From Russia With Love:

What We Learned from the

Russian Accident History

Jim Baker

August 21, 2025

Page 1 of 36

Objectives:

• Compare/contrast what we “know” about criticality accidents by adding the Russian experience to ours.
• Three editions of the Accident Report:
• LA-3611, Review of Criticality Accidents, 1967 by W.R. Stratton
• DOE/NCT-04, A Review of Criticality Accidents, 1989, by D. R. Smith
• LA-13638, A Review of Criticality Accidents, 2000 Revision, by T.P. McLaughlin, et al
• Highlight key lessons learned and one case study

​

​

​

​

​

​

Page 2 of 36

From the 1989 Revision of the Accident Report:

• Accident features and causes:
• Unfavorable geometry vessels
• All occurred in solution
• Workers not following procedures
• Activity or procedures not reviewed by NCS
• Non-routine operations
• Poor communications, including material labeling/posting

Page 3 of 36

Parallels in History: The process accidents

​

• ~1 to 2 per year, mid 50’s to mid 60’s
• Height of the Cold War production ramp up
• ~1 per decade thereafter

​

​

Page 4 of 36

Deliberate Parallels in Facilities

• Mayak = Hanford
• Tomsk = Savannah River Site / Y-12
• Arzamas = Los Alamos
• Semipalatinsk = Nevada Test Site
• Etc.

​

• But: Russia did not separate military and civilian fuel processing and power reactor facilities.

Page 5 of 36

Page 6 of 36

A Historic Opportunity

• In mid 1980’s “détente” turned to “perestroika” and “glasnost” (openness).
• By mid 1990’s, FSU collapsed, and Russians announced that they had 13 process criticality accidents.
• LANL led an effort to reach out to Russian scientists to include their information and publish a revision to the old accident report (Stratton/Smith).
• One of a number of cooperative projects done at that time
• In late 1990’s a number of visits and information exchanges occurred to facilitate documenting the information
• Primary authors: T. McLaughlin, N. Pruvost, S. Monahan (US),V. Frolov, B. Ryazonov, V.Sviridov (Russian Federation)

Page 7 of 36

Challenges for the 2000 Revision to the Accident Report

• Intense work, challenging travel conditions
• Accurate translation
• Lack of good documentation of the events at the time they occurred (sometimes classified reports)
• Not inventing new facts
• Culling the available info to the essentials
• Getting the Russian version published (w/ errata)

​

Page 8 of 36

13 Russian Process Criticality Accidents

• Mayak Production Association (comparable to Hanford)
• 7 Accidents, 5 Fatalities, 2 Amputations
• Tomsk (comparable to Savannah River/Y-12)
• 4 Accidents, One Amputation
• Elektrostal
• 1 Accident, Insignificant Exposures
• Novosibirsk
• 1 Accident, Insignificant Exposures

​

Page 9 of 36

Special Features

• Multiple terminated by active intervention
• Largest estimated first spike yield (~2x10¹⁷ fissions): Mayak PA, 2 January 1958
• Very small excursions (~ 10¹⁴ fissions) and multiple unit accident: Novosibirsk, 15 May 1997
• Metal accident: Tomsk, 13 December 1978
• Two low enrichment slurry accidents
• Tomsk, 14 July 1961, U(22.6) Oxide–Oil
• Electrostal, 3 November 1965, U(6.5) Oxide–Water

​

Page 10 of 36

Notable Characteristics

• Several apparently had no first spike (a couple are unknown)
• Some had very low yield (< 10¹⁶ fissions)
• Problematic for CAAS detection and initiating timely emergency response
• Small enough mass/system that the accuracy of the data was questioned

​

Page 11 of 36

Page 12 of 36

Page 13 of 36

Shielding made the difference

Page 14 of 36

20-cm thick

Mayak Production Association, 16 December 1965

• Residue recovery process area
• Residues generated from dissolution/precipitation/reduction processes
• Difficult to recover (dissolution)
• Dissolution vessel
• Uranyl Nitrate solution
• Facility was equipped with a criticality accident alarm system

​

Page 15 of 36

Page 16 of 36

The Setting

• Residue dissolution glovebox
• Three identical sets of process equipment, operated by vacuum transfers
• Parallel operation
• Cylindrical dissolution vessel
• Holding vessel, Filter vessel & Filtrate receiving vessel
• Dissolution vessels
• 45 cm in diameter, 100 liters in volume
• Pulsating mixer
• Feed hopper with sealable flat cover plate
• Pressure relief valve
• 2.5 cm, steam water jacket for heating

​

Page 17 of 36

Dissolution & Filtration Process

• Residue introduced via feed hopper
• Acid and heat were added to the system
• Normal process
• 100 ºC
• 1.5 hours
• constant mixing
• Transferred to hold vessel
• Filtered

Page 18 of 36

Residue Dissolution Glovebox

Page 19 of 36

15 December 1965

• In a related operation
• A supervisor ordered burning of batch #1726
• #1726 had >1% by weight U
• Violation of criticality safety limit for the furnace
• After burning batch #1726 was
• Sampled and sent to a dissolution feed staging glovebox with other batches
• Before receiving sample results

