International Atomic Energj Agency IAEA-TC-427.6 Division of Nuclear Safety (TC-SR-2)
TECHNICAL COMMITTEE ON THERMAL REACTOR SAFETY RESEARCH
HYDROGEN BEHAVIOUR AND CONTROL AND RELATED CONTAINMENT LOADING ASPECTS
PROCEEDINGS OF A SPECIALISTS1 MEETING ORGANISED BY THE INTERNATIONAL VTOMIC ENERGY AGENCY 1 A H L.D IN SUZDAL, USS' ' .U SEPTEMBER 1983
INTERNATIONAL ATOMIC ENERGY AGENCY, VIENNA, 1984 HYDROGEN BEHAVIOUR AND CONTROL AND RELATED CONTAINMENT LOADING ASPECTS
PROCEEDINGS OF A SPECIALISTS' MEETING ORGANIZED BY THE INTERNATIONAL ATOMIC ENERGY AGENCY AND HELD IN SUZDAL, USSR, 19-23 SEPTEMBER 1983
Chairman: 0. Kovalevich The Kurchatov Atomic Energy Institute, Moscow, USSR
Scientific Secretary: H. Andres International Atomic Energy Agency, Vienna
INTERNATIONAL ATOMIC ENERGY AGENCY, VIENNA, 1984 HYDROGEN BEHAVIOUR AND CONTROL AND RELATED CONTAINMENT LOADING ASPECTS IAEA, VIENNA, 1984 IAEA-TC-427.6
Printed by the IAEA in Austria September 1984 CONTENTS
Introduction
HYDROGEN PHENOMENOLOGY RESEARCH
The formation of hydrogen in the radiolysis of water in closed volumes 7 S.A. Kabakchi, I.E. Lebedeva, USSR The concentration limits of hydrogen ignition in air in mixtures with non-combustible gases under normal circumstances 10 A. V. Ivanov, A. Ya. Korol'chenko, Yu.N. Shebeko, USSR Detonation characteristics of hydrogenous mixtures (A review paper) 14 A.A. Borisov, B.E. Gel'fand, S.A. Tsyganov, USSR Theoretical evaluation of critical gas layer thickness in relation to detonation wave propagation 33 Yu.N. Shebeko, A. Ya. Korol'chenko, USSR Determination of flame propagation limits in stoichiometric oxyhydrogen mixtures with steam 37 S.M. Kogarko, A.G. Lyamin, O.E. Popov, A. Yu. Kusharin, A. V. Dubrovin, USSR
REACTOR-SPECIFIC HYDROGEN RESEARCH
Hydrogen Production in a PWR during LOCA 42 P. Cassette, France Analysis of hydrogen distribution in containments under accident conditions 52 P. Papadimitriou, H.L Jahn, T. V. Pham, Fed. Rep. of Germany
RISK EVALUATION, PREVENTION, MITIGATION
Hydrogen safety in nuclear power plant reactors 66 A. V. Dubrovin, V.A. Ermakov, USSR Assessment of hydrogen risk in French pressurized water nuclear reactors 70 J. Duco, L. Rousseau, J.M. Evrard, France Analysis of the effects of hydrogen burning and measures taken for their mitigation at the Loviisa nuclear power plant 83 B. Regnell, S. Helynen, Finland The basis for safety standards aimed at averting fire and explosion hazards during work involving hydrogen 90 AM Baratov, USSR
STATEMENT OF AN INTERNATIONAL ORGANIZATION
Statement on current and proposed activities of DG XII of the Commission of the European Communities 98 B. Tolley (CEC), presented by J. Duco (France)
List of Participants 100 INTRODUCTION
This Specialists' Meeting was organized by IAEA In Its efforts to promote worldwide exchange of informâtion In the area of reactor safety research, an activity which Is guided by its Technical Committee on Thermal Reactor Safety Research (TC-SR). The meeting was hosted by the USSR State Committee for the Peaceful Uses of Atomic Energy.
The major concerns regarding hydrogen are that important safety systems may be damaged due to either pressure loads or high temperatures. In order to assess the possible threats, and in order to select appropriate preventive or mitigating measures, it is necessary to understand how hydrogen is produced, how it is transported and mixed within the containment, and how it combusts.
Such hydrogen phenomena as concentration limits for combustion or detonation, or the combustion and detonation characteristics and the influence of diluents, are not particular to nuclear power plants, but they are of concern wherever the presence of hydrogen must be considered. Respective research results and experiences can therefore be of value also for the particular conditions prevailing in nuclear power plants and are referred to in some of the papers preser.".?d at the meeting.
In discussing general conclusions and recommendations with respect to hydrogen research, the participants agreed that the severity of risk posed by hydrogen depends largely on the reactor design and in particular on the containment characteristics. However, it was felt that more work needs to be done in the following areas:
- concentration limits for inflammability detonation characteristics - migration behaviour - influence of the composition of containment atmosphere - detection and control in compartments preventive and mitigating measures - evaluation of the ultimate strength of the containment - equipment survivability - regulatory activities relating to hydrogen danger
The above items reflect the fact that reactors are designed and operated as to avoid conditions in which sizeable amounts of hydrogen can originate. It is important, however, to understand the complex mechanisms governing Its behaviour in order to assess the possible threats to safety systems and to provide appropriate measures to ensure their proper functioning.
THE FORMATION OF HYDROGEN IN THE RADIOLYSIS point one is interested in the maximum hydrogen concentration found in the OF WATER IN CLOSED VOLUMES free space inside a closed container of air-saturated water kept for an unspecified period of time in a field of ionizing radiation. The doses S.A. KABAKCHI, I.E. LEBEDEVA absorbed by the water in such cases are very high, reaching hundreds or Institute of Physical Chemistry of the Academy of more Mrad. At such doses what is known as a steady radiation-chemical state Sciences of the USSR, is produced in the water. This state constitutes a dynamic equilibrium Moscow, in which the rate of formation of the molecular products of water radiolysis Union of Soviet Socialist Republics (H_, H.O, and 0.) is in the primary event equal to that of their dissociation in secondary reactions. After the stationary state has been established, Translated from Russian the concentration of the products remains unchanged, no matter how long irradiation continues. The relationships of the steady-state concentrations Abstract of hydrogen, hydrogen peroxide and oxygen to various parameters (dose rate, By applying the sum total of the elementary reactions involving temperature etc.) have been investigated quite extensively with regard to short-lived particles it is possible to fairly accurately calculate the the radiolysis of aqueous oxygen solutions. The relevant data are kinetics of hydrogen formation and of its separation from water, and also to set out in our paper [l]. We shall use these data again in fact in calculate the accumulation of hydrogen peroxide and oxygen during elaborating a methodology for the present case. At this point we radiolysis of pure water and water solutions at room temperature. This would like to draw attention to one particular feature of the radiolysis paper describes a semi-empirical method to calculate the kinetics of of oxygen-containing water: in the irradiation of water in an enclosed hydrogen formation for certain cases encountered in nuclear power production. space the absolute quantity of oxygen in the system remains unchanged [2].
As mentioned above, the rate at which hydrogen builds up in radiolysis The radiation chemistry of water and aqueous oxygen solutions is a branch can be calculated by means of a system of differential equations showing of chemistry which receives considerable attention nowadays. By applying the change in concentration of all the particles involved in the chemical a well-known mechanism - the sum total of the elementary reactions involving reactions in the system under study. It can be shown that, for a short-lived particles - it is possible, using a computer, to make fairly given initial concentration of oxygen, the accumulation of hydrogen accurate calculations of the kinetics of hydrogen formation and of its can be expressed by a single equation: Separation from water, and also to calculate the accumulation of hydrogen peroxide and oxygen during the radiolysis of pure water and oxygen solutions at room temperature. At temperatures above room temperature, however, it is not possible to attain a sufficient degree of accuracy in the calculations, u> ** since not all the temperature relationships of the kinetic parameters for elementary reactions and short-lived particle yields are known. Vet it is at temperatures above room temperature that the radiolysis of water is where I is the dose rate, G(H,) is the molecular hydrogen yield equal most often of interest in practice. to 0.45 molecules/100 eV over the temperature range 20-250 C; In certain cases encountered in nuclear power production, the kinetics B is the ratio of reaction rate constants of hydrogen formation can be calculated by means of the semi-empirical method = HO, H 0, (2) described below. The problem is that very often from the practical stand- OH 2 OH » H2 = H » H,O, (3) At a given temperature, this partial pressure can be calculated equal at room tenoeracure Co 1.00 and dependenc on cemperacure in by means of the following equation: accordance with the relationship
B » exp(1.25xlO3/T - 4.26), (4) (6) T is temperature in °K and, finally, A is a parameter calculated from A - the experimental data. In physical terms, the parameter A denotes the effective yield of hydroxyl radicals involved in the dissociation where a is the coefficient of Henry's law for hydrogen. Fig. 3 shows of molecular hydrogen in accordance with reaction (3). In the steady a graph from which it is possible to calculate the critical partial state, d[H-]/dt s 0 and Eq. (1) is transformed into Che algebraic relationship: pressure of hydrogen circulating in the free space of a closed container with air-saturated water and irradiated at the above-mentioned dose rates and temperatures. The graph was plotted on the basis of the (5) temperature relacionship [H,0,] and a, together with Eq. (4) and the daca given in Figs I and 2. d We stated above chat, in the radiolysis of oxygen-bearing water in a closed system, the quanticy of oxygen is noc affected by Irradiation. In conjunction with the experimental data on the dependence of stationary Consequently, ic is easy Co determine the partial oxygen pressure (P. ) concentrations of hydrogen and hydrogen peroxide on dose rate and temperature, in Che system. If ?#- and PQ and the total pressure in the system Eq. (S) provides a means of calculating the dependence A on these parameters are known, the exploslveness of a gas mixture circulating freely in which are what, essentially, also determine the hydrogen accumulation the space can be assessed. in the system in question. Figs 1 and 2 shov the dependence of parameter A on dos* race and temperature in the radiolysis of air-saturated water.
From Fig. 1 it can b« seen that A does noc depend on the dose rate 19 3 when I)ii 10 tV/(dm • s). The relacionship of the parameter to cemperacure is Che same for dose rates in the interval 5 x 10 - 1.2 x 10 eV/(dm • s). This result is precisely whac one would expect from the theoretical point of view, since, for a given dose rate and fixed iniclal oxygen concenCracion, the parameter A is determined REFERENCES by the rate conscancs for the elementary reactions which are independent of the dose race. ' , [1] SHUBIN, V.N., KABACHKI, S.A., Theory and methodology of the radiation chemistry of water, Nauka Press, Moscow (1969) (in Russian). From the results obtained, it can be inferred chat Eq. (S) used in calculating the critical partial pressure of hydrogen at doses greater [2] KABACHKI, S.A., PIKAEV, A.V., Methods of calculating gas generation 23 3 than 10 eV/(dn • s), i.e. doses ac which a sceady state is produced and evaluating the explosiveness of radiation-chemical apparatus in the radiolysis of air-sacuraCed water irradiated at doses greater equipped with a water coolant or biological shield, Ehnergoizdac, Chan 5 x 10 eV/(dm • s) at cemperatures of 2O-15O°C. Moscow (1981) (in Russian). 0,5 r 7.5
/ 5,0 V / 010 I 2,5 / 0 2 4 Dose race, 1019 eV/(dm3 • s) rat / Fig. 1 Dependence of parameter A on dose race /
005 /
3 g
1 : i 273 323 3Ï8 '3?3 398
operating T in K
300 320 340 360 380
Température in °K Fig. 3 Graph for decermining the critical partial pressure o£ hydrogen for Che radiolysis o£ wacer in a closed container. Fig. 2 Dependence of paramecer A on cemperacure THE CONCENTRATION LIMITS OF HYDROGEN These limits were determined by measuring the upward propagation of a flame IGNITION IN AIR IN MIXTURES WITH in a glass tube 5 cm in diameter and 1.5 m long with open lower end and closed NON-COMBUSTIBLE GASES UNDER upper end. The methods sec out in Ref. [3] were employed and substances of high NORMAL CIRCUMSTANCES purity were used. The ignition limits of mixtures containing up to 30*'. vol. hydrogen were studied, these concentrations being of che greatest practical A.V. IVANOV, A.Ya. KOROL'CHENKO, Yu.N. SHEBEKO interest. All-Union Fire Fighting Scientific Research Institute, The results of the experiments are set out in Figs I and 2. Comparison of Moscow, the curve for desensitizacion of the oxyhydrogen mixture by nitrogen with the Union of Soviet Socialist Republics results of Coward and Jones [l] shows good agreement, a face which attests to the reliability of the values obtained for the ignition concentration limits. Translated from Russian Earlier, we demonstrated [4] that the concentration of fuel on cne lower branch Abstract of che desensIcization curve when the iiixture is diluted with a chemically inert phlegmatlzacion agent can oe expressed as follows: The paper describes an experimental investigation cf concentration limits at whicn hydrogen and various diluents (nitrogen, carbondioxide, helium, argon, chloropentafluoroethane.1,2 - dibromtetrafluorethane) ignite in oxinitrogen oxidizing media. It is shown that the fuel and desensitizer concentrations can be calculated at desensitization points for cnemically inert diluents. where •H„ is che lower concentration limit at which che fuel ignites, *$ is the concentration of diluent m the mixture (in volume "'.) and y denotes the desensitizing capacicy of che diluent. For nicrogen and carbon dioxide this value can be cal- culated by means of che formula: Practical work with hydrogen has certain difficulties, cne notable problem being to offer sure guarantees against the danger of explosions during production, (2) transport and combustion of hydrogen. This, in turn, involves a number of theoretical and practical problems, the most urgent of which is the investigation of concentration limits for hydrogen ignition in mixture'; such as hydrogen * where H and '•!, are the absolute molar enthalpv values of the desensitiier and of » b r' oxygen diluent. Such research is of theoretical interest also, offering as it does a way of gaining a more thorough understanding o£ the chemical kinetics of hydrogen combustion in mixtures with various diluents, especially as the data in The indices ° and ' refer to the adiabatic temperature of combustion and the ' the literature on concentration limits for ignition have mostly been obtained in initial temperature of che fuel mixture. Using the expression set out above, we work with air as the oxidizing medium [l, 2]. calculated the hydrogen concentration on the lower branches of che desensitizacion
The present paper reports on experiments in which ignition concentration curves plotted for phlegmatization of hydrogen by nicrogen and carbon dioxide. limits were studied for hydrogen mixed with various diluents, and for combustion The adiabacic temperature of combustion was taken as 1000 K. Ic should be noted occurring in air and in oxygen. The diluents studied were nitrogen, carbon here that the results of the calculation are not very sensitive to this tempera- dioxide, helium, argon, chloropentaf luoroethane -r.1 1, 2-dibromtetraf rjoroethane . ture. The calculated results were then compared with those obtained experimentally. The relative mean square error was found to be 6% (for 16 experimental poincs). chloropentafluoroethane or 1,2-dibroratetrafluoroethane. The flame front is
A linear dependence of the fuel concentration * on desensitizer concentration not solid but rather a composite formation of small foci [7, 8]. Because of its higher diffusion coefficient, more hydrogen burns up per unit volume of » is also observed in the case of diluents such as helium, argon and chlorpenta- flame front than is contained on average per unit volume in the reaction vessel. fluoroethane, to which formula (1) is therefore applicable. The values of y for One would therefore expect that, the greater the difference between che diffusion these agents, determined on the basis of experimental data obtained in che presenc coefficients of hydrogen and the diluent, the more liable the lean mixtures are study, are 2.19, 0.84 and 6.0 respectively. The values of y for helium and argon, Co ignice, i.e. che more likely ic is chac the high values of che oxidizer excess calculated from expression (2), turn out to be substantially lower than those coefficient a will correspond to che desensicizacion poincs. Where che diffusion obtained experimentally. This is probably because helium and argon have high coefficienc of che diluent is higher chan chac of hydrogen, mixtures close co thermal conductivity coefficients. The high -f value obtained for chloropenca- stoichiomecnc should correspond no che desonsLCizacion poincs. fluoroethane is due to its thermal instability as a desensicizer at flame temperature, such chac additional energy is lost on partial dissociation of che Figure 3 shows the dependence of the quantity a for che desensicization diluent and heating of its decomposition products. poincs on che paramecer X as defined by the relationship
The ignition limits for the mixture H, * 0, - C.F^Br. (Fig. 2) have certain (D • n. :.f D - D > 0; distinctive characceriscics. It is known that vapours of 1,2-dibromtecrafluoroechane x . ) ° ° K) burn in oxygen in a cube 5 cm in diamcer and 1.5 m long with che flame propagating | 0, if DQ - D £ 0; upwards [5]. For vapour mixtures from a number of fuels the Le Chacelier rule where D is che diffusion coefficient of hydrogen in air and D is the diffusion holds: coefficient of the diluent in air.
As Fig. 3 shows, the values of a obtained experimentally lie with satisfactory it . (3) 1-1 1/71- accuracy on che universal curve, which can be used for calculating a priori the concentrations of fuel $ and diluent $ at the desensitizacion points for r $ where 4. is the concentration of the i-th combustible component of the mixture combustion in oxidizing media consisting of nitrogen-oxygen mixtures. and * . is its lower (upper) concentration limit for ignition. The quantity a can be expressed in terms of fuel and desensitizer concen- Equation (3) in the plane * - • (Fig. 2) should be valid for che segments trations by means-of che relation of the straight lines connecting the corresponding ignicion limits. However, it follows from this figure that in our experiments there is a fundamental violation of the Le Chatelier rule. The reason for this according to Bunev and Babkin [6] (5) probably lies in the preferential burnout of hydrogen when a flame propagates where 3 is the stoichiomecric coefficient of oxygen in the combustion reaction in an 0, + C.F^Br. mixture. (for hydrogen a =0.5) and •„ is the concentration of oxygen in the oxidizing medium in vol.7.. Solving Eqs (1) and (5) for • and • , we find It is interesting to noce that in our experiments the lean mixtures are those which correspond to the desensitization points (except where the agents are argon and helium, in which case the mixtures are stoichiometric). This is (6) probably because of the blister-like way in which hydrogen burns, which is due, in turn, to the face that hydrogen has an appreciably higher diffusion coefficient in air than do other gases such as oxygen, nitrogen, carbon dioxide, 100 12 100 - REFERENCES (7) COWARD, H.F., JONES, G.W., Bulletin 503, Bureau of Mines, Washington (1952) 144 pages.
The following method is used in applying Eqs (6) and (7). From the diffusion [2] ZABETAKIS, M.G., Bulletin 627, Sureau of Mines, Washington (1965) coefficient of the diluent in air (assuming that it is close to the diffusion 121 pages. coefficient in a given nitrogen-oxygen oxidizing medium), we use the graph in [3] MONAKHOV, V.T., Methods of investigating the fire hazard presented by Fig. 3 to derive the corresponding value of a. We then calculate * and » from various substances, Khimiya [Chemistry Press], Moscow (1979) 423 Eqs (6) and (7). (in Russian). The feasibility of this" calculation was verified on the basis of the experi- •W KOROL'CHENKO, A.fa. , TSAP, V.N. , SHEBEKO, Yu.N., E0BKOV, A.S., mental data obtained in our present work (see table below). The relative mean IVAN0V, A.V., Zh. Fiz. Khim. j£ 4 (1981) 928. square error in the calculation of • _ and ». is 3 and 137., respectively. [5] CHEBHUSHKIN, Yu.N., SARATOV, A.N., POLOZNOV, N.M., in Combustibility of Substances and Chemical Means of Fire-Fighting, Moscow VNI1P0 4 Concantration, vol.% (1973) 30 (in Russian). •„, Fuel Diluent 2 ~1 02 (From curve BUNEV, V.A., BABKIN, V.S., FGB 9 4 (1973) 605. Diluent D, cnt * s vol.7. in Fig. 3) Experi- Calcu- Experi- Calcu- [6] ment lation ment lation [7] BREGEON, B., GORDON, A.S., WILLIAMS, F.A., Combust. Flame, 13 1 (1978) 33. [8] MUTANI.T., WILLIAMS, F.A., Combust. Flame, 39 2 (1980) 169. Nitrogen 0,28 100' 2,0 4,3 91,3 91,8 Carbon dioxide 0,13 100 2,75 5,8 6,1- 86,4 85,4 Helium .>I,00 20,6 1,0 • 8,0. 7,7 73,0 73,7 • ioo 8,0 8,4. 88,0 87,4 Argon >I,00 20,6 1,0 3,8 3,5 82,5 88,0 • ioo 3,2 3,5 93,2 94,8 Chloropenta- fluoroethane • 0,08 . 20,6 3,5 9,0' 8,4 22; I 20,8 100 n,o 15,7. 67,5 .56,7
Noce: The diffusion coefficient of hydrogen is taken to be 0.78 en -s" .
