The Strategic Offensive Reductions Treaty signed in May cuts the U.S. and Russian deployed strategic nuclear arsenals to between 1,700 and 2,200 warheads. Although a significant step forward, the long-term impact of the treaty remains unknown because the accord does not require the destruction of warheads or delivery vehicles removed from service. If the United States and Russia are to cut their nuclear arsenals further and fully develop their emerging strategic partnership, predictability and irreversibility of arms reductions will be essential. Neither country must be capable of gaining an overwhelming unilateral advantage by quickly reconstituting its nuclear stockpile—a threat that concerns both Washington and Moscow today.
The substantial capabilities of the Russian warhead production complex have been of concern to some U.S. policymakers and nuclear weapons experts. The Russian complex is very large relative both to the U.S. complex and to the projected size of Russia’s nuclear weapons stockpile. Because of manufacturing and technology problems that limit the lifetime of Russian warheads to 10-15 years and because of stockpile management practices that emphasize routine rebuilding of nuclear warheads, the Russian complex also maintains high levels of production. The fear, then, is that Russia could use its large and active production infrastructure to offset arms reductions by quickly building new nuclear warheads in large numbers—that it has a “breakout capability.” Russia’s secretive nuclear weapons policies and uncertainties about the technical capabilities and operational practices of its nuclear weapons complex further exacerbate these fears.
The Department of Defense is using Russian warhead production capacity as part of its justification for maintaining a large stockpile of nondeployed warheads in the United States even as it reduces
its deployed arsenal under the StrategicOffensive Reductions Treaty. Appearing before the Senate in February, Undersec-retary of Defense Douglas Feith explained the U.S. rationale for storing, rather than destroying, warheads removed from service:
Russia has a large [nuclear weapons] infrastructure. They have a warm production base capable of producing large numbers of new nuclear weapons annually. The United States has not produced a new nuclear weapon in a decade, and it will take nearly a decade and a large investment of money before we would be in a position to produce a new nuclear warhead. So the issue of…whether we choose to build up a large infrastructure that would put us in a position to create new nuclear weapons if circumstances in the world changed and warranted it, versus taking weapons and rendering them unavailable for use in the near term by putting them in storage is an issue that…needs to be examined….1
Conversely, this plan to maintain a large stockpile of stored warheads—the so-called responsive force—is of great concern to Russian experts and officials who believe that these warheads provide the United States with a significant breakout capability.
Producing new warheads requires a source of electronic and mechanical non-nuclear components, tritium and a capability to load tritium reservoirs, fissile material components, and facilities to conduct final assembly of nuclear warheads. Only a few factors, however, are likely to limit the U.S. and Russian capability to “surge” production. Non-nuclear warhead components can be stockpiled, cannibalized from inactive or retired warheads, or procured off-the-shelf. Secondary explosives, composed of thermonuclear fuel and possibly highly enriched uranium, also have a long shelf life, and each country could retain substantial reserves. The availability of tritium, which radioactively decays to helium, could be a limiting factor in new warhead production, but for the foreseeable future both the United States and Russia are expected to maintain significant stocks, as well as the ability to produce tritium and load it into reservoirs.
New warhead production is most likely to be limited by the capacity to actually assemble new warheads and, to an even greater extent, by the availability of stored or new plutonium pits, the explosive cores of nuclear weapons. Analysis of the U.S. and Russian ability to build new warheads and the availability of plutonium pits shows that the United States has a sizeable warhead production capability and that the concern about Russia’s production capacity is not warranted at present. In addition, a close look at each country’s surge nuclear production capacity suggests ways in which the United States could configure its nuclear stockpile to facilitate further arms reductions and reassure Moscow while still hedging against a possible Russian breakout.
