The nuclear testing moratorium has driven a fundamental shift in the process for maintaining the safety and reliability of the U.S. nuclear weapons stockpile. The science-based stockpile stewardship program has replaced the process of designing and testing that originated with the Manhattan project. This has affected many aspects of weapons design activities.
Because it is no longer allowable to conduct full weapon tests, the United States can no longer rely on new nuclear-test data to inform questions such as: (1) How, if at all, have the properties of a particular warhead design changed as it has aged? (2) What are the effects on weapon performance of changes such as replacement of aging components and upgrades to enhance factors such as safety, security, and reliability? (3) What are the effects of LEPs, upgrades, or new design approaches? and (4) How well are design codes able to predict the behavior of design elements? Accordingly, design code models that require calibration based on data from full nuclear tests are being upgraded with models that build on an improved understanding of the fundamental weapons’ physics, and which therefore reduce the need for parameter fitting that attempts to link the codes to performance data from un-aged and subtly different systems.1 Although the body of data from tests conducted prior to 1992 continues to be mined, the basis of generating new data has shifted from full weapons testing to above-ground experiments and subcritical tests.
The design codes used by the weapons design community in the past—largely developed and deployed at LANL and LLNL—have been extensively validated by a variety of physics experiments and by comparison with data obtained from actual weapons tests (both above and below ground). These codes relied on approximate phenomenological models to address physics that was not yet fully understood or too costly to model (either of which may still be the case). By calibrating these phenomenological models using test data, it proved possible to use these design codes in a predictive fashion, that is, for predicting the performance of devices for which the codes had not been calibrated.
Following the testing moratorium and signature of the comprehensive test ban treaty, the design community was faced with the conundrum of how modeling and simulation could continue to be used reliably as weapons continued to deviate from the precise form for which validation data from full-system tests are available. That deviation is inevitable because components age, weapons are refurbished using materials that differ from those originally used, and changes are made to improve device safety and reliability. There is no guarantee that the model calibrations derived from the original device designs are applicable to the aging stockpile, or to the refurbished portion of the stockpile. It, therefore, became urgent to find an alternative plan that would allow weapons scientists to continue the weapons
1 The computer codes in question rely on phenomenological models that describe processes known to be important to weapons design but for which “first principles” models are not yet available. These phenomenological models contain parameters that must be adjusted so that the outputs of the simulations align with corresponding experimental data. This adjustment of model parameters is sometimes referred to as “tuning” or “calibration,” and the parameters that are adjusted are often referred to as “knobs.” For more detail, the reader is referred to National Research Council (NRC), Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile, The National Academies Press, Washington, D.C., 2009.
certification Process with the necessary confidence. Furthermore, the weapons design community has had the additional challenge of ensuring that the U.S. government understands the potential performance of novel nuclear weapons designs created abroad by established or aspiring nuclear weapons states or by non-state actors who have interests in designing “improvised” nuclear devices. Making such predictions requires a highly refined understanding of the capabilities and limitations of the models and experience derived from understanding the data from underground tests.
The science-based Stockpile Stewardship Program (SSP) was designed to support inferences about the performance of aging weapons and novel designs. In addition to weapon designers, the SSP requires systems engineers, the development of understanding of the effects of aging on existing weapons, a vibrant science base to provide the needed data, and the development of higher-fidelity design codes. The fundamental idea of the SSP was to systematically upgrade phenomenological models that required calibration from full nuclear tests, moving to higher-fidelity models based to an increasingly greater degree on improved understanding of the fundamental physics and which, therefore, do not depend as much (ideally, not at all) on the parameter adjustment characteristic of the phenomenologically oriented design codes. However, all codes require verification and validation (V&V); that is, they require tests that determine how accurately the codes solve the equations of the mathematical models (verification) and tests that determine the degree to which the models are accurate representations of reality (validation). Furthermore, it is not enough for a given science-based code to predict; it is also necessary to understand how well the code predicts, that is, to determine the uncertainties of the predictions (i.e., uncertainty quantification, usually referred to as UQ).2 While the transition to higher-fidelity codes has made progress, it has not yet been possible to eliminate all of the phenomenologically based components of design codes, and in addition, the questions that must be answered continue to change with time. Thus, the codes will need to continue to evolve, and V&V and UQ are essential parts of that ongoing process of improvement.