In conclusion, Che present paper describes an experimental investigation of the concentration limits at which mixtures of hydrogen and various diluents ignite in oxynitrogtn oxidizing media. It is shown that the fuel and desensltizer concentrations can be calculated at desensitization points for chemically inert diluents. - \V—s'
vo l \\ . 1 y- • \\ \ It • A 8
M 40 SO 40 JOO
n £0 —too
Fig. 1». Fig. lb.
Fig. \. Concentration limits for hydrogen ignition as a function of diluenc composition:
a. 1. H2 - N - 02; 2. -02; 3. Hj - Ar - air;
4. H2 - Ar - 02.
b. 1. H2 - CjFjCl - air; 2. H2 - C^F ; 3. H2 - He - air;
4. H2 - He - Oj.
13 vol.1 14 DETONATION CHARACTERISTICS OF HYDROGENOUS MIXTURES (A review paper)
A.A. BORISOV, B.E. GELT AND, S.A. TSYGANOV Institute of Chemical Physics of the Academy of 20- Sciences of the USSR, Moscow, Union of Soviet Socialist Republics
Translated from Russian Abstract ." Xi) 40 IQ &B fcvol.% The paper reviews and summarizes the résulta contained ia 48 Fig. 2. Concentration limits for hydrogen igr.Jtion as a function of che concent referenced publications related :o the subject and evaluates their of 1,2-dibromcacrafluoroethane (combuscion in oxygen). sLgnificance for nuclear power plant safety. The main sections of the paper deal with: parameters of the combustion snd detonation process ia hydrogeneous mixtures; critical conditions for activation of combustion and detonation in hydrogeneous mixtures, and paramerers of pressure waves during explosions of hydrogeneous mixtures. The text is supplemented by 6 tables and 32 figures.
I. Introduce ion
The possible formation of hydrogen-air or oxyhydrogen mxtures inside industrial equipment and also the possible discharge of hydrogen in;o the environment requires careful analysis si the operational saiatv j£ various types of equipment, including nuclear power plants, in view of the highly reactive potêncial 0Ê systems containing hydrogen.
Ofi X US f"' Although research on combustion and explosion of hydrogenous mixtures
F1«- 3- Oxidizer îtcess coefficlsnc a at the desensitization points as a has oeen carried out for quite a few years, many of the proolems concerning function of the parameter x: che safety ni nuclear power olants nave not yet been resolved. In fact, analysis oc research on comDUStion and detonation processes snows chat 1. H2 - Ar - air; 2. H2 - He - 02; 3. H, - He - air; 4. H - N - 0 ; ;he greac •najority of studies nave oeen carried out using -nainly oxygen c 5. H2 - C02 - 02; 6. H2 - 2FjCl - air; 7. H, - C,F Ci - 0 ; mixtures ind at very low pressure. Iz L5 racher difficuic "o use che resulcs ?i chij r«saarcn in analysing actual situations that occur wnen operating with hydrogen-air or oxyhydrogen mixtures. Further difficulties in pre- Ref. [5j, the vast majority of other measurements suggest aigner normal races. dicting explosion parameters arise «hen steam, carbon oxides and nitrogen Measured normal rates differ by as much as a factor o£ 1.3. rig. Ib shows are added to the combustible mixture. che extreme values for the normaL combustion r-_ita given in Ref. _5\ Evidently, che dependence in Ref. .3], representing an average of :he data Ic is not easy for researchers involved in improving che operational given in Ref. ,3^ is the more reliaDle. For praccical purposes ic is safety of nuclear power plants and formulating operational regulations important to know, apart from the normal combustion race, che visible propa- to find their way about the diversity of material availaDle on evaluation gation race, which is greater tnan the normal race as a function of che of explosion characteristics of nyarogenous mixtures. Zn formulating sets degree of density reduction of che gas. Some oc che parameters for the of preventive measures ensuring the safety of hydrogenous mixtures, che combustion process in scoichiometric mixtures of H,-0, and H.-air are given following initial parameters should be known: in Table 1. 1. The dependence of normal flame propagation and detonation rates Table 1 shows che pressure after comouscion of the mixture in a constant on initial temperature, pressure and mixture composition; volume P and oressure in the wave ? which may be transmitted to the v • s 2. The denendence of explosion and detonation parameters on initial surrounding soace; che degree of expansion 3; and che visible flame temperature, pressure and mixture composition; propagation race U . From Lee's calculated daca in Ref. .3;, Fig. 2 shows 3. The flame propagation and detonation limits for Che mixture che temperature distribution 3, the proportion of ournc material \, and composition and their dependence on pressure and cemperature: che oressure distribution in che vessel as a function of che flame posicion
4. Critical conditions for initiating combustion and detonation R_ in a closed spherical vessel of radius R . The calculation was made for a 2H.-0., mixture at initial cemperature and pressure = 298 S and on the basis of the composition of the mixture for confined, ? = C.I HPa, where p and T are the pressure ana temperature at an arbi- semi-confined and unconfined volumes; trary momenc in time. The calculations in Fig. 2 are made without allowing 3. Potential propagation regimes for the combustion front in blocked for the dependence of che normal flame velocity in the mixture (2H--0,) spaces and basic properties of these systems; process rate and on cemperacure and pressure. In Ref. [2] ic is suggested that the following pressure in the wave generated by the flame front; semi-empirical ratios for che normal flame velocicy (U.) should be used 6. Parameters of blast shock waves occurring as a result of explosion for evaluacions: of finite mixture volumes in terms of pressure, pulse and com- pression phase time. U, = 1150 + 2- Parameters of the combustion and detonation process in hydrogenous mixtures -bars .
A considerable amount of varied data is available concerning hydro- en To = 298 K genous mixtures from which it is possible to establish the dependence of the normal flame propagation rate on the composition. Figures la and lb, Calculations, allowing for the dependence of normal velocity on temperature based on data contained in Réf. [il, provide a fairly full summary of data and pressure (Fig. 3), show that the normal velocity increases by a factor on the dependence of the flame velocity m hydrogen-nitrogen-oxygen mixturss. of 1.3-1.5 during combustion and ^hat che visible flame rate decreases. There is quite a lot of scatter in the measurements of the normal combustion The broken lines in Fig. 3 show the variation in normal velocity when the rates for hydrogen-air mixtures. The measurements in Ref. [l] shown in cemperacure dependence is taken inco account and che continuous lines indi- 15 Fig. la apply to the lowest known combustion rate. As is pointed out in cate wnen che dependence on pressure is alone canen into account. 16 The cases examined apply Co flame propagation in a closed vessel. regime of this type is possible when the mixing cimes are short. In certain During combustion o£ a fuel-air mixture change in an open space, che pressure situations known from data published in che literature, mechanical failures variacion in front of the flame from, is different. The rapid fall in have been observed when there is a long de Lay between che start of gas pressure in front of the flame front occurs only for flames with a propa- leakage and the moment of ignition. If che time between che formation gation rate higher than a certain limit. According to the calculated data of the mixuere and moment of ignition is short, a fire occurs or a fireball in Ref. 1.2], for hydrogen-air mixtures chis rate ('.,',) should be > 50 vn/s. is formed as a result. However, even wich delayed combust:ion of che cloud, and for 2H,-0., mixture > 150 m/s. Figure 4 shows values taken from Ref. '"2] a blast wave in front of the flame front is rarely observed during combustion for the drop in pressure at the leading shock front ap and at ;he flame m a free volume. The combustion regime in a constant volume has more
front lp£ for 2H,-O2 (line 1) and for air - H, (a = 1) (line 2). In none bearing on che safeguarding of reactors and vehicles c-ansporcing spontaneously of the ^-N,""? syscams does the flame velocity in unblocked space actain decomposing gases and liquids from explosions. In ensuring nuclear power the values required to create shock waves. For a complete description plant safety the dependence of the pressure of the explosion on the mixture of che conditions ac which formation of blast waves becomes possible, lee composition and initial conditions is very relevant. This dependence makes ic us examine the conditions under which there is combustion wave propagation possible to evaluate dangerous load levels on neighbouring equipment if in gaseous mixtures. chere is sudden destruction of the object inside which che explosive mixture Depending on the dimensions of the volume of the gaseous mixture, has formed. In theory, the parameters of blast waves when pressurized its geometry, whether there are walls confining the volume, che normal com- vessels explode can be evaluated using the method described in Sef. [<*]. bustion rat» and ch« mode of initiation, the following combustion wave The chird of che combustion regimes mentioned is similar to comouscton propagation rtgimts are possible: at constant pressure, but differs in that it has higher flame velocities, (a) Explosion (combustion of the mixture) at virtually constant which are characteristic only of turbulent flames. For chese to occur, pressure; curbulators of unburnt material must be present.
(b) Explosion (combustion of the mixture) ac constant volume; The fourch combustion regime is non-steady, directly borders on detona- (s) Rapid, usually accelerating, combustion with formation of tion and includes so-called partial detonation regimes, which may occur compression waves of finite amplitude (sometimes called in practice during explosion Initiation of combustible mixture clouds with "deflagration combustion" in Russian publications); unevenly dijeriubeed fuel.
(d) Sapid combustion with formation of compression waves and salf- The combustion regimes described can, as a first approximation, be lgniclon of the mixture when these waves interact with obstacles represented in a pressure (p) against density (p) diagram, such as the iS Otc
— acmospheres
ïj^and Q are che ratios of che thermal capacicy of che combustion products Table 2 shows che influence of various dilucencs on che deconacion race and the heacing effecc per unit mass of combustible mixture. These ratios for oxyhydrogen mixcures from Ref. [l]. Calculations and measurements can be obcained from thermodynamic calculations. For the system H • air these indicace chat replacement of nicrogen in an H.-N, mixture by either argon 17 calculations are described in Ref. [3] and the results shown in Fig. 6. or helium does not appreciably change the pressure ac che Chapman-Jouguec aoint ar the température of che comoustion products. The 3i:ierence in As a rule, there are or there can be m react ing mixtures feedback rates for mixtures witft helium and argon added is linked to che soeed •necnanisms ensuring interaction between motion Jt the flame and che unournt of sound in the Live -nixture. «"hen excess oxygen is added co tne 2H-, - 0- Tuxcurs and che compt ass ion Javes emitted dur^.^f? cne comousciûn process.
•nixture the detonation parameters are recucea in tne same wav as vnen nl :r.e Several such mechanisms are
Detailed measurements of detonation rates for nydrogen-air mixtures The instability of the interface for gases with different densities are given in Re£. [-']• These measurements are represented in rig. 11 is considerable in closed vessels, channels and cuoes. Flame acceleration
Line 1 is the r,/drogen-air -nixture and lir.e 2 is cne mixture if -.ydrogen as a lesulr. of che SayLei^n-TayLor instability hapoens during intersection
and oxygen. An increase in initial temper a Lure or cr.3 mixture leacs "3 3 £ che flame front ov che pressure wave from tne come use ion product 5ide
reduction in the detonation parameters through a drop m pressure as a aens 1 tv is Lower than -.;i a 1 tve inxture ) when all the unevennesse5 3r che
result of the decrease m the mixture's relative specific calorific flame rronc grow rapidly in amplitude and che sur:ace of :ae flame ii increase
power Q • C . Table '•*•give s some calculated figures for deccnation in In closed vessels f Lame accélération up to the point of deconation is often
a 2H? * 0^ mixture as a function of initial temperature •" L3 ] - The resuLts caused bv che tnscabiliry ot zhe flame front: when inceracciag with com-
of calculations of the dependnence of detonation race on temperature and press ion waves. the results of measurement s in Ref. _ 19 _ show chat \ ne cer:-r ::n race .Amp 1-ifi.cation or compress ion waves when interact ing with the flame m the range of initial pressure p < 10 Mri. wnich is J: r^rojr for -racledr 1 rsne is The 'iomir.anc fact or àt cambuse ion stages close to the occurrence po«er plants, scarcely depenc-s on tne mic lai parameters 7. in i o,. of deconarion, vnen che pressure waves in tne channeLS or closed vessa13
Tîve tii^"i imp L1 c.irlo . Cl ose co the i an it ion source ih* variations m : IJTC This -neans cnac there is ne àuoscar.c.jL mange in - --2;ipe récure -p res surs _ i shape ind cunuia::.on ;f zee -nixture enraagh Tio::;r. of the gas in :ronc detonac ion characterise ics and olast vo Lume of rr.e vqr iao les D„ v. v , • /. 3 Jt cpe f L jre :' r in;, are -no re i.tiportanc - and p x p dependent on che compos it ion. The int roc-c : i :n or 5cearn rv 0 into the mixture composicion, as 15 shown by therraodvnaniic calcuLatuns I n .1 r G n /o L unie s £ 1 ame acceleration as a result of 1C5 in.scaDi.li.ty
for an H-, - air system, when 3 = -U.5 and y.5 and p = 5.3 MPa and 7, - -*2? '< jue :J '^V« .cmureasion is noc so Large. Theretore in large vo Lûmes, even
i_ 20 J * leads to a reduce ion in blast pressure and detonacion wave pressure. : or -Tuxt:res vich high reacr we potential there is no z lame acceleration.
The reduction of detonation and blast parameters in this case is proportion ai. So in che exoeriments described in Ref. [2\] for mixtures of echylene oxiae
to che quantity of steam introduced. Jich air in hemispherical voLames with a radius up to 5 m, no flame accelerac;:n
•-as detected. As is shown in Ref. [ L1 j 1 the dependence of combustion rate So far we have examined the propagation rates of combust ion reg:mes ,n Tempe racure and pressure has just as liccle influence on f Lame accélérât ion. which can be cal led ideal since they occur in an unolockact snace in smooch "lame acceleration as a resulr. ot a rise in temperature and pressure of zuDes or in an unconf ined vo lume - In reality, c.iere is a L wav s i m^n •.he -nixcurs generally plays a subsidiary role in observaole rlame acceleration probaoxlity of volumes blocked by various technological devicps and systems in A -sed volumes, while m ipen sapces it LS msignifleant. In analysing becoming filled with an explosive mixture. In sucti cases chp ooraire-er-» "he : ;i f LM.Tire ?f :*»mp*»ncure 2E che medium on f 1 ame acceleracicn, it shouli of the explosion process may differ from the "ideal". be recalled chac the increase in the speed of sound m the Live mixture Dy a jail containing the orifices d, = 5, 10, 20 and 30 im. Ignicion of when the temperature rises, inhibits formation of shock waves and cherefore che mixture cook place in che cuoe 1 1 . These experiments for hydrogen- reduces the likelihood of detonation. Reference .11. shows that heating air mixutre are summarized in Ref . '23] and one set of results is shown 2H., - 0., mixtures led to an increase in pre-detonation distance or completely in Fig. 12. This figure shows che pressure recorded in che second cube
prevenced detonation in chs experimental conditions igven in Sef. [ll]. wicn dn and *•-, as che flame passed through the orifice. The cube was open at che eno opposite che wall. As can oe seen, there were j set of orifices Turbulacion of the flow, ay speeding up the transfer process and ex- d = 5-20 .Tim for which passage chrougn co the second cavity created the cenoing the combustion surface, is the mosr. effective flame accelerator. highest pressure. Reference !_22] shows that as a result of natural curoulacion che flame velocity may increase by a maximum factor of 25-30, wnile 3n increase m The aggregate results obtained bv researcher m Vesc 3ernany i. 22-24] the flame propagation race of 5-10 times is more realistic. More serious show that in certain condicions special combustion regimes can occur, for consequences result from artificial turbulation of comouacion caused sy wnich richer cne pressure and velocity or one of che parameters mdicaced placing turbulating obstacles of differenc shape and aesign in the flame differs appreciably from both detonation and deflagration regimes. As pat.n. These may include partial channel constriction, lattices, grids, in Ret. .11,, we will refer co chese comouscion regimes as quasi-deconacicn ridges etc. [22,23,24,27]. regimes. Sucn combuscion regimes normally occur in rougn-surface or arti- ficially slocked cubes. These combustion regimes were first discovered Flame acceleration as a resulc of turbulating obstacles is particularlv by K.I. Shcnelkm, who ficced spirals in tubes filled with gaseous tmxeures. effective in tubes or closed vessels. In an open space it is usually impracti- In rough-surface pipes, combustion regimes with races of 1.5 to 1.6 times cal to place obstacles in the flame path. In industrial construction, less chan deconacion races are possible. Regardless of the type of mixcure, r.cwever, the formation of powerful shock waves has seen observed during it is observed chac che effect oc artificial curbuUcors is particularly comouscion of a fuel-air mixture, which indicates the possibility of local nigh for borderline mixtures. As will be shown below, tne introduction rapid combustion of part of or the whole of the fuel-air mixture volume. of curoulaccrs helps to expand deconacion limits co some axcenc and cherefore Usually after emergence from the blocked space, the combustion rate for is of some practical importance. most fuel-air mixtures with a content of N > 60% falls to the initial low level. This means that the high rate of motion of the combustion process Fo.- hydrogenous fixtures, che occurrence of specific characteristics is linked to flame distortions at the obstacles, as described for example for che quasi-detonacion process is observed in combustion of che H - 0 in experiments in Refs -.25,27;. The observed weax dependence of flame velocicv L mixcure when 3 = 0. 