1

CHAPTER 1

INTRODUCTION AND LITERATURE REVIEW

1.1 INTRODUCTION

The use of compressed natural gas (CNG) as a vehicle fuel has been

growing worldwide due to the advantages of significant reduction of exhaust

emission and lower fuel cost compared with gasoline. Composite materials are

increasingly being used for the construction of high-pressure CNG fuel

containers owing to their high strength to weight ratio and corrosion resistance.

Composite CNG cylinders are typically constructed with inner metallic liners in

order to arrest the gas leakage at high operating pressures. The liner also serves

as a mandrel for the filament winding operation and shares 15-20% of cylinder

loading.

Another important advantage of composite cylinders is that these

cylinders provide safe burst behavior compared to that of metallic cylinders.

The typical mode of failure of a metal lined composite pressure vessel is

described as ‘leak-before-break’, in which the leakage starts from the liner

fracture and passes through the composite overwrap progressively. In this mode

of failure, sufficient leakage of the gas takes place before the full burst happens

and thus it offers greater safety in the event of a failure. This is considered a

very significant advantage of CNG fuel storage cylinders for automobiles.

2

As a part of the manufacturing process, all metal lined composite

cylinders undergo a process known as ‘autofrettage’ or ‘sizing’ which creates a

state of pre-compressive stress in the liner. This is done by subjecting the

finished composite cylinder to a pressure sufficient enough to cause plastic

deformation in the liner while the composite is in its elastic range. The

plastically deformed liner is consequently placed under a compressive stress on

depressurization and the composite overwrap is subjected to a slight tensile pre­

stress. In normal operating cycles, the liner material operates between tension

and compression. The net result is a reduction in the operating strain range of

the liner material with a consequent increase in fatigue life. The operation of a

CNG cylinder constitutes a low cycle fatigue due to the periodical expending of

gas and re-pressurization and hence fatigue becomes an important design

criteria.

Metal lined composite pressure vessels are constructed in developed

countries using the carbon fiber-epoxy composite system with aluminium liner,

which offers the maximum weight reduction. However these cylinders are

costlier. Steel as liner material and glassfiber-epoxy as the composite overwind

can offer the combined overall benefit, considering weight saving, cost and the

local availability of glass fiber in most countries. Hence, with the objective of

developing cost effective and technically qualified composite CNG cylinder, a

steel lined glassfiber-epoxy reinforced composite construction was selected for

this work.

1.1.1 Types of Vessel Construction

There are two types of vessel construction normally being adopted for

the metal lined composite cylinders. They are called ‘fully wrapped

3

construction’ and ‘hoop wrapped construction’. In a fully wrapped construction,

the composite is wound over the cylinder as well as over the domes and the

winding consists of both helical and hoop layers. In a hoop wrapped

construction, the composite is wound over the cylindrical portion with hoop

layers (90°) only. Fig. 1.1 and 1.2 show the fully wrapped and hoop wrapped

constructions.

Fig. 1.1 Fully Wrapped Cylinder

Hoop Layers

Fig. 1.2 Hoop Wrapped Cylinder

The optimum winding angle for composite cylindrical vessel

subjected to internal pressure is given by netting analysis as 54.7°. The fiber

orientation at this angle exactly meets the 2:1 ratio of hoop verses axial loading

in composite cylindrical structures subjected to internal pressure. However, it is

4

not possible to use this angle for composite pressure vessels, since the fibers do

not stay in place over the dome surface when wound at 54.7°. Only with low

helical angle it is possible to wind over the dome. Hence the fully wrapped

pressure vessels are typically constructed with low angle helical layers to take

care of longitudinal loading and hoop layers to take care of hoop loading.

The standards for composite pressure vessels for CNG storage

applications are normally government-controlled regulations that define the

technical and safety requirements. The cylinder must fulfill the above

requirements, while at the same time being capable of mass production at a

competitive price. For composite CNG cylinders, the important regulatory

agencies are:

DoT (Department of Transportation, USA)

NGV2 (Natural Gas Vehicle Standard, USA)

TUV (Germany) and

HSE (UK)

ISO has recently published a standard for composite CNG cylinders

for vehicles (ISO 11439, First edition: 2000).