Warhead Assembly Facilities
Under routine, nonemergency conditions, warhead assembly plants perform a wide range of activities. For example, they dismantle retired warheads, produce new ones, refurbish and modernize old warheads, and disassemble and inspect warheads. The capacity of a warhead assembly facility can vary considerably depending on what type of work it is doing, what kinds of warheads it is working on, and the availability of support infrastructure and personnel. For example, the Pantex plant in the United States can disassemble and inspect only 250-350 warheads per year, but it can dismantle 3,500-4,500 warheads per year.2
However, a number of the tasks that such facilities perform are similar in nature, and for the purpose of estimating warhead assembly capacity, they can be generalized as “warhead operations.” In particular, the operations of warhead refurbishment, warhead dismantlement, and new warhead assembly each take roughly the same amount of time and require comparable specialized resources such as personnel, warhead assembly bays and cells, storage facilities, and transportation assets.
In a national emergency, a warhead assembly facility would presumably de-emphasize routine stockpile maintenance and dismantlement operations and instead concentrate on producing new warheads. This “surge production capacity” could thus be higher than regular production capacity when a facility is performing its regular mix of activities. As discussed below, although the U.S. warhead assembly capacity is significantly smaller than that of Russia, each country is capable of fielding new warheads in large numbers.
The United States
The United States has two facilities certified to work with uncased high explosives and nuclear materials. The Pantex Plant near Amarillo, Texas, is its primary warhead assembly plant. The plant has not produced a new warhead since 1991 and in recent years has been primarily involved in warhead dismantlement, refurbishment, and inspection activities. During the Cold War, Pantex performed 2,000 warhead operations per year. The projected production requirements for 2015 are approximately 1,000 warhead operations per year, most of which would be warhead refurbishments.3 Pantex’s surge production capacity would depend on specific warhead systems and is difficult to estimate, but this analysis conservatively estimates it to be 3,500 warheads per year, a rate consistent with the lower estimate of the stated dismantlement-only capacity of Pantex. (Because the process of dismantling a warhead is essentially the reverse of assembling one, a facility’s capacity for producing new warheads is likely comparable to its capability to dismantle them.4 )
The second facility, the Device Assembly Facility (DAF) at the Nevada Test Site, is a compact, state-of-the-art operation that was designed to support nuclear testing. At present, DAF supports subcritical experiments and is designated to handle unusual cases (such as disassembling damaged or foreign nuclear explosives). It has five warhead assembly cells and seven bays, as compared to the 13 cells and 60 bays at Pantex.5 DAF’s production capacity can therefore be estimated at 10-20 percent of Pantex’s, or roughly 500 warheads per year. (Such a massive warhead production effort at DAF would likely require additional support infrastructure and personnel, however.) The total U.S. warhead assembly capacity can thus be estimated to be 4,000 warheads per year.
Estimating Russia’s warhead assembly capacity is considerably more difficult because open official information does not exist. U.S. practices and facilities are probably a poor guide in analyzing the Russian complex as Russia’s stockpile management practices (and its technology for designing and manufacturing warheads) appear quite different from those in the United States. Limited (though highly uncertain) information on the size of the Russian stockpile in the 1980s, shelf life of Russian warheads, and stockpile management practices is, however, available:
- It is believed that in the mid-1980s Russia had an operationally deployed stockpile of some 30,000-35,000 warheads.6
- Russian warheads are reported to have a shelf life of approximately 10 years (with newer warheads having a life of 15 years), presumably because the warheads’ conventional high explosives degrade and their fissile components deteriorate.7
- The deployment cycle for Russian warheads is reported to be three years long.8 After three years of deployment, warheads are removed from their delivery systems and shipped to a serial production facility for modernization and refurbishment. Refurbished warheads are placed in storage prior to a new cycle of operational deployment.