4-1. i in a tube wich a 124 mm diamecer and langch of a slocked space on kinetic properties and on the normal flame velocity points co 1330 Tim, filled with steel or ceramic spheres l^ and 38 .Tim in diamecer. The the need to carry ouc a series of experiments aimed at discovering the link experimental resales [2b, are shown m rig. L3. The continuous lines 1, between combustion rate and the parameters of obstacles. The conclusions 2 and 3 correspond to races at p = 0.1, 0.2 and 0.5 MPa in che tube with drawn from the use of specific designs are net universally applicable ana steel spneies, and che lines •*, i and c co deconacion races ac che same can only be qualitatively transposed to other conditions. pressure levels in che cube wi:n ..eramic spheres. Lines ~, 3 and 9 show :he fall m pressure in che combustion wave ac 3.1, 0.2 and j.5 MPa in For hydrogen-air mixtures, a number of experiments have now been conducted the cube wich sceel spheres, and lines 10 and 11 at pQ = 0.1 Jnd 0.2 MPa in which give some indication of the effect of obstacles on the explosion tne cube with ceramic spheres. All :ne experiments, che results of which are characteristics. Experiments were carried out at j = 1, 3 = 0.1 MPa. shown in Fig. 13, were carried ouc vic.n 33 mm diameter spheres. Figure 14 gives TQ = 293 K. Two cubes of diameter d and d,, where d = 40 mm and some information on expenmencs in 1 cube filled with 19-mra diamecer sceel spheres. 19 dj = 10 and 30 ram, and of length 4 = 500 mm and 1, î 2500 ran, were joined 20 Lines 1, 2, 3 and 4 correspond to races «hen p =0.1, 0.2, 3.5 and N or Ar ignite most easily. Wider flame propagation limits characterize these p 0.9 MPa, and lines 5, 6, 7 and 3 correspond co the fall in pressure in mixtures with excess fuel. Table 5 shows ignition limits 'or the composition che wave for the same values of p . Vhen p < 0.5 MPa, the insertion of where E = I «J in Ret. [1]. Let us compare che data on minimal spark ignition foreign bodies inco che cube does noc interrupt che blast process, although energy in Fig. 15 wich data on initiation of spherical detonation. Figure 1b ics race and developing pressure level are somewhat lower chan luring deto- shows the change in critical detonation initiation energy of oxygen mixtures ac nation in an unolocked tube. It is also characteristic that in a blocked p =0.1 MPa from data given in Rets ,12, 29], while Fig. 1? shows it for cube, che pressure level of the blast process is virtually independent hydrogen-air mixcures from Refs [12, 2S, 29, 30]. It can be seen ;hac if for i of che mixture composition within very wide limits. It is interesting to spark source energy of 10 J, ignition ot che whole spectrum of mixtures noce chac the nature of the blast process in H_-N -0, mixtures has low capable of combustion is guaranteed, then for activation of spherical detona- sensitivity - up to a certain limit of dilution o£ the mixture with inert tion of oxygen mixtures a blast source wich an energy E > 1 J is necessary, and gas (or with excess fuel and oxidancs) -> co che scruccure and chermophysical for hydrogen-air mixtures the source must have an energy E > 10 J. This corre- properties of the inserted material. The authors cogecher with 3.1, Panamarchuk lation means chac there are significantly stricter limits for detonacion acci- and E.I. Timofeev conducted experiments to study combustion regimes for vacion than for ignition. Table 6 gives ;he detonacion limits tor initiation H,-U',-O, mixtures in hydro-mechanical foam with a water concentration of by explosive material charges of different energies in hydrogen-air mixture. 3 up Co 10 kg/m . Ic was demonscraced chat even ac weak ignition in a During ignition of a charge of 100 g of Tecryl, che detonation limes from 2H. • 0, mixture, crapped in foam cells, the blase process spread ac a Ret. [29] are 14.4—76Ï H- in the mixture. In Sef. [28] for hydrogen-air mixture rate of about 2500 m/s and ac a froncal pressure of 1.0-1.5 MPa. Vhen che the limics are 13-707». Deconacion propagacion measuremencs in tubes [33 j have been used co sstaolish a lower detonation limit of 15.37. H m air. This mixture «as diluted to 2H2 «• 0, •* X, the velocity of the blase process fell from 800 to 500 m/s and there was a gradual fall in velocity for the closely coinciaes with the daca in Ref. [34], where che lower limit is 15% H same composition, with a reduction of the dimensions of the gas occlusions, in air. The spread of mixture concencracion limics can be partly explained by i.e. with an increase in surface to which heac is transferred. In the the fact chat m many experiments the roughness of the tube walls was noc 2H^ • 0^ » 2N^ system, activation of explosive combustion in hydromechanical checked and che presence of various sources curbulizing the flow, such as cracks foam from a powerful ignicion source (a pyrotechnic suspension) does noc and ledges on che inner surface of the tube, was not taken into account. occur in the greac majoricy of cases. It vould be useful to conduct extensive Reference [32] shows thac che attachment of a spiral to the inner surface of research on combustion regimes for 2H, - O2~2Î<2 mlxI:ures Ln *oam ac ill8h the tube leads co expansion of che deconacion limits. Figure 18 shows the initial pressure and temperature. Research of this kind could throw some dependence of deconacion limics on cube diameter in a smooth cube (branch 1) light on the nacure of the explosion processes to be expected during the and in a tube with a spiral (branch 2). Reference [35] also points ouc that I operacion of accual nuclear power planes. in rough-surface cubes the deconacion limits may differ from those in smooth tubes. The data in Refs [31, 35] complement each-other and show thac in H^-iir mixcures where d > 50 mm, che roughness of che tube does not influence 3. Critical conditions for activation of combustion and detonation in che detonation limits. hydrogenous mixtures
Reference [l]summarizes che relationship for che variacion in minimal As Ref. [34] shows, some blurring of the data on the detonacion limits of energy required for spark ignition of che hydrogenous mixcures Hj-CO.-O,, hydrogen-air mixcures can be explained by che face chac near che limits there H2"N2"°2' H2~Ar~°2' H2"He"°2 in an oxidizing acmosphere wich 217. 0, and 797. gas. occur what are called "galloping regimes". According to Ref. [35], ac the rich- The hydrogen volume concent - 307. corresponds to a stolchiometric mixture. The mixture limit, che area where galloping regimes occur lies ben«n 68 and 717.. and ac minimal spark ignition energy depends on che type of gas. Mixtures diluted with the lean level in the range 12.5-14.3". Comparison of Che daca obtained from research on flame propagation Limits indicates che nearness of deconacion limits spectrum of mixture composiCions than detonation propagation. Unlike ignition co combustion limits in mixtures with excess H,. In dealing Mich proolems o£ ana flame propagation Limits, addicion of argon and helium does noc aftecc che safety from explosions, che area of unscaDle deconacion regimes with i pulsed '.over deconacion Limic, i.a- deconacion propagation snly slighcly depends jr race should also be examined. During these processes, local increase in loes noc depend ac all on the thermal conductivity D: the mixture. pressure (particularly m dead-end seccions) may be higher chan cor deconacion. References [37, 38] point ouc chac che deconation risk of rich hydrogen mixtures
Reference ^35] shows chac Che mosc destructive combustion regimes usually is lower chan in Lean luxnures due co the high speed of sound in che Live begin with simple flame propagation. The limits necessary for these processes mixture. Increase in cne speed of sound inhibits wave formation m front of are close to the flame propagation limits. With regard to safety proDlems, the accelerating wave Eronc. The injeccion of small quancicies of hydrogen che question of possible transition from combustion to detonation in specific mco che air leads co significantly more dangerous consequences Chan the injec- conditions is best solved experimentally, for hydrogen-air mixtures of tion of small quantities of air inco hydrogen. scoichiomecric composition and for a hydrogen volume concent up Co 5C7., as in If Che deconacion process nevercheless occurs in some pare of che system, Ref. [7], che pre-deconation section in 100-500 mm diameter cubes can be the deconacion may be Cransmicced chrough the narrow orifices into che broaaer established for weak spark ignition. Figure 19 shows pare of che Length of neighbouring seccions. Figure 21 shows che dependence of che deconacicn Cransicional seccions from che cube diamecer caking into account daca from prcpagacion limics in an '' -air mixture on the initial pressure and diamecer Refs [7, 35], Ac more powerful iniciacion Levels, as for example jets of com- }i: the cube (tube diameter in mm is shown for the corresponding curves) _32]. bustion produces injecced chrough che orifice in che end of che cube, the pre- A similar result was obtained in Ref. [35j when aeconacion was observed at detonation section is shortened and relationship 2, shown in Fig. 19. can be ?3 = 0.1 MPa. established from data in Ref. [361. As can be seen, a reduction in cube diamecer leads to reduccion in che In concluding our review of che research conducced on deconacion limits daconacion limits. Whereas in cuoes with a diameter greater than 28 mm the for hydrogenous mixtures let us look ac che resulcs of Refs [37, 38] in which detonation Limics no Longer depend on che dimensions of che transverse cross- che authors examine the influence of initial temperature and pressure and also section, in a 2-mm-diamecer cube che deconacion limics are 20-^.2% H,- Outside the effect of the addicion of steam on che detonation limits. Summarizing che the L'-shaped region formed by che curve in Fig. 21, deconacion is noc possible. data in Refs [37, 38], we can state that the detonation limit for the H.-air The data in Fig. 21 actually show che cricical diamecer for detonacicn propa- mixture changes, little as pressure increases from 0.05 MPa co 0.4 MPa. These gacion d = d . However, disrupcion of deconacion (even a shorc-cerm data are shown in Fig. 20(a) where the broken lines (2) indicate air mixtures disrupcion) is also possible if che deconacion wave formed passes chrough an and che continuous lines (1) indicate oxygen mixtures. The detonation limits orifice with a diamecer less Chan cricical d = d . Figure 22 gives che of H2O2 mixtures range from 16.17. Hj Co 93.37. H7 in the mixcure. As tempera- dependence of che cricical diamecer for passage of che deconacion wave in che cure increases, che upper detonation limit for air and oxygen mixcures remains case of hydrogen-air mixtures ac P = 0.1 MPa. Figure 23 from Ref. [31j shows virtually unchanged (Fig. 20(b)) and che lower Limit falls from 19-207. ac che reduccion in d during transition from air to oxygen mixcures ac
T = 200 K. to l^-l^S ac T s 400 K. Reference [37] points out chac addicional p = 0.1 MPa. The daca for B = 3.76 apply Co air mixtures and for 0 < 3 < 3.76 steam in the mixture leads to a definite reduction of deconacion Limics which co 2H., • 0, - BN mixtures. Line L in Fig. 24 shows che dependence of d on is, however, of litcle practical significance (the experiments in Ref. [37] pressure in 2H- * 0, mixcures from Ref. [14], Line 2 indicacing che dependence were carried out uith steam contenc of 50-500 ppm). for air mixcures from Ref. [la], and Line 3 applying co measuremencs in 2H, - 0, mixcures from Ref. [29]. So far we have been discussing limitacions on che Comparison of the combustion and detonation Limits shows that these two occurrence of deconation in Cubes. The advances made in scudying deconacion v«ry important parameters of hydrogenous mixtures coincide for systems with have made ic possible Co evaluace also che cricical dimensions of semi-confined excess fuel. Ac the lower limit, flame propagation is possible for a wider 22 and unconfined H, * 0, Tuxcures and H? - n: -n wr.icr. jeconaciùn ac:ivat;:n •_ s ci transmission of :he explosion :r^m me comparcnent ;o anocner fiLIea witn
rtoc possibLe. H,, - air .Tiixcure is dealt wicn m Ret. 'lO]. ReCârence "LO] escablisnes cne dependence of orifice dimension linking the neignoourmg volumes on mixture Figure 25 shows che denender.ce of che critical thickness of the semi- compositicn and che D Locking oi .he blast orocesô Jichm trie LL^ICS of Jne concined layer of hydrogen—oxygen mixture on. composition from Sees .39, ^0 • . section. Figure 27 shows che critical DriUcd mansions and che bLast pressure "or unconfirmed layers the critical iimenà.ans vill oe close co 2d in me 4onor vo Lume ac wtii:n :r.e exolosion 15 nor. :ransmitted CD che nei^n- "igure 26 shows the dependence on tMe H,-air T.ixcure composition oc t.ie Dour; ng vo lume. I r me data m Fig. 2" are tomoared "Ji:.n tne data in Fig. -1, dimensions of the system m whicn detonation for cuoes is possiole •' 2 ) , tor Li emerges in at cne .; r 1 : ice dimension : 0rr2spor.es t^ cne :.nannel diamecer a L jng semi-confmed volume (2), and for unconfmed •/oL-j.-ne • I,1, ~ •* . ; wnicn cne letonatiûn cannot propagate. This iimension 13 considerably Less cnan ft The data given in Figs 2j and 26 were obtained ac ? =0.1 MPa. "or che critical orifice dimens ion d , necessary for detonation cransm155 ion co predictions at higher pressures, it should be considered thac detonacion Limits cne free volume. According to data in Ref. [lO], wnen 1 = 1, the oLast process
escaDlished at p =0.1 MPa apparently do noc change as pressure increases. in an H^-air Tiixture does noc pass mLD che neignoouring vo lume en rough an
However, increased pressure oiay reduce the critical initiation energy m accor- orifice d = ;.2 mm. In mixtures wich 1*-L5% H? per voLume, this dimension is
dance vith the decrease in self-ignition delav. Although it has already oeen i*0-50 mm, and for mixtures with 60-65% H1 the dimension is 30-50 mm. These •nencioned that for an H. * air fixture the dependence t = T(p) is not such thac -Jata complemenc and agree with experimental data in Ref. ] 23 ].
- - p , vitil increased pressure the conditions for deconacion activation in
terms of the energy of the mitiacor will apparenciy still oe less severe. -*. ?ar a::ie cers ot oressur? wave s -luring expLQSLons Jt hydrsggnous mixtures This reduction in the conditions for detonation activation at hign pressure -s \ series or papers r.ave recently appeared LH che worLc Litaracure on :ne indicated by data from Ref. [1.8] on the dependence o£ the critical diameter for problem of evaluating the affect of blase wave s from gaseous expLosions passage of the detonation from a tube into the volume for che case of 92 "^1,^2, t+2 t •*••!+, -.3,-0 an j «8 ] . In view 0 :' che face chat imtial dimensions of 2H2 . 02 : d" . p-°- . fjel-air mixture clouds are ;omoarable co aistacciS over which che evaluation Practical recommendations which emerge from research findings on quer.cnmg of blast eifecc cakes place and chac in some :ases chere is no deconacion at or régénération of deconation during passage through a constriction in che all, LC is usual co differentiate fuel-air .tiixture explosions from explosions cross-section are that in designing constrictions in che cross-sec:ion che DÊ explosive naceruls by ce £ erring to them as "non-ideal" explosions. The dimensions of blow-off discharge membranes should be chosen near to che cr-cical pressure field in an ideal expL^sion - wmch includes point explosions and dimensions. The main difficulty in prescribing specific channel dimensions explosive material explosions - has been studied in detail and modelling and is che strong dependence of che critical dimensions on che mixture composition calculation methods ha-.e been developed as well as methods of evaluating the in near-limit compositions. According to calculations and measurement s made destructive effect on the cargec. in view of the fact thac che initial con- in Kefs [14 and 31] for air mixtures where 3 = 1, d % 200 mm ac p =0.1 MPa; ditions tot shook wave format ion during g35eous explosion m terms ot pressure
and d * 100 mm at p * 1 MPa; and d - 10 mm at a =10 MPa. As it passes Level and geometrical iimensions are very different from those during explo- 0 • 0 sions of explosive material, different principles govern che change in through the orifices the detonation will decay on entering a larger volume. amplitude and wave pulse oi gaseous explosion chan ;n che case of explosions However, if the acceptor volume is confined by wails or is simply a larger j : axe los ivf» laCeridl. denci», comparison it tuel--air mixcure and explosions diameter cube then a second detonation activation is possible according co >r explosive Tiater id i and atcsmpr.3 *J rind equivalent ratios, particularly in Refs 7, 36] and Fig. 19. In order to Localize che bLast process within :ne L c ne area jf wave :.; :m,ic •. on , jc;~e L ITIPS I a ad co er~oneous conclusions ina t unit s of one section of a few connected compartment s the question ^i permissible recoinmendnr ions. dimensions of channels linking neighbouring volumes muse be soived. The proolem Lac us turn co factual data on measurements of snock wave parameters ^uring The blast energy Z = ^ï, wnere Y is tne quantity 3t rueL spenc in r'orrung :he
aetonation ot 2H-, • 0? fuel-air mixtures. These data are given :n ?.et's _l. + .l, fue L-a LC -nixcure, Q_ = I " "'CO
Less work has been done on measuremtinc of che pressure pulse in the --rave exsL^sicn ,lines 2,3 and -•. Lines 2, 3 and 4 are plotted at p^p^ = 3-7, after the explosion. Here it should be noted -hac che dependence of pulse .5 D ~: = 3.:. p,p "' = 6.9, 3 a,"L =• 0.35, y = 1.^, T , ~ 1.15. As is shown DV 1 1* -5 1 o 0 I n 1 pressure in che wave on che race of blase crans format ion given in Ref. "^o * cha disrate R * IR , there is no difference in pressure fields between a deccna- 0 appears erroneous. In accordance with che basic principles of shock wave tion expLosnn and cne suddenly exploding vessel. Ac the same blase energy, the formation [^7] and che calculated data in Rets [4,6 and 48] as well as expen- pulse Joes net depend on che cyue of explosion ana can be calculated by the mencai data '42], when chere are no energy losses the pressure oulse ir. the rac105 given above. wave should not depend on che blast transformation rate. In view of this, tne The method given in Ref. _•*" natces it possible co c ale u lace also che ; me resulcs of blast wave pulse decertninacion during rapid combuscion, given m jf irrival of the wave ac a specified point in soace, the phase and rarefaction Ref. . *7Jt should be checked and corrected. From che daca in Sets ,]•*!,-2, the cimes, cne pulse of compress ion and rarefaction pnases. i.e. co fully charac- connection can be shown between blase wave parameters in tne torn terize the explosion process. Smce all the ratios m Ref. :. •*] are written in Sacns Jimensionless values, :he -necnoo used in Ref. ' •* ] is fairly universal and I = (0.026 - O.OO2)ip°*388 E°'33 can be appLied co caLculacions at any va Lues of p , p, a^ . a , Y , Y, where ip > 0.05 MPa.
