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

Report page

III. THE ACADEMIC STRATEGIC ALLIANCES PROGRAM

A. Program Goals and Structure

Under the Academic Strategic Alliances Program DOE's Defense Programs has contracted with leading U.S. academic institutions to conduct research and development activities jointly with DOE's nuclear weapon laboratories -- Lawrence Livermore, Los Alamos, and Sandia National Laboratories. According to DOE, the Alliances Program has five major goals (or program objectives).[46]

  • Establish and validate the practices of large scale modeling, simulation, and computation as a viable scientific methodology in key scientific and engineering applications that support DOE science-based stockpile stewardship goals and objectives.

  • Accelerate advances in critical basic sciences, mathematics, and computer science areas, in computational science and engineering, in high performance computing systems and in problem solving environments that support long-term ASCI needs.

  • Leverage other basic science, high performance computing systems, and problem solving environments research in the academic community.

  • Establish technical coupling of [Academic] Strategic Alliances [Program] efforts with ongoing ASCI projects in DOE laboratories.

  • Strengthen training and research in areas of interest to ASCI & SBSS [Science Based Stockpile Stewardship] and strengthen ties among LLNL, LANL, SNL and Universities.

The first three goals above generally describe the science, mathematics, computer science, and engineering research which the universities will perform in support of the U.S. nuclear weapons program. The fourth goal states that this contract work in the university community will be "technically coupled" to the nuclear weapons program and implies increased interaction between members of the university (Alliances Program) community and the nuclear weapons community. The fifth goal states that the Alliances Program will train and facilitate the recruitment of future nuclear weapons specialists. Note that this training and research objective is not directed toward the civilian applications emphasized in the research proposals.

In order to inform universities about the Academic Strategic Alliances Program and thereby encourage submission of research proposals that would be useful to the nuclear explosive simulation effort, DOE held an "ASCI Alliances Pre-proposal Conference" in Dallas, Texas on 5-6 December 1996. A total of 134 faculty, staff, and graduate students attended the conference from 47 universities.[47] The agenda for the conference included a presentation on Science Based Stockpile Stewardship by DOE Assistant Secretary for Defense Programs, Victor Reis, a presentation on the Accelerated Strategic Computing Initiative by DOE Deputy Assistant Secretary for Strategic Computing and Simulation, Gil Weigand, and presentations on specific research areas of interest to DOE's Stockpile Stewardship and Management Program (SSMP) and ASCI in Computational Physics, Materials, Energetic Materials, and Computational and Computer Science Infrastructure. Additionally, DOE Defense Programs (DOE DP) researchers prepared a set of background papers "to assist in understanding what is technically relevant for the ASCI program."[48]

A total of 48 'pre-proposals' -- presumably short descriptions of proposed research, and formal expressions of interest in the Alliances Program -- were submitted by universities to DOE DP by mid-January, 1997. Of these, 20 elicited a request by DOE for full proposals due in mid-March, from which seven finalists were chosen and site visits were conducted. Representatives from the national nuclear weapons laboratories, the U.S. government, industry, and academia participated in the review process. The DOE announced on 31 July 1997 five winning universities: California Institute of Technology (Caltech), Stanford University, University of Chicago, University of Illinois, Urbana-Champaign (UIUC), and University of Utah. The programs at each of these universities are discussed separately below.

The winning universities were chosen based on "Select[ing the] Best Overall Combination of Disciplines to Meet ASCI Goals."[49] Insight into what DOE meant by this can be had from a vu-graph presented by DOE Deputy Assistant Secretary Weigand to the attendees at the Pre-proposal Conference, shown below. In this figure the explosion of a nuclear weapon is divided into eight sequential processes beginning with the detonation of the high explosive (HE) surrounding the fissile core of the thermonuclear primary and culminating in the weapon effects.

Read vertically, the figure shows the role played by (2) simulation (computation & modeling) in the integration of (1) SSMP nuclear weapons experimental capabilities (i.e., facilities such as the National Ignition Facility which can approach some of the physical conditions occurring in a nuclear explosion) with (3) scientific research, including "academic & lab scale scientific studies & experiments."



Figure 3.1: A DOE viewgraph presented at the ASCI Alliances Program Pre-Proposal Conference to 137 members of the academic community from 34 universities. The integration of university Alliances Program research into the "Virtual Testing" SSMP effort is illustrated.


The vu-graph illustrates the incorporation of Academic Strategic Alliances Program research into the "Virtual Testing" of nuclear weapons. It is logical to assume that DOE would select a set of Alliances Program research proposals which would span the phenomena delineated in the figure (and also include manufacturing issues, not shown here). This is in fact what occurred in the selection of the winning research proposals.

Prior to the preliminary submission of Alliances Program research proposals two frequently asked questions of DOE were, "Must nuclear issues be included to get consideration?" and "Will these be the determining factor?" The DOE response in part was: "Direct nuclear issues are not the focus nor are they a required or determining factors for selection [of a winning Alliances Program proposal], as such a focus would likely result in an undesirable number of proposals that duplicate ongoing efforts at ASCI laboratories." And elsewhere: ". . . [I]t is important to specify and pursue an ASCI-relevant physical science simulation problem so that the computer science and infrastructure research is directed towards enabling and supporting ASCI-relevant problems. . . . [A] fundamental principal of this [Academic Strategic Alliances] program is intellectual independence and creativity, while remaining relevant to the ASCI program. Proposed work should not, therefore, merely furnish extra labor to accomplish laboratory programmatic work (emphasis added)."[50]

The Academic Strategic Alliances Program is now structured to have three 'Levels,' as listed in the table below. Only Level 1 funds have been dispersed pursuant to negotiated contracts, establishing the five university 'Centers of Excellence.' A request for proposals was issued by the DOE on 7 November 1997 for Level 2 research. Several research contracts to universities were awarded prior to the announcement of funding for the five Centers (i.e., prior to 31 July 1997).[51]


Table 3.1: The three-level structure of the DOE Academic Strategic Alliances Program.

Program LevelNumber of Current Contracts in Program LevelFunding RangeBrief Description
Level 1:
Strategic Alliance "Centers of Excellence"
5$1.5-$2.0 million first year; growing over a 2-3 year period to $4-$5 million annually"long term, critical mass, multi-disciplinary university centers"
Level 2:
"Strategic Investigations"
TBD$200-$400 thousand per year (total budget of $3-$5 million) "aimed at individual university departments"
Level 3:
"Individual Collaborations"
TBD$50-$100 thousand per year"primarily between one DOE nuclear weapons laboratory and one researcher"


Historically, nuclear weapon computer codes themselves and even descriptive data about them were highly sensitive and classified, both to prevent them from falling into the hands of potential proliferants and to prevent the Soviet Union and other weapon states from improving their weaponry, possibly by ascertaining what weapon physics topics were the focus of U.S. research.[52] DOE described some of its nuclear weapons science and computing research requirements to the universities in order to solicit useful grant proposals. And as seen below, Level 1 of the Academic Strategic Alliances Program involves an historically unprecedented degree of access by university researchers to supercomputers at the national nuclear weapons laboratories.

Despite Secretary of Energy Federico Peña's renewed public commitment to "openness" during the very period covered by the research and writing of this report (August 1997 - January 1998), the DOE declined to make available to NRDC the Academic Strategic Alliances Program pre-proposals, full research proposals or contracts. Little information was provided by the DOE's Office of Defense Programs beyond what was available on the official Alliances Program web site.[53] Hence one policy recommendation offered in this report (see Chapter V) is that the DOE publish the Alliances Program contracts in order to better inform the affected university communities and the public in general. NRDC obtained copies of the Chicago, Illinois, and Utah research proposals directly from the Center Directors. Caltech's proposal is available on its Center's web site. No additional documentation was received by NRDC from the Stanford Center beyond what is exhibited at the CITS web site, despite repeated requests.

The Academic Strategic Alliances Program has been evolving over the last year: definitive contracts were only recently concluded between the universities and the DP; the "Short-Term ASCI Alliances Platform Access Policy" appeared on the Alliances Program web site this past November, as did a request for "Level 2" proposals; and a long-term policy regarding the status of foreign nationals in the program is still under review (DOE anticipates that a long-term policy will be developed by 1 May 1998).

In preparing this analysis, it was necessary to assume that DOE agreed to fund the broad outlines of the winning university research proposals, which media reports largely confirm. However, as will be analyzed in the section on Caltech below, its research proposal to DOE (which is publicly available) discusses research on beryllium, uranium, and actinides. Shortly before going to press NRDC learned that DOE has declined to fund this open research on nuclear weapons materials subjected to shock waves, perhaps in part because the research would overlap areas of classification (as also discussed below). DOE's action is appropriate, but begs the question why Caltech ever believed that such a sensitive area of investigation could or should be pursued in the unclassified domain. The figure below gives a timeline of significant events in the Academic Strategic Alliances Program to date.


