapes aredesigned and manufactured annually, in Turkey, based on ECE-R67 (EN 12805) [1] in Europe and TS 12095-1 [2] in Turkey.They are also called as low-pressure cylinders since their service(working) pressure is lower than 3.44 MPa (500 psi) [3,4]. Theyare equipped with a refillable two-way hermetic valve and are pro-duced as LPG containers having water capacities ranging from 32to 80 l.

The specifications for these tanks are generally defined by theTS and ECE-R Codes. The SP is the working (or operating) pres-sure where the tanks are filled and used in related applications.The test pressure (TP) is a given design pressure by the Codesthat is applied and removed without any design failure or plasticdeformation [1–3]. Finally, the burst pressure (BP) is the maxi-mum pressure where the tank can hold without burst. Accordingto the rules, the total burst volume expansion must exceed 20%of the initial volume [1,2]. Based on these circumstances, the SPand TP for these tanks are defined by the regulations, but thefatigue performance under repeated cyclic load should be alsoclarified.

The available literature is so limited regarding the design andanalysis of the vehicle LPG fuel tanks, but typical studies areappeared on the topic of buckling and stability optimizations forthe shells and relative components using numerical methods. Sev-eral papers may be found on fatigue life of the cylindrical shells

by analytical formulations considering ideal shells with perfectgeometry and specific boundary conditions. The burst pressuresand failure locations of DOT-39 refrigerant cylinders [4], vehicletoroidal LPG fuel tanks [4], vehicle cylindrical LPG fuel tanks[5], and burst parameters of the cylindrical LPG fuel tanks used inhome applications (12 kg capacity) [6] were presented using bothexperimental and FEA approaches. The FEA of a fuel-cellvehicle’s composite high-pressure hydrogen storage vessel [7],optimization of the fiber trajectories and suitable winding patternsof filament-wound toroidal storage tanks [8], and effect of com-posite flaws on fatigue life of the high pressure vessels for naturalgas vehicles [9] were investigated.

Optimum end-closure design for DOT-39 cylinders [10], effectsof composite flows on fatigue life of high pressure vessels for nat-ural gas vehicles [11], and fatigue life of pressure vessel with noz-zles designed based on the ASME and VSR 1995 codes [10] wereinvestigated numerically. Fatigue analysis of nonwelded pressurevessels with screwed in ends used for gas transportation [12] anddesign cycles for the wire winding of high pressure vessels toimprove the fatigue life [13] were calculated. Fitness-for-serviceassessment of pressure vessels and piping systems based on API579 code [14] and statistical lifetime of pressure vessel [15] weredeveloped using FEA method. The failure modes for the design ofthe hydrogen storage pressure vessels [16] and thermo mechanicalcyclic loading properties for polymer composite pressure vessel[17] were proposed. The vibration analysis of toroidal shells todetermine the free vibration characteristics [18], development ofnongeodesic equations for an axisymmetric body of overwoundtoroidal hydrogen storage tank [19], and solution of the statics anddynamics problems of a metallic ovaloid toroidal tank pierced bya circular nozzle [20] were studied using analytical and FEMsimulations.

The fatigue strength of automotive engine cylinder heads con-sidering the Dang Van failure criterion [21] and the structuralstress analysis of pressure vessels subjected to real cyclic load toobtain the stress field [22] were calculated. Effects of microstruc-ture fatigue crack growth in a thin-walled composite-pressure ves-sel [23] and fatigue damage and fatigue curves for weldedpressure vessel components [24] were investigated numerically.

1Corresponding author.Contributed by the Pressure Vessel and Piping Division of ASME for publication

in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received December 13,2015; final manuscript received December 25, 2016; published online March 17,2017. Assoc. Editor: Hardayal S. Mehta.

Journal of Pressure Vessel Technology AUGUST 2017, Vol. 139 / 041402-1Copyright VC 2017 by ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/936116/ on 04/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

The fatigue behavior of high strength tube steel under strain-controlled conditions by applying a cyclic strain energy density[25] and fatigue damage and life of engineering components [26]were presented. A numerical algorithm for the fatigue resistanceof hydraulic cylinders was presented to characterize the specificWohler curve [27]. Based on the proceeding review, it can be con-cluded that, no similar body of knowledge appears to be availablein the current literature for the fatigue performance life and itsfailure location for the vehicle toroidal LPG fuel tanks.