This information makes rough estimates of Russia’s warhead production capacity possible. Assuming that the warheads are refurbished and returned to their delivery systems one year after removal, one-third of the operational stockpile would have to be replaced every year if the stockpile was to remain constant in size. Such an equilibrium stockpile could be imagined to consist of four equal parts. At any given moment, three parts would constitute the operational stockpile and one part would be cycling through the warhead production and storage complex. This suggests the Russian actual stockpile could be more than 30 percent larger than its operational stockpile. This also means that the Russian warhead production complex maintains a significant level of warhead refurbishment activities.
The following equilibrium model of the Russian warhead stockpile and complex could then be considered for the mid-to late-1980s. The operationally deployed stockpile consisted of 30,000 warheads, and the size of the actual stockpile was one-third larger, or 40,000 warheads. An estimated 10,000 warheads were removed from deployment and sent to the warhead production complex annually. Assuming a warhead life of 10 years, 4,000 warheads (out of 10,000) were retired and fully dismantled; and 4,000 new warheads were produced to keep the stockpile at 40,000 warheads. The rest (6,000 out of 10,000 warheads) were going through scheduled maintenance and modernization. The warhead assembly complex thus was performing an estimated 14,000 warhead operations per year, of which 8,000 were full warhead assemblies or disassemblies.
It is likely, however, that the Russian stockpile has never been in equilibrium. It was expanding rapidly until the mid-1980s and began to decline shortly after reaching its peak. This suggests that the complex was producing more warheads than it was dismantling before the mid-1980s. In subsequent years, stockpile reductions have been implemented by keeping newer warheads and retiring older ones without replacement. The production capacity of the Russian warhead assembly complex therefore could be considerably less than that estimated in the equilibrium model.
The Russian warhead production infrastructure has undergone considerable reductions after the Cold War. Russia had four so-called serial warhead assembly/disassembly facilities in the closed cities of Sarov, Zarechny, Trekhgorny, and Lesnoy. An additional, albeit relatively minor, warhead production capability is associated with the two pilot production plants at the warhead design institutes in Snezhinsk and Sarov. The facilities at Sarov and Zarechny have ceased warhead assembly work and are to stop warhead dismantlement by 2003. The nuclear weapons workforce at the four assembly facilities is projected to decline to 8,000 workers by 2005 from approximately 30,000 workers employed during the Cold War.9
Therefore, a worst-case analysis (from the U.S. point of view) is that Russia could make 7,000 new warheads per year in a surge production, approximately half of the Cold War peak capacity estimated above, because of the projected closure of two out of four serial warhead assembly facilities. In reality, however, Russia’s manufacturing capacity is likely to be considerably lower because funding limitations will allow the nuclear complex to assemble enough warheads only to support the active stockpile. Excess workers will be let go and unneeded equipment allowed to become inoperative, reducing surge capacity well below 7,000 warheads per year.
Availability of Plutonium Pits
The United States and Russia are in dramatically different situations concerning the production and storage of plutonium pits. The United States currently does not have an industrial facility to produce pits, but its pits are durable and it has a large number in storage. Russia has a substantial manufacturing capacity but must remanufacture its pits on an ongoing basis because they have a limited shelf life. For both countries, the availability of pits could be the primary limitation on the scale of a hypothetical surge warhead production effort.
It is believed that the shelf life of plutonium pit assemblies in Russia is limited to about 10-15 years because of the way the pits and warheads are designed and manufactured. In particular, because the fissile components are not completely isolated from the surrounding environment as they are being made (perhaps due to weld defects in pit casings), they are subject to corrosion and swelling. Russia therefore needs to remanufacture plutonium pits continuously.
Russia has two facilities that can manufacture fissile components: the chemical and metallurgical plants at Ozersk and Seversk. Assuming a stockpile of 40,000 warheads and a pit life of 10 years, as discussed in the previous section, the two facilities used to produce 4,000 new pits per year. The pit manufacturing facility in Seversk reportedly is no longer involved in defense work, and the Russian government has announced plans to consolidate all defense-related fissile material processing operations to Ozersk.10 If the two facilities have comparable capabilities, the remaining pit manufacturing capacity at the Ozersk plant will be 2,000 pits per year. It is likely, however, that the pit capacity would be lower and more consistent with stockpile maintenance requirements of approximately several hundred pits per year.
