Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
- Background Presentations This section of the report summarizes the seven invited presentations that were given during the course of the workshop. Five presentations were given during the introductory session on the first morning. Two were given prior to the subgroup discussion sessions on life-prediction issues (on the first afternoon of the workshop) and accelerated-testing (on the second morning of the workshop). Materials used during the presentations are found at the end of this report. FIVE OVERVIEW BRIEFINGS The first morning of the workshop was devoted to five overview presentations that were intended to convey the breadth of applications for accelerated-testing and life-prediction modeling in predicting systems performance. The first presentation was by John M. Hanson of North Carolina State University, who reviewed the aging and deterioration issues associated with civil infrastructure in the United States and current life-prediction methodologies and accelerated-testing methods applied to these systems. To encourage discussion between members of the various application communities present and to review the state of practice in other fields that could potentially be applied to civil infrastructure, the next four presentations focused on materials durability, reliability, and degradation issues in other fields. Jack E. Lemons of the University of Alabama and John Anderson of Case Western Reserve University spoke on the durability, reliability, degradation, and life-prediction issues associated with surgical implant devices. Richard Wachnik of IBM Microelectronics Division gave a presentation on the reliability and testing of high-performance integrated circuits. Carol M. Jantzen of the Savannah River Technology Center discussed vitreous materials for the long-term storage of hazardous and radioactive waste. The final presentation of the morning session, given by John Stringer of the Electric Power Research institute, focused on life- cycle performance in the electric utility industry. 12
BACKGROUND PRESENTATIONS Infrastructure Aging and Deterioration John M. Hanson, North Carolina State University 13 Professor Hanson began his presentation by stating that many people currently believe that practicing engineers and the construction industry in the United States have been slow to change and have failed to take advantage of innovations in technology that could improve the nation's infrastructure. He believes that this view should be put into proper perspective by acknowledging the significant advances that have been made in construction in the past two or three decades. For example, a review of the record height of concrete buildings in the past two decades shows how advances in technology have been translated into the construction of taller buildings. Professor Hanson pointed out that the buildings, bridges, and tunnels constructed in the United States are comparable to similar structures around the world and that U.S. engineers and contractors are frequently sought as consultants in large construction projects around the world. Professor Hanson also reminded the group that problems have often resulted from infrastructure materials and products being introduced before adequate experience with them had been gained and before their responses to the environment were fully understood: . Field welding led to many of the fatigue and fracture problems that plague steel bridges. The introduction of Sarabond, an additive in masonry mortar, caused so many problems that it was withdrawn from the market, and hundreds of buildings required recladding. The epoxy coating of embedded reinforcements for concrete was supposed to prevent corrosion, but extensive use revealed that corrosion still occurred at pinholes or where the coating had been damaged during construction. Glass-fiber reinforced concrete for wall panels was subject to the unanticipated problem of bowing with exposure to sunlight. The marble panels used as cladding on the Amoco Building in Chicago had to be replaced with granite because of excessive bowing and degradation of strength from aging. The widespread use of timber treated with fire retardent in roofs has required major repairs because of unanticipated fractures after the materials had been in · ~ service tor many years. Stucco applied over insulation on thousands of homes and buildings has had to be removed and replaced because the wood framing underneath has rotted; thousands of other buildings will probably have to be repaired for the same reason. The defining conditions for a good structure, Hanson said, are adequate strength, acceptable serviceability, and long-term durability. Mechanical response factors, (i.e. load resistance, stability, fatigue resistance, and fracture resistance)
14 RESEARCH AGENDA FOR TEST METHODS AND MODELS must be addressed to ensure structural integrity. According to Professor Hanson, the mechanisms of deterioration of the primary construction materials (i.e., concrete, steel, masonry, and timber) are quite well known, at least at the level required of an engineer. Concrete materials, as well as masonry materials, may be subject to scaling due to freezing and thawing, chemical attack, alkali-silica reaction, and corrosion of embedded reinforcements. The deterioration of structural steels is mainly due to corrosion, and the deterioration of timber is due mainly to decay. The rate of deterioration (i.e., durability) is greatly affected by environmental factors, as well as the details of construction. However, our understanding of these mechanisms is currently not sufficient to enable us to make quantitative life predictions. Additional research to enhance understanding of the basic mechanisms is critical to improving the durability of materials and structures. Professor Hanson emphasized however, that reviews of many infrastructure failures have shown that very few occurred because of deterioration, except when the structure or system had an underlying design or construction defect. Some forms of deterioration seem to slow down or even stop after a period of time. Thus, the deterioration of a material or structure does not necessarily affect safety (See, for example, NRC, 1997 and Levy and Salvadori, 1992~. Proper maintenance can prolong the life of materials by slowing their rate of deterioration. Professor Hanson concluded his presentation by stating that, although the development of a concrete that does not shrink or creep or a steed that does not corrode would be of great benefit to the construction industry, the likelihood of such a material being developed is considered to be very small. Of course, advancements have been, and continue to be, made, but accurate tests for assessing the effects of environmental conditions on their lifetimes will require a much better fundamental understanding of the damage process before they can test results could be used for making life predictions. The adoption of these materials will also depend on their economic advantage in a highly competitive market. Thus, Hanson believes that the lifetimes of structures are more likely to be extended by improvements in the quality of construction, than the use of new materials. References Levy, M.P., and M. Salvadori. ~ 992. Why Buildings Fall Down. New York: W.W. Norton and Company. NRC (National Research Council), ~ 997. Nonconventional Concrete Technologies: Renewal ofthe Highway Infrastructure. Washington, D.C.: National Academy Press.
