Filament-wound composite cylinders are used in the marine and transportation industries for storing breathing gases (SCUBA, firefighter tanks) and gaseous ...
Proceedings of the Eleventh (2001) International Offshore and Polar Engineering Conference Stavanger, Norway, June 17-22, 2001 Copyright © 2001 by The International Society of OJ~hore and Polar Engineers ISBN 1-880653-51-6 (Set); ISBN 1-880653-55-9 (VoL IV); ISSN 1098-6189 (SeO
Fiber Optic Sensor System (FOSS) for Filament-Wound Gas Cylinders t~ H. Knapp
Structural Solutions/University of Hawaii, HI, USA T.A. Shimabukuro
Structural Solutions, Aiea, HI, USA LN. Robertson
University of Hawaii, Honolulu, HI, USA
ABSTRACT Filament-wound composite cylinders are used in the marine and transportation industries for storing breathing gases (SCUBA, firefighter tanks) and gaseous fuels (vehicles). These cylinders offer light weight, corrosion resistance, dimensional stability, and the ability to store more air than equivalent metal tanks. The design methodology currently used for composite tanks, however, cannot yet guarantee their safe operation. Accordingly, the U.S. Department of Transportation (DOT) is unable to issue full certification of filament-wound tanks. Rather, some types of composite pressure tanks currently are manufactured under DOT Exemption, TC Regulation 3FCM. The composite tank industry would benefit by improving the safety of operating these tanks. The interest in developing composite pressure tanks is here and the manufacturing technology is mature. What is needed, however, is a means of insuring that composite tanks are as safe to operate as metal tanks. This will facilitate DOT certification and appeal to consumers. This paper discusses a fiber optic sensor system embedded into the composite shell wall as a structural health monitor. Using a simple, low-cost optical fiber sensor and a modified commercial connector, "smart" tanks can be monitored continuously for structural integrity. The opportunity to provide such continuity in structural health monitoring should have a significant positive impact on obtaining DOT certifications and extending product useful life. This paper presents the results of a design program to develop a new filament-wound composite cylinder containing a fiber-optic sensor system (FOSS). The purpose of FOSS is to monitor the structural integrity of the composite material each time the tank is refilled with pressurized gas. Continuous monitoring for material degradation will be an important factor that determines tank recertification intervals and useful tank life.
One of the greatest challenges of this design program has been to develop an optical connector that allows external instrumentation to be "plugged into" the composite wall to read optical signals that correlate with structural health. "While embedding fiber sensors has become routine, ingress to and egress from the embedded units remain a major stumbling block" (Spillman, 1995). A simple modification to a commercial connector is proposed for this application. The design and manufacture of the prototype FOSS cylinder shown in Figure 1 are described in this paper. Connector performance and the overall performance of a prototype Type Ill tank consisting of an aluminum liner, an E-glass/epoxy filament-wound overwrap, and an embedded FOSS are discussed. Results of a pressure test to assess FOSS performance are presented.
Figure 1. FOSS Cylinder
KEY WORDS: Smart structure, optical fiber, sensor, filamentwound, pressure vessel.
191
CONCEPT Knapp and Robertson (2000) describe the FOSS cylinder concept. In this paper, the implementation of this concept including the design, manufacture and testing of a prototype cylinder is described. Among the more successful composite gas cylinder designs are the Type III tanks as defined by the U.S. Department of Transportation. This design includes a thin aluminum liner as the gas seal and several layers of composite overwrap for structural strength. Contiguous with the aluminum liner is an internallythreaded stem at one end of the tank that accepts a pressure regulator. It is desired to add a fiber-optic sensor system to a Type III tank to monitor the structural health of the cylinder without damaging the structural integrity of the composite and without adding substantially to the cost of manufacturing. The fiber optic sensor should be embedded into the cylinder wall without modification of current manufacturing techniques. To eliminate the need to interrupt the filament-winding process when the fiber sensor is installed, the entire FOSS system is preattached to the aluminum liner. Prior to filament-winding, the optical fiber is helically-laid onto the surface of the tank using a wind mechanism. By providing two helical wraps in opposite directions, the optical fiber overlaps itself a number of times dependent on the lay angle of the wrap. These overlapping or pinch points create microbends that diminish light transmission according to the magnitude of radial pressure acting on the pinch points. With a sufficient number of pinch points to cover the surface of the tank, any deviation of radial pressure caused by a fault in the composite overwrap might be correlated with a deviation of the light power transmitted through the optical fiber. Of course, it is possible that the radial pressure deviation might not detect a localized fault located away from the pinch points. Nevertheless, the FOSS might still be able to detect faults globally by its measurement of tank diameter change.
