Positive SCRAM in RBMK Reactors and the Reliability of a Safety Function
“You are driving a car and press the brake. Instead of slowing down, the car accelerates.”
Anatoly S. Dyatlov, Chernobyl: How It Was, Chapter 8. Excerpt translated from the Russian original.
Anatoly Dyatlov used this analogy to describe a problem that challenged the very expectation of what reactor protection should do. When an operator initiates a shutdown, they expect the system to suppress the chain reaction. In the original RBMK reactor design, however, the initial insertion of safety rods could, under certain conditions, produce the opposite effect locally.
Dyatlov was a participant in the events and defended his own account of the accident in his memoirs. The car analogy forms part of his argument about responsibility. Its technical premise therefore requires independent verification. That verification is provided by INSAG-7, published by the International Atomic Energy Agency in 1992, which discusses the positive scram effect as one of the significant design deficiencies.
To understand this phenomenon, we need to examine the entire rod assembly and its movement through the core. What mattered was which material displaced the water at a given moment, and where in the core this occurred.
What shutdown must achieve
SCRAM is a term for the rapid shutdown of a reactor. In an RBMK, this was intended to be achieved by inserting neutron absorbers into the active core, the fuelled region in which the chain reaction takes place.
An absorber captures some of the neutrons that could otherwise cause further fissions. It thereby inserts negative reactivity and brings the reactor to a subcritical state. Fission power begins to decrease, but residual heat must still be removed.
Reactivity describes the departure from the critical state, in which the chain reaction is self-sustaining. For the purposes of this article, a basic distinction is sufficient: a positive contribution to reactivity promotes growth in the neutron population, while a negative contribution suppresses it. The protective action must be sufficiently strong and sufficiently rapid to shut the reactor down safely, including under the conditions that triggered the protection system.
The rod was more than an absorber
The RBMK is a graphite-moderated, water-cooled reactor. Graphite slows the neutrons, while water removes heat as it flows through the fuel channels. The control and safety rods move in their own channels, separate from the fuel channels.
In the original design of the top-entry rods considered here, a graphite displacer was attached below the absorber section by a connecting element. With the absorber withdrawn, the displacer occupied a substantial part of the channel within the active core. Because graphite absorbs fewer neutrons than water, this arrangement reduced absorption in the withdrawn position and increased the difference in reactivity worth between the withdrawn and inserted positions.
However, the displacer did not extend over the full height of the active core. The active core was approximately seven metres high, whereas the graphite displacer was approximately 4.5 metres long. With the absorber fully withdrawn, the displacer was positioned approximately centrally within the active core, leaving a water-filled section of channel approximately 1.25 metres long both above and below it.
The term “graphite tip”, often used in popular explanations, therefore does little to convey the actual geometry. The assembly comprised a displacer several metres long, connected to the absorber, with two water-filled sections remaining as a consequence of its geometry.
One movement, opposing local effects
During the initial stage of insertion, rod movement produced opposing local reactivity contributions in different parts of the core.
The absorber began to enter the upper part of the active core. There, it increased neutron absorption and contributed negative reactivity. In the lower part, however, the graphite displacer replaced the water beneath it. Because graphite absorbed fewer neutrons than the water, this substitution made a local positive contribution to reactivity.
The absorber did not remain outside the core until the graphite had completed its movement. Both effects developed during the same movement. This is precisely why the explanation requires a section through the full height of the active core.
Source: explanatory schematic based on IAEA INSAG-7 (1992), Fig. 1(a) and Fig. II-10.
The net effect depended on rod positions, the operating state and the spatial distribution of the neutron field. A reduction in power in the upper part did not, by itself, ensure that power in the lower part could not rise dangerously. In the large RBMK core, the upper and lower regions could respond differently.
Positive SCRAM therefore does not mean that every press of AZ-5 always increased power. It refers to a positive contribution during initial rod insertion, made possible by the design, which could compromise the shutdown function itself under adverse conditions. The problem involved both the insertion speed and the spatial distribution of the effects of insertion.
The effects of steam formation and rod movement
The positive effect of rod insertion must be distinguished from a positive void coefficient of reactivity.