BACKGROUND PRESENTATIONS Infrastructure Considerations: Surgical Implant Devices Jack E. Lemons, University of Alabama, and John Anderson, Case-Western Reserve University 15 Dr. Lemons began by stating that the quality of life continues to be an important aspect of human welfare. esnecialiv as the nonulation ares Thins two , ~ , ]7 ~ ~ , ,, _ ~ , · ~ · . · . . ~ important issues tor surgical implants and reconstructions are longevity and quality of function. The primary issues related to the longevity of implant devices include time of implantation, patient age and level of activity, and systemic health and level of function. Many devices constituted from synthetic materials (biomaterials) can function for decades, affording complete freedom from chronic pain and the continuation of normal activities. Some surgical implant reconstructive systems are known to have shorter lifetimes, however, and, in some situations, successive revisions may have even shorter lifetimes. Dr. Lemons explained that devices for surgical reconstruction of the musculoskeletal system are subjected to high-magnitude forces (up to seven times body weight) and, because of the dimensions of the anatomical sites, device components can be subjected to mechanical stresses approaching the strength limits of the biomaterials. The biochemical environment is a harsh organic and salt-containing solution (saline), which can cause corrosion and reduce the stability of interfaces for attachments between devices and supporting tissues. Using joint and tooth-root replacement systems as examples, Lemons and Anderson then reviewed the research, development, and application experience required for the introduction of a new materials design. Both speakers stressed the need for basic research and development, an understanding of the biomechanical and biochemical properties of the biological host, and the types and sources of degradation of biomaterials. They also explained why standardized methoclolo~ie~ would be helpful for monitoring device-related outcomes. _ ~ As High Performance Integrated Circuits Enter the National (and International) Infrastructure, How Do We Know They Are Reliable? Richard Wachnik, IBM Microelectronics Division Integrated circuits are critical parts of the data and telecommunications of effort has been exceeded in trying to infrastructure. Although a good deal , , ~ understand the detailed phenomena underlying the degradation of integrated circuits, determining their stability or reliability is often largely empirical. This is partly a reflection, Wachnik said, of the pace of change demanded by the economics of the industry. For many products, conservative design practices are used to minimize the risk of being left behind in performance or function. The rapid obsolescence of new products makes many aggressive practices conservative in hindsight. Thus, there is tremendous leverage in building reliability into the process and ground rules, and less in accurately describing degradation phenomena. Wachnik believes that this paradigm will change as
16 RESEARCHAGENDA FOR TESTMETHODSAND MODELS integrated circuit technology matures. Barring unforeseen innovations, future competitive performance and density will depend on pushing reliability to its limits and providing product designers with easy access to the tools required to implement the most aggressive possible designs. Integrated circuit technology can be divided into three subdisciplines: (~) devices; (2) chip-level interconnections; and (3) higher level interconnections and product reliability. The assessment and understanding of the reliability of integrated-circuits draws on many subdisciplines, including physics, chemistry, and materials science. The most critical degradation mechanisms of silicon devices are hot carriers; dielectric degradation and breakdown; and radiation effects. Dielectric degradation includes polarization, which affects performance, and leakage, which affects dynamic circuits and potentially affects static circuits. Radiation effects include single-event upsets as well as radiation damage. Assessing the reliability of silicon-based devices depends on knowing the locations of the energy carriers and the kinetics of defect formation by energetic carriers. Accelerated testing can be done by increasing the applied voltage and internal fields to heat the carrier distributions. Temperature accelerates dielectric breakdown and polarization, and increased radiation flux accelerates radiation effects. The details of operation can then be used to scale the actual stress time for actual products. The understanding of hot-carrier degradation has progressed the most rapidly because of its close relationship with device design and the ease with which degradation kinetics can be examined using electrical characteristics of the device. Advancing the understanding of radiation effects, and especially dielectric breakdown, has lagged behind, partly because of difficulties in studying the mechanisms in detail. The reliability or stability of the chip-level interconnections, or wires between the transistors, depends on assessing the kinetics of diffusion that are driven by stress gradients or momentum transfer from the carriers in the circuits. Electromigration failure can be accelerated by increasing both temperature and current. The acceleration of stress migration is problematic, however, because the increases