The simulation codes for the future must be able to inform judgments about the performance and reliability of warheads that have changed due to aging and rebuilding without reliance on nuclear test data. Continued progress in the SSP requires the best science available, and, thus, requires a partnership among theory, experiment, and simulation and the leadership and participation of highly competent scientists and technologists. Stockpile stewardship relies on experts in modeling and simulation, design teams, systems engineers, ongoing experimentation, and a process that ensures continuation of these capabilities and passing of skills and knowledge to new generations of scientists and engineers.
Experiments are essential for making discoveries, developing understanding of physical phenomena, and testing hypotheses. Experimental data are important for building confidence in designers’ simulation codes—to compare code predictions with reality, thereby learning how well they do and do not match—and for strengthening the insight of those who rely on the codes. The ability to certify the stockpile depends on this coupling of simulation and data. Scientific experiments are used to obtain basic physics data, and also to provide integral data in limited parts of the nuclear weapon space.3 High-quality design work requires experiments to compare with code predictions. The NNSA and the laboratories have long recognized the need for non-nuclear explosion experiments and have invested in some major experimental facilities. The principal facilities are the National Ignition Facility (NIF) at
2 NRC, Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile, 2009.
3 Much of the complication of weapons design is related to it being “multi-physics” and multi-scale, specifically the fact that (for example) it deals with a combination of hydrodynamics, strength of materials, and radiation transport. Basic science activities typically focus on one of these disciplines. When investigating complex phenomena that extend across multiple disciplines, the only experimental venue is something like NIF (where one can “mix” hydro with rad transport), or DARHT (where one mixes hydro with materials).
LLNL, the Z Pulsed Power Facility (Z) at Sandia National Laboratories,4 the Dual-Axis Radiographic Hydro-Test facility (DARHT) at LANL, and the underground hydrodynamic facility at the Nevada National Security Site.5 These are very powerful facilities that address different parts of the nuclear weapon S&E space.
However, getting relevant experimental data is more difficult than it should be. Several weapons designers who met with the committee reported that the experiments they need are very costly and take an excessive amount of time. Their diagnosis is that increases in the formality of operations—the many steps and approvals that must be accomplished before an experiment may proceed6—has contributed to driving up the time and cost for conducting experiments, which has contributed to fewer experiments being done.7
This issue was discussed in some detail, with illustrating examples, by Siegfried Hecker, a former director of LANL, in testimony before the Strategic Forces Subcommittee of the House Armed Services Committee at a hearing on February 16, 2012.8 In his statement, Dr. Hecker discussed specific examples from his long career at LANL, contrasting conditions he encountered when he arrived at the laboratory as a young man with more recent conditions and elaborating on the role of increasing oversight in driving up cost and time to conduct experimental work. What the study committee heard at meetings at all three laboratories is consistent with Dr. Hecker’s testimony.
Two areas of particular concern are (1) experiments that use radioactive or otherwise hazardous materials and (2) high explosive-driven hydrodynamics experiments (“hydro shots”), a key part of the primary design and certification process.9 The timescales involved in planning and executing a hydro shot can run from months to years, and the costs run into the millions of dollars. Dr. Hecker’s testimony includes discussion of his experiences over the years with experiments that use plutonium. The high cost of experimental work is due in part to excessive formality of operations and duplicative oversight of environmental health and safety with essentially no risk-benefit analysis to create a balanced safety program.
From the committee’s discussion with multiple weapons designers at both LANL and LLNL, a culture of risk avoidance has grown in the oversight levels of the laboratories to the point where there appears to be little distinction made among levels of risk. The following four organizations outside the
4 Also known as the “Z machine” and the “Z-pinch facility.”
5 Formerly the Nevada Test Site.
6 As discussed in the phase 1 report, NRC, Managing for High-Quality Science and Engineering at the NNSA National Security Laboratories, The National Academies Press, Washington, D.C., 2013.