The primary aim of this study is to determine the fatigue lifeand failure location of the toroidal LPG tanks using both experi-mental and numerical techniques. The experimental studies wereperformed by an accelerated hydraulic fatigue test unit in whichthe tanks were pressurized internally with hydraulic oil to predictthe fatigue life and failure location. The cycling FEA modelingprocess was also developed to predict the fatigue life and failurelocation. The computer-aided modeling was performed and simu-lated in both nonuniform and nonhomogeneous conditions consid-ering the stress life option. The shell and weld zone materialproperties including weld zone thicknesses were investigated andused in the modeling processes. These properties were used nonli-nearly in the simulations to achieve the experimental results.

2 Design of Toroidal Liquefied Petroleum Gas Tanks

Guidelines for the design of toroidal shells can also be found inmany international codes such as BS-5500 (British Standard), theASME Boiler and pressure vessel code (Section VIII, Division 1),and DOT-39 (Department of Transportation’s, Washington, DC)[4]. These rules are restricted mostly to the load carrying capacityunder internal pressure. The production process utilizes some of themost modern machinery and testing technology, e.g., consideringIndian, British, ISO, DOT standards, based on customer

requirements. However, the LPG tanks are usually designed as thin-walled vessel and manufactured according to the restrictions [1,2]considering the SP and TP specifications. Based on these regula-tions, the BP is at least (9/4)�TP which is set at between 1.2 and2�SP [1,2]. These pressure specifications are generally used for theLPG tanks as shown in Fig. 1. As seen, they are also equipped withan inlet nozzle (kloret) and a tank label welded to the tank body.

The torispherical LPG tanks are usually manufactured withintwo different groups, which are classified by their water capaci-ties; 32 l and 40 l and having nominal wall thickness of 3 mm. Inthe present study, the LPG tanks with a 40 l water capacity weresubjected to accelerated fatigue tests. The toroidal tanks are con-sisting of two main parts, two torispherical shells and welded cir-cumferentially at inside and outside of the toroidal geometry asshown in Fig. 1. As seen from the figure, some design parametersare also shown such as wall thickness (t) and (tr), knuckle radii ofcurvatures (Ri), and (R2), and toroidal radius of (R).

The toroidal LPG tanks were manufactured from Erdemir-6842steel using the welding and spinning processes for the toroidalshells. Erdemir Steel Company is an international steel manufac-turer in Zonguldak, Turkey. The Erdemir-6842 steel is hot-rolledand is of low carbon content (0.18% C); steel making is a ductilematerial suitable for a cold forming process used to fabricate thesetanks.

3 Experimental Fatigue Failure Tests

The experimental fatigue life investigations of the torus tankswere carried out at the R&D laboratory of the vehicle LPG tankmanufacturer, Step LPG Inc., Konya, Turkey. In order to performthe tests, a programmable logic controller (PLC) controlled servo-hydraulic fatigue test unit was designed and established as seen inFig. 2(a). The tank specimens were selected randomly from the

Fig. 1 The vehicle toroidal LPG fuel tank and its design parameters

041402-2 / Vol. 139, AUGUST 2017 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/936116/ on 04/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

manufacturing stacks and completely filled with hydraulic oil.The oil pressure was managed by a double-acting hydraulic pumpwith the power and frequency of 3.5 kW and 5 Hz, respectively.These tanks were placed horizontally in the experiments and airwas vented during first filling. The hydraulic circuit diagram ofthe test unit is also given in Fig. 2(b). As seen, a servocontrolled4� 3 way hydraulic valve shown schematically with the numberof 6, was used to pressurize the multiple tanks in parallel.

The fatigue tests were performed at room temperature andatmospheric conditions with a constant hydraulic oil temperatureusing a cooling fan-system. Based on the Codes [1,2] require-ments, the fluid to be used in the fatigue tests should be noncorro-sive and nonabrasive, so that the hydraulic oil was used to meetthen requirement. In addition, the hydraulic pressure was appliedwithin the frequency lower than 0.25 Hz. Three tanks can beattached in parallel to the hydraulic system in the cabin and pres-surized at the same time as seen in the circuit diagram of thesetanks which can also be assembled or disassembled individuallyfrom the test in case fatigue failure happens during the test.