In the first case, the change was caused by rod movement: graphite replaced water in the lower section of the rod channel. This effect did not require the water to boil.
The second concerns a change in the steam void fraction of the coolant in the fuel channels. Water also absorbs neutrons. As steam replaced some of the liquid water, neutron absorption by the water decreased, while the graphite moderator remained in the core. Under the conditions prevailing at the time, an increase in the steam void fraction could increase reactivity.
Source: explanatory schematic based on IAEA INSAG-7 (1992), Sections 2.1–2.2 and Fig. II-10.
This made a positive feedback loop possible. An increase in power intensified boiling; more steam contributed additional positive reactivity, which in turn promoted a further increase in power. At the time of the accident, the overall power coefficient of reactivity, which describes the combined reactivity response to a change in power, was also positive.
The local effect of the graphite displacers and the feedback effect of steam formation were different mechanisms, but they could reinforce each other in an adverse core state. Understanding the event therefore requires the rod geometry, neutron field and coolant conditions to be considered together.
The moment protection was initiated
The chronology in Annex I of INSAG-7 records the pressing of the EPS-5 button, also known as AZ-5, at approximately 01:23:40 on 26 April 1986. The rods began to move into the core. At 01:23:43, protection signals were recorded for the rate of power increase and for an excessive power level; the reported power exceeded 530 MW thermal.
This figure of 530 MW is not an estimate of the maximum power reached during the accident. It is a value reported in the sequence of recorded signals. Nor can it be used to reconstruct the precise power history in every part of the core.
Before protection was initiated, the reactor had a low operating reactivity margin, many rods were substantially withdrawn, and the power distribution was spatially non-uniform. The operating reactivity margin, or ORM, expressed the reactivity worth of the inserted rods in units of equivalent rods. It must therefore not be understood simply as a count of the rods in the core.
INSAG-7 considers insertion of the safety rods to have been the most likely final event that escalated the already adverse conditions to the point of reactor destruction. At the same time, it explicitly retains uncertainty about a definitive reconstruction of the final moments. The computational models presented in the annexes did not reproduce the recorded sequence in every detail either.
A technically sound explanation must preserve both findings: the positive effect of rod insertion was a real design deficiency, while establishing the precise accident sequence requires more than an explanation of a single mechanism.
The phenomenon had been identified before the accident
INSAG-7 states that the positive effect was identified at Ignalina in 1983. Annex I describes its experimental identification during the reactor physics commissioning tests at Ignalina Unit 1 and Chernobyl Unit 4 in November and December of that year.
The chief design organisation announced its intention to restrict full rod withdrawal and modify the rod design. However, the planned design changes had not been implemented before the accident, and the recommended procedural measures had not been incorporated into the operating instructions. INSAG concluded that there had apparently been a widespread belief that the conditions under which the effect would become significant would not occur in practice.
This is where the technical deficiency connects with organisational action. A finding documented in a report or communicated between institutions does not, by itself, change how a plant behaves. Its consequences must be assessed, responsibility for action assigned, and implementation of the action verified.
For the operator, it is particularly important that safety-significant limitations are clearly incorporated into instructions, training and the information available during operation. If knowledge of a hazardous system characteristic is lost in that transfer, a gap remains between what the designer knows about the plant and what the person in the control room can rely on.
The geometry of the protection system was also changed
The measures introduced after the accident directly addressed the design. Annex II of INSAG-7 describes improved rods that no longer left the problematic water column at the bottom of the channel when withdrawn, and that had a larger absorber section. This eliminated the water-displacement mechanism described above, which had produced the initial positive contribution.
Source: explanatory schematic based on IAEA INSAG-7 (1992), Fig. II-13(a1, c1).
Modified drives reduced the time required for full rod insertion from 18 to 12 seconds. A separate fast-acting emergency protection system was also developed. During tests in 1987 and 1988, its 24 fast-acting rods inserted a substantial amount of negative reactivity in less than 2.5 seconds. This figure describes the effect of the fast-acting system and must not be confused with the full insertion time of the conventional rods.