and has a 160 mm (6.3 inch) outer diameter. The nominal wall thickness of the liner in the cylindrical region is 2.39 mm (0.094 inches). In the end sections, the wall thickness increases to 5.84 mm (0.230 inches) at the thickest section on the bottom of the liner. The function of the liner is to contain the pressurized gas and to provide a contiguous threaded fitting to attach a pressure regulator. The wall is too thin to contain the internal pressure, so an overwrap of composite is provided to develop structural strength. To prevent damage to the acrylate coating that protects the fiber optic sensor, a room-temperature curing epoxy was selected as the matrix material for the E-glass/epoxy composite. Acrylate begins to soften at about 120°C (250°F), the minimum temperature needed to thermally cure epoxy. Following conventional design practice, two axial wraps cover the entire surface area of the liner and one outer hoop wrap is wound over the cylindrical area only. Thus, where the liner is thinnest in the cylindrical region of the tank, the composite overwrap contains two axial and one hoop wraps. Where the liner is thickest over the polar ends, only two axial wraps are required. The lay angles and thickness of these layers must be determined to satisfy two conditions; viz., that stresses must not exceed material strengths and the radial stress acting on the sensor microbend pinch points is sufficiently large to induce an adequate optical signal. From fiber microbend experiments (Knapp, 2000), it was determined that the maximum pressure that can be applied to the crossed fiber is 2.4 MPa (350 psi). Also, for manufacturability purposes the FOSS is attached directly to the aluminum liner. Thus, finite element analysis must ascertain that at the maximum internal operating pressure, the radial compressive stress at the aluminum/composite interface is limited to 2.4 MPa (350 psi). Also, a factor of safety of at least two for aluminum yield or initial matrix cracking must be provided by the design. The finite element model of the composite tank shown in Figure 2
As the t,qnk is pressurized, the tank diameter increases. In this case, the light power transmitted by the optical fiber is affected by fiber stretch and microbending. If structural damage to the composite overwrap should occur away from a pinch point, a deviation in the diameter change still should produce an abnormal change of light power transmitted that could signal structural degradation. A description of the design, manufacture and testing of a prototype FOSS cylinder follows. DESIGN Tank The aluminum liner provides the basis of the geometry of the tank. For the prototype tank, the liner is 556 mm (21.9 inches) in length
Figure 2. Tank Finite Element Model 192
was used to find a suitable stacking sequence of E-glass/epoxy that overwraps the aluminum liner. The ANSYS Shell 99 layered element (ANSYS, 1999) was used to model the aluminum and composite plies that make up the tank wall. The analysis considered interlaminar sheafing stresses and the Tsal-Wu Inverse Strength Ratio. A summary of the design results is given in Table 1.
the filament-winding process and simplicity of operation (Knapp and Robertson, 1999). The proposed sensor uses numerous optical fiber microbends evenly distributed over the outer surface of the tank. As the tank is pressurized, its diameter increases which induces fiber microbending and an attenuation of light power transmitted through the fiber. Tank dilatation and the attendant light attenuation form the basis of the sensor. By correlating the difference between light entering and leaving the fiber with internal pressure, any significant departure from the light attenuation response of a newly manufactured hull can be used to signal structural degradation. Thus, general creep, fatigue or direct mechanical damage to the composite material might be monitored continuously during the service life of the tank.
Region Bottom End Center Cylinder Top End
Material Thickness Lay Angle
AI 4.8-5.8
Material Thickness Lay Angle
AI 2.4
Material Thickness Lay Angle
A1 4.8-7.9
E/E
E/E
1.0
1.0
12 °
.12 °
E/E
E/E
1.0
1.0
12 °
_12 °
WE
E/E
1.0
1.0
E/E 4.0 85 °
Fiber Sensor Wrap A Coming SMF-28 telecommunication fiber was selected for the sensor. Its low cost and availability, availability of optical fittings such as ceramic ferrules, robust handling characteristics, and satisfactory optical performance as a microbend sensor makes this fiber a good choice to demonstrate technical feasibility. Since this fiber has an acrylate coating that softens at the temperature to thermally cure epoxy, however, a room-temperature curing epoxy system has been used for the composite tank. Future development efforts will identify alternative optical fibers with better microbending sensitivity and thermal insensitivity using a polyimide coating.
