While great progress has been made recently in the development of mechanically satisfactory lightguide fibers, detailed fundamental understanding in a number of areas seems lacking. Although in some cases this understanding may seem practically unimportant, in others it is critical to our ability to improve our product as well as to predict its reliability confidently.
Proof testing is the principal technique for revealing the largest fiber flaws that reduce fiber strength below some definite level. At the same time the proof test does not guarantee long-term (25-50 yr) reliability of optical fibers due to the growth of surface defects during proof stress unloading. In the present paper we analyze the possibility of reducing unloading effects by means of shortening the unloading time or proof testing under inert conditions. We have also estimated the reliability of hermetically coated proof-tested optical fibers and obtained simple relations that take into consideration the proof test and service conditions when evaluating the fiber reliability.
The theory behind experiments used in evaluating fiber parameters needed in the prediction of fiber lifetime is given and the theory is put into a general and unified perspective. Demonstrations are given of how measured parameters may be utilized in new equations to calculate lifetime for a fiber under constant stress. Then an experimental program is outlined in which both static and dynamic fatigue data are analyzed for commercially available fibers. The resulting parameters are obtained for several environments and geometries and are used for lifetime predictions for various deployments.
An engineering methodology for the mechanical reliability of optical fiber is developed within a fracture-mechanics framework. The model expresses allowable in-service and installation stresses as a fraction of fiber strength in a fatigue environment for a range of n values and fiber types. Failure probability is incorporated into the model by the measurement of the fiber-strength distribution appropriate to the application. For long-length applications, strength distributions of hundreds to thousands of kilometers of fiber are needed. A 400-km strength distribution captures the beginnings of the truncated portion of the distribution.
Two approaches are compared for estimating static fatigue lifetime at low cumulative failure probability. Both approaches use equations based on linear elastic fracture mechanics, but one uses the distribution of fiber strength to estimate the lifetime at low failure probability, and the second uses the distribution of time to failure. The two approaches give the same result, within experimental error, even though different sets of data are used in the calculation. Both approaches have advantages. Use of the distribution of time to failure is simpler experimentally because it does not require measurement of inert strength. However, this approach is limited to cases in which the strength distribution is unimodal. Use of the strength distribution is more generally applicable because it does not require that the strength be unimodal. Experimental data are presented for fiber tested in two-point bending at 80°C and 60% relative humidity. Lifetime predictions for a bending application are made using both approaches of extrapolation to low failure probability. The uncertainty in the calculations is estimated and the results compared. Two techniques are used to estimate the uncertainty in the lifetime calculation, the propagation of error technique and Monte Carlo simulation. A sensitivity analysis is presented that shows the sensitivity of the lifetime calculation to parameters such as fiber diameter, bend radius, and strength of the fiber.
A phenomenological formulation of Si-O bond dissociation is utilized to interpret stable crack velocity and static and dynamic fatigue phenomena. The resulting model has an exponential form and is applicable to a wide range of flaw sizes, service stresses, and test environments. Furthermore, it is readily reduced to the power law by retaining the first term of the series expansion of the exponential function. The model provides a sound physical basis for comparing different fiber compositions, service environments, and stress-time histories (static versus dynamic) from the fatigue point of view. The application of the model to silica and titania-doped silica optical fibers provides valuable insight into their relative fatigue behaviors and sheds further light on the fundamental mechanisms controlling such behavior.
Prediction of long-term static fatigue for optical fibers under stress requires a model to relate short-term accelerated test results to long-term behavior. The dependence of crack growth on stress intensity is the most fundamental model for this reliability prediction process. Statistical uncertainty for fatigue testing is shown to be significant, but typically smaller than the model uncertainty, which has been neglected in the literature. More research is needed to determine the most appropriate model. It is shown that the differences in allowed stress predictions between models become quite large at long times, especially for fiber of the same strength as that used in the fatigue test. Data-independent conversions of allowed stress from the common power model to other models provide an assessment of the difference between models for various situations. In many applications, the differences are in the range of typical safety factors. However, since model differences are quite large in other applications, universal use of the optimistic power law model is not appropriate, given the limited understanding of fatigue in optical fibers.