1.1.2 Filament Winding Process

1 Filament winding is an effective method to manufacture composite

cylinders. In a typical wet winding process, the unidirectional fiber rovings are

pulled through a resin bath and wound onto the rotating liner and then cured in

an oven to get the full strength. The filament winding also uses ‘prepregs’ in the

form of tapes. The wet winding is more commonly used for composite

cylinders owing to the advantages of low material cost and short winding time.

5

Filament winding is carried out on specially designed machines.

Precise control of the winding pattern and direction of the filaments are

required for maximum strength, which can be obtained with controlled machine

operation. Directional strength ratios can be varied in filament wound

structures. Filament winding provides high burst strengths in composite

pressure vessels. A filament wound construction provides an almost 100%

efficient strength-to-weight structure. Fig. 1.3 shows the wet winding process.

Head Stock

Fig.1.3 Filament Winding of CNG Cylinder

1.1.3 Performance Index

As far as the metal lined composite pressure vessels are concerned,

the effective weight saving that can be achieved depends on the liner material,

fiber material and the optimal design. Since the vessels can be designed for

different pressure rating and capacity with different materials, the net effect of

weight saving in pressure vessels is expressed by means of a parameter,

‘performance index’. Performance index is a number used to compare the

performance of pressure vessels and is calculated (Peters, S.T., 1994) as:

6

PVPerformance index = --------

W

Where, P - Pressure (bar)

V - Volume (liter)

W-Weight (kg)

Fig. 1.4 shows a comparison of performance index of vessels with

different material of construction (Masanori Kawahara et al., 1998).

Fig. 1.4 Comparison of Cylinder Performance Index

Higher performance index value indicates low weight vessel for the

given operating pressure and capacity. It can be seen from Fig. 1.4, that fully

wrapped carbon epoxy vessels with plastic liner (HDPE) gives the maximum

value of performance index. While using the plastic liner, the cylinder boss is

7

normally made in metal and attached to the plastic liner. Achieving a leak-proof

joint between the metallic boss and the plastic liner becomes extremely difficult

and hence the composite pressure vessels with plastic liners are not

commercially exploited, except for aerospace applications.

With respect to the vessel shape, it is well known that the spherically

shaped vessel gives minimum specific mass. However, spherical shaped vessels

are not constructed for commercial CNG and other gas storage applications due

to their constructional and mounting difficulties. CNG cylinders are typically

made as a cylindrical body with end domes. The required useful volume and the

accommodating space in the vehicle usually determine the length to diameter

ratio.

1.1.4 Material of Construction

Materials of CNG cylinder construction include fiber reinforcement,

resin matrix and liner. Most frequently used fibers are glass, aramid(Kevlar)

and carbon fibers. Typical material properties of different fibers along with the

cost are given in Tablel.l (Anthony Kelly, 2000)

Table 1.1 Fiber Properties

Property and Cost E-Glass S-Glass Aramid CarbonTensile Strength (MPa) 1700 2480 2480 2960

Elastic modulus (GPa) 70 87 131 220Density (kg/in’) 2540 2490 1440 1800

Cost (US Dollars/pound) 1.00 6.00 21.00 15.00

8

Glass fiber was chosen as the reinforcement fiber for the present

composite CNG cylinder developmental study owing to its low cost, moderate

strength and good damage tolerance and fatigue properties. Among the resin

candidates, the epoxy resin has better engineering properties in terms of good

adhesion, low water absorption and low cure shrinkage and generally used for

the pressure vessel construction.

The main selection criteria for liner materials are: high specific

strength, long fatigue life and good fracture toughness characteristics and strain

compatibility between liner material and composite overwrap. Aluminum

alloys, titanium alloys and alloy steels are the most commonly used liner

materials for the metal lined filament wound pressure vessels. The important

properties of these materials are listed in Table 1.2 (Technical Manual, PMA

Inc., 1997).

Table 1.2 Properties of Liner Materials

Criteria Titanium Alloy AluminiumAlloy Steel Alloy

Elastic modulus (GPa) 110 204 68

Ultimate strength (MPa) 900 850 350

Density (kg/m ) 4430 7920 2700

Fracture Strain (%) 8 10 12

Cost(US Dollars/pound) 21.5 3.2 4.5

Titanium alloys, due to their low density, best meet the criterion of

low weight. However, on account of their high cost, they are not used in

commercial pressure vessels. The alloy steel has high strength, better leak-

before-break criterion and lower cost compared with the aluminium alloys. On

the weight criterion, alloy steel is three times heavier than aluminium alloy.

