Structural Integrity of Offshore Wind Turbines: Oversight of Design, Fabrication, and Installation - Special Report 305 (2011)

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.

APPENDIX A Risk-Informed Approaches to Safety Regulation In risk-informed regulation, insights from risk assessment are considered together with other engineering insights. This appendix summarizes basic concepts of modern risk-informed safety regulation as they are currently used in the design of civil infrastructure, focusing on their use in the United States. RISK-INFORMED ANALYSIS AND DESIGN OF CIVIL INFRASTRUCTURE FACILITIES Risk-informed approaches to analysis, design, and condition assessment have reached a state of maturity in many areas of civil infrastructure dur- ing the past three decades, particularly in codes, standards, and regula- tory guidelines that govern design and construction. These documents are key tools for structural engineers in managing civil infrastructure risk in the public interest, and the traditional structural design criteria they contain address risks in performance as engineers have historically under- stood them. For the most part, these criteria have been based on judgment. In recent years, however, innovation in technology has occurred rapidly, leaving less opportunity for learning through trial and error (as is the case in the wind energy industry today). Standards for public health, safety, and environmental protection now are often debated in the public arena, and societal expectations of civil infrastructure have increased. Questions concerning alternative or innovative projects and structural solutions are better answered from a risk-informed perspective. Such a perspective continues to include a significant component based on judgment: the use of a 50- or 100-year mean recurrence interval (MRI) for the design wind effect is an example. However, modern structural reliability tools 137 

138 Structural Integrity of Offshore Wind Turbines have increased the contribution of risk analysis to the rational development of design criteria, which, owing to current computational capabilities, can be far better differentiated and realistic than their 1970s counterparts. This appendix summarizes basic concepts of modern risk-informed safety regulation as they are currently utilized in the design of civil infra- structure and discusses their application to structural design require- ments for mitigation of risk in the built environment. FUNDAMENTALS OF RISK ASSESSMENT FOR NATURAL AND MAN-MADE HAZARDS Risk analysis and assessment tools are essential in measuring compliance with performance objectives, in comparing alternatives rationally, and in highlighting the role of uncertainty in the decision process. This sec- tion outlines a framework for modern risk-informed decision making, providing the background for the implementation of structural design requirements for civil infrastructure facilities in the current construction and regulatory climate. Risk and Its Analysis: Hazard, Consequences, Context Risk involves hazard, consequences, and context (Stewart and Melchers 1997; Vrijling et al. 1998; Faber and Stewart 2003). The hazard is a potentially harmful event, action, or state of nature. The potential for the occurrence of a hurricane or earthquake at the site of a structure is a hazard. The occur- rence of such a hazardous event has potential consequences—building damage or collapse, loss of life or personal injury, economic losses, or damage to the environment—which must be measured in terms of a value system involving some metric. Finally, there is the context of the risk assessment, which is related to what is at risk, what individuals or agencies are measuring and assessing the risk and how risk-averse they might be, the necessity for or feasibility of risk management, and how additional investment in risk reduction can be balanced against available resources. Risk Benchmarks in Current Structural Codes Structural codes and standards and design practice historically have striven to deliver structural products and systems with risks that the pub- 

Risk-Informed Approaches to Safety Regulation 139 lic finds acceptable. In the vast majority of studies to date involving structural performance and reliability, the term “risk” is used more or less interchangeably with “probability” or is thought of as the comple- ment of “reliability” (Ellingwood 1994). Consequences (e.g., economic losses; morbidity and mortality) are included only indirectly, if at all; low target probability goals are typically assigned, somewhat arbitrarily and on the basis of judgment, to high-consequence events. While current codes and standards as well as code enforcement keep failure rates at a low level, no one knows exactly what a socially acceptable failure rate for buildings, bridges, and other structures might be, although structural engineers believe that current codes and standards deliver civil infra- structure with risks that are acceptable in most cases. At the other extreme, the de minimis risk below which society normally does not impose any regulatory guidance is on the order of 10−7/year (Paté-Cornell 1994). Failure rates for buildings, bridges, dams, and other civil infra- structure that may be calculated through the use of classical reliability analysis (Ellingwood 2000) fall in a range between 10−3/year and 10−7/year, a gray area within which risk-reduction measures are traded off against increments in the cost of risk reduction. The notion of having risks “as low as reasonably practicable” (Stewart and Melchers 1997), which is common in industrial risk management, is based on this concept. In sum, what constitutes acceptable risk is relative and can be established or mandated only in the context of what is acceptable in other activities, what investment is required to reduce the risk (or socialize it), and what losses might be entailed if the risk were to increase. The following section considers how the general concepts of risk assess- ment and management summarized above have been implemented for several types of civil infrastructure. The unique nature of each infrastruc- ture type determines how specific risk-informed decision concepts have been implemented. PROBABILITY-BASED LIMIT STATES DESIGN Load and Resistance Factor Design Structural codes and standards applicable to the design of civil infra- structure traditionally have been concerned primarily with public safety (preventing loss of life or personal injury) and, in this context, the collapse 