A technique for measuring the fiber strength distribution of many kilometers of fiber was developed. The strength distribution of 100 km can be generated in a week's time by a single operator. A strength distribution for 400 km of fiber is shown to depart from the high-strength region around the 1% failure probability level and the data exhibit the beginnings of the truncated portion of the strength distribution. Such data are believed to be useful in making failure probability predictions, process improvements, and aid in the understanding of crack-growth behavior during proof testing.
Optical fiber coatings are in continuous contact with filling cornpounds in optical cables. They are also exposed to hydrocarbon liquids (either as cleaning or lubricating fluids) during splicing operations. We
observed that hydrocarbon solvents such as toluene, acetone, ethyl alcohol,isopropyl alcohol, and light rnineral oils (in cable filling cornpounds)cause swelling and self-stripping in some dual-coated fibers. A numerical stress analysis for a swollen dual-coated fiber revealed that swelling induces
a compressive radial stress and a tensile tangential stress in the secondary coating; both stress components attain their maximum values at the primary/secondary coating interface. The self-stripping process occurs when the energy stored (due to increasing tensile tangential stress at the interface) exceeds the fracture toughness of the secondary coating. This analysis has provided quantitative measures of coating hydrocarbon
solvent/filling compound compatibility and will establish a rational basis for compatibility in optical cable filling compound design.
Low water sensitivity optical fiber coatings protect optical waveguides from the damaging effects of moisture on static fatigue. The standard test method used to determine the water sensitivity of UV-curable optical fiber coatings does not give a realistic measure of a coating's water sensitivity. The maximum absorption can be reached before 24 h, the extraction continues even after 24 h, and sample pre- and postconditioning greatly influence the results. In the proposed new test method, a number of weight change measurements of samples immersed in water are made as a function of time. Because of the dynamic nature of the test, a more realistic understanding of the water sensitivity of optical fiber coatings is obtained. The effect of different pre- and postconditioning environments is investigated and a recommendation for the best method is made.
Significant degradation of the fiber coatings occurred in an optical fiber cable that was aged for five days at 85°C. Incompatibility between the coating and the surrounding buffer tube gel was the suspected cause. To investigate this problem, fiber samples were aged in several gels, the compositions of which were determined using a variety of analytical techniques. Coating degradation occurred as cracking or discoloration in all but one of the gels. These degradations were found to have a strong temperature dependence and to be closely related to the molecular weight distributions of the gel components. Compatibility screening was identified as a necessary procedure to avoid future recurrences of this problem
Optical fiber coatings play an important role in fiber strength, fatigue, and attenuation. Polyimide coatings on optical fiber are in growing demand because of their excellent high-temperature properties. Mechanical and optical characterizations of polymide-coated optical fibers were carried out at different temperatures. The strength was virtually unchanged at room temperature and at 300°C in air. Fatigue parameters (n values) were greater than 30 when tested in water at 80°C. Optical attenuation was unchanged when the fibers were tested at 340°C for 48 h. When the fibers were subjected to - 40°C to +70°C temperature cycling, the change in attenuation was less than 0.1 dB/km at 850 nm.
During fusion splicing of hermetic carbon-coated silica fibers, the local carbon film is removed. A process to restore the carbon coating in the splice region is described. A CO2 laser is used to heat the silica surface locally, and the atmosphere of the bare section is controlled by a specially designed reaction chamber. An average strength of 2.91 GPa is realized in the carbon-overcoated region. A minimum n value of 165 is determined from dynamic fatigue measurements of carbon-overcoated splices. Furthermore, static fatigue measurements made in concentrated hydrofluoric acid solutions demonstrate the hermetic nature of this film. This process, therefore, offers a practical technique to overcoat spliced hermetic carbon-coated silica fibers.
Hydrogen-induced loss increases are known to occur in optical fibers, due both to the presence of molecular H2 and to the reaction of hydrogen with defects in the fibers. To predict the long-term loss increases expected for fibers under normal conditions, it is necessary to rely on accelerated aging experiments in which fiber loss increases are measured at high temperatures and high PH2'S. For the predictions to be reliable, the dependences must be characterized on temperature, PH2, and time for each of the hydrogen-induced loss mechanisms. In some cases, simplifying assumptions can be made in arriving at conservative lifetime predictions. Hermetic coatings can be used as one means of minimizing the amount of hydrogen seen by a fiber. The performance of a hermetic coating can often be evaluated by using accelerated experiments performed at high PH2's and moderate temperatures. The presence of reactive gettering sites can have a significant effect on the loss increases measured in high-temperature experiments with hermetic fibers. Experimental resuIts are presented for hermetic and nonhermetic fibers as examples of how long-term predictions can be made.