9

However, due to its high strength, the required liner thickness will be less and

hence the actual weight increase would be only marginally high. A comparative

performance rating evolved on the basis of design criteria for different liner

materials is given in Table 1.3 (Technical Manual, PMA Inc., 1997). Weighing

the overall benefit, alloy steel has been selected as the liner material for the

present CNG cylinder developmental study.

Table 1.3 Performance Rating of Various Liner Materials

Selection Criteria TitaniumAlloy

AlloySteel

AluminiumAlloy

Low Weight 1 3 2

Strain Compatibility between Liner and Composite

4 3 2

Leak Before Burst 3 1 2

Cost 4 2 3

Rating Order: 1 - Best 2 - Good 3 - Reasonable 4 - Poor

The liners may be either seamless or welded type. Seamless steel

liners are fabricated from seamless steel tubes by forming the end domes (with

neck and boss) using the hot spinning process. The resulting liners are heat

treated to get the desired mechanical properties and finally machined to provide

threads in the boss for fixing the valve. Welded liners are fabricated by welding

the separately formed end domes to the cylindrical portion by a circumferential

welding. Most of the pressure vessel standards do not permit longitudinal

welding. The quality of welded liner depends upon weld efficiency and this is

to be ensured by non-destructive testing and suitable quality control procedures.

10

1.2 LITERATURE REVIEW

In order to understand and assess the current status of research in

metal lined composite pressure vessels, an extensive review of literature was

carried out in the areas of composite cylinder structural analysis, winding

process optimization and performance evaluation. The main aspects of the

literature review are given hereunder.

1.2.1 Analytical Studies on Orthotropic Composite Cylinders

A fair amount of studies dealing with laminated composite pressure

vessels have been reported in the literature. However, a vast majority of studies

deal with unlined composite pressure vessels. The first comprehensive

investigation of stress distribution in a body with cylindrical anisotrophy was

made by Lekhnitskii (1968). He has developed relations for the problem of

plane stress in an orthotropic cylindrical shell subjected to internal and external

pressures. By layering a number of such shells and by matching the radial

deformations of adjacent shells at their interfaces, he developed relations

describing the stresses and strains in a multi-layer cylindrical shell composed of

a number of cylindrical layers, each with its own elastic properties.

Tsai (1971) has extended Lekhnitskii’s work to a composite filament-

wound pressure vessel where each layer of Lekhnitskii’s model corresponds to

a ‘winding layer’ of the pressure vessel. Tsai has developed the stress analysis

of thin and thick composite cylindrical vessels. He compared burst pressure

calculations from thin wall and thick wall considerations and concluded that for

outer radiiis/inner radius ratios less than 1.10, the thin wall approximation was

adequate.

11

Optimal design of unlined laminated pressure vessels for maximum

burst pressure and minimum weight has been investigated by many researchers.

Most of these analyses were based on the elasticity approach. Few important

developments have been enumerated hereunder.

Tauchert (1981) studied the optimal stress distribution in reinforced

pressure vessels for minimum strain energy. The works of Fukunaga, et al.

(1983), G.C.Eckold, (1985), C.S. Mao, et al. (1992) S. Adali, et al. (1993) and

C.W.Kim, et al.(1993) are important in the development of laminated

cylindrical pressure vessels under strength criterion. H.Fukunaga and

T.W.Chou (1998) have formulated simplified design techniques for laminated

cylindrical pressure vessels under both strength and stiffness constraints. The

analytical formulation by Alexis A. Krikanov (2000) and Viktor E. Verijenko

et al. (2001) focused on the optimization of laminated pressure vessels for

higher stiffness.

Spherical composite vessel analysis and its failure prediction were

reported by C.E.Knight (1982) and B.Mouhamath (1993). M.T.Callaghan,

(1991) and R. Heydenreich, (1998) have reported the developmental studies

carried out on composite cryogenic tanks.