in temperature that speed diffusion also tend to reduce the stress, which is the driving force of failure. The reliability of higher level interconnections or packaging is governed by mechanical factors (e.g., fatigue in particular) and electrochemical factors (e.g., moisture entering the circuit or its physical interface). Accelerated fatigue testing is done by increasing the amplitude during thermal cycling. Increasing humidity, temperature, and voltage can accelerate corrosion, as well as mechanical degradation from swelling or crack formation. Wachnik explained that much of the focus on wiring, or interconnections, has been on describing semi-empirical kinetic models of time to failure. Successful, robust processes degrade gracefully, not catastrophically. Thus, making accurate parametric descriptions of graceful degradation available to product designers will be important in the long term. Wachnik also explained that the role of the dielectric separating the wires becomes critical when new materials · 11 ~· ~. · ~a~ ~
BACKGROUND PRESENTATIONS 17 are introduced. Although this has not yet become a problem, it is a significant and practical issue in defect-controlled reliability. According to Wachnik, high-performance first-levl] and second-level packaging depends on providing high-density area connections between integrated circuits and the substrates and boards to which they are attached. The reliability of the interconnection requires that careful attention be paid to the mechanical properties of the material and solders, as well as their sensitivity to moisture. Because of trade-offs between cost and complexity, a careful evaluation of the role of moisture in the overall reliability of the product should be carefully evaluated. Wachnik noted that his presentation was focused on concerns about the product wearing out when the lifetime of the product is limited by the material and/or its application. Another critical problem is finding. analyzing. and controlling defects that can cause early failures. Wachnik concluded his presentation with a reminder that many opportunities remain for improving our understanding of the degradation of integrated circuits and improving predictive capabilities. Some areas related to process integration should be investigated in cooperation with industrial development and fabrication organizations, but some (e.g., physical and chemical investigations of degradation mechanisms) should be pursued by academic and research-oriented organizations. =7 ~- ~---= ~_ Durable Glass for Thousands of Years? That Is the Question. Carol M. Jantzen, Savannah River Technology Center Dr. Jantzen began by emphasizing that durable glasses used to stabilize a wide variety of hazardous, mixed (i.e., radioactive and hazardous), and radioactive wastes require modeling and assessments of the long-term stability of glass under a variety of environmental conditions. Because disposal scenarios vary greatly, the effects of kinetic parameters must also be modeled. Because the wastes being stabilized vary greatly, the effect of the composition of the glass on long-term stability must also be modeled. This is especially important for glasses used to stabilize highly radioactive waste. In these cases, the glass must be durable and retain radioactive species for thousands of years until they decay. The chemical durability of glass is a complex phenomenon that depends on both kinetics (e.g., temperature, length of time the class contacts a solution exposed surface area, volume of the solution, .. . . . . ... . . .. .. and glass surface) and mell~locynamlcs E.g., glass composition, 1ncluctlng me concentration of oxidized and reduced species; and glass homogeneity). Long-term durability modeling is usually based on acceleration of the dissolution process by the acceleration of one or more of the kinetic test protocol parameters. Extreme caution must be used to maintain the dissolution mechanism being modeled, to verify the long-term durability using natural analogs, and to perform service-life tests in actual disposal environments. The mechanisms modeled for glass durability are
18 RESEARCH A GENDA FOR TEST METHODS AND MODELS complex. Dissolution occurs when individual ions diffuse out of, condense in, or precipitate on the leached layer via one of four operative mechanisms: ion exchange; matrix dissolution; accelerated matrix dissolution; or surface layer (possibly of a protective or passivating nature) formation. These mechanisms control the overall durability of glass. According to Dr. Jantzen, the modeling for the past 70 years of the durability of glass as a simple function of composition has shown that the durability response is nonlinear. Consequently, most durability models are either empirical or kinetic. Early kinetic models treated glass dissolution as a simple diffusion process. More recent models mathematically describe the glass dissolution mechanisms in the form of time-dependent master equations rather than as simple diffusion processes. Although the kinetic models describe the leaching behavior of a given glass, they cannot predict which of a given group of glasses will be most durable or whether a waste glass of composition A will be as durable as a given natural analog of composition B. Dr. Jantzen concluded her presentation with a discussion of the thermodynamic hydration energy reaction mode] THERMOS. Thermodynamic energy