Laboratory studies established that corroding galvanized steel and hydrogen-producing bacteria can generate enough hydrogen to cause appreciable signal attenuation in optical fibers. Simulations of these two effects in the laboratory caused hydrogen indicators to increase up to 7.2 dB/km. The effect of hydrogen-producing bacteria can be suppressed by sulfate-reducing bacteria (SRB) in the absence or limited presence of fermentable hydrocarbons. In the abundant presence ofthese hydrocarbons, the optical fibers will attenuate, and even high concentrations of SRB cannot counteract this effect. During the fermentation process, acetic and butyric acids are produced. If these acids lower the pH of the environment below 5, the hydrogen is consumed by the formation of n-butanol and acetone. Data from fiber attenuation experiments in H2/N2 gas mixtures show that the attenuation at saturation increases approximately linearly with increasing p(H2) and decreases with increasing temperature.
Optical fiber system applications have progressed beyond massive production for long-haul telecommunications systems to special systems for military use. The environmental requirements for military systems are extreme. Aircraft systems must endure extreme shock during takeoff and landing on aircraft carriers and thermal shock during a rapid climb from the desert floor to 50,000 ft. Ship systems must weather continuous salt water spray, and space systems must survive the natural space radiation environment. All these systems as well as fixed and mobile ground-based installations may be required to endure the potential threat from nuclear weapons. Research efforts have improved the reliability and survivability of fiber optic systems under these adverse conditions. Radiation test results over extreme temperature ranges have verified that optical fibers have been developed to meet the most adverse of conditions.
The establishment of proper test procedures and appropriate evaluation techniques will provide reliable fiber optic components for the adverse nuclear environment.
A finite element model has been developed to evaluate the dynamic response of aerial fiber optic cables. The results of several dynamic excitation scenarios are presented and discussed. These include wind galloping and sudden movement of a support pole when impacted at high speed by a vehicle. Fiber optic cable splice closures attached to the support strand are potentially vulnerable under these conditions. Inertial loads derived from the computer simulations are given for evaluating the mechanical reliability of splices, splice trays, etc., contained within aerial splice closures
Service availability objectives for "fiber in the loop" telecommunication systems place limits on maximum component failure rates. For example, an optical network unit (ONU) located at or near the customer's location should have a maximum average downtime of
26 mm/yr/line, which translates to a failure rate of about 8250 FITs (8.25 ONU failures per million unit-hours). Laser modules used in this network element, therefore, need to be less than ~5000 FITs. Estimates of component failure rates, however, often have large uncertainty. For example, laser module reliability cannot be predicted more accurately than a possible range of 3000 to 12,000 FITs; the reliability of the fiber pigtail and its alignment to the laser are a major source of this uncertainty. More information is required about the laser module and other fiber optic components to increase the confidence in failure rate projections.
Many technical advances have been made in the field of optical fiber connector technology over the past 17 years, particularly with respect to single-mode fiber connectors. Millions of these connectors are now in use, increasing the need for complete understanding and resolution of reliability issues. This paper will highlight some of the important reliability-related problems in butt-joint type connectors.
The effect of dynamically changing visibility conditions on target acquisition is considered. A model is described that takes as input the statistical characterization ofthe obscurant and the physical char acteristics of the target; the output is the probability of detection as a function of time. Analysis indicates that the variation of the obscurant as a function of time greatly affects the predicted target detection times and, hence, the utility of the infrared system in realistic scenarios.
Conventional optical systems used for whole-field specklegram reconstruction are limited by the size of the lenses, the uniformity of the collimated beam, and the off-axis capability. Outside the Fresnel and paraxial regions, the fringe patterns will suffer from various aberrations. The maximum obtainable sensitivity, therefore, is severely restrained. A new optical system and instrument-the high-frequency optical Fourier transform analyzer-specifically designed for whole-field specklegram reconstruction, can extrapolate the spectrum beyond the passband. It can extract information at any point within the hemisphere at which diffraction occurs. It also minimizes aberrations, allowing the sensitivity of speckle metrology to be increased to levels commensurate with those of moire interferometry.