Netting analysis is another popular theory used in most of the cases of

unlined pressure vessels (Tew, B.W., 1995). Netting analysis is a simplified

approach to the design of cylindrical filament-wound structures under internal

pressure loading. Netting analysis assumes that all strength and stiffness

properties are derived from the fibers alone and it is considered as a

conservative approach. Apart from the popular netting theory, many composite

shell theories have also been developed for the structural analysis of composite

cylinders. These theories have addressed different features such as transverse

shear deformation and geometrical nonlinear effects and found most relevant to

the analysis of thick composite cylinders. The works of C.G.Chao et al. (1975)

K.Bhaskar and T.K. Varadan (1992) and Serge Abrate (1994) were important in

this respect. A method of evaluation of geometrical nonlinear effects in thin and

moderately thick composite shells was given by Erasmo Carrera and Horst

Parisch (1998). Claire Ossadzow and Maurice Toratier (2001) have developed

an improved shear membrane theory for multilayered shells.

1.2.2 Analytical Studies on Metal Lined Composite Pressure Vessels

The available research papers on the analysis of metal lined composite

pressure vessels are less compared to the unlined pressure vessels. An earlier

study carried out by F.A. Simonen et al. (1975) on the filament reinforced

aluminum cylinders is of importance in terms of its approach. In this study, an

existing alunimum cylinder was reinforced in the cylindrical portion with hoop

windings with the idea of increasing the load carrying capacity and reducing the

weight. Analysis was done using a commercial finite element software package

by incremental load approach and thereby predicting the pressure

corresponding to the liner yielding.

M.D. Witherell and M.A.Scavullo (1990) have derived analytical

expressions for the stress analysis of internally pressurised composite-jacketted

isotropic cylinders using the thick cylinder elasticity approach. Peter C.T.Chen

(1993) has carried out a nonlinear analysis of a steel gun barrel reinforced with

glassfiber-reinforced plastics. This structure was intended for achieving weight

reduction. Analytical solutions were obtained using thick cylinder theory for all

loading ranges up to failure.

13

David kokan and Kurt Gramoll (1994) have reported the research

work carried out at Georgia Tech, USA on the metal lined filament wound thick

composite tubes. The analysis concentrated on the stresses induced by winding

and curing. The result showed that considerable manufacturing related stresses

developed in metal lined thick composite tubes. However, the analysis did not

account interface effects between the metallic liner and the composite and

assumed perfect interface between composite layers as well as between the

metallic liner and the composite.

L.Varga et al. (1995) have developed a design methodology for CNG

tank made of aluminium liners with reinforced plastics overwrap. An analytical

method was developed considering unit internal pressure and operating stresses

were calculated for the full loading cycle of the cylinder. The paper also

reported the experimental results of burst testing.

J.M.Lifshitz and H.Dayan (1995) have outlined the basic approach for

the analysis of filament wound pressure vessel with thick metal liner. The

authors have concluded that the selection of arbitrary thick liner did not give the

optimized design; however the difference in weight reduction was found

marginal between thin and thick liners. The study has predicted an interfacial

clearance at the metal-composite interface by curing; however the analysis has

not accounted the total effect of winding and curing.

B.S.Kim, B.H.Kim, J.B.Kim and C.RJoe (1998) have reported the

studies on the effect of containing the compressed natural gas in steel cylinders

for a prolonged period (one year) at room temperature. Their study showed that

steels withstood the compressed natural gas without any chemical reaction and

adverse effects.

14

Though abundant literature is available on autofrettaging of steel

cylinders, the autofrettaging studies pertaining to metal lined composite

pressure vessels are rare. Few studies that have some relevance to composite

materials are discussed hereunder. R.S.Salzar et al., (1996 and 1999) studied

the elastoplastic analysis of layered metal matrix composite cylinders and the

influence of autofrettage on metal matrix composite reinforced gun barrels.

J.Bouchet et al. (2000) studied the static and dynamic behaviour of composite

aluminium tube for automotive applications.

Stress analysis based on finite element method has been used

increasingly for composite structural analysis in recent years. Many

formulations of the isotropic material have been adapted to composites

accounting the elastic properties and structural behaviour of composites. Wood

(1994) has given a summary on the application of finite elements to composite

parts with special reference to the type of element to be used for a particular

application. Apart from the most popular displacement formulated finite

element method, many new variational formulations with different energy

principles have been attempted for composite laminates.