Here I is in MPa.ms, up is in MPa. E is the blase energy in kcal. when p < 0.05 MPa
5 33 23 I = (0.022 - 0.0O2)Ap°' £°- en oo S 00 S in in *t CM
on vo UJ m M n i-i M
o in o in Q in 10 en o CM co te co vp ra ^ »o f- PJ «* M 327 8 331 9 Si CO
co TO
CM 17, 3 13, 7
M
o I 271 3 251 9 CM 268 8
co i> in cr» co t> co vo o CO \O co O in 39 3 M H CM CM CM CO AD 39 8 on
in o o vo co o .* in vo
O O O O O O o o c- oo CO CO CO O n CM CO IA' M f- CM H 00 t> co co cnooooQo-* S t> tn CM to o- CM MycovoinOon o o in CM t-- CM co f- •* o oo CM m in CM CMCOCOCOtHI—II-HCM H CM CM H CO CO CO in r- oo cT cT O vo —- covoncoil-tontovo co c*- cr» CM oor-ovDt^-o^voc--in Mc •X) CO £"- m OJ yo CM VD Ta b fixture Lower Lunic Upper Lim LIN, 3.O.S., McFARLANE, a.. Laminar burning velocities ot H,-air H^-air-àtaam flames, Comoust . F Lame, •* 5 i L ^ 3 3 '• 59. a2f(0,2I02+O,?9He) 10% H-, 55% 5T3EHI.0W, R.A., ADAMCZIK, A.A., et al.. The Didst Jave generated jv 7 ~%. •_ H2+(0,2I02+O,79C02) 50% spherical flames, Comoust. Flame, 35 (19~Ç; 197. H2+(0,2I02+0,79 //t 65% NETTLETCN, M.A., YOUNG. D.M., The propagation -A -.lames v^ gases -n '.arje- •iiameter pipes (?roc. 3rd Int. 5ymp. Europ. Comcust-; • 1973) 717. 5TREHL0W, R.A., LfnconfmeU vapour cloud explosions. 1-th Int. Svmp. Combust-, ComDust. tnst. '1972) 1139. Taole 6. Detonacion Limits of iniciacors with differenc energies 3UI3AO, CM., KNYSTAb'TAS. S., LEE, J.H., 3ENEDICX. ». , 3ORMAN, M.. Hydrogen-air detonations, 19th Int. 5ymp. on "omouscion, Como. Insc. (1982) 583. Veighcof charge Limits Source THI3AL'LT, P., LIU, Y.K., CttAN, C, LEE, J.K., KNYSTAUTAS, R. , GUI SAO, Z. . n 0,95+1,25 28+37 L"30] Transmission of an explosion through an orifice, I9tn Inc. 3ymD. on 10 g 0,64*2,10 • 19,2+63 Combust., Combust. Ins;, i 1982 ) 599. 100 g 0,4.8+2,58 I*,4+76 £307 LEE, J.H., MOEN, I.Û.. Mecnanism of transition from oefLagrat:on to 30 g 0,5*2,11 15*63,5 detonacion in vapour cloud explosions, Progr. Energy Ccmb. Scl. 6 (1980) 354. in pipe 3 305 mm 0,65*1,93 19,6*58 VESTBROOK, C.K., Hydrogen oxidacion kinetics m gaseous detonation, Combust. Sci. Technol. ^9 (1982) 67. in pipe 0 20 non WESTBR0CK, C.K., Chemical 'icmecvcs of hydrccaroon oxication m gaseous detonations, Cornoust. rlame, io 2 (1982) 191 REFERENCES VESTBROOK, C.K., URTIEW, ?.A., Chemical kinetics prediction of critical [l] L'YUKS, B., EL'BA, G., Gorenie, plamya i vzryv v gazakh CCombuscvon, parameters in gaseous detonation, 19th Inc. Symp. Combust., Camoust. Inst. flames and explosion in gases), Mir, Moscow (1963) 593. (1982) 615. [2] LEE, J.H., GUIRAO, CM., CHIN, K.W., BACH, G.G., Blase effects from [15] 0RAN, E.5., BORIS, J.P., YOUNG, T.R., c'-ANICAN, M. , 3URKS, T., PICONE, M. , vapour cloud explosion. Loss Prevencion, l± (1977) 59. Simulations of gas phase detonations, NRL-MR-4255 (lQ8D- L3j LEE, J.H., GUIRAO, C.ii., Gas dynamic effects of Ease exochermic reactions, [16] HUBER, P.'»1., SCHEXNAYDER, C.J., McCLINTON, C.R., Criteria for self- "Fast Reactions in Energecic Systems", D. Riedel (1981) 2i5. ignition of supersonic H,-air mixtures, NASA TR-li57 (1979). 25 26 3ORI3OV, A.A.. L33A.N'. 5.M.. Predelv leconacsv jglevoooroano-vozcjsnnvtr .30, 3UI.L. 2. C. , Concentration limits to '.ne initiation •>: jnc:n:inec :er. :ni- sroese• v :r'jDakh Detonacion Limits jf nvdrocaroon-a L r m«cures in i^:: 31^. Fiz. Goren. Vzryva 5 c?T"i -;9. I v i " V. ' •••. : 3TAL TA5, S., 1.EE. .'.-I., Jl'lRAC. :.M., The :n:ical :jO e ;une:ar :.-r 13 TCPCHIYAN, H.E., IL'YANITSKY. V.YU., 7ASIL. EV, A.A., .'i.vni; njcnai -r; Jet2nat:m ;nli^e in n.vdrociroon-s : r Tiixtir^, ComDust. "lame. -3 ,'.- tsmperaturv na paramecrv gazovov 2ec:nac^".". 'Ir.fljence ;: initia! :?nce: 3^. M^.l'ï'Jl. H., Preprint it 15tr. Jaoanese "SVTID. vn ~.:mou5r-"3n l^Bl1 oS. :ure an gas detonation parameters . riz. Goren. Vzrvvn - 1J *9 1—. 33 v 3RETCN, _'. , ^ec.Tercres 5ur la aet^nacion 3es -ne-anges g,iie^x ^esearc.T STRAUS, W.A. , SCOTT, J.N., Experimental invest:^acion 3: :ne :ecDna:.:n mtc :r.R detonation ot cas -nixtires.. Thesis, Fac. 5'iidnce. Jniv. 1 properties on H^-O^ and H^—NO-, mixtures, Comousc. Flame. ^ I *"'. 'lane ' 1-31 : . Kl'RYLO, j., OPPENHEIM, A.K., Evaluation of streets : : :e;:-.a::jn in 3 .3^, KCGARKC, 5. M. . ^EL'-CVUK, Ya.3-, D ietor.atsy gazov/KT. smesdi !n -et^n^- spherical bomb, MPI-3ench!: 23-79 (1973) IOC. Tior of gaseous -ruxtures y 3AN 'JSSR OJ 5 'l^^d'1 533. LINO, CD., VHITSON, J.C., Explosion hazards issoc:ic;c -i-_n so: ',5 •• : .3:\ 3C3I3CV. A.A., CEL'"A.ND. 3.E. LaSAN', 5.A.. MAILKOV. A.E., WAGNER, H.G., FlammenDescnleunigung-sentraLes Pro ''Res*?3rcr. on :uel-air fixture detonacion limLts in smooth âne rough t-ibes , von Explosionen, ?T3 Mitt. 91 _ 11931J 2^. ''lima. ru. n U932) i^i. Q '23' JCST, W. , WAGNER, H.G., Influence of various paramec-sri --n initiation, la] 3ARTKJECHT. 'J. , Explosions. Springer /erlag, 3emn--ieu Yorx . 1 31 ' . staDility and Limits of detonation, AFOSR 73-3jâ7 ::>'-x, ArCSR 79-Ci: "A '','] GCRCON, \.Z.. MOOPA-.IAN', A.J.. HARPER. S.A.. Lini: ino som aspects 0: (I9a:i. AFOSR 73-25M •. ;373;. J, - " , le c ::ia: 1011 s . "th tr.c. 5y?.p. lompust.. Londor 1-:- "32. DORGE, X.J., PANGRITZ, 3., .'AGNER, H.G., Expermentj ;r, velocity ausne-- "38" 3ELLES, ?.£.. Detcr.aDllitv and ;r.enical kinetics. Prîiiotion 0: Limits 1 tation of spherical flames sy grids. Acta -.strcn. £ 11-11 .l ?^; ICc*. 01 detonaci.ity of r.ydr;gen, 7:h Int. Sy*np. Conoust-, Lo^ccn ,1959) 7-:. '25' «CLANSK1, P., WOJCICK1, S., On the iiec.ianism j: ir.: 1 JÏ-.L» 0: 3o=;jci« i" .3-; D,i3rRA. =...<.. MICHCLLS . J.A.. ^.CRRISCN, 3., T'-e '."Lljence of a ton- tne flame propagation, Archivuni combust., 1, 1-Z t1931 ; 69. pressioLe baundarv on :ne propagation of gaatfous ce1. înacion, ICth _nc. Î26] KAUFFMAN, C.W., NICHOLLS, J.A., Gaseous detor.ative -rac:^! jf porous 3v-.ro. Coroo'is:. , Cotnoust. Inst. '.19Ô5'' 8 i 7 . materials for enhanced fossil fuel utilization and recoverv. IM-016r::: 3-r ^0 _ \L.iM3. T.l;., Co --Jeak detonation Javes exist' AIAA J. ^o_ 10 J19~6) 1035. U981). "-!" KOGA/KC. S.M., ADL'S"KIN, J.'!. , LYAMIN, \.Z., '.si.^ovanie ?fencheskav [27] GOREV, V.A.. MIROSHNIKOV, 5.N., L.'skoryayusne»sva gorenie v 43 ;ovvkn aetonatsy jazo"ykn smesei 'Research on spherical detonation of gaseous ob'emakh (Accelerating comoustion in gaseous volumes, , Xh'.m. Fiz. mixtures), riz. Ocren. Vzryva 2 (1965) 22. à (1982) 854. '.;] DES30RDES. D.. MA.NSCN, N., 3R0SSARD, J., Explosion dans l'air de charges [28] ATKINSON, K.. BULL, D.C., SHUFF, ?.J., Initiation of soher-.caL detonaci-'.' çphenques non confinés de mélanges réactif gazeux (Expljsion m air of in H,-air, Combust. Flame, 39 (1980; 287. spnencal unconfined charges of reactive gas mixtures), Acta Astron. 5 L29] MATSUI, H. , LEE, J.H., On the measure of the relative de-onatun harar'is 10-11 (1978) 1009. of gaseous-fuel-oxygen and air mixtures. 17th Inc. Synp. :=nbusc.. -;' 3A3CRA, E.K.. SAGi-AND. '<.«., A study of heterogeneous detonacion. Combust, inst. 1976) 1269. -*•; SEIDER, R. , OTTVAÏ. T., KNIGHT, H.T., An unconfined Urge-volume :i,-air exploMon, Pyrodynarai.es 2 2 ; 1965) 2^9. -*5 ] L£YER, J.C., Pressure effects generated ay ~ne explosion in "he atmosphere of Hydrocarbons and au mixtures, Rev. 3en. therm. 2^3 11982) 191. Ci — -o _ GCREV, V. A. , Sravnenie vozdushnykn >/o In -DC raznykh iscochnikov ; Cornoari-âon of air waves from various sources), FLZ. Goren. Vsryva 1 ^1982) Q4. -cob \U7] 3ELYAEV, A.F., SADCVSKY, M.A. , 0 prirode fugasnovo '. bnzantnovo ie1.scv1.7a 0.C? i^ vzryva ^On cne nacure of fougasse and brisance in exptosLons), Cc 11. "Fizika Vsry/a", Acad. 5ci. USSR, Moscow 11952). _ -*8 ] FTSHBURN, 3.0- t Some aspects of blase from Cue I-air explosions , Ac" a Ascron. 3 10-11 (1976) 106.9. -uC- icu goo .50c FIG. 4. I i FIG. 2. \r\h T L i/il t,i - L^ J • """7'M I //-> 5 // ! 5.0 N H •2.5 '1'W\ 3 /•J, , \/ 2. AW ! _ A 0 2o *5ô~Tû Jo lu 3o 50 ^/i p 27 FIG. 1.*. FIG. 1. fa FIG. 3. FIG. 5. O a Sri \ 6 —— — N -, 5 A / / / / o 1 \ (\ \ 3 — -—- Ci.) o o s ^C 1 \ r •" ( \ \ \ o o \ •—1 \ — \ e ?. J • L \ a -^— " • 3 y \ \ t- \ o \ X \ \ L_> ) LU —-—• o 1 L_ o '•a UJ rt ni ni n ' 7\ 7\ 3C ?. X. o o ~T5' LJ' Lu o o o A o 1 cr 7 / o O in o ft - O CO -9 T o O ô - — — - — — - - — -- O f. O O •j 1 E LJ —1^__ — o o o \ c-1 o r* o o o O In r\i ^ ^ —•—. \ ^^ — O \ o ,'l 1 \J f" 3" rA CJ LJ . O C> o —-^ —= y o r il — / [r O E Vr. T : - o C\J — — o 0> l\ j — -* — IP O V O ~T5 o o d O —-—, c 1 ' a o' P- l / —— S ,; i- o -- o * • ••* . of \ — - 1 -< — -'i — f-— % A, 'm —•- - //} / .... // / / ) J• ~" \ f / ! / / / o / CO 1 s - III 1 • 1 O 1 1 1 / - / f / / y- — — - - / > j _ / / ( » —/ /_ * •y 7- '•ft / -U.-O I i i THEORETICAL EVALUATION OF In che Literature there are works on cneoretical calculation of CRITICAL GAS LAYER THICKNESS IN RELATION critical gas layer thickness in relation to sceady--5cate propagation TO DETONATION WAVE PROPAGATION of che deconacion wave (Fig. 1). For example, m Re£- i5], estimates of the critical gas layer thickness, & , were made. These, however, Yu.N. SHEBEKO, A.Ya. KOROL'CHENKO are essentially qua Iicative since the expression for 6 obtained All-Union Fire Fighting Scientific Research Institute, includes che width ot che deconac ion wave react ion zone, which can Moscow, hardly be defined a priori. In Re£. [6] the authors devise a numerical Union of Soviet Socialist Republics model for detonation wave propagation in gaseous mixtures of a finite chickness which they use to calculate the 5 of oxyhydrogen gas. Translated from Russian Reference [7] extends model [6] co cover oxyhydrogen gas diluted wich nitrogen. These references study in detail gas.-dynamic processes and Abstract reveal che pulsaCing nature of the detonacion wave, but give only a very rough description at che chemica I kinetic processes* The present The study of detonation wave propagation in gas volumes of finite work aims co evaluate ô with a more realistic descript!on of heat thickness is important for safeguarding various technological processes from release processes resulting from the chemical reaction. explosions. Available literature includes theoretical calculation of critical gas layer thickness in relation to steady-state propagation of che The Arrhenius relationship between reaction rate and temperature deformation wave, and studies in detail gas-dynamic processes and reveals is often used in semiempirical descriptions of chemical processes in the pulsating nature of the detonation wave, but gives only a very rough detonacion waves. In order co decide whether we can use this relationship description of che chemical kinetic processes. This paper aims to evaluate for calculating 6 ^, l&t us qualicacîvely analyse che dynamics of detona- the critical gas layer thickness with a more realistic description of heat cion wave propagation in a gas layer of finite thickness. The value of <5 release processes resulting from the chemical reaction. A numerical is decermined by two competitive processes: compression in the shock simulation of two-dimensional detonation wave propagation Ï3 given in a wave and cooling as a result of lateral discharge. When en is nappensi gaseous layer of finite thickness for oxyhydrogen and for a $coicniomeccic the detonation wave can propagate through the gas Layer if che chemical hydrogen-air mixture. reaction occurs before the mixture cools, i.e. if the characteristic time for the etiemical reaction ib less than the characteristic time for dynamic cooling. The chemical reaction time is almost entirely determined by the t ime of induct ion of spontaneous combust ion in the The study of deconacion wave propagation in gas volumes of finite shock waves which, as is well known, can be found fairly accurately thickness has both theoretical and practical interest for safeguarding using che Arrhenius relationship- Hence, to calculate 6 we can use various technological processes, such as in the case of nuclear power a model for che chemical processes in the form ot the Arrhenius exponent, plants, from explosions. One of the most interesting qualitative but che empirica1 paramecers it contains must be correctly selected. characteristics of detonation processes in charges of finite dimensions The expression tor the change in concentrât ion of the combustible is che occurrence of stall phenomena caused by lateral discharge ot mixture is detonation products, which develops into a pulsating regime and eventu- ally determines the critical size of the charge. This phenomenon has 33 been disc°vered experimentally for liquids [l.2] and for gaseous mixtures 34 where C is the relative concentration of the mixture (O$C^1) conducted by the authors of Ref. [^], where 5 was found to exceed 1.0 cm. 3, T are che density and temperature of the gas, Consequently, che results of Refs [6,7] are essentially qualitative A, E are the temperature-independent factor and the activation energy. since the critical thickness of the gas Layer is dependent on the degree Under tnese conditions, che amount of energy released during the chemical to which the oxyhydrogen is diluted with nitrogen. Under these conditions reaction is given by the ratio tne 6 is underestimated and the velocity of the detonation wave D, which in Ref. [6], for oxyhydrogen amounts to 3.25 km s , is over- Cx ™ ^\O I ^""""0/ s (2) estimated. The proximity of values D and 0 for parameters A, E, Q Q O where Q is che effective thermal efficiency of the reaction used in the presenc work suggests thac calculation of 6 would yield a value quite close to a real one. C is che initial concentracion of the combustible mixture, o To calculate two-dimensional detonation wave propagation in the It is best to take parameters A and £ in Eq. (1) from an experiment geometry shown in Fig. 1 use was made of Lagrange's gas- dynamic equations on spontaneous combustion in shock waves (for example from Ref. [8]) and the chemical kinetic equation in the form of Eq. (1). Calculations and select Q in terms of its conformity to the detonation wave paraneter were carried out uich an ES-1033 computer. The program was written for which there is optimisation and which is very important for accurate in PLI algorithmic language. The computing scheme "predictor-corrector" calculation of 5 of Ref. [lOj was used. The typical longitudinal and lateral dimensions From che above analysis it follows chat we can use the propagation of the counting cell were 0.1 cm. The size of step for integration velocity of che detonation wave, which to some extent determines the with respect to time = 2 x 10 s. Development of the detonation wave chemical reaction time, as such a parameter. Parameter Q^ can be was observed until = 25-30 us. The time taken to calculate one variant selected by correlating the calculation value of the velocity of the with the ES-1033 was 8-10 hours. Two gaseous mixtures were studied: detonation wave in the centre of the gas layer thickness s » 5 oxyhydrogen C2H- * 0-) and a stoichiometric hydrogen-air mixture, which cr and the Chapman-Jouguet velocity of an ideal unidimensional detonation represents a special case of oxyhydrogen diluted with nitrogen. Allowance wave D^. Values of DQ for different gaseous mixtures are presented was made fcr the presence of nitrogen in the mixture in the following in Réf. [9J. wav . first the initial gas density and the therma 1 react ion efficiency The values of the parameters A. E, Q used in the present work per unit mass o£ rhe initial combustible mixture ^as varied, and second, were the following: A = 3 x 1011 s"1, g"1. cm3 the same chemical kinetic eauation as for undiluted oxyhydrogen was E = 84 kJ mole"1 used, y.it its initial concentration was correspondingly reduced. 3 Qo = 3.5 x 10 i The initial temperature was assumed to be 300 K. and the initial Under these conditions, the detonation wave velocity along the gas pressure 100 k.Pa. The -idiabatic exponent y was assumed to be 1.4. layer of oxyhydrogen of large thickness was 3.0 km s"1, whilst D . The detonation wave was excited by instantaneous energy release in in accordance with Ref. [9], was exactly 2.^8 km s . Parameter Ç . an area of 1 cm in Length and with thickness of 4- The energy release used in Refs [6,7] was derived by comparing 5 (0.5 cm), found lay between 10 and 120 J, and ensured initiation or the detonation experimentally in Ref. [3], with the calculated results. However, wave which then, according to 5, cither changed co the steady scate in experiments carried out in Ref. [3] the combustible mixture and or attenuated.. inert medium were separated by a thia 10 um nitrocellulose film, which Figure 2 shows typical longitudinal profiles tot hydrodynamic affected detonation wave propagation. More accurate experiments verp detonation wave parameters under steady-state condit ions. At the chemical peak, che reaction race is close to maximum. The density large thickness ot 6. The presenc paper defines che range of values of che gas reaches a peak before the scare of che intensive chemical of & over which the sizes of 5 for oxyhydrogen and che stoichiometric reaction, which begins as a result of adiabatic heading of che cambuseibLe hydrogen and air mixcure are found. As che criterion for a non-attenuating mixture by che shock wave. The temperature reaches a maximum in che detonation wave, the velocity of ics propagation 0 was assumed co be region of the chemical peak. constant over long periods of cirae. Calculations showed chat in the case of oxyhydrogen, there is attenuation at 6 * 1.35 cm and steady- Figure 3 shows typical transverse profiles for temperature and state propagation when 6 = 1.30 cm. In the case of stoichtometric density. When temperature T and density p behind che detonation wave hydrogen-air mixcure, 5 is in che range 9—12 cm. front (Fig. 3b) decrease monotonically to che periphery of the gas layer as a result of lateral discharge, the value of p at che wave It is interesting co compare the calculated results with the experiment front immediately reaches a maximum- This is due co non-uniformicy for oxyhydrogen. According co Ref. [4], for this mixture 6 > 1 cm of detonation wave propagation through the cross-section of the gas (as applied co che geometry used by us), i.e. the calculated results layer where gas regions in the centre of che layer participate initially do not contradict experimental data and chis suggescs that the detonacion in the reaction, while peripheral regions of the gaseous mixture are propagacion model in a gas layer of finice chickness can be used to only subsequently involved in che chemical transformat ion. At the determine che value of 5 end of che reaction, discharge in the cencral regions of che gas may Thus, we have given in this paper a numerical simulation of two- be both sideways and backwards, depending how much che pressure behind dimensional detonation wave propagation in a gaseous layer of finice che front has already fallen. At the same cime, nearer to the periphery thickness for oxyhydrogen and for a scoichiomecric hydrogen-air mixture. of the layer, the shock wave, somewhat delayed in comparison to the It shows that, depending on the thickness of che gas layer, the detonation .wave at che centre of che layer, scill compresses che gas adiabacically, wave may attenuate or be capabla ofsteady-scace propagacion. Here a which is what increases the density in this region. scrong pulsacing regime is possible. The inicial chickness of the gas layer foroxyhydrogen and a stoichiomecnchydrogen-air mixture is found. As in Ref. [6] we found non-steady-state phenomena during detonation wave propagation wich values of 5 near co 5 . These pheonomena are as follows. In che initial scage of detonacion propagation, a fairly REFERENCES long quasi-steady-stace system occurs, characterized by a slow reduce ion in che plane of the detonation wave front. Ac a certain moment m [l] DREMIN A.N., SAVROV S.V. ec al-. Detonation waves in condensed cime, a new zone of incense reaction, nearer co the edge of che charge media, Moscow, Nauka Press (1970) (in Russian). in the shock compressed gas occurs alongside the cencral zone of energy [2] TANAKA K., HIKITA T., Acca Astronautics 3, (1976) 1005. (in English). release. As this occurs, lateral discharge increases sharply and this leads to quenching of the reaction at the edge of the charge. This [3] DABORA E.K.. , NICHOLLS I.A., MORRISOV R.B. , 10th International effect may be repeated during further wave propagation i.e. a detonacion Symposium on Combustion. Pittsburgh: The Combustion Insticute pulsating regime is sec up which is also observed in the experiment (1965) 817 (in English). described in Ref. [4]. A detailed explanation of chis effect can be [4] VASILYEV A.A., Fiz. Goreniya Vzryva IS 3 (1982) 98. found in Ref. [6]. [5] WILLIAMS F.A., Combust. Flame 2£ 3 (l9"6) 403 (in English). As has already been mentioned, che detonation wave is capable [6] IVAN0V M.F., FORTOV V.E., 3ORISOV A.A., Fiz. Goreniya Vzryva 35 of steady-state propagation in a gaseous layer only at a relacively 17 3 (1981) 108. [7] IVANOV M.F., SHEBEKO Yu.N., LUK•YANCHUK L.N., in: Fire and explosion hazards of substances and materials. VNIIPODZEMGAZ, Moscow (1982) 33 (in Russian). [8] DIMITROV V.I., Simple Kinetics, Novosibirsk, Nauka Press (1982) (in Russian). [9] NIKOLAEV Yu.A., TOPCHIYAN M.E., Fl2. Goreniya Vzryva 12 3 (1977) 393. [10] ZHAROVTSEV V.V., Zh. Vychist. Mac. Mat. Fiz. (USSR Computational Mathematics and Mathemcical Physics (USA)) 12 5 (1977) 1320. \ \ 1 • S V z 3 1' Diagram of two-dimensional detonation wave propagation in Fig. 2. Typical longitudinal profiles of hydrodynatnic parameters a gas layer. D - wave velocity; XY - co-ordinate axes; in the centre of the layer (X = 0). Mixture TA^ * 0^, i = 1.35 cm. S - thickness of layer Initiation energy 15 J. Time t = 18 us. 1 - inert médium; 2 - combustible mixture; 3 - strong diathermic 1 - concentration C; 2 - P/PQ (P - density, OQ - initial density); wall; 4 - wave front. 3 - temperature T; 4 - P/Po (P - pressure, Po - initial pressure). DETERMINATION OF FLAME PROPAGATION LIMITS IN STOICHîOMETRIC OXYHYDROGEN MIXTURES WITH STEAM S.M. KOGARKO, A.G. LYAMIN, O.E. POPOV, A.Yu. KUSHARIN Institute of Chemical Physics of the Academy of Sciences of the USSR, Moscow A.V. DUBROV1N Î.V. Kurchatov Atomic Energy Institute, Moscow Union of Soviet Socialist Republics Translated from Russian Abstract Experiments are described and results presented oa combustion processes of mixtures of hydrogen, oxygen and steam under pressures up to 16 atm and temperatures up to 250 C. The effects of temperature (at constanc pressure) and of pressure (at constant temperature) on concentration limits are shown. During operation of water-cooled nuclear reactors, water in the reactor core 3. Typical transverse profiles for hydrodynamic parameters. often dissociates into oxyhydrogen as a result of radiolysis. The oxyhydrogen Mixture 2H2 + 0^. t = 1.35 cm. Initiation energy 15 J. may accumulace in closed spaces. In order to ensure safety from explosions it Tim* t « 18 us. was necessary to study the comDustion processes of mixtures consisting of a-Y=6.2cm(at the wave front) hydrogen and oxygen and steam. These studies had to be conducted using pressure * - Y * 5.8 cm (behind the wave front) and temperature parameters equivalent to operating reactor conditions. 