Figure 3.2: A timeline of significant events in the Academic Strategic Alliances Program to date. This NRDC report drew heavily on the information in four of the university research proposals, but NRDC was denied access to the Alliances Program contracts.



B. California Institute of Technology

Background

All nuclear weapon designs include the detonation of chemical high-explosives (HE) to produce shock waves in materials -- most significantly in plutonium and uranium, but in lighter bomb constituents as well. Both TATB-based (insensitive) and HMX-based (sensitive) plastic-bonded high-explosives are used in deployed or stockpiled U.S. nuclear weapons.[54] Experimental, theoretical, and computational research on high explosives is a core research component of the U.S. nuclear weapons program. Shock-wave induced spherical implosion and compression of fissile material is the principal method by which a fission chain reaction with significant energy yield is produced in a thermonuclear weapon's primary component. If the primary's high explosive assembly fails nothing nuclear happens, and the secondary component (the source of most of the bomb's explosive yield) does not ignite.

The DOE currently operates several so-called "hydrotest" facilities for performing implosion experiments with subscale (i.e., subcritical) or non-fissile primary assemblies. In full-scale hydrotests, a non-fissile material such as depleted uranium or plutonium-242 is substituted for the primary's fissile core, so as not to produce a nuclear chain reaction; leaving shock-wave induced spherical, or hemispherical implosion, and the properties and behavior of weapon materials in various states of compression, as the primary phenomena to be examined. Under DOE's SSMP a next generation hydrotest facility is under construction at Los Alamos National Laboratory: the Dual Axis Radiographic Hydrodynamic Test Facility (DARHT). An upgrade to Livermore's hydrotest facility (the Flash X-Ray Facility, or FXR) is being carried out in order to provide both nuclear weapon design laboratories with state-of-the-art capabilities for penetrating, time sequenced radiographic images of densely imploded objects, such as the plutonium core of a weapon primary.


The Caltech contribution.

Under the Academic Strategic Alliances Program the "Facility for Simulating the Dynamic Response of Materials" has been established at Caltech to create a virtual shock physics facility, or "virtual shock tube." A diagram of one configuration of Caltech's virtual shock tube is shown in Figure 3.3. Caltech initially displayed another configuration of the virtual shock tube on the Center's web site. This "flyer plate" configuration more closely mimicked the design of modern nuclear weapons. A flyer plate is a technique for increasing the compression of target materials using a given amount of chemical high explosive by creating a gap that is used either to increase the momentum of either the plate material striking a target, or the target material itself (as in the case of a hollow shell of material that is accelerated inward and rebounds at the center.


Figure 3.3: A diagram of Caltech's virtual shock tube, showing detonator, high explosive charge, and test materials.


In Caltech's planned virtual experiments, a computer simulation will be performed of: the detonation of a high explosive charge; the effects of the ensuing shock waves on test materials (such as fracturing and phase changes); and the shock-induced compressible turbulent flow and mixing at material interfaces. Five research initiatives have been planned at Caltech in order to create the virtual shock tube:

  • Modeling and simulation of fundamental processes in detonation;
  • Modeling the dynamic response of solids;
  • First principles computation of materials properties;
  • Computation of compressible turbulence and mixing; and
  • Computational and computer science infrastructure.

The "Facility for Simulating the Dynamic Response of Materials" differs significantly from the other four Alliances Program Centers, in that no application of the research other than to nuclear weapons is developed in Caltech's proposal to the DOE (as described below, other Alliances Program Centers are emphasizing the simulation of gas turbine engines, supernovae, solid rocket boosters, and accident scenarios involving fires and explosions as presenting physics and simulation issues that are analogous to or partially overlapping those involved in modeling the performance of nuclear explosives).

At the "Facility for Simulating the Dynamic Response of Materials," Caltech has organized an interdisciplinary team to improve upon existing capabilities to model the detonation of high explosives. A trend in this field is noted in its Alliances Program research proposal:

With increasing [length] scale, models [of the detonation of high explosives] become less rigorous and more empirical until finally, at the macroscopic level, the models are only reliable if they have been carefully calibrated against experiments with specific materials.[55]

Caltech scientists will conduct calculations of the molecular/electronic structure of high explosives, develop detailed models of the chemical reactions (for explicit explosive systems), and explore "the interaction of chemistry with mechanical deformation." The latter effort aims at a better understanding of how the chemical reactions are modified if the explosive materials are subject to high strains(at the shock front, for example, or at the interface between the crystal grains of high explosives and the plastic binder.

Caltech states that the computational capabilities of the ASCI program are expected to permit modeling the detonation of a macroscopic piece of the high explosive material using the molecular-level description of the process.[56] If successful, this would represent an advance over current capabilities, in which Caltech characterizes the molecular-level computer simulations and macroscopic computer simulations as largely distinct efforts.[57]

In terms of the detonation of high explosives, the Caltech Center's agenda extends beyond basic science and computer simulation to the production of a computer research tool advertised for use in the U.S. nuclear weapons program:

High explosives are a key component in nuclear weapons and realistic modeling of detonation in high explosives is a long-standing issue in performance, safety, and reliability studies carried out in the DP Laboratories. A major deliverable from this portion of our program [i.e., "Modeling and simulation of fundamental processes in detonation"] will be integrated into a problem-solving environment. . .Our computational environment will provide ASCI researchers a means to explore systematically chemical, mechanical, and numerical issues through high-fidelity detonation simulations.[58]

This computational environment is an evolution of the Caltech code AMRITA[59] from a two-dimensional capability to a three-dimensional one, while incorporating the molecular-level modeling research discussed above. AMRITA is one of several candidate codes which Caltech may choose to develop into the virtual shock tube. Experimental validation of the detonation simulations "will be carried out through comparison with gas phase detonation experiments carried out at Caltech and HE experiments carried out at DP Laboratory facilities."[60]

The portion of the Caltech Alliances Program work devoted to "Modeling the dynamic response of solids" has two research components related to simulating the response of solid targets to strong shocks: 1) deformation and failure mechanisms in materials (such as the fracturing); and 2) polymorphic phase changes (the change of a material from one solid phase to another solid phase, such as the rearrangement of atoms to form a new crystalline structure.) "Algorithms must be developed to describe these processes in multidimensional Eulerian and Lagrangian codes."[61] (For a definition of Eulerian and Lagrangian codes, see Box 3.1.) As in the Caltech Center's study of detonation, issues associated with modeling phenomena across many orders of magnitude in length scale is at the center of the research problem:

A key requirement in simulating the response of solid targets to strong shocks is the resolution of multiple length scales straddling the gray zone between atomistics and continuum behavior. We have developed a quasi-continuum method that seizes upon the strengths of both atomistic and continuum theories and allows for the seamless and simultaneous consideration of multiple scales.[62]

With funds from the Academic Alliances Program, Caltech will extend this method to three dimensions (from two) and to dynamic (instead of merely static) problems. Caltech advises DOE that solid-to-solid phase transitions, a common phenomenon in shocked materials, currently lack a quantitative theoretical description which can be implemented in codes. Caltech proposed to model polymorphic transitions in silicon dioxide (SiO2), titanium dioxide (TiO2), sodium chloride (salt, NaCl), iron (Fe), iron oxide (FeO), beryllium (Be), boron (B), thorium (Th), uranium (U), and zirconium (Zr).[63] Use of the highlighted materials in a shocked state is confined almost exclusively to nuclear weapons. Not coincidentally, one of the materials best known for its poorly understood polymorphic phase changes is plutonium. According to Caltech's proposal, "the installation of new DP [DOE Defense Program] Laboratory radiography facilities (such as DAHRT) provides additional opportunities for [experimental] validation [of the new modeling approach]."[64] These DOE laboratory hydrodynamic testing facilities are the only ones in the nation capable of conducting fully contained above ground high explosive tests using plutonium. An obvious inference to be drawn from Caltech's proposal is that any new techniques for modeling polymorphic transitions will be extended by national lab scientists to include plutonium.


Box 3.1: Eulerian vs. Lagrangian Based Computer Codes

When writing a computer code which will predict how a fluid will flow, for instance air flowing around the body of a speeding car, there are two options for describing the motion. For example, one can calculate how the density and velocity of air changes with time at various points around the car, or calculate the path traveled by individual air particles as they flow around the car. The two approaches use different sets of equations, but if performed correctly, lead to the same solution or consistent results.