12 ° _12 ° Note: A! = 6061-T6 Aluminum; E/E = E-glass/Epoxy Uniply; Thickness in mm; Lay Angle measured from cylinder axis.
The resulting overall size of the aluminum liner and composite overwrap is 172 mm (6.76 inches) in diameter and 556 mm (21.9 inches) in length. Several attempts were made to arrive at an optimal design. The final design is a two-ply axial wrap (±12°from the tank axis) of Eglass/epoxy unidirectional ply that covers the entire surface of the tank. The outermost layer is a hoop wrap (+85 ° from the tank axis) that is wound only over the cylindrical section of the tank. This selection resulted in a factor of safety of two on matrix cracking (governing failure mode) with a tank internal pressure of 3.45 MPa (500 psi). Also, the radial pressure at the aluminum/composite interface is the desired 2.4 MPa (350 psi).
Figure 3 shows the two-layer, contrahelical wrap of optical fiber around the tank with the fiber ends terminating at the bottom end of the tank where a connector housing is installed (not shown). Where the fiber overwraps itself, pinch points are produced that give rise to microbending and the attendant light power loss that serves as a radial pressure transducer between the aluminum liner and first layer of composite. To stabilize the fiber wrap, small dots of cynoacrylate adhesive are applied along the length of fiber at equal intervals to secure the fiber to the tank liner. The adhesive is cured rapidly by a chemical spray, thereby making it possible to install the sensor rapidly and without damage to the fiber.
Fiber Optic Sensor System (FOSS) The FOSS must provide a linear optical signal that reports tank dilatation as a function of internal pressure. Also, a robust connector that attaches to external instrumentation is required. The connector must have repeatable insertion losses so that calibrated sensors provide accurate monitoring throughout the life of the tank.
•
Three types of optical sensors were considered, including Intensity Sensing for simple, low-cost gross detection of structural faults, Spectrometric Sensing for detailed, wide-area sensing (Bragg grating) and Phase Sensing for interferometric measurements of
F
PINCH POINTS
CYLINDER VIEW
strain (Fabry-Perot) (Udd, 1995a,b). The first sensor category, intensity sensing was selected for the low cost of both sensor and external instrumentation, the practicality of its implementation into
Figure 3. Opticai Fiber Wrap
193
A length of fiber at the fiber ingress and egress points is left unsecured so that optical connectors can be attached to the two fiber ends.
Figure 7Jsa~sehematic drawing of the connector housing showing a cover and rubber boot that protect the connectors during filament winding and in service.
Optical Connectors To provide repeatable, low-attenuation connectors, modified ST connector components have been used. These connectors are bonded to the optical fiber sensor attached to the tank liner. Figure 4 shows the component parts (from left to right): optical fiber, an epoxy-bonded ceramic ferrule, a beryllium-copper split cylinder, an aluminum casing and a nylon retaining ring.
B ¢ONNf~'TOIt--
mI.E
IlWmm
op~e.aJ. MI
i
Figure 4. Connector Assembly First, the optical fiber is inserted and epoxy-bonded through the center of a ceramic ferrule. The end of the ferrule/fiber is polished for efficient light power transmission. The ceramic ferrule is pressed into a beryllium-copper split cylinder which then is locked inside the aluminum casing with the nylon retaining ring. A standard external ST connector can be inserted into the right end of the split cylinder through the retaining ring. This results in two ceramic connectors coming into contact with nearly perfect axial alignment for a very low-loss light power coupling.
Connector Housing A cylindrical connector housing with the same diameter as the valve stem is structurally bonded to the bottom of the tank (Figure 5). This part houses two internal optical connectors and serves as a "turn-around" during filament winding. The two internal connectors are potted into this housing as shown in Figure 6. The two connectors are laid into grooves that form a circular arc tangent to the surface of the tank end (Figure 7). The radius of the arc is sufficiently large so that light attenuation through the housing is negligible. The ceramic ferrules are potted into the housing with epoxy to completely fill the grooves.
Figure 7. Connector Housing FABRICATION A primary objective of this effort has been to develop a FOSS that can be economically integrated into the filament-winding process. The approach taken here achieves this objective. The fiber sensor is preattached to the aluminum liner and the connectors are potted into the end housing prior to winding. The connector housing serves the additional purpose of providing a turnaround for the filament winding. Figures 8-10 show the sequence of winding steps. In Figure 8, the double helical wrap of the optical fiber sensor can be seen. At the right end is the connector housing and protective cover that prevents intrusion of epoxy into the optical connectors and provides a rotational support for the tank end.