1 - P/PQ (e - density, Po - Initial density) No information of this kind has yet appeared in the literature. It vas 2 - temperature T. impossible to use the generally accepted method of determining flame propagation limits for gaseous mixtures, since the standard method defines the limits only 37 at atmospheric pressure. Moreover, in the literature chere is no unambiguous *jo interpretation of "flame propagation 1 i rates", since rhe 1 irait s are noc always The experiment was conducted in the fol lowing way: the first component fed determined for the same degree of burn-up ot the mixture in the experimental facility. into the reaction vessel was water. When the required temperature of the reaction vessel walls was reached, the steam pressure was measured, which corresponded to For such a crucial power facility as a nuclear reactor, one cannot allow any the tables ot data in Refs [l, 2]. Next, calculated quantities of hydrogen and increase in pressure in a closed space, that is to say even any local reaction oxygen were successively introduced, via measuring tanks, into the 'eaction between hydrogen and oxygen with a recot dab 1e increase in pressure. vessel tnrougn the apertures m the mixing pipe at the critical rate which In view of tnese considerations, we deciJe-i. -n iur research, to assume the guaranteed rapid mixing of components and ensured a homogeneous mixture. hydrogen and oxygen concent to be maximum in the case of a mixture wich steam Analysis of errors in the method of determining the concentration of in wmch burn-out of a wire by an electric current cid not cause a recordable components showed that in all the experiments the error in determining the steam increase in pressure m a closed space, i.e. .1? = 0. mass did not exceed 5%, while the relative errors in concentrât ion limits were 3-o".. Since steam under the experimental conditions (pressure up to 16 atm and temperature up to 250 C) is not an ideal gas, it seemed advisable to determine Figure 2 shows the concentration limits of oxyhydrogen nixed with steam at noc the volume, but the mass concentration, of oxyhydrogen mixed with steam. •a con cane temperature ot 200 C for a few overall pressure vaLues between <* and 20 atm. The white Cots show cne concentrations below the limit (non-comoustible In order to ascertain the possible effect of suspended minute drops of liquid niixrure anc no recorded increase in oressure); the black dots show mixtures in jater on the flame propagation 1 irait, experiments were conducted with ooth super- which the increase m pressure dur ma combustion is greater than 2. The half-oljc* nea Led and wien satura ted steam. dots s ho--' compos it îcns for which the increase in pressure lies between 0.1 and 2.0. Study of flame propagation in such mixtures with increased initial pressures The curve divides the areas where reaction in the space occurred from those and temperatures involves certain difficulties. An experimental facility had co where there was no reaction. Th small part of the volume Is burnt. Enrichment of Che -nixture witn oxyhydrogen operating conditions m power facilities. substantially accelerates the combustion process and intensifies che rise m The measurements of concentrât ion limit obtained can be used co analyse the pressure. process of emergency cooling of a reactor if, during operation, the mass concentra- Figure 4 shows the relationship oetween concentration limit and overall tion is maintained below a limit corresponding to a temperature of 200"C. As an pressure as it changes between 1 and 20 a cm. The curve also divides the areas of example let us take four (mass) concentration levé Is from 2 to 7% and ascertain concentration where reaction occurs in the space vrith appreciable increase in up to what temperature the reactor can be cooled without the oxyhydrogen concentra- preisurt (combustible mixture) from those where there is no increase in 'ion reacning the limit corresponding co r.ne temee rature level. These re su 1 : s pressure (non-combustible mixture). The increase in initial pressure in the are shown in Table II. system with saturated steam systematically lowers the concentration limit. More Analysis of the cable data shows that when :ne (mass) concentration is specifically, when che pressure increases by a factor of 20 the concentrât ion maintained from 3 to 57. during operation, the -nixture at the time of emergency i imi: decreases by a factor of 1.3. bhut-down and cool ing becomes combust ible onlv a: temperatures below 130 anc It should be noted that the concentration limits of oxyhydrogen mixed w:cn '. "0 C, respectively. saturated steam shown in Fig. U are for different températures. 0->e snould remtmoer that no concentration of Dxynydrogen in steam is The systematic fall m concentration limit of oxyhydrogen when temperature 3DsoIULCly safe, since wnsn steam condenses, zr.e -ixrjre always becomes comDustID Le. and pressure increases, which was observed in experiments both for dry and It is only a quest ion at wnat temperature. The :ef.peracure limit depends on che saturated steam, can be explained, according to the Zeldovich-Frank-Kamenetsky initial concentration, as is shown in the table. theory [3, 4], by the reduction in relative losses which are proportional to a/U , where a is the thermal diffusion coefficient and U is th?1 norma1 flame n n velocity. Given constant pressure and a rise m Initial temperature, the chemicsi Conelusions reaction rate ( flame propagation) increases. Given constant temperature and i rise in pressure (to 20 atm), the thermal diffusion coefficient of steam decreases 1. A procedure is developed and measurements made for concentration limits of wich pressure and the rate of change in combustion rate with pressure in mixtures oxyhydrcgen mixed with steam at pressures of 1-20 atm and temperatures of up near to the limit is close to U - P* " . The decrease in relative heat losses, to 200°C. n 2. For saturated steam at 200 C it was found that the mass concentration limit when the initial pressure rises, is also furthered by the fact chat the volume of oxyhydrcgen was (12." ^ 0.5)7.. As the temperature of the saturated steam density of the energy for mixtures of the same concentration increases linearly falls, rhe concentration limit increases to (16.7 ^ 0.6)% at 100 C. with pressure. The measurements of concentration limits of oxyhydrogen in dry and saturated 3. The effect of temperature on concentration limits is explained. At constant steam show that the difference between these is not very great. pressure, the mass concentration limit of oxyhydrogen in superheated steam is For actual power facilities in operation, one can always expect drops of water reduced- The pattern of reduction in the range of 140-250°C is nearly linear and to be present in the space; they can only impair the conditions of flame propagation the decrease in concentration limits is 0.27. for every ten degrees. since heat will be lost as a result of the drops heating up and evaporating, 4. It is bhown that at a constant temperature of 200 C, the mass concentration thus increasing the reflux action of the steam. limit of oxyhydrogen in steam decreases from 15.6 to 12.77. when the pressure 39 is increased from <* to 20 atm. Tab Le l (1) (2) (3) (5; T> / T3 P0,ai " 3' ' 16,6 3,8 — — 16,8 3,9 4,5 1,2 3,0 17,1 3,9 5,5 1,7 1,3 17,5 4,1 3,0 2,0 1,5 19,2 4,2 8,7 2,1 1,0 21,0 M 20 4,5 0,6 24,1 4,5 25 5,7 0,2 Table 11 Concencracion C : Normal mass concencracion limic. X 2DO 2 3 5 7 12,7 «• 190 2,5 3,7 0 .1 8,5 13,6 180 ' "3,0 5,1 7,3 10,7 15,5 1. Diagram of a vertical faciLicy for measuring the concentration limic 170 3,3 5,7 3,5 13,3 15,0 of oxyhydrogen in steam. 160 4,3 7,2 12 ,0 - 15,5 I. Oxygen cylinder. 2. Reaction vessel pressure gauge. 3. Measuring 150. €.2 9,3 15 ,4 - 15,9 tank tor oxygen and uacer. 4. Oxygen pressure gauge. 5. Hydrogen 140 . -S,0 12,0 - - IS ,3 pressure gauge. 6. Measuring tank for hydrogen. 7. Hydrogen cylinder. 130 10 ;5 15,8 - - 16,5 8. Thermocouples. 9. Mixing pipe. 10. Reaction vessel casing. IZO 14,0 -" — 15,6 II. Incendiary coil. 12. Water. 13. Additional electric heater. 14. Electric heater men heat insulated casing. 15. Vacuum pump. 16. Loop oscillograph. 17. ID-2 amplifier. 18. DDI-21 pressure gauge. t'ZOOC c_x m 15 • —«— » •ë 11.7 - 0,5 y2 - 10 o 10 - a 6 J L & .0 iO -12 ^ 16 18 20 22 p 1 f 1 1 am /. 5 10 15 • 20 nom Fig. 2. Relationship between mass concentration limit for oxyhydrogen mixed Fig. 4. Mass concentration limits of oxyhydrogen mixed with saturated sceam at with steam and pressure at constant temperature of 200 C- different pressures. C 7 nap 20 r REFERENCES 1. VIKALOVICH, M.P., Thermodynamic properties of water and steam (1968). 2. ALL-UNION HEAT ENGINEERING INSTITUTE, Tables of thernodynamic properties iO of water and steam VTI (1958). 3. ZELDOVICH, Ya.B., Theory of gaseous combustion and detonation, USSR Academy of Sciences (1944). 200 250 t C 3. Relationship between mass concentration limit of oxyhydrogen mixed 4. FRANK-KAMENETSKY, D.A., Diffusion and heat transfer in chemical kinetics, 41 with superheated steam and temperature at a constant pressure of (*• atm. "Nauka" Press (1967). 42 HYDROGEN PRODUCTION IN A PWR allowing enough time for a recombtner to be installed. Hydrogen solubility in water is not sufficient to inmbit the radiolysis of sump water or to induce an important degassing in DURING LOCA the upper portion of the containment. Two types of hypothetical accidents involving a larger oxidation of fuel cladding P. CASSETTE have been considered : a LOCA by large primary breaK vith less of electrical supply and a LOCA with small break, subsequent to the total loss of electrical supply. In both cases, Institut de protection et de sûreté nucléaire, the percentage of the Zr-H^O reaction is evaluated with a simple thermal hycraulio Département d'analyse de sûreté, codes, arieçua;!/ modified : BCIL.<. Commissariat à l'énergie atomique, In this accident type, ^ater radiolysis and aluminium corrosion are negli^iDle Fontenay-aux-Roses, France sources of hydrogen. The core melting limits the zirconium oxidation to about 6ÛSD: this represents a mass oK^fi Among the containment metals that can be corroded by the aspersion solution, only aluminium is present in significant amount. The reaction kinetics depends en the temperature according to a law of the Arrhenius type and the parameters which are used concern pure aluminium, due to the lack of knowledge on the used alloys. INTRODUCTION The hydrogen solubility in water defines its behaviour during the accident and affects water radiolysis. This solubility is evaluated by considering the gas as perfect and There are, in a PWR, three main barriers between highly the solutioution as ideal. The approximation is justified by calculations of the dev.ûT.on radioactive fission products and the environment : fuel cladding, relativeà t• ^o ideality* *4 A •^ 1 Ï •* m m. primary system and containment building. In the event of a loss of The application of these calculation methods to the evaluation of the hydrogtn coolant accident (LOCA), as the two first barriers are supposed to produced during the design basis accident shows that the main hydrogen source is fail, the containment, building remains the ultimate barrier fo. aluminium cor'osion. The radiolytic hydrogen production depends on the assumptions made on the fission product release from the fuel. The assumption made in France in the fission products. It is therefore important to study the various fundamental safety rule applying to the fission product release leads to a hydrogen possible containment failure scenario, including those involving production slightly exceeding the production calculated according to b5 assvictions. The inflammability limit for hydrogen in dry air is reached, in the worst cas?, in 15 c'ays, thus hydrogen combustion. Previous assessments on hydrogen hazards during LOCA. the classical A.N.S., standard curve are listed in table 2. The total assuming normal operation of all emergency systems, led to the integrated energy, as well as beta and gamma energies are plotted conclusion that hydrogen production is a rather slow process and versus time in figure 1. All these calculations have been made with French PWR have been provided with several features of atmosphere conservative hypothesis. control, i.e. mixing, measuring and recombining. B/ Water radiolysis : The Three Mile Island unit 2 accident produced about S00 kg of hydrogen, notably exceeding previous estimates. This acci- Following a loss of coolant accident, water radiolysis by dent highlighted the possible production of large quantities of fission products energy, takes place in three regions : reactor hydrogen that could pose a threat to the containment. core, containment sumps and containment atmosphere. In order to calculate the amount of hydrogen generated-tejL.water rîdiolysis, the The purpose of this paper is to provide information on following assumptions have been made : hydrogen generation during LOCA in French 900 MW PWR power plants. The design basis accident is taken into account as well as - core radiolysis : - 7,4% of gamma energy and no beta energy more severe accidents assuming failure of emergency systems. is absorbed by water, 1/ HYDROGEN GENERATION MECHANISMS - the radiolytic hydrogen yield is 0,45 molecule per 100 eV of energy absorbed, Hydrogen sources considered in this paper are : zircaloy oxidation, water radiolysis and aluminium corrosion by spray solu- - residual energy is computed with "PO- tion. Major accidents resulting in complete core melt and vessel TENS" code. penetration are not examined and therefore hydrogen generation by core-concrete interaction is not taken up. - Sump radiolysis - 100% of gamma and beta energy is absor- bed by water, The basis for calculation of hydrogen generated by radio- lysis and metal oxidation can be derived from the inventory of the - the radiolytic hydrogen yield is 0,45 physico-chemistry state of the reactor water, and from calculations molecule per 100 eV, of the residual heat of fission products. - residual energy of the released fission A/ Residual heat calculation and fission products inventory : products is calculated with "POTENS". In order to estimate residual heat, fission products have - Containment atmosphere been classified into eight groups, according to their chemical pro- radiolysis : - 100% of beta energy and no gamma energy perties. The fission products release fraction is listed in table 1 for of volatil fission products is absorbed by gap release or core melt release. steam, Each fission product energy has been calculated by a - the radiolytic hydrogen yield is 0,5 mole- computer code "PEPIN", developped in French Commissariat à cule per 100 eV. l'Energie Atomique, and summarized in a computer data-pack called 43 'POTENS". Comparisons with the "ORIGEN" U.S. code results and In all cases, the amount of hydrogen produced by con- - endothermic dissociation of the water molecule on the oxide tainment atmosphere radiolysis is negligible regarding hydrogen surface, produced by other mechanisms. However, the containment atmos- - oxygen interstitial diffusion through the oxide layer and zirco- phere radiolysis may produce nitric acid with a yield as high as 2,9 nium metal, leading to a stabilization of an « zirconium phase, molecule per 100 eV absorbed. - exothermic formation of zirconium oxide C/ Aluminium corrosion : The kinetics of zirconium-çteam reaction is governed by Some internal metallic structures of PWR containment the oxygen diffusion process and thus, described by a parabolic building are exposed to spray solutions after a loss of coolant law : accident. Corrosion of these metallic parts is a source of hydrogen. In the French PWR, the main problem is caused by aluminium which 2 can be corroded by the alkaline spray solution of soda and boric m = K(T).t acid (Ph= 9,3). where m is the mass of metal reacted per surface unit, K(T) The kinetics of this corrosion is calculated by the follo- the parabolic rate constant function of temperature and t wing equation : is time. V = A exp (-E/T).S H2 The parabolic rate constant is expressed in an Arrhe- 3 where Vj,2 is hydrogen released in Nm /hour nius-type equation : A is 9,5.10» E is activation energy, 9300 K K(T) = A exp (-E/RT) S is aluminium area in m2 T is temperature. where A is a constant and E the activation energy of the reac- tion. The amount of hydrogen released by aluminium corrosion is computed by time integration of this equation using two aluminium Owing to zirconium and zircone phase changes with inventory hypothesis sets, a best estimate one and a conservative temperature, three sets of kinetics coefficients are used to compute one. These hypothesis are listed in table 3. the parabolic rate constant. These coefficients are listed in table 4. D/ Zircalov oxidation : Though zircaloy-steam reaction mechanism is known, an uncertainty remains concerning radiation effects on this mechanism, The zircaloy-steam reaction can be characterized by the due to the presence of highly reactive species created by water following balance equation : radiolysis. Zr + 2 HjO •» ZrO2 + 2 H2t AH = -140 Kcal/mol 2/ HYDROGEN GENERATION DURING A DESIGN BASIS ACCIDENT This equation hides a complex process that can be divided in different steps, as follows : The design basis accident for hydrogen generation and containment pressure peak is the main LOCA assuming normal - steam gaseous diffusion through the hydrogen layer present on operation of emergency systems, including the containment spray the surface of the oxidizing clad, system. Hydrogen generation during this accident is calculated Changes from initial BOIL 2 model have been done con- with the following assumptions : cerning the following points : - 1,5% of zirconium cladding is oxidized, - residual heat : "POTENS" subroutine has been substitued to the previous ANS analytic curve, - 100% of rare gases, 50% of halogens and 1% of non-volatil fission products are released form the fuel to the coolant - a new set of thermodynamics subroutines has been added to water, calculate water and steam properties in the range 50°C-350°C and 1-150 bars, - the containment atmosphere temperature decreases from 140°C, at the beginning of the accident, to 50°C after 40 days. - zirconium-steam reaction is computed as exposed previously, using three sets of parabolic constant coefficients, function of The results are plotted, versus time, in figures 2 and 3 temperature, with the two assumptions made on the containment building alumi- nium inventory. - break flow is estimated by a correlation function of primary coolant pressure for critical flow and differential pressure for The lower fluismabillty limit for upward flame propagation, subcritical flow, i.e. 4% volumic concentration in the containment building, is reached in respectively 15 and 40 days. This delay allows the - containment atmosphers pressure is estimated by a simple starting up of the containment atmosphere monitoring system and correlation function including a constant steam condensation the installation of an hydrogen recombining system. model. 3/ HYDROGEN GENERATION DURING MORE SEVERE ACCIDENTS B/ Hydrogen generation during LOCA with emergency systems fai- lure : Two types of hypothetical accidents, involving a larger oxidation of fuel cladding have been considered : '..OCA by large This hypothetical accident supposes total loss of two main primary break with failure of emergency core coo.nig system and containment spray system, and a LOCA by small break, subsequent redundant systems : emergency core cooling system and containment to the total loss of electrical supply. spray system. It is thus beyond design basis. In both cases, water radiolysis and aluminium corrosion Computation results appear on figure 5 and 6 for core are negligible sources of hydrogen. The percentage of oxidized melting and zirconium oxidation percentages. Zirconium oxidation is cladding is evaluated with a simple thermal hydraulics code : limited to about 60% by two phenomena : steam availability and fuel BOILK. melting. The total hydrogen mass released is about 450 kg leading to an hydrogen percentage in volume of 14% in the containment A/ The BOILK code : building, assuming, homogeneous hydrogen distribution by natural convection. The boilk code is derived from a Battelle Colombus code, BOIL 2. The core is divided into 6 radial zones and 24 axial zones. 45 Mass and energy transfer models are summarized in figure 4. it Cl Hydrogen generation during small break LOCA subsequent to A major uncertainty subsisting on hydrogen hazard is the total loss of electrical supply : hydrogen distribution during the first hours of the accident. This point determines the effects and consequences of local detonation or This hypothetical accident is subsequent to a total loss of deflagration which could possibly be harmful to safeguard systems, electrical supply ; LOCA is due to to a primary pump seal break. or induce missile generation in the reactor building. Computation results, for a 27 cm2 break area, appear on As electrical supply failure are identified as an important figures 7 and 8 respectively for core melting and zirconium oxi- contributor to severe accident risk, corrective actions have been dizing percentage. taken in France to improve their reliability, including the installa- tion of a gas turbine on each site to supplement the existing Cladding oxidation is found accelerated, four hours after sources. These actions are thus contributing to hydrogen r-izard the accident beginning, due to steam flow increase by partial corium reduction. drop into vessel bottom. The total amount of iiydrog'en produced is about 500 kg and an hydrogen volume percentage of 15% is reached in the containment building. In this case, zirconium oxidation is not limited by steam supply. CONCLUSION Hydrogen generation during a PWR LOCA has been esti- mated for design basis accident and for two more severe hypothe- tical accidents. TABLE 1 Hydrogen production during design basis accident is a rather slow mechanism, allowing in the worst case, IS days to Fission products release (WASH 1400) connect an hydrogen recombining unit to the containment atmos- phere monitoring system. Hydrogen generated by steam oxidation during more core melt severe hypothetical accidents was found limited by steam availability Fission product gap release release and fuel melting phenomena. Uncertainty is, however, stilj remaining on corium-2irconium-steam interaction. 