Environmental Issues > Nuclear Energy, Nonproliferation, and Disarmament Main Page > All Nuclear Energy, Nonproliferation, and Disarmament Documents

Report page


On 31 July 1997 the Department of Energy (DOE) announced the award of multi-million dollar, multi-year contracts to the California Institute of Technology, Stanford University, the University of Chicago, the University of Illinois at Urbana-Champaign, and the University of Utah to conduct basic science, mathematics, computer science, and engineering research for the U.S. nuclear weapons program. These five universities, in affiliation with numerous other institutions, are participating in DOE's evolving Academic Strategic Alliances Program, which DOE has stated is an important component of its overall Stockpile Stewardship and Management Program (SSMP).

Beyond maintaining the reliability and safety of U.S. nuclear weapons under a Comprehensive Test Ban, the SSMP seeks to sustain and even enhance U.S. capabilities, formerly dependent on nuclear explosive testing, to design and prototype nuclear weapons and "certify" changes to existing nuclear weapon types. Some of these changes are intended to increase weapon lethality against deeply buried or hardened targets, while others are directed at other military attributes, such as integration with alternate delivery vehicles, transport safety, ease of maintenance, or improved performance margins over a longer-than-planned stockpile life.

Such nuclear design capabilities can obviously be used to study advanced nuclear weapon concepts, although the U.S. government disavows the existence of any current military requirement to do so. However, to enable future nuclear design and prototyping capabilities without resorting to underground nuclear explosive testing, the DOE's program seeks to achieve a "virtual testing" capability, whereby vastly improved computer models of nuclear explosive performance, validated against past nuclear test data as well as experiments conducted in an array of powerful new above-ground testing facilities, are intended initially to offsett, and perhaps ultimately replace, the loss of data from underground nuclear testing.

The contributions of the Academic Strategic Alliances Program to the SSMP are in basic (but largely interdisciplinary) scientific research, computer hardware and software research for computer architectures based on massively parallel processing, and the union of these in the simulation of complex systems. This research was initially billed as being entirely unclassified, but the DOE states that it may ultimately fall under Department of Commerce or Department of State export control restrictions and/or DOE safeguards and security regulations, and in reality some participants will be required to have "Restricted Data" Q clearances to get access to certain classified ASCI computer systems on which time has been allocated for university research.

Beyond these more tangible deliverables, the Academic Strategic Alliances Program contracts are intended to spawn small-scale analogs of the SSMP's integrated "virtual testing" approach that will serve as proving grounds for DOE's new "Science-Based" stewardship paradigm that seeks to replace experimental nuclear detonations with computer simulations of unprecedented scale and complexity. The universities will work directly on the supercomputers acquired for virtual testing at the DOE's nuclear weapons laboratories.

For decades, nuclear weapons codes run on supercomputers and nuclear explosive testing were highly interdependent research tools used to design, prototype, and certify the military performance, reliability, and safety of succeeding generations of U.S. nuclear weapons. Nuclear weapons codes were (and are) dependent on data from underground nuclear testing and other experiments in order to "calibrate" the calculation(to set the values of numerous adjustable parameters in the codes which partially compensate for a hitherto incomplete, inexact theoretical description of the nuclear weapon explosion.

Underground nuclear tests were likewise dependent on computer calculations for their design, analysis, and interpretation. Nuclear weapons codes establish linkages between experimental measurements (explosive tests and laboratory experiments), the current theoretical description of the basic physical processes occurring in a nuclear explosion, and the specific device design parameters. Historically, it is not only the number and quality of tests which has determined whether a nuclear weapon design enters the U.S. arsenal, but also the computer power available to the weapons designers.

The U.S. last tested a nuclear weapon in 1992 and President Clinton signed the Comprehensive Test Ban Treaty (CTBT) in September 1996. Faced with the probable loss of underground nuclear testing for the foreseeable future, DOE nonetheless seeks to maintain an extensive and vigorous nuclear weapon design and supporting research base, with a breadth of nuclear design expertise comparable to that which helped sustain the U.S. in the now defunct nuclear arms competition with the former Soviet Union. The centerpiece of DOE's effort to recapture the technological dynamism of the arms race is its drive to create a "virtual testing" capability.