Eulerian Equations specify the density and velocity of the fluid at each point in space and at each instant of time. This approach to the problem focuses on what is happening at a particular point in space and time.

Lagrangian Equations describe the motion of fluid particles (or individual points in the fluid) as a function of time. Instead of focusing on what happens at particular points in space, this approach tracks what happens to individual fluid particles. This way of describing the flow of a fluid is an immediate generalization of Newton's particle mechanics taught in high school and college.


The section of Caltech's Alliances Program research devoted to "First principles computation of material properties" describes research that touches upon nearly all the other aspects of the Center's agenda. Again, the central problem of simulating phenomena over a large range of length scales leads the discussion:

In this section, we propose a hierarchical approach to materials modeling in which parameters are derived from quantum mechanics (QM) through averaging over successively larger scales of time and length. The approach leads to a rigorous description of continuum parameters required in describing crack initiation, spallation, chemical decomposition, etc. These computational techniques will be directed toward calculating phase behavior of metals, reaction kinetics relevant for HE, and structural information for metallic alloys (including actinides) at high temperatures and pressures (emphasis added).[65]

Given that plutonium and uranium(actinides(are used in nuclear weapons, it is relevant to understand where the United States has currently drawn the boundary between classified and unclassified research. U.S. classification guidelines exist for the equation-of-state (EOS) describing the relationship between the thermodynamic state variables of a material: pressure, density, temperature, energy, or entropy.

The U.S. classification rules for equation-of-state calculations and data prior to December 1997 are given in Box 3.2. The classified versions of these classification rules are more specific. The set of elements heavier than lutetium (Z=71, where Z is the charge of the atomic nucleus or equivalently the number of protons or electrons in the atom) contains elements for which equation-of-state data is "useful for designing nuclear weapons." For neptunium (Z=93) and plutonium (Z=94), the range of pressures for which equation-of-state data and calculations are unclassified is explicitly defined. Caltech states in its Alliances Program research proposal:

We propose to apply the method we used for Fe [Iron] to study a range of 3d, 4d, and 5d transition metal equation of state to ultra high pressure (1 TPa) . . . however, some systems [to which the method is applied] lead to enormous errors and problems are encountered with excited states and with actinides. We propose two approaches to solving this problem. . .[66]

The 5d transition metals[67] span hafnium (Z=72) through gold (Z=79): within the mass range for which classification guidelines may apply. As noted above, classification guidelines apply explicitly to actinides. The pressure regime Caltech intends to explore under its Alliances Program contract(up to 1 TPa (Tera-Pascal or 1012 Pascals)(is well above the classified pressure regime for neptunium and plutonium which begins at 0.002 TPa (20 kbar = 0.002 TPa). Classifying the high-pressure properties of gold (Z=79) and platinum (Z=78) may at first seem bizarre, but these materials are used in nuclear weapons. The classified pressure range for plutonium and neptunium corresponds to the range of pressures that bomb material is exposed to during a nuclear explosion.


Box 3.2: U.S. Classification Guidelines for Equation-of-State Calculations and Data
Prior to December 1997
[68]

The following data or calculations were considered unclassified prior to December 1997:

h.  The equation-of-state (EOS) studies for all elements under conditions other than those revealing classified information. (67-1)

i.  Information on EOS and opacities of certain materials not of significance to weapon design. (72-11)

j.  The calculated EOS data from theoretical models for certain materials (for Z less than 72 all materials; but for Z=72 and higher, only materials at pressures whose EOS data is not useful for designing nuclear weapons). (83-6)

k.  Information concerning EOS:

  1. (1) Static data for Z of 93 and 94 at pressures equal to or less than 20 kb. (89-1)

  2. (2) Static data for Z greater than 94 at pressures equal to or less than 1 Mb. (89-1)"


In December of 1997 Secretary of Energy Frederico Peña held a press conference in which the results of the Fundamental Classification Review were presented. The Report of the Fundamental Classification Policy Review Group, chaired by former Sandia National Laboratory director Dr. Albert Narath, discussed the changing context of basic physics relevant to nuclear weapons:

Classification of scientific information underpinning nuclear weapons design activities must be viewed in a somewhat different fashion. Limited resources have become and will likely remain a significant constraint in managing the acquisition of necessary scientific knowledge. However, the past 40 years have seen a large and sustained growth in areas of general science closely related to nuclear weapons technology -- astrophysics, condensed matter, high temperature experiments, and computer design and applications. These resources can be leveraged by encouraging scientific exchange between U.S. researchers and the worldwide community.

With no nuclear testing, the safety and reliability assurance of the stockpile will rest on the ability to attract and retain highly skilled scientists and engineers. Their willingness to center their careers in the nuclear weapons field may be enhanced to the extent that their scientific accomplishments can be recognized and rewarded by their peers in the open and unclassified arena. [69]

Lack of access to experimental facilities which can produce conditions similar to those encountered in a nuclear weapon explosion has historically limited the amount of unclassified or non-governmental research in sensitive areas. However, the Policy Review Group stated that "general science" is expanding in bomb-relevant areas. In their judgment, retaining quality personnel in the nuclear weapons program necessitates permitting SSMP research to be published in the open scientific literature. Here one sees international proliferation concerns reflexively subordinated to an ostensible requirement to sustain and enhance unilateral U.S. nuclear capabilities. Better to disseminate our nuclear weapons science research, the Policy Review Group argues, than to compromise its quality by obscuring the technical achievements of weapons program personnel, thereby discouraging the best and brightest from devoting their professional lives to nuclear weapons work.

The Report of the Fundamental Classification Policy Review Group contains the following passages on the classification of equation-of-state measurements and theory:

The relations between the thermodynamic variables of a material(density, temperature, pressure, energy, and entropy(are referred to as equations of state. Understanding nuclear weapon performance is dependent on good equation of state information at very high temperatures and pressures.

Because of the importance of uranium and plutonium to weapon design, the equations of state for the actinides (atomic number greater than 89) should remain classified. All currently classified equation of state information used in weapons design calculations should remain classified because it may embody empirical information gained by comparisons with classified experiments. Otherwise, equation of state information for elements whose atomic number is less than or equal to 89 can be treated as unclassified.[70]

More specific information on the new classification guidelines for equations-of-state is relegated to a classified appendix of that report. The Caltech research on 5d transition metals at very high pressures may or may not overlap the recently revised classification guidelines. What should be noted, however, is that the trend towards open publication of crucial nuclear weapons data -- basic physics in this instance -- is not merely the direction "general science" is evolving towards, but a conscious process abetted by the "Science-Based" Stockpile Stewardship Program.

As was discussed above, the explosion of a nuclear weapon is initiated through the basic mechanism of high-explosive-driven spherical compression of fissile material. The Caltech research program includes a study phenomena when shock waves impinge on the interface between materials.

An important component of the research to be carried out in the [Caltech] shock physics facility is the study of the interaction of strong shocks with material interfaces . . . Upon interaction with the shock, the material interfaces are impulsively accelerated and the resulting baroclinic generation of vorticity due to the misalignment of the resulting pressure and density gradients gives rise to the well-known Richtmyer-Meshkov instability and ultimately produces turbulent mixing that can contaminate or dilute the materials bordering the interfaces. The modeling and simulation of these Richtmyer-Meshkov instabilities and the resulting inhomogeneous anisotropic turbulence is a major thrust of the proposed research. The instability process as well as the modeling of the resulting turbulence lies at the heart of many ASCI applications. An understanding of compressible turbulence and mixing is essential, for example, in important ASCI applications in which shock-driven implosion is a key step (emphasis added).[71]

Here Caltech has subdivided this modeling effort into three stages: the contact of the shock wave with the material interface, during which the initial vorticity is generated; the growth of the layer at the material interface in which vortices have formed and mixing of the materials occurs; and the ensuing compressible turbulent flow. These are the very complex phenomena, related to the turbulent mixing of plutonium and D-T in the primary (or lithium deuteride and uranium in the secondary), that had hitherto rendered inadequate computational modeling inadequate, forcing a continuing dependence on nuclear explosive testing to establish confidence in new or modified nuclear explosive package designs.

The nature of the Caltech Alliances Program research begs the question of whether or not the Caltech virtual shock tube could in fact be used as a nuclear weapons code, as it is intended to ultimately contain the combined simulations of high explosives, shocked materials(including actinides; the effects of material interfaces; and shock-induced compressible turbulence and mixing. A bomb code would additionally need to incorporate criticality, fission and fusion nuclear processes, and the energy released in the explosion. These processes and related computer coding have been developed as part of the civil fission and fusion energy programs. In addition, the University of Chicago program, discussed below, is engaged in modeling fusion ignition and burn processes as part of an astrophysics research program.