1 Xe , Kr 0,030 0, 87 2 I, Br 0,017 0, 88 3 Cs , Rb 0,05 0, 76 In worst case, calculations lead to the production of 500 4 Te , Se, Sb, As, Sn 10-" 0, 15 6 kg of hydrogen, thus leading to a volume concentration of 15% in 5 Sr , Ba 10- 0, 10 6 Rv , Rh, Pd, Mo, Tc Ag containment atmosphere, assuming homogeneous hydrogen distribu- 7 Y, Ac . . , La . . 0,03 tion within the reactor building. This concentration is within 8 Zr , Nb flammability limits but not within detonation limits. However, hydro- gen detonation due to local hydrogen accumulation cannot be discar- ded, a priori. TABLE 3 TABLE 2 Aluminium inventory hypothesis ! Mass (kg) Area m2 POTIÎNS ORIGEN 1 ANS TLMVS To ca 1 3 Y To cal 3 Y Total best estimate ! 655 129 i 1 I 75 74 68,61 144,35 .79 79.86 161.65 167. 85 81 conservative ! 60 > 41 71 45,84 87,55 44 .04 50,44 94,48 100, 87 (including electric ! 580conn0 609 500 s 26 • M 33.20 59,94 61. 14 wires) ! 600 s 23 .f.4 29.85 53,69 57. 98 7200 s II ,31 14,79 26,10 27, 74 4 I0 » II .29 13,61 24,90 25, II 1 j 5 .12 7,37 12,49 5 .17 6,95 12,12 12. 78 2 j 4 .13 6,12 10,25 10. i ; TABLE 4 <• ï ! ,50 5,08 8,58 3 .41 4,67 8,08 8 II 10 j 2.75 3,50 6,29 2,70 3,34 6,04 5 77 12 j 2.58 3.31 5,04 2 ,58 3,13 5,71 5 33 A Activation energy ! 24 j 2 ,07 2.32 4,39 2 ,04 2,18 4,22 4 10 Temperature range J Kcal/mole ! 30 j 1.91 2.03 3,94 1,83 1,86 3,69 3 74 40 j 1.72 1,69 3,41 3 îl T < 880°C 5 ,52 10* 29 ! 50 j 1 .58 1,45 3,03 1 95 880 < T < 1577 3 ,58 105 33,5 ! 60 j 1.47 1,28 2.75 1.33 1,15 2,'.8 1 7J 1577 < T < 1850 1,04 10" 78,8 47 33 o VOLUME D'HYDFOGEMS DEGAGE (129 J d' :^ CECAGE (609 m d'alU3iniu=) t VOLUtfS H2 (n3 TPN! - L VOLUME H2 I M3 TPN ) SOURCES D'HÏUKOGENE SGCÛ SOL-BCES D'HYUKOCESE ?CCC. .1- A:OXÏUATION Dt L'ALUMINIUM A:OXYL'ATION De L'ALUMINIUM Z:OXIDATION DU ZIRCONIUH Z:OXYDATIO.M UU ZIHCONIUM P:SADIOLÏSt PUISAHD P:RADIOLYSE PUISARD C:RADIOLYSE COEUR CjRAOIOLYSE COEUiî T: TOTAL T: TOTAL 2CC0. - .CCO. 1CC0. - 2000. - TEMPS (JOURS) 0. S 0. 40. 80. 0. 40. 80. TEMPS (JOURS) 49 FIG. 2. FIG. 3. MA1;-» TRANSFERS 50 a water vaporisation at interface a water u^yJ ry 7.1 rcaloy-fteam reaction c nyui '*> jert produced i^y ? :rc.i ioy-stpam react ion d steam at ~or^ c.itlet e • hyocoqen d'_ core outlet r î ,c«M .1 ": '-; rcj j'ruc'^ros one pur y ; )\ y 'i tooe ft at jp p*_r sLiT'icturos output 1 hi.-i.j-- h/ir'jqen flew ] . 1. or lit» "5 a te st->iim L:I conta inmunc b'i' 1'i ir\q 1 :tuta: icsidudL hedt 2 residual heat with fission products release 3 vuLjti1 fission prcdUTts re lease 4:water vaporisation energy !i : stean enthalpy s 6-?f.ean enthalpy t r core outli-t 7 • ît/dt j^jen en thai py at -ore outlet G : sr.ea-i enthalpy at upper structures output 9.nydrogen enthalpy at upper s tructures output 10:break steam enthalpy 11•hydrogen enthalpy 12 :water- lewer structures convec t. ion 13:water-fuel convection 14:gas-fuel convection 15.gas-upper structures convection 16:fuel radiation to water 17:fuel radiation to upper structures 18 :zircaioy-5team reaction energy 19 ; fuel roelting energy FIG. 4. FUSION FIG. 5, OXYIAT:C:>1 °/o 60 OXYDATION TE:-*s (si FIG. 8. FIG. 6. » FUSTCK '-2 '-4 1.8 2.0 Te=p5 (si .10"" FIG. 7. 37 52 ANALYSIS OF HYDROGEN DISTRIBUTION IN The RALOC /I/ code has been developed in order to provide a realistic physical model for the calculation of the time-depended thermal hydraulic CONTAINMENTS UNDER ACCIDENT CONDITIONS behaviour and the mixing process of steam with non-condensibles in a post accident atmosphere ûf a multicompartment containment. P. PAPADIMITRIOU, H.L. JAHN, T.V. PHAM This paper describes the basic physical phenomena simulated by RALOC anc discusses its capabilities. Analytical and experimental results on hydro- Gesellschaft fur Reaktorsicherheit (GRS) mbH, gen distribution and mixing in large compartmentalized containments are G arching, presented. Special consideration is given to the wall heat transfer mecha- nisms and their influence on the distribution and mixing processes. Federal Republic of Germany The generation of hydrogen and particularly chemical aspects such as radiolysis or zirconium oxidation rates are not subject of this report. Abstract BASIC PHYSICAL MODEL RALOC was developed for the analysis of thermal hydraulic transient and hydrogen distribution processes in the containment. The lumped parameter The knowledge of the distribution of hydrogen, oxygen, steam and inert gases of the post LOCA containment-atmosphere is important for the analy- code simulates the containment building as a network of nodes which are sis of a possible reaction of hydrogen with oxygen. The computer code connected by one dimensional flow equations. RALOC is developed to calculate such distributions with respect to tine and location. The containment and the thermo-fluiddynamic processes are Mass (water, steam, non-condensible gases-H-, 0., N, etc.) and energy con- realistically simulated in this lumped parameter model. Special conside- servations are considered for each node. ration is provided to the associated important questions of heat transfer The pressure at each node (control volume) is uniform and it is assumed mechanisms and transient temperature profiles due to heat conduction with- that water, steam and non-condensable gases have the same temperature in containment structures of different materials. Calculated and experi- (thermal equilibrium condition). mental results of basic (Battelle Tests) and integral DCA-tests (HEDL- It is also assumed that the non-condensible gases are ideal and homo- Standard Problems) are discussed. geneously mixed with steam in each control-volume. Gas flow between compartments is governed by pressure, gravitational forces and frictional forces. No mass or enery exchange is considered within flov connections. The multicompartment model assumes lumped parameters (i.e. the physical state in the node is represented by volume mean values) and is a system of ordinary differential equations expressing conservation of IKTKODuCTIOK mass, momentum and energy. The containment vessel represents one of the most important engineered The mass conservation is formulated for each of four gas species (steam, safetyguard systems to prevent the release of fission products to the hydrogen, oxygen and nitrogen) and for liquid water in each node. environment. The containment vessel design for a light water reactor Conservation of momentum is specified at flow junctions, which are the requires thermal hydraulic calculations for the containment-atmosphere connections becween compartments. Pressure differences and gravitational behaviour after a loss of coolant accident (LOCA). forces control the gas flow between compartments, while pressure loss is caused by friction forces. Compartment interconnections can be mo- During severe accidents large amounts of hydrogen are produced by the delled as orifices with flow resistances. core radiolysis and the oxidation of zirconium and other metals, when partial core uncovery lead to high fuel temperatures. These gases will be The conservation of energy is written for the total mixture in each node released into the containment atmosphere. under the thermal equilibrium condition. The governing time-dependent Many devices of electrical equipment are installed in the containment variables are: building which ecu act as a potential ignition source to induce a deflagra- tion or, under extreme circumstances, a detonation of hydrogen-steam-air- nixtures. Mik. G. and T, The resulting high pressures and temperatures may endanger the containment i = H , 0 , N , Steam, Water integrity. The ignitybility depends on the relative gas concentrations of 2 2 2 hydrogen and oxygen, which in turn depends on the distribution aechanisms and the generation rate. Detailled considerations need the knowledge of Where M.. is the nass of the i-th component in the k-th compartment. The the local physical state and the partial gas component concentrations for flow raèe in the j-th junction is defined by G = A.w p\ , where A. is the possible prevention of dangerous containment shell loadings. interconnection cross sectional are», w. isJ theJ gar velocityJ in the junction and p. i» the junction gas density, an averaged density value (2) Conservation of Momentum between the densities of the interconnected control volumes. T, is the mixture temperature of the k-th control volune. The flow equation which connects nodes is specified for the flow junctions The conservation equations define the derivative of each of these three as types of variables with respect to the dependent variable time. The total number of equations that are solved simultaneously is G. = -2J- (P - ) -KjCjl J I-J k Pl N = (N 2) • N N. eqs gases coopc s junc The first term on the right hand side represents the pressure difference between the nodes, the second term is the friction loss and in the third where Nc and N. are the number of compartments and the number of one the elevation effect is considered. Where A = flow cross section The conservation equations are given below: I = length of interconnection H = geodetic height . K = pressure loss coefficient (1) Conservation of Mass can be written in following form, if the solubi- k, 1 = node number lity of tbe non-condensable* is the liquid is neglected. For the pressure loss coefficient, caused by wall and eddy friction, vari- ous statements can be selected depending on the character of the inter- connection (orifice, pipe) and of the flow (laminar, turbulent). The used formulation is: "ik " J Gij Gec Ik ÏC - t 1 (5) 8 » o, for aon-conden»ablei 2 2p.Aj i = Hj, 02, N2, steam, water j = junction number where £ is the friction coefficient k = node number • • time derivative 4 = —— — for laminar flow (6) RI D where G is the volume overal evaporation or condensation rate in each 4=1 for turbulent flow compartment. The pressure loss term in eq. (4) has a damping effect on the flow mo- tion. The constitutive equation of mass transfer due to evaporation or conden- The diffusion flow rate is calculated stationary by the formula: sation inside a node volume G can be determined by a volume balance: V + V = 0 (2) water steam J The solvation of this equation with respect to G gives the evaporation or condensation rate in the node as * function of the component masses, where T. = (T. + T,)/2 mass flows, the node temperature and its change D = diffission coefficient R = Gas constant Gwvw Gsvs (3) v - v v s (2.1)Critical Flow w = water s = steam The flow of atmosphere is limited to the critical flow. This is calculated 53 as a function of total pressure, gas property and concentrations by using ' = temperature derivative the assumption of a homogeneous médium, thermal equilibrium conditions The rates of convective, radiative as well as condensation heat transfer Q and isentropic behaviour. at the node boundary are kept constant over the time intervall it. The equations, which are transformed in the form (10), are then solved using 1/2 i -1/2 ,-Vz. a linearized 0DE-so]ver for stiff problems /2/, taking time steps less or C =p A T flcL • i_i_ equal to the heat transfer time step At. At the end of the time step, the 1 (8) T" 3p 3h p 5 wall temperatures are updated by solving the appropriate time-depended heat conduction equation for each heat slab at the node boundaries. c = gas, steam concentrations Heat Removal Model R = gas constant The heat sinks or sources are treated 3S heat slabs with lumped masses. All the surface and wall properties are assumed to be constant. p. = gas specific heat The time-dependent temperature distribution m the heat sliso is the so- lution of the Fourier-equation for wall heat conduction. (3) Conservation of Energy (11) According to the first law of thermodynamics, energy conservation in each 3t pcpV compartment is given by: where (9) - I.G..1.. • ZyQ = I S h - Vp Ti a I; ij IJ i. i i * q = (12) 6 i = index of the componenc (H,, 0,, N,, steam, water) 1 D j = index of mass flow (interconnections of a node) a. n Q = heat removal by heat sinks or sources h = specific enthalpy q = heat flux T = temperature of the boundary node V = volume of node X = conductivity lia (i=mlet, a=outlet) p = pressure of node p = material density F = surface area c = material specific heat 6 = layer thickness The energy equation is defined in each nods and gives the time-dependent V*= volume n = laver number temperature change of the mixture. Three different materials can be used for each heat slab. Method of Solution Keat exchange between containment atmosphere and heat slab surfaces is The thertno- and fluiddynamical mixing and distribution problem of noc- expressed as condensatle and condensable gases in multicompartmeot containments is (Tnode - Twall> toverned by * system of partial differential equations in time ind space. Their spatial semidiscretization (nodalization) le«d« to a initial value where a = =./.a df is the averaged and a- is the local he«t transfer problem (IVF) of ordinary differenti»l equations (ODE) in time of the form coefficient at the heat slab area. Equations (11), (12) and (13) are solved for the determination of the u(t) = f(t, u(t)) w»ll temperature Ty ,, and the heat flux Q, where Q is needed for the u(») = u (10) solution of the th^rmo- and fluid-dynamic equations (1), (4) and (9) in 0 D the next time step etc. 0 StcXtR, X = [a,b], X); u(t), û(tUR Calculation of containment pressures and temperatures after a LOCA and finally the distribution of generated hydrogen in the comparuaents is IVPs of this simulation form are often characterized by stiffness behaviour dependent on the heat transfer at the wall. of oscillatory type if only eigenvalues of small modulus contribute signifi- Heat is transfered from the hot atmosphere (mixtures of air, steam and cantly to the solution u(t). non-condensables) to the containment wall or from structures by three The method of solution of the basic multicomponent equations (1), (4) and important mechanisms: (9) used by RALOC is as follows: (2) Radiative Heat Transfer (î) natural or forced convection (laainar, turbulent) (n) radiation Radiative heat transfer from or to a gas occurs only at specific wave- (iii) filmwise condensation (laminar, turbulent) length intervalls in their spectrum. Nonpolar e.g monoatomic and biatomic or dropwise condensation gases (K,, 0,, K,) are oractically transparent to thermal radiation. Polar (iv) condensation in the presence of non-condensibles e.g. trTatomlc gases like steam emit and absorb thermal radiation over a bright temperature range. If steam is considered as â grey medium, di- The heat transfer process is dependent on many variables, i.e: rection and spectral effects dimisb and the characterization of the - heat transfer mode (convection, radiation, condensation), either one emissivity and absoptivity coefficient can be displaced by only one or more together eraissivity parameter for a grey body. surface condition of heat sinks (influenced the condensation mode, The Hottel card /*/ represents experimetal grey emissivity data of water either film- or dropwise) steam in dependence of temperature, partial pressure and gas thickness. steam and noncondensables distribution and concentration The Hottel card is approximated By Schack /bl to the following steam' gas mixture velocity (turbulence of the atmosphere) emissivity formula temperature of the atmosphere - geometry and location of beat sinks ( . fCps)g(ps,T) - properties of heat sinks 1 e ) (16) temperature of heat sinks where c is the value for large steam layer thickness, a function of the Different correlations are used for the heat transfer mode and st3te con- a temperature and f and g are functions of partial pressure p, steam layer ditions detemiced by flow at the structures surfaces. thickness s, and temperature T respectively. The heat transfer coefficient for radiation used is (1) Convective Heat Transfer (17) Nu = DOT3 XI (I )4 -i Convective heat transfer is the process of energy transfer governed by convection. Convection can be "forced" or "natural". Forced convection Vs is governed by pressure and friction forces. Natural convection is caused by buoyancy forces. In eq. (4) both modes of convection are considered. where 0 is the Bolzmann constant The solution of this momentum equation gives values of flow rate due zo T, = vail temperature total convection. Using an adequate discretization therefore it is allowed T = node temperature to calculate the convective heat transfer coefficient by empirical Cn = correction factor for pressure correlations for forced flow conditions. e = wall emissivity c = steam emissivity For laminar flow conditions (Re<2320) the Sieder-Tate /3/ correlation is used (3) Condensation Heat Transfer 1/3 K (UÎ Nu= 1.86 (Re.Pr. j! ) (r,f/nu) °- Heat removal by condensation converts vapor into liquid. Condensation occurs within vapor on entrained particulates and at the wall boundaries wnere q^ = viscosity of fliud by node temperature of the system. Condensation at the wall system boundaries is considered Hw = viscosity of fluid by wall temperature here. Condensation within the thermal system i» described by eq, (3). Re = pwD/q Reynolds number There are two types of wall condensation: filmwise and dropwise. Filmwise Pr = qc A Prandtl number accurs when the condensate wets the surface and a liquid film is formed. Nu = oD?A Nusselt number Filmwise condensation is more common when dropwise condensation especially in containment buildings and has a lower heat transfer coefficient than For turbulent flow conditions (Reï2320) the Hausen /3/ correlation is dropwise condensation. used Dropwise condensation occurs on non-wetting surfaces. The vapor condenses in drops which grow by further condensation and fall down the surface 2/3 8 n leaving space for new droplets to form. Because a great part of the wall Nu = 0.027 (1 • (D/L) ) Re°- Pr (15) surface is not wetted for long time by the insulating condenstae, greater heat transfer coefficients than for film condensation are possible. It is with n = 0.37 by beating unlikely that this process will occur in containment buildings, neverthe- t = 0.30 by cooling less it can happen if the critical material surface tension Oc is lower 55 D = Diameter of the compartment than the condensable surface tension /6/. L = lenght ce (3.1) Laminar film condensation occurs if a /c <1 where 2oTo The Nusselt condensation theory is used Nu= rpMT ' Re= —r) r ; Pr= ne p M and 3 Nu=0.943 { PgA (pf-ps) [r 3/8cp&T] (18) n.=_ 2oldo/dTlî L n AT where Pp-, = fluid density The fluid properties are determined at the saturation condition. The Ps = steam density nondimensional number n, describes the influence of the capillar forces. I - surface length Values of oc for different material surfaces are tabulated in /6/. r = vaporization energy (5) Effect of non-copdensible gases K = water conductivity T = water viscosity The presence of non-condensible gases such as air or hydrogen considerably reduces the heat transfer relative to the value obtained in pure conden- (3.2) Turbulent film condensation occurs if the condensate film exceeds sable steam. a critical size. The heat transfer coefficient is larger than that of the This important effect is simulated by the following correlation which laminar one, because not only heat conduction but also turbulent diffu- approximates Rose's /&/ calculational results for filmwise condensation • sion take place. The transition of the laminar to turbulent film conden- sation it obtained by the Grigull /3/ critérium 2 2/3 1/3 3/2 Se'=0.3 10" [ \p g iT I ] i 350 (19) and (23) Nu/Nu cond. for c< 10 The related turbulent Hussein number is The correlation (23) is similar to that of Hendersor and Marchello /9/. The value Nu related to condensation of pure vapor and the value c Hu represents the concentration of non-condensables. (20) Summerizing this method for calculation of the heat transfer coefficients at containment structures utilizes well known established theories for The temperature dependent transient properties are evaluated at an aver- estimation of local condensing heat transfer coefficient. aged temperature V^sat^wall5 /2- The variables mainly taken into consideration are containment wall tempera- ture, atmospheric temperatures and the concentrations of tteam and non-con- densitle gases and flow parameters, calculated at the last time step. (4) Dropwjje eendensation occurs if the surface material has as extreme These values itselfs depend on the value of the heat transfer coefficient. low affinity to the vapor. This is valid for o./o >1. Dropwise condensa- The effect of the vapor velocity on condensation is neglected. tion on vertical surfaces is described by the following formula /7/: EXPERIMEKTAL ANALYSIS In the region: 8 1O"4< Re < 3.3 10"3 (1) Battelle Frankfurt Experiments J)u=3.2 10"* (21) The Battelle Frankfurt laboratory performed various hydrogen distribution experiments in the model containment shown in Fig- 2, where several com- partments were simulated. Test 2 was designed to investigate the effect of And in the region: 3.3 10"3< Re < 1.8 10"2 geometry (orifice) at isothermal conditions whereas Test 6 shows the effect of geometry (orifice) and coermal inversion. Test 16 shows the effect of geometric complexity (more rooms, otiliccs) (22) and thermal inversion. Tabel 1 describes the test conditions. For test 6 the important fedLui.es are Table 1: Battelle Test Conditions the warmer atmosphere in the upper room resists effectfully to con- vective mixing. The cooler atmosphere of the bottom room shows no Exp. Ho. 2 6 16 significant buoyancy. especially in the orifice the gas mixture has a density inversion H.