additivity was first proposed as a mode] for understanding the mechanistic relationships between glass structure, composition, and physical properties as early as 1945. Thermodynamic modeling was applied to medieval and Roman window glass in 1977. Modeling of complex waste glasses, initiated in 1982, resulted in the development of an improved model, THERMO_, in ~ 995. THERMO_ lineariv predicts the d~,rnhilitv of AL {mm its romnn~itinn ~ r J ~ r by mechanistically modeling both general glass dissolution and accelerated glass dissolution. These mechanisms are modeled as a function of solution pH and weak acid-strong base equilibria. THERMO_ discriminates between the durability response of homogeneous or phase-separated glasses by a compositionally dependent, phase-separation discriminator. The mode! can predict the durability of glass in environmentally specific (e.g., pH-Eh) environments. Predictions of thermodynamic reaction products derived from THERMO_ can be used as input to computer codes used for reaction-path durability assessments in a variety of disposal modeling and long-term (repository) environments. Life-Cycle Performance in the Electric Utility In(lustr~r John Stringer, Electric Power Research Institute Dr. Stringer began his presentation by stating that the electric power system in the United States is a single, very large, interconnected entity. For the purposes of this presentation, however, he divided the system into three distinguishable components: generation, transmission, and distribution. The failure of a single component in any part of the electric power system, however, can result in serious and expensive consequences, and concerns about potential failures are increasing as the system ages.
BACKGROUND PRESENTATIONS I 19 Forty percent of U.S. power generating capacity will be more than 30 years old by the year 2000, and many parts will be considerably older. The generation component involves: coal-fired thermal systems, which are responsible for approximately 55 percent of the electricity generated in the United States; natural-gas-fired systems, which include Rankine and Brayton-cycle based systems; oil-fired systems; nuclear-power thermal generators, which currently represent about 20 percent of generating capacity; hydroelectric generators; and a very small percentage of other sources, including biomass-fired systems, wind turbines, and solar photovoltaic systems. The major lifetime issues are related to the significant fraction of thermal generating plants that are more than 25 years old, which was their notional lifetime. Many hydroelectric plants are even older, some having been built before the Hoover Dam, which is more than 60 years old. Dr. Stringer said the power transmission and distribution components are also old and that the determination of their remaining life is complicated by their inaccessibility. For example, New York City has an extensive underground distribution system, some of which dates to the time of Edison. A significant part of the transmission component was put in place in the 1950s, and a second part was laid in the 1970s as the demand for power grew. Much of the transmission system consists of"cable in pipe." Examinations have shown that the cable itself may have a lifetime of more than 100 years, in the right circumstances, but that the conduit can deteriorate even if the casing (pipe) is unaffected. For example, the major failure that blacked out Auckland, New Zealand, appears to have resulted from a rise in the temperature of the cable, which caused the dielectric to fail; this failure appeared to be the result of a decrease in the thermal conductivity of the surrounding earth (the "root cause") related to weather conditions. The main technique for avoiding failure is to attempt to determine the remaining lifetime of critical components to guide so-called "runJrepair/replace" maintenance strategies. This approach involves (~) identifying the critical failure, or life-limiting, processes; (2) identifying the root causes of these failures; (3) determining whether a damage accumulation process can be identified that would allow a life fraction to be measured; and (4) developing instrumentation and inspection procedures for making predictions. EPR] has been developing models for several key components (e.g., combustion turbine blades, thick-section pressure components in steam systems, and boiler tubes) for a number of years. These are computer-based systems, in some cases resembling expert systems, to assist operators. In addition to deciding on remedial actions, these systems can also advise an operator of the effect on component lifetime of operations outside the nominal range. This approach has been advocated for dealing with infrastructure lifetime issues, Stringer concluded, and opportunities for advancement include the development of smart materials and systems. TWO FOCUS BRIEFINGS Two additional invited presentations were given during the course of the