Thus, if the Caltech research program is permitted to continue for its five to ten year course and the virtual shock tube produced, much of the work behind generating a bomb code will have been accomplished, at the same time incorporating fore-front physics and computer science calibrated against data from state-of-the-art university and SSMP facilities. Given that the Caltech program has the simultaneous missions of producing unclassified research products while remaining relevant to nuclear weapons simulation and also educating foreign nationals, it is clearly of concern with regard to the proliferation of nuclear weapons technology.


C. Stanford University

Background

Given that turbulence is an ubiquitous physical phenomenon, it is not surprising that it plays an important role in nuclear weapons. In the explosion of a nuclear weapon, the bomb material undergoes compressiblefluid flow with attendant compressible turbulence. Recall that the density of a fluid is defined as the mass per unit volume. If the density of a fluid changes insignificantly during its flow then this flow is termed incompressible, or a constant density flow. Low-speed flight through air and movement in water can usually be treated as incompressible flows. On the other hand if the density of the fluid during flow varies it is termed compressible (the fluid mass can be compressed into a smaller volume).

The equations governing compressible flow are much more complex than the equations governing incompressible flow.[72] In addition, significant portions of the compressible turbulent flow in either the nuclear weapon or gas turbine is reactive: energy is being generated by the flowing fluid either through chemical reactions as fuel and oxidizer (air) mix in the turbine or through nuclear reactions as the lithium-6, deuterium, tritium, neutrons, and other bomb constituents and thermonuclear reaction products mix in the nuclear explosion.

Unstable flows in highly compressed materials are ubiquitous in [nuclear] weapon physics. We typically must determine the thickness of the mixing layer between two materials caused by the passage of strong shock wave. Much research on turbulent flows relies on the assumption that the flow is incompressible (like water in an ocean). However, here we are interested in the situation where considerable compression and ionization can occur at the same time as turbulent, mixing motion.[73]

The importance of understanding and simulating turbulence for the U.S. Stockpile Stewardship Program is evinced by its mention in the "Green Book," the overarching plan for the future of the U.S. nuclear weapons program:

Theoretical models for specific phenomena are being developed and simulated in stand-alone codes. Often they address technical areas, such as instabilities and turbulence, which are not yet formally understood by the general scientific community. New scientific data must be obtained to guide the theoretical development and validation of these models, requiring new experimental capabilities.

When validated, the models must then be put into a form that is accurate and efficient for use in our weapon simulation codes.[74]

The field of Computational Fluid Dynamics (CFD) and DOE's Stockpile Stewardship Program are evolving towards diminished reliance on experimentation. The DOE expects to benefit from the CFD lessons in this regard, where full-system tests are not prohibited by an international treaty. At the Academic Strategic Alliances Program pre-proposal conference Charlie Westbrook from Lawrence Livermore National Laboratory gave a presentation on an "Example Center for Modeling Reactive Flows"[75] in which Professor Steve Koonin (California Institute of Technology Provost) was quoted:

You are developing computational models for nuclear weapons simulations that you can never test. Can you develop comparable models for unclassified systems that can be tested and validated?[76]

Claude Navier and George Stokes discovered the principal equations governing fluid flow approximately 150 years ago, but it was not possible to solve the Navier-Stokes equations even for some relatively straightforward problems (like the solution of planar, slowly-moving fluid flows around an object) until these calculations were performed on supercomputers in the late 1960s. Except for a few special cases there are no analytic solutions to the Navier-Stokes equations, only numerical solutions. The principal method of testing whether, for example, an aerodynamic design performs as intended has been to experiment in a wind tunnel, but supercomputing has demonstrably reduced the historical reliance on experimentation in this area:

Although both computational fluid dynamics and wind tunnels are now used for aircraft development, continued advances in computer technology and algorithms are giving CFD [computational fluid dynamics] a bigger share of the process. This is particularly true in the early design stages, when engineers are establishing key dimensions and other basic parameters of the aircraft. Trial and error dominate this process, and wind-tunnel testing is very expensive, requiring designers to build and test each successive model. Because of the increased role of computational fluid dynamics, a typical design cycle now involves between two and four wind-tunnel tests of wing models instead of the 10 to 15 that were once the norm.[77]

In a similar manner (as discussed in Chapter II), supercomputers have served to reduce the reliance of nuclear weapon designers on expensive and politically-charged nuclear tests. In the context of the Comprehensive Test Ban Treaty, improvements in the fidelity of computational fluid dynamics simulations will obviously be valuable for the nuclear weapons design establishment.


The Stanford University contribution.

As one of the five Alliances Program Centers DOE has funded the Center for Integrated Turbulence Simulations (CITS) at Stanford University. It is closely linked with the Stanford/NASA Center for Turbulence Research. The focus of the Stanford Center is on complex systems simulation for advanced system design as initially applied to aircraft gas turbine engines, in particular:

  • new turbulence models and associated numerical simulation methodologies that will enable a new paradigm for the design of advanced systems in which turbulence plays a controlling role;

  • numerical methods, compilers, operating systems, and computer architectures driven by and supporting these massively parallel turbulence simulations.

Initially the Center for Integrated Turbulence Simulations will focus on components of the gas turbine engine (compressor, combustor, and turbine) and later integrate the individual component simulations to model interactions among engine components. Future plans for the Stanford Center involve support from other sponsors on additional systems involving turbulence simulation.

The goal of the Stanford Alliances Program Center is to "develop simulation technology capable of dealing with systems as complex as a full jet engine and phenomena as complex as the plasma turbulence of a Hall thruster (emphasis added)."[78] With respect to the U.S. nuclear weapons program: "the benefits to science-based stockpile stewardship [from the Center for Integrated Turbulence Simulations at Stanford] are improved understanding of compressible flow computations, turbulence and transport modeling."[79]

Parviz Moin (currently Directory of the Stanford Center for Turbulence Research) and John Kim wrote a January 1997 Scientific American cover story: "Tackling Turbulence with Supercomputers." Moin and Kim define and give examples of turbulence in their article:

Practically all the fluid flows that interest scientists and engineers are turbulent ones; turbulence is the rule, not the exception, in fluid dynamics [both gases and liquids are classified as fluids: fluids are regarded as any substance which cannot remain at rest under a sliding, or shearing, stress].

But what exactly is turbulence? A few everyday examples may be illuminating. Open a kitchen tap only a bit, and the water that flows from the faucet will be smooth and glassy. This flow is known as laminar. Open the tap a little further, and the flow becomes more roiled and sinuous--turbulent in other words. The same phenomenon can be seen in the smoke streaming upward into still air from a burning cigarette. Immediately above the cigarette, the flow is laminar. A little higher up, it becomes rippled and diffusive.

Turbulence is composed of eddies: patches of zigzagging, often swirling fluid, moving randomly around and about the overall direction of motion. Technically, the chaotic state of fluid motion arises when the speed of fluid exceeds a specific threshold, below which viscous forces damp out the chaotic behavior. [80]

Between July and December 1997 the Alliances Program contracts were negotiated and signed between DOE and the five universities which won the Alliances Program competition. At least in the case of Stanford, the DOE insisted on a modifications to the research proposed earlier in the year. In an E-mail message dated 1 December 1997, CITS director William C. Reynolds responded to NRDC's request to obtain a copy of Stanford's original proposal:

I was reluctant to share the proposal because it contained some things that were not funded (a study of Hall Thrusters), and did not include some things that were added at DOE request (more computer science). We have just completed a 90th-day Project Plan Reprise that is up to date, and I will be happy to share that with you.

Thus, while Alliances Program centers do build upon established research bases at the universities, and may have important research emphases outside of nuclear weapons, DOE clearly has the capability to influence the Center agendas to conform them to the research priorities of its nuclear weapons program.


D. University of Chicago

The Center on Astrophysical Thermonuclear Flashes has been established as an Alliances Program 'Center of Excellence' at the University of Chicago. The Chicago center is the only one in which the simulation of a system involving thermonuclear reactions is being studied, albeit in stars rather than in nuclear weapon systems.

The relationship between the Chicago Center research and SSMP objectives is succinctly summarized early in the research proposal:

Thus, the astrophysical thermonuclear flash problem has many of the same features that the Stockpile Stewardship program faces: The physics is highly varied and complex; the range of spatial and temporal scales that must be treated is huge; and the ultimate physical system being modeled is not directly accessible to experiment, or even to observation. The daunting computational challenges are also similar, and validation closely coupled to the computations is essential in order to know whether the ultimate results are at all correct.[81]

More significantly, the physical conditions, and many of the physical phenomena, are similar to those confronted by the Stockpile Stewardship program. The (fully ionized) plasmas are at very high temperatures and densities; and the physical problems of nuclear ignition, deflagration, or detonation, interface dynamics, and turbulent mixing for complex multicomponent fluids are common to the weapons program.

