Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 43
Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile 5 QMU and the RRW Program Task 4: Evaluate whether the application of the quantification of margins and uncertainties used for annual assessments and certification of the nuclear weapons stockpile can be applied to the planned Reliable Replacement Warhead program so as to carry out the objective of that program to reduce the likelihood of the resumption of underground testing of nuclear weapons. From a historical perspective, there is evidence that some new nuclear warhead designs could be certified without nuclear testing. The Hiroshima bomb (Little Boy) was never tested before the military used it. The weapon tested at the Trinity site and used at Nagasaki (Fat Man) was of a quite different design. Designers could not accurately predict the yield to within limits acceptable at the present day, but they were within a factor of two of the actual yield, which is remarkable given that there were no test data on which to base their calculations. Today’s historical archive of approximately 1,200 U.S. nuclear tests provides an extensive database. Only a small fraction of the test archive is directly relevant to a particular warhead design, but the whole set has informed the design labs’ understanding of the physics involved.1 Further, that small fraction provides established reference points for performance (including tests with unexpected or even no nuclear yield). 1 The entire archive provides the basis for what we have called “designer or expert judgment” in this report.
OCR for page 44
Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile Given this knowledge and the archive of tests, a new nuclear weapon that is nearly identical to a tested existing design (such as the heavy B83 gravity bomb) and that needs only some relatively insignificant modification could be certified. All existing U.S. missile warheads, however, have smaller M/U ratios than the B83, because they were constrained to maximize yield while minimizing weight and size. The first project in the RRW program was for a replacement missile warhead. The requirement for yield-to-weight ratio was relaxed considerably for the RRW competition, which enabled a design with substantially greater M/U ratio; this is the WR-1 design of LLNL. Also relevant to the RRW program is legislation passed by Congress2 that directed NNSA to begin a new Science Campaign called Advanced Certification, saying that “[Congress] believes the recent findings of the JASON group revealed significant systemic gaps in NNSA’s stockpile certification process.” The findings referred to are in a JASON report on the RRW.3 In the legislation, NNSA is directed to report to Congress on Advanced Certification within six months of enactment; at this writing, the mandated report has not yet been issued. To give the reader an idea of what Advanced Certification might embrace, the study committee summarizes here Congress’s direction for Advanced Certification: Improvement of the weapons certification process through expanded, independent peer review mechanisms and refinement of computational tools and methods. Advancement of the physical understanding of surety mechanisms. Further exploration of failure modes. Manufacturing process assessments. The study of strategic system-level requirements. Finally, Congress calls for NNSA to state in its report “progress [NNSA has] made in implementing the JASON’s recommendations and improving the stockpile certification process.” The JASON report specifically concerns RRW, but the committee believes that the intent of Congress is that Advanced Certification should apply to life-extension programs and annual assessments as well as to RRW. This observation is supported by recent Senate action on the FY2009 Energy and Water Development appropriations bill. In the report accompanying its bill, the Senate Appropriations Committee notes its continued support for the Advanced Certifi- 2 U.S. Congress, Consolidated Appropriations Act, 2008, Division C—Energy and Water Development and Related Agencies Appropriations Act, 2008, House Appropriations Committee Print to accompany P.L. 110-161 (2008), p. 583. 3 JASON, Reliable Replacement Warhead Executive Summary.
OCR for page 45
Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile cation effort “to increase the safety and reliability margins of the stockpile without underground testing.”4 Finding 4-1. QMU can be applied to the evaluation of any new designs, including the RRW, to contribute to enabling certification of those designs without nuclear tests. QMU was used extensively in the design and certification plan for the WR-1. The approach taken in the WR-1 was to begin with tested designs and to then increase the primary yield to increase margin for driving the secondary while controlling uncertainties. (More information on this topic is included in Note 6 in the classified Annex.) Analyses to date have focused on two main performance issues: (1) primary boosting and (2) primary yield to drive the secondary. The secondary design is based on solid principles of physics and engineering. Its ultimate certification will depend on further development, further analysis, and nonnuclear tests. (More information on this topic is included in Note 7 in the classified Annex.) The current certification plan for WR-1 includes some key experiments to test and prove the design. These include three core-punch hydrotests and three pin dome shots. About 6 to 10 smaller hydrotests are for the fill tube and for parts of the radiation case. (More information on this topic is included in Note 8 in the classified Annex.) The committee observes that the schedule is success-oriented in that an unexpected failure in any of the major hydrotests could substantially delay the project. Confidence could be generated by designing an experiment that should lead to abnormal behavior and then actually seeing that behavior in the test. Which data are needed most can be determined using the QMU methodology to identify the physical models that engender the largest uncertainty. The experimental plan in support of WR-1 is of minimal extent, and because it assumes a high probability of success it does not adequately provide for resolution of test anomalies. Finding 4-2. Any certifiable RRW weapons design will have to be “close” to the archival underground nuclear test base, while meeting reasonable criteria for adequate margin. The design and certification of new nuclear weapons that are sufficiently “close” to particular legacy designs could, in principle, be accomplished without nuclear tests, based on the existing nuclear test archive, on new experiments with no nuclear yield, and on modeling and 4 U.S. Congress, Senate, 2008. S. Report 110-416, pp. 121, 146.