140 Structural Integrity of Offshore Wind Turbines of a structure or a large portion of it. The probability of structural col- lapse is a surrogate for all other metrics, and limiting that probability addresses the fundamental goal. Most first-generation probability-based structural design codes focus on that performance objective. Other per- formance metrics—direct economic losses from structural damage, indirect losses due to interruption of function, forgone opportunities, and loss of amenity—have not been addressed in current construction regulations but may be of concern to certain stakeholder groups in cer- tain types of infrastructure facilities. The use of classical structural reliability principles and code calibra- tion has historically formed the basis for the development of load com- binations in American Society of Civil Engineers (ASCE) Standard 7-10, Minimum Design Loads for Buildings and Other Structures (ASCE 2010); Eurocode 1, Actions on Structures (CEN 1994); and structural strength criteria found in most standards and specifications (e.g., AASHTO 2007; ACI 2005; AISC 2010). Such codified procedures gloss over the issue of consequence and context by presuming that “risk” and “probability of collapse” are identical. However, these procedures avoid the difficulty of selecting appropriate risk (loss) metrics and transform the analysis of risk into a problem amenable to solution by principles of structural reli- ability theory (Ellingwood 1994; Melchers 1999), which is an essential step in first-generation probability-based structural design. In modern probability-based limit states design codes, the require- ment that the reliability equal or exceed a target reliability is transformed into a traditional safety-checking equation: Required strength (Qd ) < design strength ( Rd ) (A-1) The required strength to resist loads, shown on the left-hand side of the equation, is determined from structural analysis by using factored loads, while the design strength (or factored resistance) on the right-hand side is determined by using nominal material strengths and dimensions and partial resistance factors. The load and resistance factors are functions of the uncertainties associated with the load and resistance variables and the target reliability index. The target reliability index, in turn, may depend on the failure mode (e.g., brittle or ductile) and the consequences of a member failure (e.g., local damage, possibility of global instability). 

Risk-Informed Approaches to Safety Regulation 141 The most common representation of Equation A-1 in the United States is as follows: ∑γ Q < ϕRn (A-2) i ni where Rn is a specified nominal (characteristic) strength, ϕ is a resistance factor, Qni is the nominal (characteristic) load, and γi is the associated load factor for load type i. The design format suggested by Equation A-2 is transparently deterministic, but the load and resistance factors are in fact based on explicit reliability benchmarks (reliability indices) obtained through a complex process of code calibration. Existing Implementation of Load and Resistance Factor Design; Measures of Reliability Buildings The first probability-based design specification in the United States [denoted as load and resistance factor design (LRFD) for steel structures] was introduced in 1986 and has since been followed by several other specifications. LRFD is now a mature concept and has been widely used in structural design practice for the past two decades. The required strength, Σγi Qni, is determined, in all cases, from the set of load combinations stipulated by ASCE Standard 7-10. In first-generation LRFD (Galambos et al. 1982; Ellingwood et al. 1982), the benchmark target reliability index (β) for a member limit state involving yielding of a tension member or formation of the first plastic hinge in a compact beam was set equal to approximately 3.0 for a service period of 50 years, corre- sponding to a limit state probability of approximately 0.0013 in 50 years; annualized, this probability is on the order of 10−5. The value of β equal to 3.0 was selected following an extensive assessment of reliabilities asso- ciated with members designed by traditional methods and is applicable to load combinations involving gravity loads but not wind or earthquake loads (Galambos et al. 1982).1 Reliability indices for other limit states were set relative to 3.0 (e.g., reliability index values for connections are on the order of 4.0 to ensure that failure occurs in the member rather than 1 The annual probability of partial or total collapse of a properly designed redundant structural frame is approximately one order of magnitude less, or on the order of 10−6/year. 