The torus tanks were subjected to zero-based repeated cyclicpressure load. It was assumed that these tanks were completely

filled up with the LPG and used until fully empty regularly usedfor private and/or commercial purposes in daily life. The appliedcyclic pressure loading procedure complied with the Codesrequirements. The maximum (fully pressurized until SP) and min-imum (fully empty, pressure free) cyclic pressure values wereapplied lower than 2/3 and 1/10 times the TP, respectively. Basedon these circumstances, they can be subjected to internal pressurefrom 0 to the SP value. Therefore, the loading type for the fatiguetests can be described as a zero-based constant amplitude cyclicload. According to the fatigue test requirements by the Codes, thetanks under cyclic pressure load of 60,000 cycles, any design fail-ure or plastic deformation should not be experienced. The testedtanks in this experiment did not show any design failure until80,000 cycles.

For the tanks in the accelerated fatigue tests, distribution of themeasured number of loading cycles for total 13-tank is shown inFig. 3 as a function of test frequency histogram, where the fre-quency refers to the number of specimens. The measured end-of-fatigue life loading cycles were obtained different ranged80,000–92,000. The Erdemir 6842 steel sheet is produced in vari-able thickness tolerances that may cause the different loading

Fig. 2 (a) Fatigue test equipments and (b) circuit diagram of the hydraulic system. 1, filter; 2,AC motor and hydraulic pump; 3, pressure gauge; 4, pressure relief valve; 5, check valve; 6, ser-vohydraulic valve; 7, cycle counter; 8, the LPG tanks.

Fig. 3 Results of the accelerated fatigue tests

Journal of Pressure Vessel Technology AUGUST 2017, Vol. 139 / 041402-3

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/936116/ on 04/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

cycles. Therefore, the mean value of loading cycles was calcu-lated about 90,000 for the torus LPG tank having water capacityof 40 l.

The fatigue failure locations on the tested 13 LPG torus tankswere mostly obtained at the inside junction of circumferentialweld line and base torus drawn with chalk as seen in Fig. 4. Asshown, the 12-tank out of 13 were failed at around the circumfer-ential weld line inside the torus shown with point “o.” A tank outof 13 was failed at the junction of the inlet nozzle and tank bodywhere the number of cycle was calculated about 80,000.

Those regions where the fatigue failure occurs can also bedefined as heat affected zone in which the grain size of the struc-ture and material properties are changing due to welding process.

A microscopic picture is used to observe the fractured regionenlarging 200 times using a basic laboratory microscope (Fig. 4).The fractured region shows ductile behavior rather than brittlethat also complies with the requirements of the Codes [1,2]. Asseen from the figure, the fatigue fracture started from the placeshown with point “o” (Fig. 4) and propagates usually in circum-ferential direction. The observed fatigue fracture occurred nearthe weld line on the tank is not usually visible with the naked eyesince a very fine crack as a line was developing.

4 Computer-Aided Modeling of Fatigue Test

Computer-aided investigation of accelerated fatigue tests werecarried out by finite element based computer code, ANSYS

WORKBENCH, to predict the fatigue life and failure location of thetanks. To do this, solid modeling of the tank was created in 3Dusing SOLIDWORKS software and imported into the ANSYS WORKBENCH

to convert the finite element model as seen in Figs. 5(a) and 5(b),respectively. The FEA models were created in nonhomogeneousand nonuniform conditions, using the material properties and thegeometrical dimensional variations. To create these models andsimulate the fatigue tests, the toroidal shell material properties andthickness variations due to welding and spinning processes weremeasured and used in the modeling. Due to the spinning process,the material properties of the tank were also investigated. Afterselecting an appropriate finite element, loading, and boundary con-ditions, the nonlinear 3D FEA models were generated and simu-lated within stress life option. The stress life approach comparesthe stresses in the model to the fatigue limit of the material’s S–Ncurve. In stress life approach, fatigue failure parameters have beenproposed by many methods such as Goodman and Soderberg con-sidering the mean stress effects on the fatigue life.