OCR for page 46
Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile simulation tools supported by a QMU methodology more mature than at present. For a certifiable RRW, the design labs will have to make the case that a new design is “close enough” to tested designs. The case would depend on establishing that the design is based on well-understood principles of nuclear warhead physics and engineering, that the design is related in key ways to designs that were successful in archived historical nuclear testing, and that any gaps between the knowledge of physics and engineering and the archival underground nuclear test base are bridged by experiments. Interpolation is highly preferable to extrapolation. Recommendation 4-2. The design laboratories should lay out in detail their arguments for the relevance and closeness of archival underground tests to any proposed RRW design. These laboratories should investigate methodologies for helping address the problem of quantifying closeness. How to transparently define and quantify “closely related” is a difficult issue to which the labs should devote sufficient effort. “Close enough” depends on the direction of the change as well as the magnitude—the direction should be away from “cliffs,” and expert designer judgment must go into assessing “close enough.” Prior warhead anomalies and their “fixes” should be used to validate the definition of “close enough.” The goal is to increase the critical margins while controlling the uncertainties so that M/U ratios are greater than 3 or so. The margins and cliffs here are intentionally spoken of in the plural because there are multiple failure modes, and increasing one margin might decrease another—for example, increased Pu mass might endanger one-point safety, so all must be considered together. A primary lying between two successfully tested designs (i.e., interpolated rather than extrapolated) can provide additional confidence. The design and certification of new nuclear weapons that are sufficiently “close” to particular legacy designs could, in principle, be accomplished without nuclear tests, based on the existing nuclear test archive, on new experiments with no nuclear yield, and on modeling and simulation tools supported by a more mature QMU methodology. It must be noted, however, that there is no commonly accepted quantification of closeness in the laboratories. While closeness will always have a substantial qualitative component based on expert judgment, a quantitative measure is clearly needed. This is not a trivial problem. While this committee is not in a position to offer a credible solution, it believes that any such solution will involve both QMU methodology and expert judgment. It also believes that it is possible to devise simulation tools that can
OCR for page 47
Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile help materially in quantifying closeness. There are presently many probabilistic approaches to closeness, such as Mahalanobis5 or Bhattacharyya6 distance, which could be modified and used by the laboratories in their search for such a definition. Finding 4-3. The relevant performance gates might be different for different designs. There could be new failure modes if new features are added. The M/U values of the old subsystems might change and new subsystems might then dominate the M/U. Further, the incorporation of surety features could create a situation in which new performance metrics would be needed to establish confidence that the design is not near a failure threshold. Recommendation 4-3. The design labs should carefully examine all of the failure mechanisms for new RRW designs, criteria for the RRW to pass all performance gates, and the methodology used for these analyses. As an example of problems that might arise in new designs, the addition of Pu mass to a primary design in order to increase the margin of the primary yield, Yp, performance gate might decrease margin at the one-point safety gate. (More information on this topic is included in Note 9 in the classified Annex.) Finding 4-4. A higher level of peer review, documentation, and experimentation without nuclear testing are essential to a credible RRW certification process. For credible certification, the labs need to document the design and its analysis thoroughly and transparently, via QMU methodology, so that outside experts can independently judge its correctness and evaluate its credibility. The labs also need to track and document changes in design and reasoning by a version-control process. The evolution of the design is important evidence for the viability of the design. Peer review is essential to credibility. Assessment and certification, with or without nuclear tests, are based on credibility. A strong peer review process, however, is not simply a step tacked on at the end of a design process. It is built into the process by taking steps throughout that 5 Available at http://eom.springer.de/M/m062130.htm. 6 Available at http://eom.springer.de/B/b110490.htm.
OCR for page 48
Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile make the data and reasoning more transparent. QMU shares similar goals and, once implemented thoroughly, will much more readily support peer review. The laboratories have long practoied lab-vs.-lab review of designs. The committee suggests a stronger and more independent review process than that used for previous nuclear weapons, by engaging experts not directly involved in the project—that is, it remains lab vs. lab but now includes outsiders such as retirees and, perhaps, British experts—who can knowledgeably assess the design process and who can use their own simulation codes and analysis methods. For this peer review to affect the design, it must be timely. Recommendation 4-4. The NNSA and the design laboratories should ensure that the certification plan for any RRW is supported by strong, timely peer review and by ongoing, transparent, QMU-based documentation and analysis in order to acheive a confidence level necessary for eventual certification.