All of the Center's activities are tightly-coupled and necessary to the ultimate solution of our overarching problem area. Further, the results of every significant activity. . .will be either directly applicable to similar problem areas in the ASCI program, or have direct analogs in that program. [82]

White dwarfs, neutron stars, and black holes are compact objects, formed when normal stars have consumed most of their nuclear fuel. Due to their large mass stars would collapse inward from the force of gravity were it not for the heat and countervailing pressure generated by thermonuclear reactions. White dwarfs and neutron stars are supported against further gravitational collapse by what is called degeneracy pressure: the 'pressure' produced by squeezing electrons or neutrons into the same physical space. Black holes are unsupported against gravitational collapse. Compact objects are much smaller than normal stars of comparable mass and thus have a much greater density.

Part of the Chicago Center research will focus on a binary system comprised of a star and a compact object, where matter from the star is sucked away by the gravitational force of its companion and is captured by or "accretes" onto the compact object. Another component of the research will focus on thermonuclear reactions in the cores of stars or compact objects. Table 3.2 lists the phenomena to be studied and ultimately simulated with nuclear weapons program funding at the University of Chicago's Center on Astrophysical Thermonuclear Flashes.

The research program at the University of Chicago has been partitioned into two "Paths," to be pursued simultaneously.

Path-1 is an evolutionary approach to gain early experience on ASCI machines by re-engineering and porting existing methodologies. In Path-1 we will develop, over a period of three years, a simulation environment to describe all of the physics listed above [hydrodynamics, reaction kinetics, radiation diffusion, and magnetohydrodynamics], within the framework of well-understood numerical methods and static grids. . .Furthermore, this will ensure that we will have a working production application by the end of the first year to begin work on the physical modeling and to make contact with the validation experiments.[83]


Table 3.2: Astrophysical systems to be simulated at the Chicago "Center on Astrophysical Thermonuclear Flashes" under the Academic Strategic Alliances Program.

The University of Chicago proposal states: "all have in common the ignition of a nuclear fuel under degenerate conditions, followed by the propagation of thermonuclear burning laterally via a convective or turbulent flame front (or deflagration wave), or radially outward via a strong shock front (or detonation wave)." [84]

SystemPhenomenonDescriptionAssociated Problem
Accreted envelopes of neutron starsX-ray burstsCombined hydrogen-plus-helium and pure helium flashesMasses and radii of neutron stars
Accreted envelopes of white dwarfsClassical novaeHydrogen flashesAbundance of intermediate-mass elements in the galaxy; how masses of white dwarfs change with time in a close, binary system
Core of low-mass asymptotic giant starsHelium shell flashesHelium flashes
Core of a white dwarfType Ia supernovaeCarbon flashesBirth rate and masses of neutron stars; abundances of intermediate and heavy elements in the galaxy; "standard candles" in determining the Hubble constant
Base of a deep, accreted envelope on a white dwarfType Ib supernovaeHelium flashes"


Path-2 involves novel approaches to increasing the speed and accuracy with which the thermonuclear flash simulations can be performed.

The motivation behind the Adaptive, Multi-Scale Dynamics (Path-2) strategy is to improve further the resolution capabilities of the simulations by introducing both temporal and spatial adaptivity in the application codes. In particular we will consider three areas, interface and front tracking, local mesh refinement, and timestep adaptivity. The purpose of this type of adaptivity is to resolve efficiently highly localized structures by concentrating the computational resources in those regions where high spatial and temporal resolution are needed. The algorithmic bases for the Path-2 applications are in general well-known on non-scalable architectures, but little understood for application to the ASCI-type machines.[85]

The activities of the Chicago Center are divided up into 33 percent astrophysics and validation (i.e., physics experiments) and 67 percent computational and computer science activities.

Experimental validation of the Chicago computer simulations will be obtained not only from astronomical observations, but also from weapons laboratory experimental facilities.

"At the University of Chicago, we propose specifically to do "table-top" research in the areas of sonoluminescence, interface motion, and mixing. . . In addition, we intend to collaborate with similar experimental and theoretical studies conducted at the DOE DP Laboratories, and to expand significantly our validation studies at high energy densities by collaborating with work carried out at the DOE DP Labs in problems related to, for example, mixing at accelerated interfaces, externally-driven implosions, and mixing in the presence of magnetic fields. Particularly relevant will be experiments conducted at the laser and pulsed power facilities."[86]

The Chicago Alliances Program Center defines its "deliverables" to DOE DP under its ASCI contract as 1) the codes themselves and 2) lessons learned from such an interdisciplinary collaboration "(viz., how one can get physicists and computer scientists to work together for a common goal, and do so productively and with a minimum of friction)."[87]


E. University of Illinois, Urbana-Champaign

The Academic Strategic Alliances Program Center established at the University of Illinois at Urbana-Champaign (UIUC) is focused on computer simulations of solid propellant rockets: "The goal of the proposed Center is the detailed, whole-system simulation of solid propellant rockets under both normal and abnormal operating conditions . . . The activities and resources of the Center will be organized to support simulation of rocket systems as the central objective."[88]

The director of Illinois' Center for Simulation of Advanced Rockets stressed to one of the authors of this report that the Center is intended to make a substantive contribution to the space-launch industry. One problem faced by engineers and scientists who work with solid rocket boosters is the difficulty of performing experiments that provide comprehensive, detailed data -- hence the importance of accurate computer simulations of some key performance and safety issues. We will not attempt to evaluate here the quality of the proposed rocket science or its impact on the space-launch industry; rather, we will focus on why DOE"s nuclear weapons program would enter into a multi-million dollar contract with UIUC to support this work.

Those involved in preparing the research proposal for the Illinois Center assert its value to the U.S. nuclear weapons program. In the Illinois research proposal submitted to Defense Programs, the issue of its relevance to the Accelerated Strategic Computing Initiative is specifically addressed (as was required by the DOE):

This [University of Illinois, Urbana-Champaign] proposal is fully responsive to the scientific and technological needs of the United States Department of Energy posed under the Accelerated Strategic Computing Initiative/Academic Strategic Alliance Partnership Program (ASCI/ASAP). The outstanding quality of the faculty and staff, facilities, and research infrastructure offered by UIUC will enable a unique partnership between university researchers and the DOE Defense Program [nuclear weapon] laboratories. State, regional, and university resources are committed to the program, and an experienced research team is dedicated to fulfilling the mission of the Center. . . The simulation of advanced rockets is an outstanding focus for an ASAP center.

It includes almost every technical area of interest to the ASCI program and enables the execution of a wide variety of performance, safety, and reliability simulations.[89]

The Chancellor of the University of Illinois, Michael Aiken, wrote to DOE:

The national importance of simulation to the science-based stockpile stewardship program cannot be overestimated, and it is an area in which we [the University of Illinois at Urbana-Champaign] are willing and highly able to contribute. . . I believe that multidisciplinary research and training partnerships embodied in the DOE Academic Strategic Alliance Program and our [Illinois'] proposed Center are critical to the development of national defense policy and preparedness, and we are excited about the prospects of working in this enterprise.[90]

Crucial linkages between the Illinois Center and the U.S. nuclear weapons program are broadly imputed to UIUC: "University leadership has placed a high priority on the success of this Center, fully cognizant of the national importance of simulation to the science-based stockpile stewardship program."[91] In an afterthought to a list of five Illinois Alliances Program goals, the research proposal states: ". . . another important goal of the center is to provide relevant physical modeling input and computational tools developed for the rocket simulation to the DOE DP labs for the pursuit of their missions. (emphasis added)"[92]

Despite the fact that most of the United States' strategic nuclear weapons are deployed on solid propellant missiles, the Department of Energy does not list rocketry as one of its "scientific and technological needs" in this context. However, as with all of the Alliances Program Centers, the issue of experimental validation of the computer simulations arises, and in the Illinois case experimental validation will be performed at the UIUC Center for Novel Energetic Materials to Stabilize Rockets (CNEM). CNEM was recently established by the U.S. Ballistic Missile Defense Office and the Office of Naval Research.

