This group brings together the best thinkers on energy and climate. Join us for smart, insightful posts and conversations about where the energy industry is and where it is going.


Did Graphite in the Chernobyl Reactor Burn?

Charles Barton's picture
Nuclear Green

I am a retired counselor. My father was a nuclear scientist and I have had a life long interest in and fascination with his work.

  • Member since 2018
  • 654 items added with 103,524 views
  • Apr 14, 2011

In two previous posts, ” Does Nuclear Grade Graphite Burn?,” and “Did the Graphite in the Windscale Reactor Burn?” I reviewed a number of reports and other information sources on Nuclear Graphite Flamibility. Although I did not come to a firm conclusion, i did find strong evidence that Nuclear Graphite does not burn under many conditions in which one would expect fire. There is also startling evidence that at least one of the the two reactor fires which are attributed to graphite, the Windscale accident, appears to have not involved a graphite fire. I concluded my Windscale review with the statement,

Given these facts, the assertion that there was a core graphite fire at Chernobyl ought also to be revisited.

This post considers several reports that are relivant to an evaluation of the role of graphite in te Chernobyl fire.

In the wake of the Chernobyl Reactor fire, the United States Department of Energy had a serious concern. The DoE operated a reactor that was similar to the Chernobyl reactor, the N reactor at Hanford, Washington. The N reactor, like the Soviet RBMK-1000, had graphite in its core. The DoE wanted to know if a Chernobyl type accident would be possible at Hanford. The DoE commissioned a review of N Reactor safety in light of the Chernobyl accident. The researchers asked

What is the potential for obtaining conditions conducive to a graphite fire in N Reactor?

And answered,

The graphite stack is protected by a helium cover gas contained within the shield structure. Combustion cannot occur unless the shield structure is sufficiently damaged to leak inert gas faster than available makeup supply. Should that occur, the rate of oxidation would be very slow because graphite temperatures would remain below the threshod for rapid oxidation because of heat removal from the stack by the ECCS [Emergency Core Cooling System] or the GSCS [Graphite and Shield Cooling System], The GSCS alone is capable of removing both decay heat and any heat load from graphite oxidation, stabilizing temperatures in a range which ensures control.

In the Chernobyl accident sequence, the plant was effectively destroyed and conditions for exothermic chemical reactions involving a number of core materials were present before graphite fire made any contribution. It is likely that the major contribution from graphite was to serve as a refractory container for decay heat buildup, zirconium oxidation along with carbothermic reduction of the UO2, and complex gas producing redox reactions. For any N Reactor accident where the GSCS and biological shield are intact, there is no way to achieve ignition of the graphite. It has been demonstrated experimentally that oxidation nuclear grade graphite takes very high temperatures to initiate, and the contribution to total heat load is only a small fraction of the decay heat.

They also reported finding that

Detailed reaction rate models have been developed to analyze graphite oxidation. These models tend to show that graphite oxidation in N Reactor would be limited both by available oxygen and the requirement that a high-temperature source (>1100°C) be available to drive a significant reaction. The analyses have effectively shown that graphite will not con- tribute significant accident heat loads.

Why then did the Chernobyl reactor graphite burn? According to the N Reactor review,

The Chernobyl release must be viewed as resulting from both very high temperatures in the core rubble, extensive mechanical disruption and dispersal of core material and the large draft “chimney effect” that followed the total disruption of that particular reactor configuration. There is no accident sequence that could produce an equivalent disruption of N Reactor; there would be some confinement even in the lowest probability event sequences. Because of the horizontal arrangement of pressure tubes, Chernobyl fission product release rates and magnitude are not pertinent to N Reactor accident scenarios with mechanistic initiators.

In 1987 the NRC did its own safety assessment of the Graphite Reactors it licensed. The NRC report described the limitations of graphite fires,

For reasons that are well understood, graphite is considerably more difficult to burn than is coal, coke, or charcoal. Graphite has a much higher thermal conductivity than have coals, cokes or charcoals, making it easier to dissipate the heat produced by the burning and consequently making it more difficult to keep the graphite hot. Concomitantly, coals, cokes and charcoals develop a porous white ash on the burning surfaces which greatly reduces radiation heat losses while simultaneously allowing air to reach the carbon surfaces and maintain the burning. In addition, coals, cokes and charcoals are heavily loaded with impurities which catalyze the oxidation processes. Nuclear graphite is one of the purest substances produced In massive quantities.

The literature on the oxidation of graphite under a very wide range of conditions is extensive. Effects of temperature, radiation, impurities, porosity, etc., have been studied in great detail for many different types of graphites and carbons [Nightingale, 1962]. This information served as a foundation for the full scale detailed studies on graphite burning accidents In air-cooled reactors initiated and completed at Brookhaven National Laboratory [Schweitzer, 1962a-f]. After British experimenters at Harwell confirmed the results obtained at BNL [Lewis, 1963] there appeared to be no new conclusions from additional work in this field. The aspects of the work pertinent to evaluating the potential for graphite burning accidents are described here In some detail.