Huang et al. (1987) have developed a three-dimensional finite element

formulation, which takes into account the interlaminar shear stresses. The

theoretical aspects of finite element formulation as applicable to composites

were discussed by G. Duvaut et al. (2000) and Maenghyo Cho et al. (2000). A

finite element formulation developed for crashworthiness studies for a multi­

layered multi-material (steel and composite) was reported in a recent paper

(D.Coutellier, P.Rozycki, 2000).

15

1.2.3 Studies on Dome

Filament wound pressure vessels always require some type of end

closures or domes. Pressure vessels constructed out of isotropic materials, such

as steel or aluminum, utilized either an ellipsoidal or torispherical shape for the

end caps to reduce the critical stresses within the structure. The ideal dome

profile for the composite cylinder is an ‘isotensoid’ profile that provides

constant fiber stress along the dome profile.

A photoelastic investigation of stresses in torispherical drumheads

was done by H.Fessler and P.Stanley (1965). L.Younsheng and L. Ji (1992)

have used the sensitivity analysis in shape optimization design for pressure

vessels. Dome profile design equations for polymeric composite pressure

vessels were given by M.Hojjati et al. (1995). Chen et al. (1996) studied the

stress concentration at the round comers of flat heads in pressure vessels

subjected to internal pressure. The effect of localized plastic deformation in the

dome especially at the knuckle region was studied by J.Blachut (1997) and

Kalnins et al. (1998). Lei Zhu and J.T. Boyle (2000) have given a brief

overview of the optimal shapes for axisymmetric pressure vessels.

1.2.4 Studies on Winding Process Optimization

Many studies were reported on the filament winding process in an

effort to optimize the process variables and most of these studies were carried

out on composite tubes and pipes.

Several analytical models have been developed in recent years to

model the winding process. Lee et al. (1982) were the first to develop a fiber

16

motion model, namely sequential compaction, in which consolidation is

assumed to take place beginning with the outermost layer. The fiber

consolidation model described how the wound layers were influenced by the

external compaction pressure when a new layer was wound onto the surface of

existing layers. Dave (1987) and Gutowski (1987) have proposed a ‘squeezed

sponge’ model. In this model, compaction was not sequential and the applied

pressure was shared by both the fiber bed and resin.

Cai et al. (1992) described a fiber motion model that combined

Darcy’s flow and nonlinear spring compaction. They have concluded that the

deformation of the fiber bed defined the final cylinder dimensions and that the

fiber bed stiffness was a function of the fiber volume fraction. Susan C. Mantell

et al. (1994) has summarized the different filament winding process models.

These models have been developed for the general case of filament winding and

found to provide good results for prepreg winding than wet winding.

The influence of fiber waviness on the mechanical properties of

unidirectional fiber composites was reviewed by M.R.Piggot, (1995). Hung-

Chung Chen and Shu-Min Chiao, (1996) have used the undulating channel

model for fiber consolidation in the filament winding process. G.J.Dvorak

(1996) has derived analytical expressions for the stresses due to winding

tension using the three dimensional elasticity approach. He has considered the

mandrel as another layer with its thickness and the isotropic elastic properties.

D.Cohen (1997) has studied the influence of filament winding

parameters on composite pressure vessel strength. This study was done

primarily for rocket motor casing using the design of experiments (DOE)

approach.

17

Analytical and experimental study of fiber motion in wet filament

winding was done for the general case of filament winding by Banerji et al.

(1998). A.S.Hadi and J.N.Ashton (1998) have studied the influence of pre­

stress on the mechanical properties of unidirectional composites. The influence

of winding patterns on the damage behaviour was analyzed by Rousseau

(1999). Liyang Zhao et al.(2001) has attempted to model the winding process

using finite element approach.

1.2.5 Life Prediction Studies

Safety considerations impose severe testing conditions for composite

CNG cylinders before approved for commercial use. Composite cylinders are

subjected to full-scale burst testing and fatigue testing to assess the long-term

durability under simulated service conditions. Considering that the cylinder is

subjected to two repressurizations (refueling) a day, 18,000 cycles are

experienced by the cylinder during its expected life of 15 years. Such a frequent

refueling cycle is found to be typical for a taxi.