-Input time 3h 47' 18'i-lh 48' 7h 20' caused by temperature and K^-concentration gradients The gas Gasraixture H X/H 5 66/34 66/34 66/34 oscillates in this location*'because of the competing effects of buoyancy and concentration gradients tending to destabslieed the flow, Input rate mJ/h 1.19 2.35/0.98 1.31 although the basic experimental trends are predicted reasonably Mean gastem.'C 17 19 23 well, RALOC underpredicts the K^-concentration m t',e lower room by approximately 0.5 volume percent. This discrepancy is unresolved, Temper.upper/lower 17/17 35/19 33/22 but it is possible that the reported and the measured injection rates room °C are somewhat different. Orifice area m2 1 1 1 Height of (1.2) Battelle Test 16 source above base ID 0.6 0.6 0.6 A schematic of the test geometry of the Battelle test facility is shown ic Participated compartments Rl, R2 RI, R2 Rl, R2, S3, Fig. 1 ana perspectively in Fig. 2. During the test the rooms Rl, R2, R5 , R6, R7 and R8 were used. All other rooms were sealed off. The hydrogen R6, R7, ?.S source was m room RI. Tue ùocializatioa used for the calculation with RALOC is shown in Fig. 2. Two different calculations were done, one vi*n- out consideration of the heal stored in the walls and the other with con* sideration of the energy transport to or from the walls. It is predicted that this stored heat energy is sufficient to maintain toe (1.1) Battelle Teat 2 and 6 temperature of the atmosphere and sustain a thermal stratification. All compartments except Rl and R2 of the Battelle model containment, were The results cf this two calculations are shown in Fig. 5, 6, 7 and 8. For sealed off during test 2 and test 6. The test geometry and the discreti- test 16 the important features are zation scheme are shown ia Tig I. For cesc 2 Che rooms were isothermal ac 17°C, whereas for test 6 the upper room was heated to 35°C and the lower the calculation^ results without heat slabs, show a good homogeniza- room temperature was still held at 19°C. The results of the KALOC calcu- tion of the H.-concentration in the containment after approximately lation are compared with the experimental data measured at different 20000 s, whicfi was not observed during the experiment, positions within the compartments. the calculation with simulation of the walls by heat slabs maintain the temperature inversion within the containment. The heat transfer Fig. 3 and 4 show the comparison of the predicted and experimental results from the walls effectively blocks the nomogenization of the gas mix- for test 2 and 6 respectively. The code predicts correctly the phenomena ture. such as the almost homogeneous concentration distribution in test 2 and - no stabilizing effect was observed if the structures were neglected, the significant concentration difference between the lower and the upper the homogenizacioa concerns temperatures and concentrations, room in test 6. the simulation of the structures can recreate the atmosphere tempe- However, the predicted concentrations are higher for the end phase of test rature inversion. However this stabilizing effect may be overpredicted, 2 and generally lower for test 6 compared co Che measured values. The if the starting cemperature profiles of the structures are not accurate- reason is uncertain, however, it is possible that the saturation condition ly known. assumed in the model was not valid for the experiment, or more likely the convection and heat transfer are interdependent. The sensitivity of injection rates were not given quite acurrate. the heat transfer on the physical state conditions is demonstrated by Fig. 9 and 10. For test 2 the important features are (2) KEPL-Tests A and B /IP/ the concentration in each compartment is fairly uniform The HEDL-tests A and B are experiments on hydrogen (test A) and helium the concentration difference between the two compartments is small (test B) distribution in a simple containment. The HEDL facility consists 57 of a large steel vessel, which is 20.4 m high and 7.6 m in diameter. Aa annular test compartment was constructed in the bottom half on the vessel 43 CO with a 300 degree segment 7.62 r» O.D. , 3.0 m 1.0. and 4.72 m high. Air CONCLUS ION'S recirculation between both compartments was provided by a blower. Tbe air flow passages consisted of 24 vertical sloes cut into the wall of circular Tbe muldnodal thermal-hydraulic computer code RA1OC was extended by duct segments at the ceiling of the annular test compartment. a heat transfer model Measurements of gas concentrations and temperatures were made ac the bottom, middle and top elevations at 5 circumferencial locations. The Comparisons of calculationed and experimental results show that tem- upper compartment was also connected with the eavmroment (pressure boun- perature and concentrations histories are in good agreement, if dary condition 1 bar). A fan heater had been turned on (in zone 20) to sufficient information of the experimental conditions are given simulate the beat up process, before the steam/and H, (or He) was injected (prehistory, history, measuring) (prehistory) The horizontal jet for test A and the vertical jet for test B are shown in The quantitative understanding of heat transfer to or from the walls Fig. 11 with a schematical figure of the test section and the RALOC discre- is important for exact pressure, temperature and cencenvration rise tization. estimation. The test flow rates of hydrogen of helium and steam and the jet tempera- tures for test A and S are shown in Table 2. - Steam superheating is not considered in this RALOC calculations and may have a certain effect, under special conditions. However this is not expected to be of great significance for the investigated HEDL- Table 2: KEEL-Test Condition tests as saturated steam was injected. Experiment - RA10C can predict the thermal-hydraulic history and H^-concentrations Gas input sbort and long term considerations of severe accidents. However for He highly transient processes a thermal nor-equilibrium model should Initial atmosphere Temp./Gas component 65/>f2 65/AlJ? be developed. Steam/Gas input time (mis) 10/10 10/10 - In order to understand the H,- and 0,-distribution in containments, Steam/Gas input rate (Kg/mm) 24/0.25 12/0.4 which is important for the detection of deflagrative or detouative Steam/Gas initial Temp. CO 150/150 170/170 gas mixtures, further analytical and experimental studies should be performed under more realistic accident specific conditions. The specific blower inlet temperature and concentration were given for users of finite difference codes. All gas concentrations are measured on a "dry" basis (i.e., the concentra- REFERENCES tion after condensing the steam). The gas concentrations calculated by RAXOC on "dry" and on "wet" basis are presented for the 125 degrees loca- tion. The calculation on "dry" basis was blind, the "wet" basis calcula- /I/ Jahn H.L. , "Hydrogen Distribution after a Loss-of-Coolaot Accident tion, done later for explaining the discrepancy between the predicted con- in the Subdivided Containment of Light Water Reactors Transla- centration and the experiments. The Fig. 12 and 13 show temperatures and tion" NUREG/CR-1831, SAND80-6031, Sandia National Laboratories concentrations for test A and B, whereas Fig. 16 and 18 show typical pre- (November 1980) dicted heat transfer coefficients for these tests. For the HEDL tests the most important features are that HI Hofer E. , "An A(a)-Stable Variable Order ODE-Solver and its Appli- - the "blind" calculations show a good agreements for the temperatures cation as Advancement Procedure for Simulations in Thermo- and - the obtained concentrations on "dry" basis are overpredicted Fluid-Dynamics" Presented at the ANS/ENS International Topical - the obtained concentrations on "wet" basis are in good agreement Meeting "Advances in Mathematical Methods for the Solution of with the experiments Nuclear Engineering Problems", Munich, FRG, April 1981 - the discrepancy between the calculated concentrations on "dry" basis and the experimental data is unresolved, but it is possible that the /3/ Grober/Erk/Grigull, "WarmeUbertragung" Springer Verlag, Berlin 1963 steim was superheated, which may lead to significantly lower "dry" con- centrations (Fig. 16). However the comparison of the given measured /4/ Hottel H.C., Egbert R.B., "the Radiation of Furnace Gases", Trans. hydrogen inventory and the specified injection «tes also offer some Amer. Soc. Mechan. Eng., 63,297/357, 1941 discrepancy of the same order of magnitude. - Fig. 16 shows that with increasing steam superheating the concentra- /5/ Schack K, "Berechnung der Strahlung von Wasserdampf und Kohlendioxid" tions on "dry" basis approacbe to those on "wet" basis. Chemie Ing. Technik, 42,53/58, 1970 /6/ Herte H., "Condensation Heat Transfer", Advances in Heat Transfer, Vol. 9, Acadeaic PreJJ., 1973 PI Isachenko V.P., Osipova V.A., Sukoœel A.S., "Heat Transfer", MIR-Publishers, Moskow, 1980 /8/ Rose J.W., "Condensation of Vapour in the Presence of a Non-consen- ding Gas", Int. J. Heat Mass Transfer, 233/237, 1969 /9/ Henderson C.L., Marcbello J.M., "Film Condensation in the Presence of a Non-condensable Gas", J. Heat Transfer, 447/450, 1969 /10/ Bloom G.R. et al. "Hydrogen Mixing out Distribution in Containment Atmosphere", EPRI NP-2669, March 1983 FIG. 2: BATTELLE CONTAINMENT AND RALDC DISCRETIZATION FOR EXPERIMENT 16 - • — • 2 3 R2 ORIFICE 5 G r 6 I •> ! i 9 10 o ," CM 1» It —LRAUOC WATER SOURCE ... -ÊXP. u A <^ m 20CO «XXD 6CC0 SEC FIG. 1: RALOC DISCRETIZATION FOR BATTELLE EXPERIMENTS 2 AND 6 FIG. 3: HYDROGEN CONCENTRATIONS BATTELLE EXP, 2 o j 60 R5 m .——• R5 rsi o. R6 f ( ^ j p?^ — Calculator Pi — Exponmçn R7 RMÏÏê; — Colcuialio IRC OH 0 10 000 20 000 30000 sec. o 2000 TOO 6000 sec R2 o. 1 Fi -. 4: HYDROGEN CCNCEMTRATIONS BATTELLE EXP. 5 RI sr- • —= i- O- — Exoerimen , , IR7| R2 Ipt; — Calculation — Expenmen fëo — Calculator1" 0 10 000 20 000 30000 sec. o R7 •^ __ ro R8 •^ Expenmtn O. |R7' R Caiculolio I_JR5 -— Expenmon 1 iR6 — Colculatio > lg_R p r o- ] 10 000 20 000 30000 sec. FlG. 5: GAS TEMPERATURES 1 EXP. 16 (WITHOUT HEATSLABS) *— Experiment p.i—- ^r^-_ R Calculation — Calculolicn 5 experiment RGI Rl — CQÎCUIQIICP E& 10 000 20 000 30000 sec. ID o ! o > L • t$ r P: _) i — Expenmen — Exparirnent «. R7L R2_ 'p- —— Calculation — Cclcvjlc'ior • ,/y — Experiment- Rl RÔp I. — Calcul Glior Ik — Calculctton ^ r,. 10 000 20 000 30000 10 000 20 000 30000 sec. sec. c Experiment o. — Calculction — Experiment — Calcolalion 10 000 20C00 200C0 10 000 20 000 30000 El sec. sec. FIG. 6: HYTROGEN CONCENTTÎATICNS BATTELLE EXP, 15 (WITHOUT HFATSLABS) FIE. 7: GAS TEMPERATURES BATIELLE Ex~. 16 (WITH HEUTSLABS) 62 300C0 sec. 0 10000 2C0C0 SEC 300X FIG, 9: TYPICAL HEAT TRANSFER COEFFICIENT FOP. BATTELLE EXP, 16 —— Experiment _ — Calculation "•' 3OOCQ sec. R6, R8 f R 2- R 5, R7 10 CCO 20 000 300C0 0 10CO0 2CO00 SEC 30000 5ÛC. FIG. 8: HYDROGEN CONCENTRATIONS BATTELLE EXP, 16 WITH HEATSLABS) FIG. 10: TYPICAL HEAT TRANSFER COEFFICIENT FOR BATTBLE EXP, 16 NOZZLE A IJ.-I..IP1I. FIG. 11: IEDL-TEST VESSEL AND RALOC DISCRETIZATION 63 i -—- ExrtHIHNT — prediction blown —*rr BASIS — prediction trpfier 10am I FVlEDICTiOT ,*'?.. . . r-Pd Howe* FV1 urpci 'orm position: blnwcr / /.0 60 BO 20 1,0 GO 00 min min '•Experiment •*• ExrtRiitNi — Predklion — «ET BASIS II 'A — CRY BASIS nIII n g •VS- — position I?S W position: I?1 lower foom lowei loom i | 1,0 GO no 20 20 80 min min Cômporison of Atmosphere Temperatures Comparison of Concentrations TIG. 12: IICDI. TEST A TEMPERATURES /Vm HYDROGEN CONCENTRATIONS Experiment btowcr "*- EXPEDIANT ptetliclion blower — «I BASIS i (Mediation upptr room _ • — DRY BASISJ VICTIM _ position '. blower 20 1.0 CO 00 min 1 1 • • —— EXPERIMENT ••penmenl — 1er BASIS ) - H» Racic / ftlEBICTION It /'' pastitonTTi: - posilion I2S* - lowtr rooiT lowci room 20 60 00 20 60 00 min min Comparison of Atmosphere Temperatures Comparison of Concenirations FIG. 13: IEI1 TEST B TEMPERATURES (Vm IIELIIM CONCENTRATIONS 0 20 tO 50 MIN FIG. M: TYPICAL HEAT TRANSFER COEFFICIENTS FOR HEDL TEST A 1 i | i i 1 1 5 S la L.22 ,13NE 10 j ZONE 7 i / Z» r 3 &=» .fa., r — __ [ 20 60 MIN 80 FIG. 15: TYPICAL HEAT TRANSFER COEFFICIENTS FOR HEDL TEST B 1. ••••• ' J N C 4TOOK H «et \\ \\ \ «\1 n-.iQK iT = 5K p=l bar STrOk- 20 iO 60 80 100 STE*M TEMP. °c FIG. 16: SENSITIVITY OF HYDROGEN CONCENTRATIONS DUE TO STEAM SUPERHEATING 66 HYDROGEN SAFETY IN NUCLEAR For example, under normal opérât ing conditions the concentrât ion of POWER PLANT REACTORS hydrogen m che water of water moderated and cooled (WVER) reactors is kept at a levé 1 cf 60 nml/kg, while the coral quantity of hvdrogen in A.V. DUBROVIN, V.A. ERMAKOV the circuit is 18 nm . I.V. Kurchatov Atomic Energy Institute, Under normal operating conditions there are two -nain reasons for Moscow, the presence of hydrogen in the primary circuit: it can either be olown Union of Soviet Socialist Republics off from the channe1led-flow tanks or else IC can result from conden- sation of steam in single-circuit units with boiling-water reactors. In Translated from Russian the first case, for channelled blow-offs occurring at a rate of 20 t/h, Abstract the hydrogen yield is 1.2 nm /h [ 1 ]. In the second case - chat of steam condensation - there is a considerable increase in the concentration of The paper examina various aspects of nydrogen safety for WWER the gas, i.e. the radiolytic products of water or hydrogen. If the reactors during normal operation and for loss of integrity of the primary situât ion is rot rectified the concentrât ion of radio 1 ytic gases m -he circuit and outlines a comprehensive research programme to solve open steam will reach a Level of ^0-60 nml/kg and give rise to oxyhydrogen questions, with a view co defining criteria and developing standards on generation at a rate of 70 nm per MW-hour of heat energy. nyarogen safety. In emergency situations involving loss of leaktightness in the primary circule there may be a release of the accumulated hydrogen or oxyhydrogen into the buildings of the plant where, ultimately, it will accumulate in The intensive development of nuclear power has brought with it an the leaktight areas of the facility or in the leaktight containment system. increased need to ensure the safe operation of nuclear energy sources- In aceiderits invoIvmd disrupt ion co the reactor core heat-rernova 1 system One particularly important aspect of safety at nuclear power plants is it is possible for a reaction to Lake place oecween the seeam and the that of hydrogen safety» since, generally speaking, cne presence of metal housing of the cjre such that hydrogen LS formed in a quant, icy hydrogen in such facilities can result from oocn tne no ma I operation considerably in excels oi che limit permitted under normal operat ing of the nuclear energy source and from accidents. Furthe rmore, che candi t ions• uncontrolled leakage or combustion of hydrogen leads co further compli- The well-known accident ar. the Hamsburg plant demonstrated the cation of the radiation conditions and 'he force of powerful explosions need for me r s» exrer.s ive research into hydrogen safety ac nuclear power can damage the plant equipment and buildings. planes and oc-ûs-oned a number of expérimenta I studies concerning, Hydrogen at nuclear power planes during normal operating and emergencies chiefly, tt.e processes at work in reactor containment systems. Through the action of ionizing radiation the reactor water in a In the USSR one of che questions examined m che design of reactor water-cooled reactor undergoes radiolysis, the end products of which facilities is hydrogen safety. In che construction of a single-c îrcuic *r- molecular hydrogen and oxygen. The well-known methods of suppress- unit with a boil ing-water reactor research has been conducted into the ing radiolysis entail maintaining a sufficiently high concentration of conditions governing oxyhydrogen ignition in a turbine capacitor and hydrogen in the reactor water to bind the free oxygen and prevent oxy- data have been obtained on concentration limits for oxyhydrogen ignition . hydrogen (i.e. oxygen and hydrogen linked in a stoichiomecric relation- at low pressures. Special systems have been devised for reactor facili- ship) from forming in the gas chambers of the primary reactor ;;rcu:t. ties co ensure thac gases containing hydrogen are b Lown off and pro- cps^ed in accordant*" vrh £jre-s3fery requirement s. In the planning of a nuclear power plane designed to supply heat The process of hydrogen combustion and the consequences various aspects of hydrogen safety are singled out for special attention. of its ignition (degree of pressure increase, develop- Analysis of the operating conditions of one such reactor facilicy has ment rate of processJ; revealed that in order to solve problems relating Co hydrogen safety m The conditions governing the transition from rapid com- a logically consistent and reliable manner it was necessary to develop bustion (deflagration) to detonation. systematic research programmes in a number of directions. Considered 2. In order to ensure the safe operation of the facilicy it is froo a more fornalistic poinc of view, the problem of hydrogen safety necessary to overcome a number of problems, in particular: can be said to consist of a sec of individual aspects. Monitoring of the hydrogen content of the reactor v.iter Aspects of the problem at nuclear power plants (see Fig. 1) and steani in the primary circuit and space protected by The problem of hydrogen safety presents itself in one form or another the containment system; as a function of the possibility of approaching the reaccor unit and Regulation of gaseous conditions within the reactor; power plant as subjects of research. Prevention of the formation of local zones with a high 1. Examination of the technological process in operation at hydrogen concentration in engineering components or in nuclear power plants shows that the following questions call the reactor buildings, including the containment system; for more detailed study: - Systematic elimination of hydrogen in the reactor and 1.1. Under normal operating conditions: buildings of the plant (catalytic combustion, systematic Mechanisms whereby radiolytic gases are generated in the deflagration); reactor core, methods of suppressing radiolysis and Exclusion of the possibility of detonation processes quantitative indicators for gas conditions in the reactor; from the equipment components and building of the plant. Distribution of the products of radiolysis (hydrogen) in 3. Sound decision-making in the planning of power planes with the reactor circuit where there is a volume filled with nuclear water-cooled reactors is dependent upon the establish- steam and areas of steam condensation present; ment of technically and scientifically sound safety standards 1.2. Where normal operating conditions are disrupted as a result which meet the requirements of expediency and economy while of loss of leaktightness in the primary circuit: taking account of the actual conditions prevailing at nuclear Possibility of and quantitative characteristics for power plants. hydrogen formation in steam/metal reactions; Leakage of a steam-gas mixture from the reactor in the Safety-related tasks (see Fig. 2) event of the I1" iktightness of the primary circuit being Sufficient expertise has been acquired in practically all areas of disrupted and the possibility of localized volumes with an hydrogen safety to identify the problems which need to be resolved; enhanced hydrogen concentration forming as a relilt. Che quantitative characteristics of the processes have been more or 1.3. In normal and emergency situations: less defined and steps are being taken to ensure safe conditions in each of the individual cases examined. However, a certain breach seems 67 The concentration limits for the ignition of hydrogen in a medium containing oxygen; to have occurred in the co-operation between specialists in the field of power engineering and chose in ocher fields relating co hydrogen To devise and introduce efficient, non-samD L:ng methods or safety. It would therefore be useful to co-ordinate che efforts ot the monitoring the hydrogen content ot water and 3a se s, construe: different specialists within one comprehensive programme. Such a pro- monitoring devices on :he oasis of che protan conductivity jf gramme would have to cover all possible facets of the problem, from the solid electrolytes and devise automated hydrogen-detect ion processes involved in hydrogen generation m the reactor to rhe conse- systems ; quences of its combustion. Its aims should include the fol lowing: To devise and introduce electrochemical methods for régula:ing To determine quantitative indices for the hydrogen content of reactor gas conditions; reactor-circuit water under various water-gas conditions, To investigate and introduce chenna I-engineering and hydrodynamic particular attention being given to the consequences of condi- means of prevent ing a build-up of hydrogen co dangerous concen- tions under which the coolant boils; trations (temperature conditions in areas of steam condensation, - To determine quantitative indices for che process of hydrogen methods of mixing gases); formation in steam-metal reactions and to formulate measures To investigate and determine the most acceptable and potentiaLIy to prevent such reactions; useful means (both chemical and electrochemical) of control led "o determine quantitative indices for che processes by which hydrogen combustion; hydrogen builds up in the reactor and reactor circuit when To devise a system for suppressing processes leading co detona- there are areas of steam condensation and to investigate che tion inside the reactor, in spaces in the engineering equipment, influence of che physical properties of various gas mixtures in the reactor building and containment systems; on the accumulation and distribution of mixtures containing To devise a systematic approach to the evaluation of hydrogen hydrogen; safety under normal operating conditions and in emergency sit- - To investigate che discharge of screams of steam-g^s mixtures uations at nuclear power plants with a view to defining criteria Into cramped spaces in reactor equipment and free spaces in and drafting standards on hydrogen safety. the service rooms and containment: system; Conclusion To study the inTluence of various desensitizers (steam, nitrogen Many of the aspects of hydrogen safety and research aims listed above and helium) and environmental parameters (pressure, temperature have already been raised on previous occasions and successful practical and humidity) on hydrogen ignition concentrât ion limits; solutions have been found in the chemical industry in general, in fuel- To study the process of hydrogen combustion (degree of pressure gas production and in other areas of Industrial activity. Nevertheless, increase, flame propagation rate) and che influence of its as in the operation of nuclear power plants, there arise accual condi- volume and configuration on the way in which the process tions and factors of a highly specific nature which may either complicate develops; or facilitate the solution of the problem and which cannot be disregarded To determine che critical detonation parameters and the condi- in any practical analysis of potential situations. Accordingly, the tions under which the transition is made from deflagration to problem of ensuring hydrogen