The overall breakout capacity of the Russian complex then would be limited by its pit production capabilities of up to 2,000 pits per year. Because of resource limitations and short pit life, Russia does not have a large stockpile of reserve pits outside of nuclear weapons that could be used in massive surge production of new warheads. Indeed, by 2015 all pits existing today would be well beyond their design life and most would likely have been destroyed.
The United States
The United States currently does not have a pit manufacturing facility—the Rocky Flats Plant near Denver stopped producing new pits in 1989 and has since been fully dismantled. A new, fairly small capability to produce 50 pits per year (80 in a sprint mode) is being established at the Los Alamos National Laboratory, but clearly that facility would not be able to contribute greatly to a U.S. surge production effort.
U.S. pits are considerably more durable than their Russian counterparts, lasting an estimated 40-60 years. But because a significant fraction of pits in the active stockpile warheads were built in the late-1970s to mid-1980s, many U.S. pits could become unusable and would have to be remanufactured starting around 2020. For a stockpile of 10,000 warheads (including operationally deployed, responsive, and reserve),11 pit retirement and replacement would likely occur over a period of 20 or so years at an average rate of 500 pits per year.
The Los Alamos facility will not be able to support such a significant pit replacement process, and the Department of Energy is currently in the process of designing a plant with a manufacturing capacity of 500 pits per year. But the facility, which may be built at the Savannah River Site near Aiken, South Carolina, will not be ready for more than a decade. General John Gordon, formerly the head of the National Nuclear Security Administration, said, “We need to begin thinking seriously about a modern pit production facility” but added that he does not “foresee a need for such a facility for at least 15 years.”12
However, despite its inability to produce large numbers of new pits, the United States’ breakout potential remains quite large because it maintains a stockpile of approximately 5,000 stored reserve pits. It could therefore support a surge warhead production effort for at least a year.
Despite Russia’s larger nuclear weapons infrastructure, the data presented show that its breakout capability is not greater than that of the United States. Russia’s warhead assembly capacity is considerably greater than that of the United States, but this is not particularly important because it can only produce an estimated 2,000 pits per year at Ozersk, provided the Seversk chemical and metallurgical plant is fully shutdown and dismantled. By contrast, the United States can assemble 4,000 warheads in one year, although such production would have to rely on stored pits. Furthermore, the United States plans to maintain a large number of stored strategic warheads in the inactive stockpile and responsive force (perhaps totaling more than 5,000 in 2012) that could be returned to the operational stockpile.13
To illustrate the strategic significance of these stockpile reconstitution capabilities, imagine that in 2015 the United States and Russia have fully implemented the Strategic Offensive Reductions Treaty. The United States, as is expected, deploys 2,200 warheads allowed by the treaty’s upper limit. In addition, the United States maintains 2,400 warheads in the responsive force, as Pentagon officials have indicated it will; 3,000 warheads in the inactive stockpile; 5,000 reserve pits; and a large number of delivery vehicles. It also builds an industrial pit manufacturing facility to replace aging pits. In contrast, Russia deploys only around 1,700 warheads as is currently expected. Its fleet of strategic delivery systems has declined because of block obsolescence of missiles and submarines and is not able to accommodate reserve warheads. Warheads from the retired delivery systems have been deactivated and, because of their short life, dismantled.
If, in this scenario, Russia wanted to break out, it could produce 2,000 warheads in one year, but unless it made a significant effort to build new strategic missiles and bombers, which in itself would be an advance indicator of its intentions, its delivery systems would not be able to accommodate more than 500 new warheads. The introduction of these additional warheads would probably require conversion of the currently single-warhead SS-27 strategic missiles to a three-warhead configuration.14 (Russia would presumably be less limited by the availability of delivery systems for tactical nuclear weapons.)