The proposed [UIUC Alliances Program] Center will be a microcosm of a DOE DP laboratory in that it will have designers/customers (in this case the "rocket scientists" of CNEM) and interdisciplinary teams of scientists and engineers whose goal is to provide the designers with integrated simulation tools to evaluate the various options in design space.[93]

The word "customers" above also refers to the common characterization of the Department of Defense (DOD) as the customers of DOE for their commodity, nuclear weapons. Collaborations between the national nuclear weapons laboratories and the Ballistic Missile Defense Office (BMDO) are currently under review. In 1996, Congress directed Los Alamos, Livermore, and Sandia National Laboratories to report on how they could further help the BMDO. Outgoing LANL director Sig Hecker stated that the Accelerated Strategic Computing Initiative (parent program to the Academic Strategic Alliances Program) could be leveraged at missile defense issues, as elaborated on by Tom Meyer (of LANL's DOD Program Office):

This [ASCI Program] is far beyond the state-of-the-art of what exists now, and I think it is a computational capability that BMDO could use. . .That's one of the things that we're interested in, doing the campaign-level decision-making kinds of tools where you fight the entire war game of a full-up ballistic missile defense, or a theater missile attack depending on which one you want to look at, and you do a complete simulation of all the scenarios involved.[94]

Los Alamos National Laboratory's incoming director, John Browne, had managerial responsibility for all of the Strategic Defense Initiative Organization (SDIO, commonly referred to as "Star Wars") work at the lab.

The Illinois' Center research involves not only improving upon simulations of individual solid rocket booster components but also -- importantly -- integrating the component simulations into a simulation of the entire solid rocket booster.

Full simulations of such complexity will require a sequence of incremental developments -- in engineering science, computer science, and systems integration--over an extended period of time. From the outset, however, the emphasis will be on system integration rather than separate threads of development that eventually come together at some point in the future. Rapid exploration of critical system integration issues will entail the use of simplified -- but fully integrated -- models and interfaces initially, to be followed by successively refined models. . .[95]

The decision to emphasize system integration up front is one of the most important determinants for the Illinois Alliances Program agenda and highlights the importance of this issue for the success of the SSM Program's technical strategy. System integration entails correctly interfacing the output of component simulations which may use different computer languages, grid structures and resolution, or coordinate systems (e.g., Eulerian vs. Lagrangian).

The five principal Center research areas are: structural and solid mechanics; fluid flow simulation; combustion and energetic materials; interface code (computer science research on component integration); and computational infrastructure (programming environment, data management, I/O). The UIUC Alliances center has proposed to devote 60 percent of its research efforts to science and engineering and 40 percent to computer science (an analysis of this mix for the overall Stockpile Stewardship Program would be interesting).

The generic goal of both DOE's Stockpile Stewardship Program and the Academic Strategic Alliances Program Centers is the "modeling and simulation of complex, multiscale systems." Solid rocket boosters and nuclear weapons are complex systems in the sense that they contain numerous subsystems and components which interact in a variety of ways. These components are integrated in order for the system as a whole to produce the desired effect. "Multiscale" systems are those in which important phenomena occur over different scales in length, time, energy, and other parameters.

For example, the physical dimensions of nuclear weapon components are macroscopic: readily visible to the human eye and on the order of millimeters to meters in length. The nuclear and chemical reactions occurring in the bomb during an explosion are explicitly described by the physics and chemistry of interacting atoms, molecules, and subatomic particles; that is on a microscopic scale on the order of nanometers in length (10-9 meters or 1/1,000,000,000 of a meter). Bridging these two extreme length scales is what is sometimes referred to as the mesoscopic length scale: on the order of a micron in length (10-6 meters). Some of the structural details of bomb components, such as the irregularities in a machined surface, fall within the mesoscopic length scale. Nuclear weapons and solid rocket boosters are multiscale in that important phenomena occur at the microscopic, mesoscopic, and macroscopic length scales.

Ideally, a multiscale system would be comprehensively simulated at the smallest scales involved. It is logical to assume that a correct simulation at the microscopic level would reproduce mesoscopic and macroscopic phenomena, since these longer length scale phenomena have their ultimate origins at the atomic level. Certainly the existing nuclear weapons codes are a marriage of macroscopic, mesoscopic and microscopic modeling. Issues for the ASCI Program are the accuracy of current microscopic models of individual phenomena and the integration of numerous fragmentary microscopic and mesoscopic models into the reliability, safety, and performance modeling of the entire nuclear weapon system for particular geometries and at longer length scales.

One example of UIUC multiscale solid rocket booster accident simulation is the mechanical failure of the solid propellant due to a crack. The crack in the propellant is believed to grow during booster firing due to mechanical stress and the combustion process itself. Specifically, the propellant is composed of explosive material in the form of closely packed 30-300 micron-sized grains held together by a polymeric (plastic) binder material. Cracks propagate due to failure of the binder between the grains. The simulation aims "to capture effects of the granular response on the macroscale failure process of the solid propellant."[96]

A volume on the order of several cubic centimeters is simulated at the granular level, and the rate of propagation of the crack is simulated at "subscale" or at the molecular level. "The rate-dependent, nonlinear constitutive model used for the cohesive elements derives from subscale finite element analyses involving a crack propagating between adjacent cracks. Such simulations include features of the subgrain microstructure (voids in the binder, adhesive vs. cohesive failure), and draw upon even more refined, discrete treatment of material at the molecular scale."[97] This is an example of a simulation which incorporates phenomena over many length scales -- a key research interest of the SSMP.

The Alliances Program Center at UIUC is tackling the broad simulation issues of multi-component, multi-phenomena, multi-scale systems for solid rocket boosters: "Although the individual [solid rocket booster] component technologies are reasonably well understood, what has been lacking is high-resolution, fully three-dimensional, integrated modeling and simulation of their complex interactions. . ."[98] A key question for the DOE is whether the ASCI Program can deliver simulation capabilities sufficient to design and engineer new nuclear weapon components, modifications to existing nuclear weapon types, and new nuclear weapon types. How far analogous rocket simulations can go in this respect will be explored by the Illinois Center's program:

Moreover, the use of new, higher performance or more environmentally benign propellants will require systematic redesign to account for higher structural loads and temperatures, and resulting changes in system instabilities. Similarly, in retrofitting ICBM [Intercontinental Ballistic Missile] motors to launch commercial payloads into Earth orbit, relatively minor alterations in structural materials and payload mass may significantly degrade performance and safety.

Comprehensive simulation will provide a much safer and less expensive way to investigate technical issues in rocket design than traditional methods based on experimental trial and error.[99]

DOE has launched five experiments in teraflop-level simulations in the Academic Strategic Alliances Program, and the successes or failures of the Illinois center will, like those at the other four Alliances Program Centers, inform the analogous efforts at the nuclear weapons laboratories to replace nuclear testing with computer simulation.


F. University of Utah

The "Center for the Simulation of Accidental Fires and Explosions (C-SAFE)" has been established at the University of Utah under the Academic Strategic Alliances Program. C-SAFE is a multidisciplinary collaboration between the University of Utah, Brigham Young University, Utah State University, and the Thiokol Corporation. U. of Utah's research proposal submitted to the DOE described three accident scenarios which would be computer simulated as the initial focus of the Alliances Program Center:

  • rapid heating of a container with conventional explosives in a pool fire (e.g., an atomic bomb involved in an intense jet-fuel fire after an airplane crash);

  • impact and ignition of a container with subsequent explosion and firespread (e.g., shelling of a mine storage building by terrorists); and

  • heterogeneous fire containing a high energy device (e.g., ignition of a containment building in a missile storage area.[100]

The C-SAFE work would produce computer tools for the numerical simulation of these and eventually other potential accidents, incorporating fundamental chemistry and engineering physics at an unprecedented level to "help to better evaluate the risks and safety issues associated with fires and explosions."[101]

Given DOE Defense Program's responsibilities for assembling, disassembling, storing, and handling nuclear weapons and their components, the relevance of C-SAFE's research to the SSMP may seem the most obvious among the five Alliance Program Centers. The Utah proposal states: ". . .the resulting simulation will have direct value to the ASCI program in its efforts to mitigate collateral damage and death due to accidental events involving the stockpile."[102]


Table 3.3: A listing of the U. of Utah C-SAFE research areas and component events of computer-simulated accident scenarios.

C-SAFE Research AreasC-SAFE Accident Scenario Components
  • Fundamental gas and condensed phase chemistry

  • Structural mechanics

  • Turbulent reacting flows

  • Convective and radiative heat transfer and mass transfer

  • Computational engineering

  • Computer Science
  • Ignition

  • Fire Spread

  • Container Dynamics (physical and chemical changes in containment vessels and structures; mechanical stress and rupture of the container)

  • High Energy Transformations (Deflagration-to-detonation transitions of any energetic material in the fire)


But in Utah's discussion of the value of C-SAFE to ASCI, a greater emphasis is given to the idea of supporting an interdisciplinary effort aimed at achieving "a quantum step forward in the ground-up integration of atomisitic, non-equilibrium chemistry and state-of-the-art engineering with advanced computational techniques." Thus (at least Utah) conceives of the Alliances Program contract with DP not just as a work order for computer codes to be used in accident analyses, but as a test of the "ASCI paradigm:" interdisciplinary collaboration, the integration of scientific models at molecular, mid- and macroscopic scales, and the incorporation of this work in a user-friendly simulation on 1-100 teraflops-speed, massively parallel computer systems at the national nuclear weapons laboratories.