4.1 The Material Properties of the Liquefied PetroleumGas Tanks. Two different types of material properties, truestress–strain and S–N curve were obtained and used in the simula-tions. The true stress–strain data of the tanks as seen in Fig. 6were investigated using the tensile test technique and recollectedfrom our previous work [4]. The tensile test specimens cut out pla-ces on the tank were cut out from the tank in different orientations

Fig. 4 The fatigue failure locations of the torus tanks

041402-4 / Vol. 139, AUGUST 2017 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/936116/ on 04/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

as shown in Fig. 6. As seen, the test specimens were named as A,B, C, and D. The specimens B and C were cut out from the torusshell in both circumferential and toroidal directions to measurethe properties. The material properties in toroidal direction werefound lower that was used in the simulations. Mechanical proper-ties of the tank material measured with tensile test are given inTable 1.

In Fig. 6, the elastic range is not shown exactly. The linear por-tions of the curves in Fig. 6 are not showing exactly the elasticmodulus for the materials. The values of the stress–strain dataincluding modulus of elasticity for the inner ring are lower thanthe others since the inner ring material is the raw material that iscoming from the steel manufacturer (Erdemir Co.).

The S–N data of a low carbon structural steel used in the FEAfatigue simulations having similar contents and properties withthe Erdemir 6842 steel were obtained from the ANSYS materiallibrary as shown in Fig. 7.

The toroidal shell thickness varies due to the spinning processmentioned above so that the thickness variations were also investi-gated by measuring from the full cross-sectional geometry of the

tank as shown in Fig. 1. The measurements were done using amicrometer having a precision of 0.001 mm in both “point-by-point” and “by-sliding” on the surfaces. Totally, nine differentthicknesses were measured from nine different points of the toroi-dal tanks (see Fig. 1). The measured thicknesses were obtainedslightly different from the nominal thickness of Erdemir 6842blank steel sheet [28]. In fact, the steel is manufactured within thethickness tolerances of the sheet as well. Therefore, the thicknessvariation of the tank was revealed as a function of shell regions asillustrated in Fig. 8. In addition, the weld deposits were generallyformed quite uniform during the tank assemble so that the averagethickness value of the weld deposit was measured about 6.35 mmwith the micrometer from the full cross section [5].

4.2 Development of Finite Element Analysis Model. Bothnonhomogeneous and nonuniform FEA models were generated inthis study. Three different types of material properties obtainedfrom the toroidal shell, inner ring, and weld zone (Fig. 6) wereapplied to construct the nonhomogeneous FEA model (Fig. 5(b)).

Fig. 6 Orientations of the tensile test specimens and materialproperties of the toroidal LPG tank

Fig. 5 The torispherical LPG tank: (a) 3D solid model and (b) FEA model

Table 1 Mechanical properties of tensile test specimens of the torus tank material [4]

Mechanical properties of torispherical LPG fuel tanks

Specimen Tensile yield strength (MPa) Ultimate tensile strength (MPa) Elongation (%)

A (weld zone) 408 690 18B (toroidal shell) 296 410 20D (inner ring) 301 413 25

Fig. 7 The S–N curve of the torus tank material

Journal of Pressure Vessel Technology AUGUST 2017, Vol. 139 / 041402-5

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/936116/ on 04/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

As seen, true stress–strain data were converted from the engineer-ing stress–strain data and applied to associated regions of the FEAmodel. The thickness variation of the tank and weld depositthickness were used in the relevant regions of the tank geometryto create the nonuniform FEA model. In order to adapt thethickness variations (Fig. 8) during the FEA modeling process,the wedge function [5] were employed. Preliminary investiga-tions were carried out to select the most suitable 3D solid ele-ment, SOLID186, from the ANSYS element library and used toform the mesh generation. The generated 3D FEA model has237,727 nodes and 66,500 elements (Fig. 5(b)). The SOLID186is ten-node brick structural element with tetrahedral options andhaving three translational degrees-of-freedom at each node ineach direction [29,30].