​

Page 20 of 36

The Lead-Up Events

• An operator, preparing material for dissolution discovered that batch #1726 had no analysis results
• Phoned the analytical laboratory
• Informed batch #1726 was 0.32% by weight uranium
• In reality:
• The operator had been given the analytical results of batch #1826
• #1726 was 44% by weight uranium, 138 times greater

​

Page 21 of 36

The Lead-Up Events on 16 December 1965

• 5 kg of #1726 were loaded into dissolution vessel #1
• 2.2kg of U(90)
• 1980 g of ²³⁵U
• Criticality safety limit was 300 grams
• Unknowingly the limit was violated by > 6
• After 40 minutes the process was shutdown
• Accommodate cleaning for shift change

Page 22 of 36

The Accident

• 10 minutes after shutting the system down the nearest criticality alarm sounded momentarily
• The operator reported to the control room per procedure
• While at the control room the alarm again sounded
• Shortly thereafter ~22:10 additional more distant alarms began sounding
• The building was evacuated (tunnel muster point)

​

Page 23 of 36

The Response

• Dissolution vessel #1 or its holding vessel was determined to be the accident location based on
• Personnel interviews
• Accountability records
• System schematics
• Surveys indicated
• ~8 R/h @ 2m from the glovebox during minima

​

Page 24 of 36

The Response (cont’d)

• Accident monitored from a nearby building (~50m)
• Between ~22:10 and 23:00 four additional power peaks were observed
• The accident was clearly on-going and cyclical
• 23:00 emergency response personnel arrived
• Criticality safety, facility management, health physics
• Determined the control room could be safely re-entered and moved their recovery operations

​

Page 25 of 36

The Response (cont’d)

• After the 9th excursion, Cd poisoned solution was remotely pumped to the holding vessel
• Arguments between the criticality safety representative and process supervision developed
• Process supervision wanted to re-occupy the area and begin recovery/clean-up
• After ~20 minutes, while the argument was still on-going, a 10th excursion occurred
• Firmly established that the accident site as dissolution vessel

Page 26 of 36

The Response (cont’d)

• Draining or adding Cd solution to the dissolution vessel was considered too dangerous
• Multiple operators
• Time intensive valve manipulations
• The recovery team decided to send personnel in separately to:
• Remove the gloves from two glovebox ports
• Unlock and open the feed hopper lid
• CAREFULLY insert a wadded-up piece of Cd foil into the nitric acid solution

​

​

Page 27 of 36

The Response (cont’d)

• After preparation
• Tasks 1 and 2 were accomplished separately
• Two specially chosen operators
• ~30 and ~60 seconds, respectively
• Recovery was then halted as radiation reading indicated that an 11th excursion was underway

​

Page 28 of 36

The Response (cont’d)

• After this excursion and appropriate waiting period
• A senior engineer (NCS specialist) entered the area
• Placed the foil on top of the solution through the hopper
• The foil dissolved and the excursions were terminated

Page 29 of 36

Consequences

• 5.5 x 10¹⁷ fissions total, 26 liters, 69 g²³⁵U/l
• Exposures
• Recovery team members limited to 0.3 rem
• 3 between 0.2 and 0.27 rem
• 7 between 0.1 and 0.2 rem
• 17 @ 0.1 rem or less
• Operations resumed the next day
• 94% of process equipment was replaced with favorable geometry over the next 2 to 3 years

​

Page 30 of 36

Causality: The New View

• Prior to 1990, accident discussion often focused on “the cause”.
• The 2000 report emphasizes “no single cause”. Each accident resulted from a series of conditions or events that, in combination, set the stage.
• Consistent with Sidney Dekker’s book:
• Excellent resource to understand and address failures in complex systems

​

Page 31 of 36

Notable Differences in the 2000 Revision

• Importance of sharing the information
• Less emphasis on non-routine operations
• Highlights lack of familiarity with the operations
• Integrating NCS with material accountability
• Recent emphasis on holdup monitoring
• NCSP training class has a module on NDA
• Operations should know how to respond to malfunctions and mistakes
• Avoid unapproved actions after evacuation
• Awareness of the hazard and stop work authority

​

​

​

Page 32 of 36

Notable Differences in the 2000 Revision

• Making ease of operation a goal
• Support for self-reporting of upsets (more learning, less punishing)
• Key conclusion: All accidents have been dominated by design, managerial, and operational failures. The focus for prevention should be on these issues.

Page 33 of 36

Follow-Up Efforts

• In early 2000’s more meetings were held to promote further collaborations between US & Russian Labs
• Technical representatives from Arzamas met with us at LANL
• Held a review of the Zhakarov accident 17 June, 1997

​

Page 34 of 36

Conclusion

• Including the Russian experience makes the 2000 Report a much more valuable resource.
• Particularly the section on Observations & Lessons Learned
• Further illumination of “the human element”
• Collective experience is key to doing a thorough hazard analysis (or safety case) for new or modified processes

​

Page 35 of 36

Q&A Time

36