20 RESEARCH A GENDA FOR TEST METHODS AND MODELS workshop. The first focus briefing, by Professor Kenneth Reifsnider of Virginia Polytechnic Tnstitute and State University, was a discussion of life-prediction issues associated with infrastructure applications. The talk was given before the subgroup discussions, which took place on the first afternoon of the workshop. The second briefing, by Dr. Jonathan W. Martin of the National Tnstitute of Standards and Technology, focused on accelerated testing. This talk was given on the second morning of the workshop, prior to the second subgroup deliberations. Life-Prediction Approaches for Infrastructure Applications Kenneth Reifsnider, Virginia Polytechnic Tnstitute and State University For the purpose of life prediction, Professor Reifsnider described "life" as a function of time, cycles, or history to the "failure" of a "component." Failure was defined as unsuitability of service based on measurements of stiffness, strength, properties, appearance, and other factors. He defined a component as a structure, element, joint, bond, or sub-element. Life prediction for infrastructure is complicated by complex environments, long service lives (often exceeding 100 years), dynamic and stochastic applied conditions of load, strain, temperature, and moisture, and the quasi-brittle, reinforced materials of which infrastructure is constructed. The basic issues in life prediction are understanding physical degradation processes at the basic or constituent level; modeling physical rate processes and the evolution of material states; establishing independent physical observables that track the processes; modeling the effects of combined processes; and validating models on "real" structures. He then described the four elements of life prediction: the need to describe the physical behavior, i.e., damage and failure modes modeling the behavior, including discrete events and multiple processes identifying measurable independent observables as inputs to the models actual life predictions, which are extensions, generalizations, and accelerations of laboratory experience Professor Reifsnider explained durability and the damage-tolerance approach to life prediction, described relevant environmental factors, and described the specific mechanism of polymer degradation. He then presented data on a number of physical processes affecting service life and showed how these relationships could be incorporated into models of physical processes, and, ultimately, into more complex, predictive models. He described a damage accumulation approach for modeling the combined interactive effects of fatigue, creep, stress rupture, environment, and microdamage. This mode! has been applied to life prediction of a buried multilayer composite pipe and a highway bridge incorporating composite elements.
BACKGROUND PRESENTATIONS 21 He then offered some observations on life prediction of composites. Changes in composite properties may not follow changes in matrix properties; property evolution may be substantial; and mechanical/thermal/chemical coupling may be significant. He noted the need for analysis and experiments on changes in material states; a materials science base for dependence relationships with time, temperature, and environments; and data with which to populate models. Accelerated-Testing Approaches for Infrastructure Applications Jonathan W. Martin, National Institute of Standards and Technology Dr. Martin described a reliability-based methodology for predicting the service life of infrastructure materials. The current durability methodology has been used in the construction and other industries for at least SO years to simulate natural outdoor weathering factors in the laboratory. Improving the predictive capabilities of this methodology, however, has proved to be difficult. The problems have generally been ascribed to inadequacies in laboratory-based aging tests, especially the difficulty of isolating the ideal "balance of weathering factors." According to Dr. Martin, however, the failure of the current methodology can be attributed to faulty premises, inadequacies in experimental design, and the difficulty of replicating weather conditions realistically over time. Dr. Martin then described an alternative reliability-based methodology that has a strong mathematical and scientific basis and a long history of successful applications in the electronics, medical, aeronautical, and nuclear industries. A number of experiments with coatings and other construction materials have already been conducted using this method, and the results indicate that this methodology will be generally applicable to a wide range of infrastructure materials, components, and systems (Martin et al., ~ 996~. He noted that implementation of a reliability-based methodology will require substantial changes in the current experimental procedures including: (~) the design of improved exposure equipment; (2) the systematic characterization of the initial properties of coating systems, (3) the quantitative characterization of each weathering variable in the in-service environment; (4) the quantification of macroscopic degradation and relating submacroscopic to macroscopic measures of degradation: (51 the use of experimental design techniques in Knin ~nr1 _=,__ ~- of-- --- r~-~~~~~= A .. . . . ~ . . ~ _. _ executing short-term, laboratory-based experiments, and (6) the development of computerized techniques for storing, retrieving, and analyzing collected data. Dr. Martin believes that these changes will be justified by greater reliability of the models and the speed of obtaining results. Reference Martin, I.W., S.C. Saunders, F.~. Floyd, and I.P. Winburg. 1996. Methodologies for Predicting the Service Lives of Coating Systems. Blue Bell, Pa.: Federation of Societies for Coatings Technology.
Appendixes