Burning, as used here, is defined as self-sustained combustion of graphite. Combustion is defined as rapid oxidation of graphite at high temperatures. Self-sustained combustion produces enough heat to maintain the react- ing species at a fixed temperature or is sufficient to increase the temperature under actual conditions where heat can be lost by conduction, convection, and radiation. In the case where the temperature of the reaction Increases, the temperature will continue to rise until the rate of heat loss Is just equal to the rate of heat production. Sustained combustion is distinguished from self-sustained combustion when, in the first case, the combustion is sustained by a heat source other than the graphite oxygen reactions (e.g., decay heat from reactor fuel).

Early attempts to model the events at Windscale [Robinson, 1961; Nairn, 1961] were followed by the BNL work described here.

Some 50 experiments on graphite burning and oxidation were carried out in 10-foot long graphite channels at temperatures from 600°C to above 800°C. To obtain a lower bound on the minimum temperature at which burning could occur, the experiments were specifically designed to minimize heat losses from radiation, conduction, and convection.

The objectives of the full scale channel experiments were to determine under what conditions burning might initiate in the Brookhaven Graphite Research Reactor (BGRR) and how it could be controlled if it did start. Channels 10-feet long were machined from the standard 4 in. x 4 in, blocks of AGOT graphite used in the original construction. The internal diameter of the BGRR channel was 2.63 Inches. Experiments were also carried out on channel diameters of one to three Inches on 10-foot long test channels In order to obtain generic Information. The full length of the channels was heated by a temperature controlled furnace and was Insulated from conductive heat losses. At intervals along the length there were penetrations in the furnace through which thermocouples used to read the temperature of the graphite and air were introduced, and from which air and air combustion products were sampled. A preheater at the inlet of the graphite channel was used to adjust the air to the desired temperature. The volume of air was controlled and monitored by flow meters to allow flow measurements in both laminar and turbulent flow conditions.

In a typical experimental run the graphite was first heated to a preselected temperature. The external heaters were kept on to minimize heat losses by conduction and radiation. The temperature changes along the graphite channel were then measured for each flow rate as a function of time with the heaters kept on. It was observed that below 675°C it was not possible to obtain temperature rises along the channel if the heat transfer coefficient (h) was greater than 10~ cal/cm-sec-°C. Below 650°C it was not possible to get large temperature rises along the channel with 30°C inlet air temperatures at any flow rate. For h values lower than 10~ cal/cm-sec-”C maximum temperature rises were 0-50″C and remained essentially constant for long periods of time (five hours). For h values greater than 10~ cal/cm-sec-°C the full length of the channel was cooled rapidly.

There were two chemical reactions occurring along channels. At low temperatures the reaction C + O2 to form CO2 predominated. As the temperature Increased along the channel CO formed either directly at the surface of the channel or by the reaction CO2 + C. At temperatures above 700″C, CO reacts in the gaseous phase to form CO2 with accompaniment of a visible flame. It was observed that the unstable conditions which were accompanied by large and rapid Increases in temperature Involved the gas phase reaction CO + O2 and occurred only for h values below 10~ cal/cm-sec-°C below 750″C. Temperature rises associated with the formation of CO2 from C + O2 were smaller than those due to CO + O2 and decreased with time. They too occurred at h values below 10″ cal/cm-sec-°C.

In a channel which was held above 650°C there was an entrance region running some distance down the channel which was always cooled. A position was reached where the heat lost to the flowing gas and the heat lost by radial conduction through the graphite was exactly equal to the heat generated by the oxidation of the graphite and of the CO. This position remained essentially constant with time. Beyond this point rapid oxidation of graphite occurred with the accompaniment of a flame (due to the CO-0 gas phase reaction). Under conditions of burning, the phenomena were essentially Independent of the bulk graphite chemical reactivity. Rate controlling reactions during burning were determined by surface mass transport of reactants and products.

The experiments were used to develop an equation which expressed the length of channel that can be cooled as a function of temperature, flow rate (heat transfer coefficient), diameter and reactivity of the graphite. It was found that the maximum temperature at which thermal equilibrium (between heat generated by graphite oxidation and heat removed by the air stream) will occur in a channel can be predicted from the heat transfer coefficient, the energy of activation and a single value of the graphite reactivity at any temperature. Above this maximum temperature the total length of channel Is unstable and graphite will burn. The studies show that the bounding conditions needed to initiate burning are:
1. Graphite must be heated to at least 650°C.
2. This temperature must be maintained either by the heat of combustion or some outside energy source.
3. There imist be an adequate supply of oxidant (air or oxygen).
4. The gaseous source of oxidant must flow at a rate capable of removing gaseous reaction products without excessive cooling of the graphite surface.
5. In the case of a channel cooled by air these conditions can be met. However, where such a configuration is not built into the structure it is necessary for a geometry to develop to maintain an adequate flow of oxidant and removal of the combustion products from the reacting surface. Otherwise, the reaction ceases.