As now being implemented by the Accelerated Strategic Computing Initiative (ASCI), "virtual testing" incorporates the goals of calculating nuclear explosions: 1) in three instead of one or two physical dimensions; 2) with finer spatial and temporal resolution; 3) with a fuller description of the underlying physics; and 4) including all or a large subset of the components in a nuclear weapon.

To enable the achievement of the desired nuclear weapons software advances, the ASCI program is funding development and acquisition of the next generation of supercomputers based on massively parallel processing, initially through contracts with Intel, IBM, and Silicon Graphics/Cray. The ASCI Program Plan calls for a hundred-thousand fold increase in nuclear weapons code performance between 1996 and 2003.

The ASCI agenda for supplying "virtual testing" capabilities within the next decade is problematic. It has been recently demonstrated that by improving one piece of a nuclear weapons code(for example by putting in more precise measurements of a material's "opacity" or ability to absorb radiation(one can cause the overall calculation to disagree with underground test data. The most probable interpretation of this result is that the numerous approximations and "fudge factors" which appear throughout the lengthy nuclear weapons codes are correlated in as yet unknown ways. Implementing just two of the four ambitious ASCI software goals -- adding a third dimension to the calculation while simultaneously refining the theoretical description of the nuclear explosion -- will undoubtedly raise similar problems.

Two SSMP experimental facilities intended to calibrate the new ASCI codes(notably the National Ignition Facility (NIF) and the Dual-Axis Radiograph Hydrodynamic Test Facility(are still under construction and one, the NIF, is not likely to achieve its design goal of ignition and modest gain within the decade planned for bringing the new codes up to routine "production" status (i.e., qualified for use in the nuclear weapons certification process). Indeed, the science and technology undergirding the NIF is sufficiently speculative that it may fail to achieve ignition altogether.

Equally fundamental to a fair appreciation of the difficulties inherent in the virtual testing approach is that many hardware and software issues associated with massively parallel processing at ASCI target levels for computer speed, memory, and data communications are still at an early stage of research. Given the difficulty of the SSMP/ASCI "virtual testing" effort, one likely result is that the next generation of codes will raise a host of questions which, paradoxically, can only be answered by conducting carefully designed underground tests. It is this daunting array of challenges that have spurred the DOE to request the aid of universities in the ASCI program, and even to characterize it as "critical."

Under the Academic Strategic Alliances Program, five university "Centers of Excellence" have been established with an initial five year grant of roughly $20 million apiece. Each of the five universities will focus on an interdisciplinary, full-physics, multi-component, three-dimensional computer simulation to be performed on the ASCI supercomputers. The California Institute of Technology is creating a "virtual shock tube," a computer simulation of the detonation of high explosives and the effect of the ensuing shock wave(s) on test materials which is intended ultimately to serve as a "virtual experimental facility." Two universities are concentrating on simulating the performance of complex devices: Stanford on gas turbine engines and the University of Illinois at Urbana-Champaign on solid rocket boosters. The University of Chicago will simulate the thermonuclear processes involved in certain astrophysical events, and the University of Utah will conduct research on simulating accident scenarios involving fires and explosions.

In terms of basic science, several areas of commonality exist across the five Academic Strategic Alliances Program Centers and are of key interest to the U.S. nuclear weapons program: high explosives; material properties under extreme conditions; hydrodynamic instabilities; compressible, reactive, turbulent flows; thermonuclear reactions; material aging and defects; and the integrated component simulation of an engineered device. In addition, the universities have contracted to perform computer hardware and software research, with a focus on the ASCI computers on which the universities are programming their simulations and the nuclear weapons specialists are programming their improved 3-D nuclear weapon codes.

Importantly, the DOE has expressed to the universities its need for a future generation of nuclear weapons specialists who may never participate in an underground nuclear test but must be suitably trained in other ways. Given that Academic Strategic Alliances Program research is an integral part of the nuclear weapons "virtual testing" effort at the national laboratories, it is logical to conclude that from among the hundreds of young scientists participating in this program, the DOE expects to identify and steer recruits to the nuclear weapons laboratories who can be induced to devote a significant fraction of their professional careers to the design and maintenance of weapons of mass destruction.