The United States, however, could produce 4,000 new warheads. In fact, the United States would not even need to produce new warheads because it could simply supplement its operationally deployed warheads with the thousands of warheads it is keeping in the responsive force and the inactive stockpile. It would also be able to deploy most of these warheads quickly because U.S. delivery vehicles are newer and the United States can afford to maintain them in large numbers. The U.S. stockpile reconstitution capabilities in this scenario are consistent with the recent nuclear posture review, and they are clearly of concern to the Russians.
However, the United States could be threatened by a Russian breakout capability if both sides were to reduce their stockpiles to 500 operationally deployed warheads, the United States eliminated all stored pits and inactive warheads, and the United States did not build a new industrial pit production facility. In that situation, Russia could increase its stockpile to 2,500 warheads in one year, while the United States could increase the number of warheads it deployed only slightly. (Again, Russia’s ability to deploy new strategic warheads would be severely limited by the availability of delivery systems.) Indeed, the U.S. ability to expand its stockpile would probably be limited to a few hundred warheads, including spares and warheads based on new pits from Los Alamos. This scenario would be unacceptable to the United States, but it could be easily prevented if the United States were to retain a stockpile of 2,000 reserve pits.
This analysis assumes that neither country will be involved in multiyear, secret preparations for a breakout and that a strategic balance depends on the relative sizes of stockpiles one year after the beginning of breakout production. A considerably more conservative methodology from a U.S. security standpoint would be to consider a multiyear surge production effort in Russia until the time (for example, five years) when the United States is capable of constructing new or expanding existing production facilities. Such sustained surge production, however, would probably not be possible for economic reasons in the case of Russia and because of limitations imposed by the availability of non-nuclear components and/or tritium.
The analysis of surge warhead production capacities in the United States and Russia and the above scenarios suggests the following:
- The risk of a breakout posed by the Russian warhead production complex will remain minimal unless the United States reduces its stockpile to a few hundred warheads.
- Limits on the number of strategic delivery systems remain important for limiting stockpile reconstitution capabilities in the case of Russia.
- The proposed 500 pits/year manufacturing facility in the United States would be of limited value in the context of surge production unless its capacity could be expanded rapidly by a factor of four. The primary utility of such a facility would simply be to rebuild aging pits in stockpiled warheads and in the reserve pit stockpile (if the Los Alamos plutonium facility is not capable of meeting these requirements).
- The United States could eliminate its stockpiles of stored warheads but keep a stockpile of approximately 2,000 reserve pits. Maintaining stored pits (as opposed to warheads) would be less threatening to the Russians and would facilitate deep arms reductions. At the same time, a pit stockpile would minimize the breakout risk posed by the Russian production infrastructure. The shutdown and dismantlement of the pit production line in Seversk is critical to limiting Russia’s surge warhead production capacity.
- A more effective approach to addressing the breakout production concern would be to further downsize and reconfigure the U.S. and Russian nuclear warhead production complexes. In particular, all fissile component manufacturing and warhead assembly operations could be consolidated to pilot-scale facilities associated with nuclear weapons research and development centers. In Russia, they could be transferred to the pilot plants of the Sarov and Snezhinsk warhead design centers. In the United States, pit manufacturing and warhead assembly could be conducted at the Los Alamos plutonium facility and DAF, respectively. The existing industrial facilities (including Ozersk, Seversk, Sarov, Zarechny, Trekhgorny, and Lesnoy in Russia; and Pantex in the United States) then would be shut down and irreversibly dismantled or converted.
- Concerns about production asymmetries could also be reduced by cooperative transparency measures. Initially, such transparency measures could include warhead stockpiles and manufacturing declarations, and monitoring of the production facilities that no longer manufacture new warheads. Eventually, transparency arrangements could be implemented at the remaining active warhead production facilities as well.