The report went on to discuss the potential contribution of Wignarian energy to a graphite reactor fire, and found that if a reactor operated at a high enough temperature to preform Wignerian annealing its graphite would not accumulate Wignerian energy. The report also stated that,

The factors needed to determine whether or not graphite can burn in air are the graphite temperature, the air temperature, the air flow rates, and the ratio of heat lost by all possible mechanisms to the heat produced by the burning reactions [Schweitzer, 1962a-f]. In the absence of adequate air flow, graphite will not burn at any temperature. Rapid graphite oxidation in air removes oxygen and produces CO2 and CO which, along with the residual nitrogen, suffocate the reaction causing the graphite to cool through unavoidable heat loss mechanisms. Self-sustained rapid graphite oxidation cannot occur unless a geometry is maintained that allows the gaseous reaction products to be removed from the surface of the graphite and be replaced by fresh reactant. This necessary gas flow of Incoming reactant and outgoing products is Intrinsically associated with a heat transfer mechanism. When the incoming air is lower in temperature than the reacting graphite, the flow rate is a deciding factor in determining whether the graphite cools or continues to heat. Experimental studies on graphite burning have shown that for all the geometries tested which Involved the conditions of small radiation and conduction heat losses, it was not possible to develop self-sustained rapid oxidation for graphite temperatures below about 650*’C when the air temperatures were below the graphite temperature. At both high and low flow rates, the graphite was cooled by heat losses to the gas stream even under conditions where other heat loss mechanisms such as radiation and conduction were negligible.

At temperatures above about 650°C, in realistic geometries where radiation is a major heat loss mechanism, graphite will burn only in a limited range of flow rates of air and only when the air temperatures are high. At low flow rates, inadequate ingress of air restricts burning. At high flow rates, the rate of cooling by the flowing gas can exceed the rate of heat produced by oxidation.

Studies have shown that burning will not occur when there is no mechanism to raise the graphite temperature to about 650°C [Schweitzer, 1962a-f]. If the temperature is raised above 650°C, burning will not occur unless a flow pattern is maintained that provides enough air to sustain combustion but not enough to cause cooling. Since the experiments were designed to minimize all heat losses other than those associated with the air flow, 650°C can be considered a lower bound for burning.

Thus the NRC’s answer to the original question which I asked at the beginning of this series is “yes, graphite does burn” but only under a very limited set of conditions.

The NRC report simply assumed that those conditions had been meet at Windscale and Chernobyl. We now know what the NRC did not know in 1987, that the Windscale fire was not a graphite fire. Neither report reviewed here offers conclusive evidence that the Chernobyl fire was a graphite fire. A major conclusion of the report draws a big question mark over the Chernobyl graphite fire hypothesis,

in order to have self-sustained rapid graphite oxidation in any of these reactors certain necessary conditions of geometry, temperature, oxygen supply, reaction product removal and favorable heat balance must exist.

Yet the Soviets claimed and American nuclear safety experts like H.J.C Kouts accepted the notion that Graphite could burn like charcoal.

The emission of radionuclide continued for about nine days, aided by burning of the graphite. It is estimated that upwards of ten percent of the graphite in the core burned, in a manner similar to the rapid oxidation of charcoal.

We know that Kouts view cannot be correct, nuclear graphite does not burn like charcoal, and the assertion that only 10% of the Chernobyl core graphite burned does not suggest graphite was the major source of the Chernobyl fire. The question is were the conditions conditions that are conducive to a graphite fire present at Chernobyl, and if so how? In answers to these questions, and without other evidence we must consider the claim of a graphite fire at Chernobyl to be unconfirmed.

As we have seen, the use of graphite in a reactor core is consistent with safe reactor operations. The danger of a core fire due to graphite burning is quite limited. The time has now arrived to ask the question, is it dangerous to use graphite in the core of a Molten Salt Reactor.

We have already noted that the possibility of graphite fires in a reactor core can be eliminated by core design. In the case of Molten Salt Reactors, the possibility of a core fire is eliminated by the two modes of MSR operation. A MSR is only active if liquid salt is present in the core of the reactor. But if liquid salt is present then air cannot be. In the case of the presence of molten salt in the core, the presence of salt would prevent air from reaching the graphite. If the salt is drained, either deliberately, by accident or by operation of the freeze valve safety system, then the heat producing fission products will be drained from the core as well. The absence of fission products in the core would mean that a high enough temperature required to trigger a graphite fire would not be possible. Thus the use of graphite in a Molten Salt Reactor core would be inherently safe.

Willem Jan Oosterkamp's picture
Willem Jan Oosterkamp on Apr 13, 2012

The conditions during the Chernobyl accident were exceptional. Fuel temperature must have reached 3000 degrees centigrade. Graphite in contact with fuel will certainly be heated up sufficiently to burn when in contact with air

Charles Barton's picture
Thank Charles for the Post!
Energy Central contributors share their experience and insights for the benefit of other Members (like you). Please show them your appreciation by leaving a comment, 'liking' this post, or following this Member.
More posts from this member

Get Published - Build a Following

The Energy Central Power Industry Network is based on one core idea - power industry professionals helping each other and advancing the industry by sharing and learning from each other.

If you have an experience or insight to share or have learned something from a conference or seminar, your peers and colleagues on Energy Central want to hear about it. It's also easy to share a link to an article you've liked or an industry resource that you think would be helpful.

                 Learn more about posting on Energy Central »