Figure 5. Housing & Connector (see arrow)
Figure 9 shows that the entire surface of the tank is Covered by the axial wrap. The hoop wrap in Figure l0 covers only the cylindrical portion of the tank. After completing the winding, the part is vacuum-bagged and allowed to cure at room temperature. A slight rise in temperature to about 100°F is created due to the exotherm of the curing resin. The installed fiber optic sensor proved to be useful in monitoring the buildup of radial pressure as winding progressed. Radial pressure was limited to prevent damage to the fiber optic sensor.
Figure 6. Installed Connectors 194
The fiber sensor produced an optical signal that varied nearly linearly with the radial pressure. TESTING The prototype tank was hydrostatically tested according to the diagram in Figure 11. A manual hydrostatic test pump pressurized water from a low pressure hose inlet to an internal tank pressure of 4.8 MPa (700 psi). This produces the desired radial pressure of 2.4 MPa (350 psi) acting directly on the fiber optic sensor located on the aluminum liner. A 1310 lam wavelength light source was connected at one end of the fiber sensor and a power meter to the other end of the sensor. As pressure was increased, light power transmitted through the sensor decreased nearly linearly as shown in Figure 12. The data presented in this figure represent the mean values of six load cycles. As Figure 12 shows, the response for both the loading and unloading cycles is nearly linear.
Figure 8. Initial Axial Wrap
/ - L ~ P o ~ b~ter
~ T ~ J
Figure 11. Pressure Test Figure 9. Completed Axial Wrap
Prototype 1 ,
~=
I=
5
Unload
-J
,
~ -10
._~ ,- -15
t1=
I
~ -2o
1
2 3 Pressure, MPa
4
Figure 10. Completed Hoop W r a p Figure 12. Mean Light Power Loss
195
.5
This result clearly indicates the potential of the embedded optical sensor as a structural health monitor. Although not yet tested, it is proposed that a structural fault in the composite such as a delamination or broken fibers would cause a noticeable deviation in the pressure curve shown in Figure 12. CONCLUSIONS
Smart Structures, John Wiley and Sons, pp. 121 - 153. Udd, Eric (1995a). "Fiber Optic Smart Structures Technology," Fiber Optic Smart Structures, John Wiley and Sons, pp. 5 - 21. Udd, Eric (1995b). "Fiber Optic Sensor Overview," Fiber Optic Smart Structures, John Wiley and Sons, pp. 155 - 169.
The proposed fiber optic sensor system suggests a robust and economical approach for integration with conventional filament winding. Optical connectors are protected during part processing and in service. The proposed system would add very little cost to the production of filament-wound pressure vessels, would improve the safe use of these tanks since slructural integrity can be checked at any time, and would facilitate DOT certification of composite tanks. Additional gas cylinders currently are being fabricated for testing the ability of the embedded FOSS system to detect structural faults. Tests to be performed include cyclical pressurization and localized damage to fibers. Also, connector performance in terms of repeatable insertion loss during the cyclical tests is being evaluated. ACKNOWLEDGMENTS This work was funded by the National Defense Center of Excellence for Research in Ocean Sciences (CEROS). CEROS is part of the Natural Energy Laboratory of Hawaii Authority (NELHA), an agency of the Department of Business, Economic Development & Tourism, State of Hawaii. CEROS is funded by the Advanced Research Projects Agency (ARPA) through grants to NELHA. This report does not necessarily reflect the position or policy of the Government, and no official endorsement should be inferred. REFERENCES
ANSYS (1999). ANSYS User's Manual (version 5.5), Ansys, Inc., Houston, PA. Knapp, RH and Robertson, IN (1999). "A New Concept for Smart Composite Pressure Vessels," Proc. 9~ (1999) International Offshore and Polar Engineering Conference, Brest, France. Knapp, RH and Robertson, IN (2000). "Fiber Optic Sensor System for Filament-Wound Pressure Vessels," Proc. 10th (2000) International Offshore and Polar Engineering Conference, Seattle, Washington. Knapp, RH (2000)."Embedded Fiber Optic Sensor System," Patent Pending. Spillman, William B. Jr., and Lord, Jeffrey R. (1995). "Methods of Fiber Optic Ingress/Egress for Smart Structures," Fiber Optic 196