Organizational issues associated with a large, interdisciplinary science/computing effort are clearly of interest to the DOE. C-SAFE's organizational structure is shown in Table 3.4.


Table 3.4: C-SAFE organization by teams.

The C-SAFE is organized by two sets of teams. One set of teams is the "Discipline Teams," responsible for the fundamental science and engineering of each discipline. The other set of teams is the "SDRM [Simulation Development RoadMap] Step Teams," each of which is responsible for one step of the overall simulation. Participants in C-SAFE will be members of each team. Such organizational issues are of interest to the DOE in implementing the SSMP/ASCI nuclear weapons "virtual testing" and "virtual prototyping" efforts.

C-SAFE Discipline TeamsC-SAFE SDRM Step Teams
Molecular Fundamentals Team: molecular dynamics, electronic structure, and statistical mechanics in an integrated fashion to dynamically obtain properties for all materials (condensed phases, vaporized phases, and structures) in the fire and explosion

Computational Engineering Team: develop meso-scale models that bridge the ranges of length and time scales between microscopic and macroscopic properties. . .large-scale Eulerian and Lagrangian models to describe structural and transport processes with geometric and mechanistic fidelity

Computer Science Team: system development framework which combines target architecture performance analysis tools at the lowest level with an integrated, higher level scientific problem solving environment to provide interactive computational steering, visualization and large data set analysis capabilties

Ignition Team

Fire Spread Team

Container Dynamics Team

High Energy Transformations Team

Each C-SAFE participant will be a member of a Discipline Team and of a SDRM Step Team.

Rationale:

  1. Common objective of developing a verified, fire and explosion simulation system achieved
  2. Modern scientific & computational techniques used throughout


The C-SAFE proposal (on which the present discussion is largely based) was submitted to the DOE by Utah in anticipation of a $25 million grant over five years, with a high probability of renewal at this level of funding for an additional five years. That the DOE has an interest in an Alliances Program Center devoted to accident simulations can be inferred from a Sandia presentation on an "Example Center for Modeling Accident Environments and Events" at the ASCI Academic Strategic Alliances Program Pre-proposal Conference. Utah's request in May of 1997 totaled $26.78 million. It appears, however, that the DOE is only giving the five Alliances Program Centers $20 million apiece over five years, or 25 percent less that Utah expected.

Only $3 million of the $20 million will reach Utah during the first year of the five-year grant -- an average of $115,000 for each of 26 participating faculty members.

The computer scientists "can get much bigger grants than $115,000 on their own," [University of Utah Vice President for Research Richard] Koehn said. "Individual faculty members were questioning their participation in this program based on the reduced budget.[103]

This state of affairs is reflected in an E-mail message to NRDC from Utah regarding the original C-SAFE research proposal, dated 14 November 1997:

One fact you must take into consideration as you read it is that we did not receive full funding for the proposal (we received $20 million, not the 26.1 million originally bid) so we have had to downscale the scope of work substantially. We will be putting the revised implementation plan on our web site soon, which was what I was referring to in my first message.

As of the date of this NRDC report, the University of Utah has not updated its web-page with a revised C-SAFE implementation plan.


G. SSM Research Agenda of the Five Alliances Program Centers

Table 3.5, below, lists the areas of physics, chemistry, and engineering in which the Academic Strategic Alliances Program university Centers are engaged in research with DOE Defense Programs funding. Commonality across the university agendas is found, particularly for high explosives, material properties under extreme conditions, and compressible, reactive, turbulent flows, all of direct relevance and interest to the SSMP. A common theme underlying all five university research agendas is the construction of a computer simulation which predicts the behavior of a macroscopic system based on the mid-scale and microscopic analyses of the fundamental physical processes. A second common theme is the extension of prior computer simulation efforts to permit three-dimensional, integrated component simulations.

For Caltech and Chicago, validation of the simulations will involve experimental research at SSMP nuclear weapons facilities (DAHRT and NIF, respectively). In general, the DOE has stated that the successes or failures of the five university Alliances Program Centers to simulate a variety of complex phenomena will inform its parallel efforts to simulate nuclear weapons testing and production under the constraints of the Comprehensive Test Ban Treaty.


Table 3.5: A listing of the areas of scientific research for the five Academic Strategic Alliances Program university Centers of Excellence.


CaltechStanfordChicagoUIUCUtah
High Explosives X

XX
Material Properties under Extreme Conditions (High Temperatures, Pressures, Densities, Particle Velocities) XXXXX
Hydrodynamic Instabilities X
X

Compressible, Reactive, Turbulent Flows XXXXX
Thermonuclear Ignition and Burn

X

Material Aging and Defects X

XX
Integrated Component Simulation of an Engineered Device
X
XX
MPP Computer Science; Adaptive Mesh Refinement; Data Visualization; Problem Solving Environments XXXXX



Notes

46. "ASCI Academic Strategic Alliance Program: Major Goals and Objectives," http://www.llnl.gov/asci-alliances/asap/goals.html (last modified November 27, 1996).

47. The list of attendees to the ASCI Alliances Pre-proposal Conference is given on the world wide web at http://www.llnl.gov/asci-alliances/asap/attendees.html. "Last modified December 20, 1996."

48. ASCI Strategic Alliances Program Research: Overview of the Background Papers, www.llnl.gov/asci-alliances/bb/001toc-bb.html (pg. 1). "Last modified November 27, 1996."

49. At the Alliances Program Pre-proposal Conference, Derrol J. Hammer (HPCC Group Leader, Procurement & Materiel, LLNL) listed "Relevance and practicality to ASCI goals" first on a vu-graph entitled "Preliminary Proposal Review." Hammer's vu-graph entitled "FINAL PROPOSALS" states that the "Unweighted evaluation will focus on strengths and deficiencies," but that the DOE will "Select Best Overall Combination of Disciplines to Meet ASCI Goals (We are not selecting "best proposals") (emphasis in original)."

50. "ASCI Strategic Alliances Program Research: Overview of the Background Papers," http://www.llnl.gov/asci-alliances/bb/001toc-bb.html (Last Modified November 27, 1996).

51. Defense Programs would not make any information on these contracts available to NRDC, beyond confirming that the University of California, Berkeley, had received Academic Strategic Alliances Program funding.

52. "Computer codes are special because they are central—they have it all together. It's obvious that we deny outsiders access to our [nuclear weapon] design information and [nuclear] test data; perhaps it is less obvious that for the same reasons we have to restrict what is said about the codes. Let me explain why that is and the kinds of questions an outside party, if he knew the evolution of our codes, would look at and exploit. He could ask, 'What are they working on new?' Or, 'What do they plan to develop?' He would infer that by which parts of the codes are markedly better than they were previously and which parts are the same. 'What physics topics are being improved and where is something being modeled not quite correctly and being improved?' This is important information to the other side, from which he might anticipate improvements in our weaponry. Finally, 'Which thing are they not doing correctly, and what does that imply in terms of design weaknesses of their weaponry that they may not even appreciate?;'" Henderson, "Computation: The Nexus of Nuclear Weapons Development," p. 143.

53. Secretary Peña conducted an "Openness Press Conference" on December 22, 1997. NRDC is representing 38 organizations nationwide in a lawsuit against DOE for failing to comply with a 1990 Federal District Court Order requiring the preparation of environmental impact statement covering the activities of two broad federal programs conducted by the Department of Energy -- "Stockpile Stewardship and Management", and "Waste Management and Environmental Restoration." DOE officials contend that the lawsuit has had a "chilling effect" on their willingness to provide the 38 plaintiff organizations (i.e. the public) with information about DOE programs connected in some way with the lawsuit.

54. "PBXs [Plastic Bonded Explosives] are typically used as maincharge explosives in nuclear weapons. PBXs in current use include LX-10 and LX-17. LX-10 is a very energetic explosive based on HMX that uses Viton A (vinylidine fluoride/hexafluoropropylene copolymer, made by DuPont) as a thermoplastic binder; it is used in the W68 and W79 weapons. LX-17 is a less energetic TATB-based explosive that uses thermoplastic Kel-F 800 (chlorotrifluoroethylene/vinyliding fluoride copolymer, made by the 3M co.) as a binder. LX-17 is considered an IHE [Insensitive High Explosive] because TATB is a highly insensitive explosive that does not readily detonate accidentally. It is used in several modern nuclear weapons, including the B83, W84, and W87." From "Formulating High Explosive Materials," Energy and Technology Review, Lawrence Livermore National Laboratory, February 1988, p. 25.