Research proposals submitted by the universities to the DOE during the competition for Academic Strategic Alliances Program grants contain passages on weapons science education and training activities and interactions with the nuclear weapons laboratories. This may be the most controversial aspect of the program for members of the academic community, an issue which may be exacerbated by the substantial matching funds (or other services) universities have offered these Centers. At the very time in their careers in which students are exploring and seeking to establish the dimensions of their personal and professional moral universe, the Academic Strategic Alliances Program forces an early choice between, on the one hand, involuntary utilization of one's work product for nuclear weapons work, and on the other, forgoing graduate research support in one's chosen field.

The issue of the participation of foreign nationals in the Academic Strategic Alliances Program is difficult for both parties to these contracts. The tension between open academic research and the secrecy of nuclear weapons work is evident in the "Short-Term ASCI Alliances Platform Access Policy,"[1] in which the Department of Energy has chosen to deny foreign nationals (students, postdoctoral associates, and professors) accounts on ASCI computers, or access to ASCI computer software which currently falls under export control restrictions.

To the extent that these limitations constrain or effectively shut out foreign students from meaningfully participating in the core Alliances Program research, long standing academic traditions of openness and non-discrimination are compromised. To the extent that weapons related research is declassified and the Academic Strategic Alliances Program limitations do not discriminate against foreign nationals, the Academic Strategic Alliances Program will provide citizens from non-nuclear weapon states with many of the scientific skills and even some of the critical data needed for nuclear weapons design and engineering.

Policy Recommendations

Given the controversial nature of the Academic Strategic Alliances Program, NRDC offers the following recommendations to the U.S. government, to the universities, and to individual participants in this program:

Recommendation 1: The U.S. government should conduct a broad-based interagency review of the Academic Strategic Alliances Program to determine:

  1. if both its political and technical non-proliferation risks are commensurate with its purported benefits to national security through its role in the SSMP as currently designed; and

  2. if there are more restrained, technically conservative approaches to stockpile stewardship and management(more in keeping with the objectives of the nation's nonproliferation and test-ban commitments(that do not require a "Strategic Alliances Program" to "accelerate" development of a "virtual testing" capability for the nuclear weapons stockpile.

Recommendation 2: The Department of Energy should publish the Academic Strategic Alliances Program research contracts and any subsequent task orders, e.g., on the World Wide Web.

Recommendation 3: University endorsement of the Academic Strategic Alliances Program should be re-evaluated in light of:

  1. the nature of the SSMP as an explicit effort to mitigate and perhaps nullify the restrictions on nuclear weapons design imposed by the Comprehensive Test Ban Treaty;

  2. DOE's efforts to use the on-campus programs as the induction, training and recruiting ground for the next generation of nuclear weapon specialists;

  3. the unavoidable contradiction between nonproliferation concerns and the nuclear weapons simulation objective of Academic Strategic Alliances Program research, resulting in the current discriminatory policies of the Alliances Program against foreign nationals; and

  4. the involuntary association the Program creates between a graduate student's (or faculty member's) research funding and work product, and the design and reliable operation of weapons of mass destruction, to which an otherwise qualified researcher may harbor deep moral objection.

Recommendation 4: Individual scientists and engineers should reevaluate their participation in the Alliances Program, given their lack of control over the fruits of their research and its immediate known application to the task of simulating and maintaining the performance of weapons of mass destruction.


1. "ShortTerm ASCI Alliances Platform Access Policy," http://www.llnl.gov/asci-alliances/9711policy.html (Modified on 11/11/97).

Back to contents page

last revised 1/22/1998

Get Updates and Alerts

See the latest issue >

NRDC Gets Top Ratings from the Charity Watchdogs

Charity Navigator awards NRDC its 4-star top rating.
Worth magazine named NRDC one of America's 100 best charities.
NRDC meets the highest standards of the Wise Giving Alliance of the Better Business Bureau.

Donate now >

Share | |
Find NRDC on