4.3 Loading and Boundary Conditions. The geometricalstructure of the LPG tanks was considered as in 3D volume forloading and boundary conditions. They are subjected to two dif-ferent types of loads, static and cyclic, in the simulations. Thereason of the static loading application was to find out thestress–strain distributions of the tank. The SP value was alsoapplied as a repeated cyclic load in the simulations for thefatigue performance. These tanks have commonly been used inservice between the pressure values of 0 and 1.75 MPa in indus-trial applications. The meaning of zero (0) and 1.75 MPa pres-sures can be defined as the tank is fully empty and pressurizeduntil the SP, respectively, that is called constant amplitude zero-based cyclic load as seen in Fig. 9. As seen from the figure, thevertical axis represents the pressure values in MPa. In zero-based cyclic loading, the stress and amplitude ratios are takeninto account as 0 and 1, respectively. Based on these pressurevalues, they were subjected to the zero-based cyclic load in theFEA fatigue simulations. For the boundary conditions, top sur-face of the inlet nozzle (kloret) were fixed in all directions, andthe rest of the entire model was taken as free in the simulations(see Fig. 5(b)).

4.4 Fatigue Failure Analysis and Failure Locations. Theequivalent stress approaches considering the maximum shearstress, von Mises, and maximum principal stress theories wereconsidered for the fatigue based on the static yield criteria. Thezero-based cyclic pressure load was applied to the tank, and thecalculated maximum stress as von Mises stress (157.73 MPa) wasshown in Fig. 10 [31]. From the static analysis, the maximumequivalent stress was calculated as 157.73 MPa as given in Fig. 11applying the static pressure load, SP¼ 1.75 MPa. The calculatedstatic stress was obtained lower than the yield stress, around300 MPa, of the Erdemir 6842 steel (Fig. 6).

The fatigue failure location of the tanks is well known by theexperimental results explained above. The fatigue damageoccurred at the junction of toroidal shell and circumferential welddeposit shown by point “o” (Fig. 4). The failure was also pre-dicted by the FEA simulations at the same location by calculatingthe fatigue life and safety factor (SF). That location can be definedas the fatigue failure location for the torispherical LPG tanks sincethe maximum equivalent stress (von Misses) and strain were alsoobtained at the same point. The calculated stress value had fallenin unsafe region of the S–N curve (Fig. 7). On the other hand,the obtained failure location in this study complied with theburst failure location for these tanks were obtained in the previouswork [5].

The fatigue life of the tank was calculated as available life,95,431 cycles, and shown within contour plots in Fig. 10(a) that

Fig. 8 Thickness variation of the toroidal LPG fuel tanks

Fig. 9 The applied zero-based repeated pressure loading Fig. 10 (a) Fatigue life and (b) safety factor for the LPG tank

041402-6 / Vol. 139, AUGUST 2017 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/936116/ on 04/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

represents the number of cycles until the tank will fail due to thefatigue failure before reaching the defined design life designatedas 106 cycles in the fatigue simulations.

As seen from the figure, the minimum life or first failure loca-tion of the torus tank is shown with “max” sign which is insidethe tank at the junction of the weld deposit and tank body. Themaximum life of the tank is shown with “min” sign which is onthe midpoint of the weld deposit. In addition, the maximum safetyfactor of the tank with respect to the fatigue failure for the defineddesign life is displayed as 15, illustrated in Fig. 10(b). As seen,the safety factor values were calculated lower than 1 for theregions where the fatigue failure happens.

To examine the fatigue damage occurred on the tank, the calcu-lated damage value was obtained greater than 1 which indicatesthat the failure also occurs before reaching the design life.Besides, the structural behavior of the tank under repeated cyclicload was observed by the maximum deflections in the simulations.To illustrate this, four nodes were selected from the model asshown in Fig. 12. As seen in the figure, the nodal displacementsfor the selected nodes of the model were plotted as a function ofloading cycles.