142 Structural Integrity of Offshore Wind Turbines in the connection; because the cost of connection design is determined primarily by fabrication rather than materials, providing the additional conservatism has little economic impact). Similar benchmarks have been adopted for most other building construction materials. Bridges The American Association of State Highway and Transportation Offi- cials (AASHTO) LRFD Bridge Design Specifications dates from 1994, with the 2007 edition being the latest. The probabilistic design methodology adopted there is essentially the same as that used for building structures. The supporting study (Nowak 1995) focused on the strength of individ- ual bridge girders, with truck loads applied to the individual girders through empirically derived girder distribution factors for moment and shear. AASHTO uses essentially the same LRFD format as is used for ordinary buildings and other structures. The load and resistance factors in the LRFD Bridge Design Specifications (AASHTO 2007) were devel- oped in such a way that bridge girders achieve a reliability index, β, equal to 3.5 at the inventory or design level for a service period of 75 years. No distinction is made between steel, reinforced concrete, and prestressed concrete girders in terms of their target reliabilities, nor is the target reli- ability index dependent on the girder span or on whether the girder is simply supported or continuous over internal supports. Offshore Platforms Formal design guidance for offshore structures originated in 1967 with the release of American Petroleum Institute (API) RP 2A (API 1967). This standard used a working stress approach, consistent with the pre- vailing steel design practice for land structures. In 1979, work began on development of an LRFD version of API RP 2A. The format was parallel to that developed by Galambos et al. (1982). The calibration strategy focused on developing partial factors for identified components that would yield a platform design having members and connections equiv- alent to those resulting from use of the existing working stress code. This approach was summarized by Moses and Larrabee (1988): The traditional one-third allowable stress increase for environmental loading found in working stress design (WSD) has been replaced in the Draft RP2A- 

Risk-Informed Approaches to Safety Regulation 143 LRFD by separate load factors (γ) for dead load, live load, wind–wave–current load, earthquake load and wave dynamic load. Resistance factors (ϕ) vary for pile capacity, beam bending, axial compression, hydrostatic pressure, etc. Together, these load and resistance factors provide a level of safety close to present practice, yet provide more uniform safety and economy. Calibrated β-values ranged from 2.0 to 2.8 for a 20-year service life with a 100-year loading event used as the reference load level. Similar val- ues for the North Sea were developed by Turner et al. (1992). Recently, International Organization for Standardization (ISO) 19902:2007, Fixed Offshore Steel Structures, which was based on API RP 2A-LRFD and expanded to include loading specifics for international locations, became available and is referenced in the International Electrotechnical Commis- sion (IEC) offshore wind turbine design standard (i.e., IEC 61400-3) as the offshore structural guidance document. Other Civil Infrastructure Applications As noted above, probability-based design of buildings and bridges has focused on member or component limit states and has measured relia- bility by making use of the reliability index β. More recent applications of risk-informed decision making to civil infrastructure, brought about in part by the move toward performance-based engineering, have con- sidered system behavior and expressed performance through limit state probabilities rather than through use of the reliability index. These devel- opments have been made possible through advances in structural com- putation, which now make nonlinear dynamic analysis of complex building and bridge structures feasible in design. Several standards and guidelines have begun to adopt such concepts. ASCE 7-10 Commentary 1.3.1.3 ASCE Standard 7-10 has implemented a new general design requirement for performance-based procedures. The commentary to these procedures contains two tables with acceptable reli- ability levels: the first stipulates annual limit state probabilities and relia- bility indices for nonseismic events, and the second provides anticipated probabilities of structural failure for earthquakes. These acceptable relia- bility levels are dependent on the risk category of the structural facility and the nature of the structural failure involved. In nonseismic design situa- tions, the acceptable annual probability of failure ranges from 3 × 10−5/year 