The fatigue life of the reinforcing fibers is much higher than fatigue

life of steel or aluminium liners (Anthony Kelly et al., 2000). The fatigue

characteristics of composites are different in many respects from those of

metals. The fatigue damage in FRP tends to be progressive and extends

throughout the stressed region of the material. The progressive manner of the

failure of composite material can be broadly classified as debonding, resin

cracking and delamination (Owen and Duke, 1987). Composites can absorb

more energy before failure. The comparative fatigue behaviour of composites

and metals (Salkind, 1972) is shown in Fig. 1.5.

18

Fatigue Cycles ----------- p-

Fig. 1.5 Comparative Fatigue Behaviour of Composites and Metals

Some recent studies pertaining to the failure and life prediction of

pressure vessels are given hereunder.

X.B. Lin (1998) studied the fatigue growth prediction of internal

surface cracks in pressure vessels. The burst behaviour of a welded pressure

vessel was studied by G.S. Bhuyan et al. (1999). ‘Leak-Before-Break’ failure

mode was described by Gery Wilkowski, (2000). W.X.Yao and N.Himmal,

(2000) have given a new cumulative fatigue damage model for fiber reinforced

plastics. S.Michael Spotswood and Anthony N. Palazotto (2001) have given a

progressive failure analysis of a composite shell. Christos C. Chamis and Levon

Minnetyan (2001) have analyzed the Defect/Damage tolerance of pressurized

fiber composite shells.

Dam

age S

ize --------

-►

19

1.3 SCOPE AND OBJECTIVES OF THE THESIS

From the literature review, it can be observed that the previous studies

concentrated on aluminium lined carbon composite cylinders. Previous

analytical studies reported for steel with glassfiber composite overwrap were

relevant to non-autofrettaged composite tubes and cylinders and the analyses

have not addressed combined autofrettage and low cycle fatigue. The

performance characteristics of steel as liner in composite cylinders are quite

different from aluminium liners. Due to its high strength and high modulus, it

becomes necessary to take into account the fracture toughness characteristics of

steel also to meet the fatigue life of liner. Hence, there is a strong need for

theoretical and experimental research studies for developing glassfiber

reinforced composite CNG cylinders with steel liners.

Two other aspects that generated considerable interest during the

literature review are the winding parameter optimization and the effect of

winding and curing at the metal-composite interface. A winding parameter

optimization study reported by D.Cohen (1997) for rocket motor casing used

the conventional design of experiments (DOE) approach and the results were

found to have a wide range. Also, the selected parameters were more relevant to

the winding of rocket motor casing. Hence, a need was recognized for winding

parameter optimization study for the cylinder using a better experimental design

technique.

With regard to the winding and curing effect at the liner-composite

interface, most of the previous studies have assumed a perfect interface between

all layers including liner-composite interface. Though, a previous study on

metal lined composite cylinder by J.M.Lifshitz et al., (1995) has predicted an

20

interfacial clearance at the end of curing, the analysis has not accounted the

total effect of winding and curing. Hence, it is planned to study the winding and

curing effect at liner-composite interface in order to account for the major

autofrettage prestrain effects. Based on the above discussion, the objectives of

the present research work are listed below:

1. Development of analytical formulation incorporating the plastic yielding

of liner and fracture mechanics based fatigue life design of steel lined

glassfiber epoxy composite cylinders.

2. Development of a computer program based on the analytical formulation

to compute the stresses and strains in liner and composite in the hoop

and longitudinal directions for various pressurizing conditions such as

zero pressure (after autofrettage), service pressure and burst pressure.

3. Winding parameter optimization studies for the composite cylinder using

the ‘Taguchi Experimental Design Technique’: Taguchi technique is a

recent experimental design method increasingly being used for discrete

process optimization and is employed for the filament winding parameter

optimization in the present investigation.

4. Autofrettaging and prestrain evaluation studies: Experimental studies are

planned to evaluate the autofrettaging effect on the residual stress

development in the liner. Experimental studies are also planned to

evaluate the prestrain induced in the liner due to winding.

5. Performance evaluation studies by subjecting the prototype cylinders to

burst testing, cyclic testing, drop testing and bonfire testing and to carry

out failure analysis based on the failure pattern of the tested cylinders.