7 Of course, simply counting the number of experiments provides only a coarse measure, as other factors can contribute to reductions in the numbers. For example, costs can increase because of a determination that a more complex measurement is required than was originally planned; and schedules can be delayed for reasons related to availability of equipment. Fewer experiments might be needed because of greater-than-anticipated success of the earlier experiments in a planned series. If a particularly type of experiment is conducted less and less often over time, that could indicate operational constraints, but it might also be indicative of funding problems, or it could indicate a successful program that has accomplished what it set out to do. For these reasons, the committee found the most compelling input to be the judgment expressed by many experts at the laboratories that the number is inadequate.
8 See, for example, discussion and examples cited in testimony before the Strategic Forces Subcommittee of the House Armed Services Committee, Siegfried Hecker, February 16, 2012.
9 NNSA online definition: “Hydrodynamic testing (hydrotesting) is the execution of high-explosive driven experiments to assess the performance and safety of nuclear weapons. Under test conditions the behavior of solid materials is similar to liquids, hence the term ‘hydrodynamic.’ These large scale hydrodynamic experiments utilize test assemblies that are representative of nuclear weapons but with the fissile material in an actual weapon altered or replaced with surrogates.” The NNSA Quarterly SSP Experiment Summary-FY12-2Q (Final) further defines: “Subcritical Experiments: High explosive driven experiments to obtain information critical to certifying weapons performance in the absence of underground testing while still employing nuclear materials. No critical mass is formed due to the amount and quality of the nuclear material. As such, no self-sustaining nuclear chain reaction can occur in these nuclear experiments.”
laboratories impose constraints on the conduct of operations: (1) the NNSA field office for each laboratory10; (2) NNSA Headquarters; (3) the DOE Office of Health, Safety and Security; and (4) the Defense Nuclear Facilities Safety Board (DNFSB). The DNFSB is an advisory body that does not directly impose regulations, although DOE and NNSA usually accept DNFSB recommendations. Moreover, the laboratories themselves sometimes contribute to increasing cost and decreasing throughput of experiments involving hazardous materials by imposing even more stringent regulations as pre-emptive defensive measures. The result is a layering of constraints that reportedly inhibits the experimentation needed for high-quality S&E.
Most of these restrictions appear to be based on an analysis of risks alone; the important benefits of executing experiments also need to be taken into account.
Finding 2.1. Experiments that support the nuclear weapons programs often involve hazardous materials or otherwise carry safety risks. Assessing and controlling those risks is necessary, and mechanisms have been put into place to do so. However, this process necessarily adds to the cost of conducting experiments and can slow or deter experimental work, particularly when the process involves multiple overseers (e.g., NNSA, NNSA field offices, the DNFSB, etc.) with overlapping safety responsibilities. Moreover, these assessments generally focus on the safety risks associated with particular experiments rather than weighing those risks against the benefits to be derived from the experiments and the risks to the nuclear weapons program from not conducting the experiments.
In addition to obtaining needed data, experiments provide exploitable experience to those who participate in them. Maintaining opportunities to participate in experiments, and comparing their outcomes with simulated data, contributes to building and maintaining a work environment that attracts new high-quality staff and encourages experienced staff to stay. Reducing the amount of experimental work, therefore, has multiple adverse consequences, all of which can have major impacts on the quality of S&E.
Current Design Capabilities
Nuclear weapon design capabilities at the NNSA national security laboratories extend beyond the expensive facilities and extensive S&E research base to include the people who maintain the ability to apply these impressive tools to national defense issues. This critical laboratory resource consists of trained, qualified, and experienced personnel. The staff members involved with weapons design who interacted with the committee certainly met all of these characteristics. The weapons design communities at both LANL and LLNL appear to be composed of high-quality people who are highly motivated and view their work as vital to the national interest. They have done high-quality work in the face of major obstacles, including complex layers of oversight, weakened funding, and impediments to experimental work.
The design program activities include monitoring and analysis of the current weapon stockpile, refurbishing selected portions of the stockpile through LEPs, and developing better understanding of (and computational models for) weapons design physics problems. The committee’s discussions with the designers focused on weapon design physics issues, especially boosting. This work is very good—it is well thought out, and they have been making progress.