The nodes N1 and N2 selected from the location where themaximum deflections of the tank were obtained at midsection ofthe torus geometry since the tank is greatly expanding due tomaterial ductility. The deflections of the N1 and N2 were plottedin radial direction. The minimum deflections of the tank wereobtained at inside the torus geometry as shown with selected

nodes, N3 and N4 (Fig. 12). The deflections of nodes, N3 and N4,were also plotted in radial direction of the torus. The maximumdeflection was calculated about 2.32 mm in radial direction at themidpoint of torus shell.

5 Life of the Vehicle Toroidal Liquefied Petroleum

Gas Tanks

The life of the vehicle toroidal LPG tanks generally used forprivate or commercial purposes can simply be calculated andgiven in Table 2 considering the number of cycles obtained in thisstudy. As seen from the table, these tanks in vehicles can be usedin different frequencies such as one or two-cycle in a week andone or two-cycle in a day for the use of private and commercialpurposes, respectively. The frequency means that tanks are filledup fully with the LPG and used it empty completely in daily orweekly life. The LPG tanks of the commercial taxies in metropoli-tan cities, for instance, are mostly filled up fully at least once ortwice a day (24 h), but the tanks in the private vehicles are usuallyfilled up completely once or twice in a week.

In order to calculate the life of the vehicle toroidal LPG fueltanks, it was assumed that some inevitable factors are not affect-ing the tanks’ life. However, these factors are extremely influen-tial for the tanks such as corrosion, humidity, temperaturechanges, vibration, noise, chemical compositions of LPG, manu-facturing, and assemble processes of the tank products. Based onthese ideas, the life of the tank can be calculated basically takingthe design SF as 5 and 10 for the tank considering the 90,000cycle as seen in Table 2. Designers usually take the SF> 1.5–2 todesign the mechanical components, but they consider the SF> 10for mechanical systems that are extremely important for humanlife safety, such as LPG, CNG, chemical, flammable tanks,

Fig. 11 The equivalent static stress distributions

Fig. 12 Structural behavior of the LPG tank for selected nodes

Table 2 Life of the vehicle toroidal LPG fuel tanks

Tank life/90,000-cycle

Frequency Cycles/365-day SF¼ 5 SF¼ 10

Private use 1 cycle/week 52 346/year 173/year2 cycles/week 104 173/year 86/year

Commercial use Cycle/day 365 49/year 25/yearCycles/day 730 25/year 12/year

Journal of Pressure Vessel Technology AUGUST 2017, Vol. 139 / 041402-7

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/936116/ on 04/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

elevators. For example, if a commercial taxi fills its LPG tanks upfully twice a day, the tank works 12 year without failing with aSF¼ 10 (see Table 2). In most private use of vehicles, drivers filltheir-tanks up fully twice a week so that their tanks work 87 yearwithin the SF¼ 10 considering the assumption aforementioned.Therefore, in Turkey, the vehicle LPG fuel tanks are used maxi-mum for 10 year, and they must be replaced in every decade basedon the rules of the regulations, TS 12095-1 and ECE-R67 (EN12805).

6 Conclusions

An accelerated fatigue experimental unit managed with PLCservohydraulic valve was designed and established successfully todetermine the fatigue life of the torispherical LPG fuel tanks andcompared with the numerical results. To predict the fatigue lifeand its failure location, the LPG tanks were pressurized internallywith repeated cyclic pressure load from 0 to 1.75 MPa. On theother hand, the computer-aided modeling was performed andsimulated in both nonuniform and nonlinear conditions consider-ing the stress life option. The shell and weld zone material proper-ties including weld zone thickness were investigated and usedsuccessfully in the modeling processes to achieve the experimen-tal results. Based on the generated results, some conclusions canbe made as follows.

(a) A good agreement between the accelerated experimentalfatigue and corresponding nonlinear 3D FEA simulationswas found for the fatigue performance. The number offatigue cycles obtained from the FEA simulations (Fig.10(a)) was about 96,000 cycles. This result is slightlyhigher than that of the mean cycle value of 90,000 obtainedfrom the experimental studies (Fig. 3).

(b) A good agreement was also found about the fatigue failurelocation between the accelerated fatigue test and FEA sim-ulation. The failure location was obtained at a place shownwith point “o” in Figs. 4 and 10 from the experiment andFEA methods, respectively. The safety factor was also cal-culated as zero at the same point using the simulations.