U.S. concerns about the breakout potential of the Russian warhead production infrastructure are not justified at present and threaten to hinder further easing of the Cold War nuclear threat. In the future, as the two countries continue to slash their nuclear stockpiles, the United States could hedge against the risk of a Russian breakout by maintaining a modest stockpile of nuclear components. In parallel, the United States and Russia should work to eliminate any breakout potential by drastically consolidating their respective nuclear weapons production infrastructures and increasing transparency of nuclear operations.
1. Senate Armed Services Committee, Hearing on the Results of the Nuclear Posture Review, February 14, 2002.
2. Department of Energy’s Office of Defense Programs, FY 2000: Stockpile Stewardship Plan, March 1999, (sanitized version).
3. FY 2000: Stockpile Stewardship Plan.
4. In some cases, new production capacity could be higher than that of dismantlement. According to Yuri Zavalishin, former director of the Avangard warhead assembly plant in Russia, “If we compare assembly and disassembly of nuclear warheads, the latter is more difficult…. During assembly operations, every part and every subassembly, everything is visible. The technology is highly optimized…. The reverse process is a different story. There is a certain unpredictability because a lot depends on the condition of what is inside. Both explosives and uranium change during the years in storage.” Yuri Zavalishin, Atomic “Avangard,” (Saransk: Krasny Oktyabr’, 1999), p. 272-273.
5. George West, United States Nuclear Warhead Assembly Facilities (1945-1990), (Amarillo, TX: Pantex 1991). Gravel Gertie assembly cells, constructed at every U.S. warhead facility handling high explosives and fissile materials together, are designed to vent explosion gases and filter radioactive dust after a high explosive explosion. A typical cell features a reinforced concrete tube with a labyrinth entrance and has a screen-filter roof covered with a thick layer of gravel. Mechanical assembly bays are blast-proof rectangular rooms, which are used for final mechanical assembly and disassembly of intact warheads and nuclear explosive packages containing insensitive high explosives.
6. U.S. and Russian reports on the subject are discussed in Joshua Handler, Russian Nuclear Warhead Dismantlement Rates and Storage Site Capacity: Implications for the Implementation of START II and De-alerting Initiatives, Princeton University, CEES report, 1999.
7. According to General Evgeni Maslin, former head of the 12th Main Directorate of the Russian Ministry of Defense, “[A]pproximately after ten years in storage, it [high explosive] starts to crack and change its chemical and physical properties…” O. Falichev, “Who Keeps the Keys from the Nuclear Arsenal,” Krasnaya Zvezda, December 26, 1993. The problem of corrosion and swelling of nuclear components were mentioned in Stenographic Records of the Parliamentary Hearings “Safety and Security Problems at Radiation-Hazardous Facilities,” November 25, 1996, Moscow (see Yaderny Control, October-November 1997, p. 7-11.)
8. For a brief discussion of the described warhead management system, see K. Belyaninov, “Cheap Asymmetric Response,” Novyey Izvestia, April 7, 2001.
9. Oleg Bukharin, Frank von Hippel, and Sharon Weiner, Conversion and Job Creation in Russia’s Closed Nuclear Cities (Princeton University, November 2000).
10. Lev Ryabev, presentation at the Princeton Conference, March 13-16, 2000.
11. This estimate is consistent with the 2002 nuclear posture review and includes operationally deployed, responsive, and inactive warheads. See Faking Nuclear Restraint: The Bush Administration’s Secret Plan for Strengthening U.S. Nuclear Forces, (Washington, D.C.: Natural Resources Defense Council, February 2002)
12. Senate Armed Services Committee Hearing on the Results of the Nuclear Posture Review, February 14, 2002.
13. See Faking Nuclear Restraint.
14. See Oleg Bukharin and James Doyle, “Transparency and Predictability Measures for U.S. and Russian Strategic Arms Reductions” (Los Alamos National Laboratory, report LA-UR-01-5001, October 2001).