At the same time, additional absorbers, a higher required operating reactivity margin and changes to the fuel reduced the positive reactivity effect of steam void formation. The ORM indication available to the operator was improved, and the operating documentation was updated.
These measures addressed different aspects of the same problem: the hazardous initial effect of the rods, an excessively slow protection response, adverse reactivity feedback and inadequate information for operation. Eliminating the positive SCRAM effect described here does not, in itself, constitute a comprehensive safety assessment of an individual reactor.
A safety function must also work under adverse conditions
An explanation of Chernobyl should examine the decisions made by the operating personnel, the adequacy of the procedures and the state of the reactor. What the protection system physically did when initiated requires equally careful examination.
Positive SCRAM shows why the presence of an absorber and a command to insert it are not, by themselves, sufficient to assess a safety function. The geometry of the entire assembly, its initial position, its speed of movement and the spatial response of the core all matter. Protection must also be verified under foreseeable adverse conditions, when its operation is needed most.
This also marks the distinction between procedure and design. Procedures specify how the plant is to be operated and within which limits. The design must ensure that the intended protective action achieves the required safety effect in time, under the conditions for which it is intended. A known hazardous characteristic of the protection system requires corrective action whose implementation can be verified; confidence that adverse circumstances will not coincide does not change the physics of the system.
Dyatlov’s analogy does not settle the overall question of responsibility for the accident. It does, however, express very clearly an expectation that the designer must verify and on which the operator must be able to rely.
The brake must brake.
Note on Illustrations
Cover photograph: Detail of the upper surface of an RBMK reactor at the Chernobyl Nuclear Power Plant. Photograph: rootstocks / iStock.
The diagrams “One movement, opposing local effects”, “Two mechanisms, different triggers” and “Removing the lower water column” were created by Elite Studio 3D / By the Protocol specifically for this article, based on technical descriptions and figures in IAEA INSAG-7 (1992).
The first diagram shows the opposing local reactivity contributions during initial insertion of an original-design RBMK control rod. The second distinguishes the effect of graphite displacing water in a control-rod channel from the effect of increasing steam fraction in a fuel channel when the void coefficient is positive. The third compares the withdrawn positions of the original and retrofitted manual control rods, focusing on the elimination of the lower water column within the active core.
Geometry, connecting hardware and coolant representation are simplified. Reactivity signs indicate qualitative local contributions, not calculated magnitudes or the net reactor response. The diagrams do not reconstruct the precise accident sequence and do not constitute design or operating documentation for any particular reactor.
Sources and verification references
Anatoly S. Dyatlov: Чернобыль. Как это было (Chernobyl: How It Was). Chapter 8, Люди (People), section on A. P. Alexandrov. The opening quotation is a direct English translation of two sentences from the Russian text. The analogy forms part of the author’s argument; it is used as the starting point for this article, rather than as an independent technical finding.
International Nuclear Safety Advisory Group: The Chernobyl Accident: Updating of INSAG-1. Safety Series No. 75-INSAG-7, IAEA, Vienna, 1992. The principal source for verification of the physics, historical sequence and corrective measures. The page numbers below refer to the report’s printed pagination, not the sequential PDF page numbers.
Rod geometry and insertion speed: Sections 2.2 and 2.3, p. 4, and Figure 1, p. 5. Positive void and power coefficients: Section 2.1, pp. 3–4. Spatial effects, ORM and prior knowledge of the phenomenon: Sections 4.1 and 4.2, pp. 13–15. Assessment of the most likely final event and the limits of certainty: Section 5.1, pp. 16–17.
Annex I contains the 1991 report of the Soviet supervisory commission. Findings from the 1983 commissioning tests appear on pp. 43–45, and the chronology of signals on pp. 53–55. Annex II contains a separate technical report from 1991. Computational reconstructions and their limitations are discussed in Section II-2.5, pp. 115–123; post-accident measures are described in Section II-3, particularly pp. 124–125. Findings from these reports are attributed to the corresponding annex in the article.
Chernobyl control rods nuclear safety positive SCRAM positive void coefficient RBMK Reactor Physics reactor protection safety culture
Last modified: October 15, 2026