(c) The actual shell elastic–plastic material properties of thetank and relevant S–N curve values were well specified inthis study and successfully adapted into ANSYS software.Additionally, the behavior of internal pressure loading wasapplied successfully as a zero-based constant amplitudecyclic load.

(d) The life of the LPG tanks produced and used in similar con-ditions is sufficient and complying with the regulationrules. It may be concluded that the replacement of the LPGtanks in every decade is a reasonable decision.

Acknowledgment

The authors would like to thank Step, Inc.1 to perform theseexperimental tests, providing LPG tanks and related expenses.Special appreciation is expressed to Faruk Guraksu, GeneralDirector; Aydin Karateke, Engineering and Manufacturing Direc-tor, and other lab technicians.

References[1] Japan Office of Trade and Investment Ombudsman, 1999, “Acceptance of Pres-

surized LP Gas Fuel Supply Apparatus for Automobiles,” Resolution to Com-plaint to the Japan Economic Planning Agency, OTO No. 589.

[2] TSE, 2004, “Automotive LPG Components-Containers,” Turkish StandardInstitute (TS), Ankara, Turkey, Standard No. TS 12095 EN 12805 (in Turkish).

[3] Tchoupou, K. M. T., and Fotsing, B. D. S., 2015, “Fatigue Equivalent StressState Approach Validation in Non-Conservative Criteria: A ComparativeStudy,” Lat. Am. J. Solids Struct., 12(13), pp. 2506–2519.

[4] Kisioglu, Y., Brevick, J. R., and Kinzel, G. L., 2001, “Determination of BurstPressure and Locations of the DOT-39 Refrigerant Cylinders,” ASME J. Pres-sure Vessel Technol., 123(2), pp. 240–247.

[5] Kisioglu, Y., 2011, “Burst Pressure Determination of Vehicle Toroidal OvalCross-Section LPG Fuel Tanks,” ASME J. Pressure Vessel Technol., 133(3),p. 031202.

[6] Kaptan, A., and Kisioglu, Y., 2007, “Experimental of Burst Pressures and Fail-ure Locations of Vehicle LPG Cylinders,” Int. J. Pressure Vessels Piping,84(7), pp. 451–459.

[7] Aksoley, M. E., 2004, “Examining the Burst Test Parameters Applied on LPGTanks,” M.Sc. thesis, Gebze High Technology Institute, Gebze, Turkey (inTurkish).

[8] Marzbanrad, J., Paykani, A., Afkar, A., and Ghajar, M., 2013, “Finite ElementAnalysis of Composite High-Pressure Hydrogen Storage Vessels,” J. Mater.Environ. Sci., 4(1), pp. 63–74.

[9] Zu, L., Zhang, D., Xu, Y., and Xiao, D., 2012, “Integral Design and Simulationof Composite Toroidal Hydrogen Storage Tanks,” Int. J. Hydrogen Energy,37(1), pp. 1027–1036.

[10] Kim, Y.-S., Kim, L.-H., and Park, J.-S., 2011, “The Effect of Composite Dam-age on Fatigue Life of the High Pressure Vessel for Natural Gas Vehicles,”Compos. Struct., 93(11), pp. 2963–2968.

[11] Kisioglu, Y., Brevick, J. R., and Kinzel, G. L., 2005, “Bottom End-ClosureDesign Optimization of DOT-39 Non-Refillable Refrigerant Cylinders,” ASMEJ. Pressure Vessel Technol., 127(2), pp. 112–118.

[12] Giglio, M., 2003, “Fatigue Analysis of Different Types of Pressure VesselNozzle,” Int. J. Pressure Vessels Piping, 80(1), pp. 1–8.

[13] Rauscher, F., 2003, “Fatigue of Non-Welded Pressure Vessels Made of HighStrength Steel,” Int. J. Pressure Vessels Piping, 80(3), pp. 197–204.

[14] Alegre, J. M., Bravo, P. M., and Cuesta, I. I., 2010, “Fatigue Design of Wire-Wound Pressure Vessels Using ASME-API 579 Procedure,” Eng. FailureAnal., 17(4), pp. 748–759.