144 Structural Integrity of Offshore Wind Turbines for failures that are benign to 7 × 10−7/year for failures that are sudden and lead to widespread damage or collapse. In seismic situations, the accept- able probabilities are conditioned on the design-basis event; for ordinary building structures, this conditional probability (given occurrence of the design-basis event) is 10 percent for total or partial collapse. ASCE Standard 43-05 Standard 43-05 (ASCE 2005) addresses seismic design criteria for nuclear facilities. Like ASCE Standard 7-10, it adopts a uniform risk approach to earthquake-resistant design rather than a uniform hazard approach. Table 1-2 of this standard stipulates target performance goals in terms of the annual probability of failure for facil- ities requiring different levels of protection. For facilities requiring con- finement of highly hazardous materials with high confidence, the target probability is 10−5/year or less, and the structure must be designed to remain essentially elastic under such conditions. CRITICAL APPRAISAL OF EXISTING RISK-INFORMED ANALYSIS AND DESIGN PRACTICES FOR APPLICATION TO OFFSHORE WIND TURBINES Component Versus System Reliability Analysis Most codified reliability-based design for civil infrastructure has focused on individual buildings, bridges, and other industrial facilities for which the hazard can be identified at a point (e.g., Ellingwood 2007). A distin- guishing and essential feature of risk-informed decision tools for wind tur- bine farms in coastal and offshore environments is their ability to account for the spatial correlation in the intensity of the hazard (such as from a hur- ricane) over geographic scales on the order of tens of kilometers within the region affected (Vickery and Twisdale 1995); multiple wind turbine units experience correlated risks under such conditions. In addition, the presence (or lack) of advanced warning systems and the effect on risk- mitigation options should be considered (Lakats and Paté-Cornell 2004). Design MRIs of Joint Wind Effects MRIs of design wind effects for strength design have typically been spec- ified with consideration for knowledge uncertainties. Such uncertainties 

Risk-Informed Approaches to Safety Regulation 145 influence, for example, estimates of wind effects associated with a 50- or 100-year MRI. For typical building occupancies, ASCE Standard 7-10 specifies a 700-year MRI wind speed. Similar MRI estimates are needed for wave and current effects or for combined wind, wave, and current effects. Note that the MRI is insufficient to establish the structural reliability. The associated load factor also plays a key role; for example, the probability of exceedance of some load level, 1.6W, with W determined on the basis of a 50-year MRI wind speed, is about the same as the probability of exceeding 1.0W when W is defined on the basis of a 700-year wind speed. This is also the reason why the IEC-based offshore wind turbine design procedure, which begins with a 50-year wind speed basis and applies load factors of 1.25 or 1.35 when verifying ultimate limit states, might yield the same reli- ability as the use of an alternative factored load that begins with a 100-year wind speed (as in API RP 2A) and applies a load factor of 1.0. Whereas a typical MRI for an offshore oil and gas platform design is 100 years, a 50-year MRI is commonly used for offshore wind turbines in Europe. Although the combination of the MRI and an associated load factor can lead to similar reliability levels with either the 50- or the 100-year MRI, the 50-year MRI used for offshore wind turbines in Europe partly reflects the thinking that consequences of a turbine failure typically do not lead to loss of life or grave environmental effects (see Chapter 4). The selection of MRI for the design-basis event of a facility is not sufficient to determine the risk for that facility. Finally, to account explicitly for economic consequences or the con- sequences of an unreliable energy supply, approaches similar to those presented briefly in this appendix may be used to establish appropriate alternative design MRIs, rather than an approach based on engineering judgment with regard to structural performance. Time-Domain Methods Computer-intensive time-domain methods similar to those recently developed by Simiu and Miyata (2006) and Long et al. (2007) can allow rigorous estimates of (a) combined load effects, with any mean recur- rence interval, from Monte Carlo simulations of simultaneous time his- tories of wind, wave, current, and storm surge effects; and (b) attendant uncertainties in those estimates. Such methods will help to sharpen sig- nificantly estimates of combined load effects used for allowable stress 