55. "A Facility for Simulating the Dynamic Response of Materials," California Institute of Technology, August 14, 1997, Section 2.2.1, p. 1. The Caltech Alliances Program proposal is available on the World Wide Web at: http://www.cacr.caltech.edu/~jpool/ASAP/proposal/ Note that this quote could also describe the overarching Stockpile Stewardship and ASCI problem: current weapons codes lack a full physics description and rely on test data for calibration. As high explosives play such an important role in nuclear weapons, the relevance of the quote to SSMP is more than just analogy.

56. Caltech refers to this as the "micromechanical level" of simulation.

57. A review of high explosive modeling and experimentation at Livermore is given in "The New Science of High Explosives, Science And Technology Review, LLNL, June 1997; and in "High Explosives," Energy and Technology Review, LLNL, January/February, 1988.

58. "A Facility for Simulating the Dynamic Response of Materials," Section 2.2.1, p. 1.

59. AMRITA is an acronym for "Adaptive Mesh Refinement Interactive Teaching Aid." On April 16, 1997, Professor James J. Quirk gave a lecture on AMRITA at his institution, Caltech. The talk abstract includes this descriptive information: "Although Adaptive Mesh Refinement (AMR) algorithms have matured to the point where they provide a robust, economical means of computing flows governed by disparate physical scales, because of their high development costs, they have failed to impact on the grass roots scientific community to the extent they deserve. In an effort to improve this situation, albeit in one small area of computational fluid dynamics (CFD), I have unified my own software tools to form an operating system (Adaptive Mesh Refinement Interactive Teaching Aid) which can be used as a jump-start by anyone interested in gaining a toe-hold in the world of AMR. AMRITA's web page is currently http://www.galcit.caltech.edu/~jjq/. Professor Quirk is a co-investigator on the Caltech Academic Strategic Alliances Program grant. In section 2.4 of Caltech's proposal ("Collaborations with ASCI Laboratories"), it states: "J. Quirk and J. Shepherd are investigating the application of AMRITA to detonation problems of interest to the DX [Dynamic eXperimentation] group at LANL."

60. "A Facility for Simulating the Dynamic Response of Materials," Section 2.2.1, p. 1

61. Ibid.

62. "Facility for Simulating the Dynamic Response of Materials," Section 2.2.2 Modeling dynamic response of solids, p. 1.

63. Caltech claims that beryllium and uranium modeling was not funded by the DOE. Even if true, however, extending the Caltech model to nuclear weapons materials is a simpler exercise than initially developing the model.

64. Ibid., p. 2.

65. "Facility for Simulating the Dynamic Response of Materials," Section 2.2.3 First principles computation of materials properties, p. 1.

66. Ibid., p. 2.

67. The group of metals known as transition metals are classified differently from other metals not because of their physical properties, but because of the structure of the electrons orbiting the atomic nuclei. The transition metals have their valence electrons in more than one shell. Regular, or representative metals have their valence electrons in only one shell.

68. Drawing Back the Curtain of Secrecy: Restricted Data Declassification Policy, 1946 to the Present, RDD-1, June 1, 1994, U.S. Department of Energy ( http://www.doe.gov/html/osti/opennet/document/rdd-1/drwcta.html ).

69. "Report of the Fundamental Classification Policy Review Group," Dr. Albert Narath, Chari, Unclassified Version (Classified Material Has Been Removed), Issued by the Department of Energy, December 1997, pp. 30-31, emphasis added.

70. Ibid., p. 31.

71. "A Facility for Simulating the Dynamic Response of Materials," Section 2.2.4 Compressible turbulence and mixing, p. 1.

72. The Navier-Stokes Equations are a set of partial differential equations that describe the motion of a continuous fluid. The set contains 5 equations: mass conservation, 3 components of momentum conservation, and energy conservation. Certain other properties of the fluid being modeled must be specificied, such as the equation of state. The Navier-Stokes Equations are non-linear and coupled. For practical problem-solving purposes, non-linear means that solutions to the equations cannot be added together to get solutions to a different problem, i.e solutions can't be superimposed. Coupled means that each equation in the set of 5 depends upon the others, so that they must all be solved simultaneously. But if the fluid can be treated as incompressible, then the conservation of energy equation can be de-coupled from the others and a set of only 4 equations must be solved.       It is generally agreed that turbulence is modeled by the Navier-Stokes Equations. However the many scales of motion that turbulence contains, especially it's micro-scales, cause the modeling of turbulent processes to require an extremely large number of grid points.

73. S. B. Libby, "NIF and National Security," Energy and Technology Review, Lawrence Livermore National Laboratory, December 1994, pp. 23-32.

74. Stockpile Stewardship and Management Plan (the "Green Book"), U.S. Department of Energy, Office of Defense Programs, February 219, 1996, p. IV-24.

75. "Example Center for Modeling Reactive Flows," Charlie Westbrook (LLNL), ASCI Academic Strategic Alliances Program: Pre-Proposal Conference, December 5-6, 1996, Presentations, http://www.llnl.gov/asci-alliances/slide/slides.html.

76. The citation on Westbrook's transparency reads: "Professor Steve Koonin, Provost, California Institute of Technology, National Security Advisory Committee, November 26, 1996, paraphrased." This point perhaps requires a little elaboration. It is certainly possible -- indeed it is already planned as part of the SSM Program -- to test the validity of ASCI simulations by "post-dicting" the results of previous nuclear explosive tests representing the existing proven spectrum of design possibilities. Koonin's remark is literally true only for nuclear explosive simulations of weapon designs that are beyond, or at least at the margins of, the proven design space.

77. Parviz Moin and John Kim, "Tackling Turbulence with Supercomputers," Scientific American, January 1997, available on the web at http://www.sciam.com/0197issue/0197moin.html

78. "Center for Integrated Turbulence Simulations," http://ctr-sgi1.stanford.edu/CITS/

79. "How the Universities Will Support DOE's Accelerated Strategic Computing Initiative," Fact Sheet for News Release NR-97-07-07, Lawrence Livermore National Laboratory (July 31, 1997).

80. Moin and Kim, "Tackling Turbulence with Supercomputers."

81. "Center on Astrophysical Thermonuclear Flashes," The University of Chicago (undated: received via fax by NRDC on 2 December 1997 from Carrie Clark of the University of Chicago, Department of Astronomy and Astrophysics), p. 2.

82. Ibid., p. 6.

83. Ibid., p. 12.

84. Ibid., p. 1.

85. Ibid.

86. Ibid., pp. 20-21.

87. Ibid., pp. 24-25.

88. "Center for the Simulation of Advanced Rockets," March 25, 1997, University of Illinois at Urbana-Champaign, Executive Summary, p. i.

89. Ibid.

90. Letter from Michael Aiken, Chancellor, University of Illinois, Urbana-Champaign, to Gilbert Weigand, Deputy Assistant Secretary, Defense Strategic Computing and Simulation, Department of Energy, dated March 21, 1997 (this letter immediately follows the cover page in the University of Illinois' proposal, the "Center for the Simulation of Advanced Rockets.")

91. "Center for the Simulation of Advanced Rockets," p. ii.

92. Ibid., p. 22.

93. Ibid., p. i.

94. "Los Alamos Offers Support for BMDO," BMD Monitor, November 14, 1997, Vol. 12, No. 23.

97. "Center for the Simulation of Advanced Rockets," p. 23.

96. "Center for the Simulation of Advanced Rockets," p. 7.

97. Ibid.

98. Ibid., p. 1.

99. Ibid.

100. "Center for the Simulation of Accidental Fires & Explosions (C-SAFE): submitted to the Accelerated Strategic Computing Initiative by The University of Utah," Principal Investigator and Director David W. Pershing, Ph.D., Executive Summary, p. i.

101. Ibid.

102. Ibid., p. 1-2.

103. Siegel, Lee, "Huge Grant Expected by U. Will Be Millions Smaller," The Salt Lake Tribune, November 7, 1997.

Back to contents page

last revised 1/22/1998

Sign up for NRDC's online newsletter

See the latest issue >

Switchboard Blogs

Feeling a Cold War Chill over Ukraine
posted by Matthew McKinzie, 7/23/14
China Environmental News Alert
posted by Greenlaw from NRDC China, 6/22/14
Senate hearing on nuclear reactor decommissioning challenges
posted by Bemnet Alemayehu, 5/17/14

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
YouTube