[15] Hossain, M. M., and Seshadri, R., 2010, “Simplified Fitness-for-ServiceAssessment of Pressure Vessels and Piping Systems Containing Thermal HotSpots and Corrosion Damage,” Int. J. Pressure Vessels Piping, 87(7),pp. 381–388.

[16] Camara, S., Bunsell, A. R., and Thionnet, D. H., 2011, “Determination of Life-time Probabilities of Carbon Fibre Composite Plates and Pressure Vessels forHydrogen Storage,” Int. J. Hydrogen Energy, 36(10), pp. 6031–6038.

[17] Chou, H. Y., Bunsell, A. R., and Thionnet, A., 2012, “Visual Indicatorfor the Detection of End-of-Life Criterion for Composite High PressureVessels for Hydrogen Storage,” Int. J. Hydrogen Energy, 37(21),pp. 16247–16255.

[18] Song, L., Xiaolong, J., Hongjie, S., Hongwei, S., David, H., and Xiaoping, Y.,2013, “Thermo-Mechanical Properties of Filament Wound CFRP Vessel UnderHydraulic and Atmospheric Fatigue Cycling,” Composites, Part B, 46,pp. 227–233.

[19] Wanga, X. H., and Redekop, D., 2011, “Natural Frequencies Analysis ofModerately-Thick and Thick Toroidal Shells,” Procedia Eng., 14, pp. 636–640.

[20] Zu, L., Koussios, S., and Beukers, A., 2010, “A Novel Design Solution forImproving the Performance of Composite Toroidal Hydrogen Storage Tanks,”Int. J. Hydrogen Energy, 37(19), pp. 14343–14350.

[21] Zhan, H. J., and Redekop, D., 2009, “Static and Dynamic Loading of an Ova-loid Toroidal Tank,” Thin-Walled Struct., 47(6–7), pp. 760–767.

[22] Augustins, L., 2013, “An Empirical Multiaxial High Cycle Fatigue Criterionfor Automotive Cylinder Head Design,” Eng. Failure Anal., 28, pp. 264–274.

[23] Santanna, H. M., and Leal, M. F., 2011, “Practical Procedure to Assess CriticalDefects in Pressure Vessels Subjected to Fatigue Loads,” Eng. Fract. Mech.,78(8), pp. 1669–1683.

[24] Spear, A., and Ingraffea, A., 2011, “Microstructurally Small Fatigue CrackGrowth in Thin, Aluminum-Alloy, Pressure Vessel Liner,” Procedia Eng., 10,pp. 686–691.

[25] Rudolph, J., Schmitt, C., and Weib, E., 2002, “Fatigue Lifetime Assessment forWelded Vessel Components,” Int. J. Pressure Vessels Piping, 79(2),pp. 103–112.

[26] Koh, S. K., 2002, “Fatigue Damage Evaluation of a High Pressure Tube SteelUsing Cyclic Strain Energy Density,” Int. J. Pressure Vessels Piping, 79(12),pp. 791–798.

[27] Varvani-Farahani, M. R., and Kianoush, M., 2007, “Fatigue Failure Assessmentof Engineering Components Under Service Loading Conditions,” Mater. Des.,28(2), pp. 575–580.

[28] Marczewska, T., Bednarek, A., Marczewski, W., Sosnowski, H., and Jakubc-zak, J., 2006, “Practical Fatigue Analysis of Hydraulic Cylinders and SomeDesign Recommendations,” Int. J. Fatigue, 28(12), pp. 1739–1751.

[29] Erdemir Steel, 1997, “Erdemir 6842 Steel Properties,” Erdemir Product Cata-logue, Erdemir Group, Zonguldak, Turkey.

[30] ANSYS, 2003, “ANSYS User Manual, Swanson Analysis System, V 8.0,”ANSYS, Inc., Canonsburg, PA.

[31] Stephens, R. I., Fatemi, A., Stephens, R. R., and Fuchs, H. O., 2000, MetalFatigue in Engineering, 2nd ed., Wiley, New York.

1http://www.steplpg.com and http://www.meridyenlpg.com/

041402-8 / Vol. 139, AUGUST 2017 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/936116/ on 04/03/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use