146 Structural Integrity of Offshore Wind Turbines design, strength design, limit states design, and design for fatigue, and to define geographical areas whose environmental conditions are compat- ible with the use of specified classes of turbine designs. REFERENCES Abbreviations AASHTO American Association of State Highway and Transportation Officials ACI American Concrete Institute AISC American Institute of Steel Construction API American Petroleum Institute ASCE American Society of Civil Engineers CEN Comité européen de normalisation AASHTO. 2007. AASHTO LRFD Bridge Design Specifications. Washington, D.C. ACI. 2005. Building Code Requirements for Structural Concrete. Standard 318-05. Farm- ington Hills, Mich. AISC. 2010. Specification for Structural Steel Buildings. Chicago, Ill. API. 1967. Recommended Practice for Planning, Designing and Constructing Fixed Offshore Platforms. Washington, D.C. ASCE. 2005. Seismic Design Criteria for Structures, Systems and Components in Nuclear Facilities. Standard 43-05. Reston, Va. ASCE. 2010. Minimum Design Loads for Buildings and Other Structures. ASCE Standard 7-10. Reston, Va. CEN. 1994. Actions on Structures, Part 1—Basis of Design. Eurocode 1, European Pre- standard ENV 1991-1. Brussels, Belgium. Ellingwood, B. R. 1994. Probability-Based Codified Design: Past Accomplishments and Future Challenges. Structural Safety, Vol. 13, No. 3, pp. 159–176. Ellingwood, B. R. 2000. LRFD: Implementing Structural Reliability in Professional Prac- tice. Engineering Structures, Vol. 22, No. 2, pp. 106–115. Ellingwood, B. R. 2007. Strategies for Mitigating Risk to Buildings from Abnormal Load Events. International Journal of Risk Assessment and Management, Vol. 7, Nos. 6–7, pp. 828–845. Ellingwood, B., J. G. MacGregor, T. V. Galambos, and C. A. Cornell. 1982. Probability- Based Load Criteria: Load Factors and Load Combinations. Journal of the Structural Division, ASCE, Vol. 108, No. 5, May, pp. 978–997. Faber, M. H., and M. G. Stewart. 2003. Risk Assessment for Civil Engineering Facilities: Critical Overview and Discussion. Reliability Engineering and System Safety, Vol. 80, pp. 173–184. 

Risk-Informed Approaches to Safety Regulation 147 Galambos, T. V., B. Ellingwood, J. G. MacGregor, and C. A. Cornell. 1982. Probability- Based Load Criteria: Assessment of Current Design Practice. Journal of the Structural Division, ASCE, Vol. 108, No. 5, May, pp. 959–977. Lakats, L. M., and M. E. Paté-Cornell. 2004. Organizational Warnings and System Safety: A Probabilistic Analysis. IEEE Transactions on Engineering Management, Vol. 51, No. 2, May, pp. 183–196. Long, P. T., E. Simiu, A. McInerney, A. A. Taylor, B. Glahn, and M. D. Powell. 2007. Methodology for Development of Design Criteria for Joint Hurricane Wind Speed and Storm Surge Events: Proof of Concept. NIST Technical Note 1482. National Institute of Standards and Technology, Gaithersburg, Md. Melchers, R. E. 1999. Structural Reliability: Analysis and Prediction. John Wiley and Sons, Chichester, United Kingdom. Moses, F., and R. D. Larrabee. 1988. Calibration of the Draft RP 2A-LRFD for Fixed Plat- forms. OTC 5699. Proc., Offshore Technology Conference, Houston, Tex. Nowak, A. S. 1995. Calibration of LRFD Bridge Code. Journal of Structural Engineering, ASCE, Vol. 121, No. 8, pp. 1245–1251. Paté-Cornell, M. E. 1994. Quantitative Safety Goals for Risk Management of Industrial Facilities. Structural Safety, Vol. 13, No. 3, pp. 145–157. Simiu, E., and T. Miyata. 2006. Design of Buildings and Bridges for Wind. John Wiley and Sons, Hoboken, N.J. Stewart, M. G., and R. E. Melchers. 1997. Probabilistic Risk Assessment of Engineering Systems. Chapman and Hall, London. Turner, R. C., C. P. Ellinas, and G. A. N. Thomas. 1992. Towards the Worldwide Cal- ibration of API RP2A-LRFD. OTC 6930. Proc., Offshore Technology Conference, Houston, Tex. Vickery, P. J., and L. A. Twisdale. 1995. Prediction of Hurricane Wind Speeds in the United States. Journal of Structural Engineering, ASCE, Vol. 121, pp. 1691–1699. Vrijling, J. K., W. Van Hengel, and R. J. Houben. 1998. Acceptable Risk as a Basis for Design. Reliability Engineering and System Safety, Vol. 59, pp. 141–150.