The laboratories’ mission focus on nuclear stockpile stewardship is well in hand. However, the committee observed—and shares—concern that the intensive focus on the analysis of a static inventory of
10 Los Alamos Field Office, Livermore Field Office, and Sandia Field Office.
nuclear weapon hardware may impede continuance of the design capabilities inevitably required to maintain the safety, security, and performance standards of an evolving stockpile. The lifetime extension programs do offer design challenges, but these may not be sufficient. The perpetual maintenance of a cadre of experienced nuclear weapon designers is no easy task, but is the central responsibility of nuclear weapons laboratory management.
One suggested approach to preserving high capability and competence for weapons design is to consider operational exercises to test nuclear weapon design capabilities in a process that mimics the phase 1 and phase 2 process used to develop the current stockpile.11 Such an operational test—based on, for example, a design-and-experiment cycle for subcritical devices—could provide metrics of the status of design capabilities to include design and engineering. The metrics are important because they help build confidence in people and their judgments.
NIF could offer some opportunities of this kind for training weapon designers. The inertial confinement fusion (ICF) program utilizes a capsule implosion in an attempt to achieve fusion. While there are some differences from weapon design processes, there are also some similarities. The opportunity to design and conduct integral implosion experiments12 at the NIF would be a great training tool for weapon designers and might even bring in some new ideas that would benefit the ICF program.13
Recommendation 2.1. In order to continue the laboratories’ good work at recruiting, training, and retaining high-quality nuclear weapons design staff, the laboratories should include opportunities for them to conduct implosion experiments at the National Ignition Facility.
A positive observation about design laboratory personnel from LANL and LLNL is the quality of the early-career designers. The intelligence and enthusiasm of this group bodes well for the future of the nuclear design enterprise and is a clear indicator of the forward-looking capability development of the management of both laboratories.
International Issues (Nuclear Proliferation)
Serious international issues call for a highly competent domestic nuclear weapons design capability: it is essential to maintain high-quality expertise in weapons design not only to steward our own weapons, but also to understand thoroughly designs that are developed elsewhere. Many countries are embarked on nuclear energy programs. Some of these programs have already involved efforts to develop nuclear weapons, and some countries have demonstrated success by underground testing of nuclear devices. Not all countries will follow established paths in their nuclear weapons designs. The U.S. intelligence community works to find out as much as possible about efforts of other countries, but to understand the significance of such information, and to assist the intelligence efforts, having and involving a group of knowledgeable U.S. designers is crucial.14 U.S. design teams are also important in assessing the significance of information concerning foreign groups’ attempts to develop improvised
11 Appendix B of the “Nuclear Matters Handbook” (http://www.acq.osd.mil/ncbdp/nm/nm_book_5_11) describes the life cycle of U.S. nuclear weapons as consisting of seven phases. Phase 1 is a concept study; phase 2 is a feasibility study followed by phase 2A, which is a design-definition and cost study.
12 That is, experimental designs that involve several different scientific disciplines. When combining codes that derive from different areas of physics, there is the huge problem that just because these codes, individually, passed verification and validation tests (these are the unit tests), it is not necessarily true that when one couples them that the resulting multi-physics code will give reliable results. So the only way to make sure that the coupled codes work is to carry out integral experiment—and in the case of the design codes, this obviously means proxy designs that exercise all of the physics components of a coupled design code.
13 “National Nuclear Security Administration’s Path Forward to Achieving Ignition in the Inertial Confinement Fusion Program,” Report to Congress, United States Department of Energy, Dec, 2012.
14 Examples exist where such has been shown to be important, but specific examples are classified.
nuclear devices. Understanding the gathered information requires substantial knowledge of weapons design.
Because such global nuclear security issues are of increasing importance, the laboratories are challenged to maintain or re-establish design capabilities from the past that may not be relevant to the modern U.S. nuclear stockpile. The overlap of design capabilities required for U.S. stockpile stewardship and those required for global security is extensive but not complete. The time urgency and precision required of the answers to questions that may come from our national leadership are quite different and may require different tools. The often abused notion of “expert judgment” plays a central role here, and this point again emphasizes the importance of maintaining the requisite human infrastructure—highly qualified and knowledgeable experts—in order to maintain mission capabilities.