safety ac such plants should be cackled by detonation in hydrogen-oxygen mixtures diluted by nitrogen, nuclear power specialists in collaboration with specialists from other sceam and helium; branches of science and technology. In this regard a great deal of atten- tion should be given Co che choice of research nethods (i.e. experimental or analytical) and also chat of experimental conditions (i.e. whether research should be based on models or full-scale experiments; as well as the degree to which the analytical models simulate actual situations. ANNEX 2 Since the problem of hydrogen safety is 3 verv real one and tine level of COMPREHENSIVE .RESEARCH PROGRAMME expertise gained so far is not sufficient to permit an evaluation D: ill the consequences of reactor accidents involving hydrogen, the undertaking of a comprehensive research programme is both necessary and full/ jus"::iea. REFERENCES [1] VERHOVETSKIJ, S.A. et al., At. Ehn«rg., 51_ 6 (1982) 373. [2] ANAN'EV, à.P. «t il., Ac. Ehn«rg., ,32 1 (1968) 10. ANNEX 1 HYDROGEN tN HYDROGEN tN FACILITY THE REACTOR BUILDINGS AND CIRCUIT CONTAINMENT SYSTEMS ASPECTS OF HYDROGEN SAFETY NORMAL OPERATION SYSTEMATIC HYDROGEN ELIMINATION HYDROGEN GENERATION ACCIDENTS LEADING TO HYDROGEN RADIOLYSIS SUPPRESSION AND COOLANT LEAKAGE METHODS PREVENTION GASEOUS CONDITIONS OF HYDROGEN FORMATION IN STEAM/ DETONATION METAL REACTIONS DISTRIBUTION OF HYDROGEN I HYDROGEN LEAKAGE INTO REACTOR - THROUGH THE REACTOR CIRCUIT I BUILDING AND CONTAINMENT SYSTEM SAFETY CRITERIA ANC STANDARDS CONCENTRATION LIMITS FOR DETONATION IGNITION Aspects of hydrogen safety Fig. 2. Comprehensive research programme 69 COMBUSTION PROCESSES 55 CONSEQUENCES 70 ASSESSMENT OF HYDROGEN RISK IN ultimate strengtn of French containment buildings have led to the ) 56 FRENCH PRESSURIZED WATER NUCLEAR REACTORS conclusion that containment integrity should not be questioned. However the problem of the likelihood of local detonations and of J. DUCO, L. ROUSSEAU, J.M. EVRARD their impact on structures is still examined : studies on hydrogen Institut de protection et de sûreté nucléaire, spatial distribution within the containment, on hydrogen detonation Commissariat à l'énergie atomique, wave modelization and on containment building response are Fontenay-aux-Roses, France underway. Another aspect of the hydrogen risk is the possible Abstract impairment of safety-related equipment in the containment during an eventual hydrogen combustion or explosion ; this point will be The provisions made to allow the French PWRs to behave probably emphasized in the future dnd some tests on materials are satisfactorily in accident situations have been based for years on already envisaged. the concept of conservative, conventional design basis accidents, and were supposed to be sufficient enough to cover the consequences Information coming from the EPRI research program on hydrogen of oil credible accident sequences. combustion and control, in which the safety body (CEA/1PSN) and the utility (EDF) jointly took a participation, is to back up our own Due to the reactor operating experience and to the development studies on hydrogen risk analysis. of the risk assessment methods, such a philosophy has heen refined and quantitative safety objectives have been set up for the plan; Up to know all the results we gathered as regards the design as regards the protection of the surrounding population and integrity of the large, dry containments of the French PWRs against the prevention of core-melt. Risks resulting from the development hydrogen explosion tend to relegate such a risk to the level of a of specific severe accident scenarios have to be first assessed, residual risk. before decisions are made as regards the setting up of Although the studies are going on, we do not expect a drastic complementary design improvements and/or the elaboration of adapted change of this trend in the future. Therefore no hydrogen specific operating procedures. design modification is presently required by the safety authorities. During the course of a severe core accident, hydrogen evolving, mainly due to zircaloy cladding-steam reaction, may form INTRODUCTION early a flammable mixture in the containment. As a general assumption one can say that the safety of a The risk of a short-term containment failure due to a hydrogen nuclear reactor is based on a gooo design, an efficient explosion, which would result in a large radioactivity release into construction, and effective0 operating rules. Nevertheless the the environment, is currently being assessed for the various types failure of major safety arrangements, even designed and implemented of large dry containments existing for French PWRs. according to the state-of-the-art rules, is not inconceivable ; U In this iramework, comparisons between pressure peaks, due to the has m fact one probability of occurrence and will result in most severe conceivable hydrogen déflagration, and the realistic radiological consequences for the general public. Ihe terms of probability and consequence have to be considered c) the third level consists of an analysis of plant behavior in jointly and compared to safety objectives ; this can result in the the event of accidents, which are supposed to cover all specification of some extra safety arrangements, for reducing the accidental sequences resulting "rom possible failures ; this probability of failure, or mitigating the external consequences of analysis must demonstrate Mat the action of safeguard systems such a failure, or both, so as to meet the aforementioned safety is able to limit the consequences to the plant and to the objectives. environment below given limits ; it must be done with Hydrogen evolving, resulting from an hypothetical severe conservative assumptions, both for the accidental scenarios and accident with an extensive core damage, may form flammable mixtures for the evaluation of plant behavior. However it raises some with air and steam within the reactor containment. When ignited, difficult questions, such as the guarantee of exhaustivity of such mixtures burn or explode, possibly threatening the containment the 11st of design basis accidents and their consistency with tightness, and/or impairing some safety equipment inside the the overall safety level of the plant. containment. The relevant risk has to be assessed 1n a best-estimate manner, as far as feasible, and confronted with the 1.2 Design basis situations safety objectives ; the decision to control hydrogen or not within the containment will be highly dependent on the issue of such an According to sucn a safety approach, a list of accident analysis. situations has been prepared l-y the designers, the accidents beeing classified into four categories, depending upon their estimated 1 FRENCH APPROACH TO SAFETY DOCTRINE (1) probabilities ot occurrence, as shown on the table below : 1.1 The defense-in-depth principle Class Annual frequency F | Radiological consequences The French safety philosophy, which was based at the origin on 2 the NRC regulation for the first PWR plants, progressively took I and II 10" < F ' authorized releases i shape as were developed the programs to extend the production of 4 2 electricity of nuclear origin. III 10" < F< 1Q" I 500 mrem whole body dose i In fact the French safety approach is based on the IV 15 rem whole body dose defense-in-depth principle, which can presently be formulated in the following terms : It should'be noticed that such s table is not actually part of any official regulation, and that the figures are to be consioered a) The first level is to provide sufficient safety margins at the as strictly indicative. design, construction and operation stages in order to guarantee a good behavior of the plant in normal operation ; The engineered safety features, which are intended to avoid unacceptable consequences both in the short term (automatic b) the second level requires to implement the protection systems actions) and in the long tern (A-type operating procedures in case with the necessary redundancy, so that they are able in all of an accident), should the worst of these situations occur, are anticipated transients and incidents to bring back the plant designed using, in particular, the single failure criterion. 71 into its normal operating domain ; 12 1.3 General design safety opjective for population protection of the plan is a basic assumption, which implies no early containment failure : this point is closely related to the In 1977, after intense discussions Between the safety assessment of the hydrogen risk ; authorities ano the utility (EDF), the need for safety objectives was recognized in terms of probability and consequences ; this b) the populations to be protected, before a major release of would solve the issue of accidents "beyono design basis", or "at radioactivity occurs, shall not be situated farther than 10 km the limit of design basis". These objectives were defined in the from the nuclear site ; more specifically, evacuation is to be most simple way ; envisaged merely within a 5 km radius from the nuclear site, and confinement beyond this limit, within the above 10 km radius "As a general objective, the design of the installations of a from the site ; single unit comprising one pressurized water nuclear reactor should be such than the overall probability that the said unit can induce c) no further protective action is required in the few days unacceptable consequences will not exceed 10 per year". following the actual release of the bulk of the radioactivity. In fact this probability level is to be considered as an order Assumptions b) and c) Imply a limitation of the activity of magnitude, an indication of the objective to be sought, rather released, according to the maximum acceptable individual exposures. than a strict limit. At this stage, let us identifiy as S, the level of such a maximum source term compatible with an easily feasible PPI implementation. 1.4 Emergency planning for the public As it will be seen below, Si, level corresponds, for core-melt accidents, to the normal leakage of the aerial part of the Later on it was specified that "unacceptable consequences" did containment, plus a delayed release through the soil, following the refer to the necessity of short-term evacuation just around the basemat melt-through. site : there 1s no corresponding regulatory limit 1n France, since the medical experts think that it must be determined on a 1.5 Corollary design safety objective for core-melt prevention cast-by-case basis, but the corresponding individual exposure is betwtm S and 50 rems, equivalent whole body dose. As a corollary to the above safety objective, a secondary objective was set up by the safety authorities as regards the rules Adapted measures are elaborated to provide protection to the for taking Into account or not, for the design of the plant, some general public in the frame of the so-called Particular Emergency ominous "groups of events" as they can ce Identified in the event Plans ("Plans Particuliers d'Intervention", in French, PPI fcr trees : short). "When a probabilistic approach is to be used to assess So as to get a good chance of success in plan implementation, the whether a group of events should be allowed for in the design of a PPIs are based on the following assumptions : unit, it should be assumed that this group of events must Be allowed for if the probability that it may lead to unacceptable a) No major release of radioactivity is anticipated during the consequences exceeds 10" per year". first 12-24 hours ; this time allowed for a safe implementation As similarly said for the general design safety objective for 1.6 The "h" operating procedures population protection, the "10" per year" level is to be considered rather as an order of magnitude than as a strict The above operating procedures, called "H" procedures, are at threshold. "Unacceptable consequences" refers here to ar extensive the limit of the design, wnich means they presently do not corply core damage resulting in reactor vessel melt-through. Strictly with normal design rules. Due to this fact and also to uncertainties linked to human As a consequence of this corollary design safety objective for factors, the probability of failure of a particular H procedure core-melt prevention, it is necessary to evaluate the consequences cannot be assessed at less than 10" per reactor-year. This means of the total loss of several systems, important for the plant that the implementation of an adapted H procedure allows the safety, for which the designers have provided some redundancy in corollary design safety objective of 10" per year to be satisfied application of deterministic safety criteria. in the case of a series of failures that would otherwise result in an extensive core damage with a probability of occurrence possibly The issue of such assessments makes it clear that, to satisfy as high as 10 per reactor-year. the preceding corollary design safety objective, some complementary safety arrangements have to be provided, particularly on systems Up to now five H procedures have been defined for the frequently in operation. following categories of events : Two directions are currently investigated : a) Loss of external heat sink H,) b) Total loss of main and auxiliary feedwater to the steam a) An increase in the redundancy and/or in the diversification generators (H2) ; of equipment of the systems concerned ; as an example, c) Total loss of off-site and on-site power supplies (H,) ; design modifications have been made on the 1300 MWe plants d) Long-term failure of safeguards in case of a LOCA (H^) ; so as to take into account the anticipated transients e) Flood exceeding the one-in-thousand-year flood level, for some without scram (ATWS) : the solution was found in the river sites (H.). diversification of the signals ordering the turbine trip and the start-up of auxiliary feedwater ; for the nuclear Procedures H,, H. and H- are presently operational ; procedure power units of the IW standard, ATWS will be taken into H, is due in one year, and procedure H^ a little later on the 1300 account in the design ; MWe reactor at Paluel. b) The elaboration of special operating procedures, aimeo at 1. i The "I!.," operating scheme for cor°-melt prevention preventing a major core degradation by stabilizing the situation at a safe level during a period of time The above procedures for accident management - either A, with sufficient for the recovery of the failed function. correct operation of engineered safety features, or H, with failure of a redundant system - are based on an analysis of clearly identified families of accidental sequences, however all such 73 measures may be inadeouate if the operator finds himself in a 74 situation where the reactor does not benave as he expected, which 1.8 The "U" operating procedures for class-9 accident is a very disturbing case, since he cannot be sure of the right way mitigation to reacn a safe state. In the treatment of very severe accidents, we have tc rely or. In oroer to attempt to stop the oeve'iopment o* potentu'ily the knowledge of the physical phenomena which govern tne serious situations which could lead to severe core dégradation if propagation of damage in core, and on the analysis of accidental the proper actions are not taken, a new approach has been proposed sequences leading to containment failure. by the utility (EDF), based on the cnaractenzation of all possiole cooling states of the core, with their stability ranges and This is a field where intense research is developed around the transitions. This approach, which requires some additional world, and where more work is still needed before we can reach instrumentation in the primary system, has been approved by the final conclusions. However our approach is funded on tne following safety authorities. assumptions : a) A steam explosion, energetic enough to break the containment and A preliminary operating scheme, called U. (U as "ultimate") lead to a very high radioactive release within a few hours after has been arawn, based on the cooling state approach, wmch reactor shut-down, is very unlikely indeed, if not completely represents the final step in the prevention of core-meltdown. In impossible ; this scheme any available means will be used to avoid core degradation or, if some degradation already occurred, to keep the b) Hydrogen explosions will not be able to give rise to large core inside the reactor vessel. breaches in the containment, at least in the case of the large dry containments used in French PWRs ; this point results from If Uj operating scheme fails, it is clear that the the analyses we have made up to now on the hydrogen risk ; containment, which is designed to withstand the consequences of nevertheless studies are going on, and special provisions to specific design basis accidents, will not be able to support all control hydrogen concentrations could be needed if the results severe conditions resulting from core melt-down. This is the reason did not confirm the present evaluations : these aspects will be why a subsidiary safety objective has been defined, which can be developed below in § 2 ; summarized as follows : "In the case of a core-melt, the containment should constitute c) Containment isolation failures, which may result from leakage at an ultimate line of defense which would reduce with a reasonable penetrations or failure of isolation systems, must be taken into probability the radioactive releases in the environment at a level account. A specific procedure -U2- has been defined, in order to compatible with a feasible off-site emergency plan". detect, localize and repair any containment isolation failure ; A reasonable probability 1s presently understood as 90-99;, d) The basemat melt-through is the less improbable failure mode of and the level of radioactive releases compatible with the the containment, but it is also the one which leads to the feasibility of a PP1 is the S3 level we have already mentioned in lowest radioactive release in the environment. So it satisfies § 1.4. the above subsidiary safety objective, except in some 1300 Mwe plants, with double containment and a drainage system in the basemat, which could create a pathway for gaseous and volatile 2 HYDROGEN RISK ASSESSMENT IN FRENCH PWRs fission products towards the atmosphere. To prevent this unwanted phenomenon, the U. procedure was set up ; 2.1 Probable containment failure modes for typical core-melt accident scenarios. Relevant radioactive releases. e) Finally, depending on the assumptions made about the erosion speed of the concrete a--' the corresponding gas production, one In order to specify consequence mitigation measures, which cannot exclude a slow increase of pressure inside the would be part of the in-plant emergency plan, a set of three containment, and after some time, a loss of integrity. typical core-melt accident scenarios has been selected, to assess The corresponding release of fission products is difficult to the delays before containment loss of integrity and the assess, and, as an extra precaution, it was decided to install corresponding releases of radioactivity into the environment. on all French PWRs a system which would make possible a controlled ana filtered venting of the containment. These scenarios are, according to the WASH-1400 terminology : The filtration efficiency is modest : 10, as an objective, and a) AB : A loss of coolant accident with a complete loss of a corresponding special procedure, Uç, defines how it should electric power. This sequence represents what would happen be used. after a design basis accident should all engineered safeguards be unavailable 1 To sum up, the treatment of class-9 accidents in our safety approach makes use of 4 ultimate procedures : b) TMLB' : Loss of electric power with no feedwater supply. This accident was found in WASH-1400 as a dominant a) U,, to cope with a containment isolation failure (pmode, contributor to the risk. Because the primary system is according to the WASH-1400 report) ; pressurized at the time the vessel fails, an extensive release of steam occurs due to depressui ization and to the b) U^, to prevent a direct pathway to the atmosphere in the event interaction between the fuel and the water injected from of a basemat melt-through (c-mode) ; the accumulators ; c) Uj, containment depressurization through coarse filters (d-mode) c) AC : A loss of coolant accident with failure of the containment spray system. In this sequence, a large a) Uj, which deals with the use of mobile units to rescue the fraction of emergency water is vaporized to the containment safeguard systems, and is a complement of the H^ procedure while no heat removal system is available. mentioned above. Finally , for the other two failure modes, we consider The BOIL code (2) has been used to treat the core heat-up, the that theo^ -mode is very unlikely, and that provisions for hydrogen generation, the fission product release and the cor» melt hydrogen control could take care of the 'Ç'-mode of failure, as far progression. Pressure and temperature responses in the containment as necessary. 75 were calculated by the PAREO code (3). The failure pressure ranges of the reactor containment were then estimated at 3 - 9 bar for tne 76 - But the core would continue to be cooled during a long time 900 MUe units, which haye a prestressed containment with an delay corresponding to the vaporization of recirculating internal steel liner (figure 1), and at 6 - 6.5 ba»- for tne 1300 water ; this time is expected to be employed to feed the MUe concrete double-containment units (figure 2). recirculation system with extra water to compensate for vaporization ; These calculations, which assumed a likely qussi-continuous burning of hydrogen as it was generated, led to the following - Early radioactive release associated with possible cladding conclusions : failures will be low. a) AB and TMLB" scenarios 2.2 Containment-scale hvdrooen deflagration - No containment failure would occur during the first stage Calculations made for the AB and TMLB1 seauences do not assume of the accidents from steam overpressurization ; significant hydrogen accumulation in the containment, which is the most probable issue. However, the hypothetical case of a - Containment failure would occur beyond 16 hours from s containment-scale hydrogen accumulation and explosion has been competition between stean-gas accumulation in the containment {Q- examined, to assess the risk of an early release of radioactivity mode) and floor melt-through (f.mode) ; into the environment by a ^-mode containment failure, which could severely hinder the implementation of a PPI. Figure 3 gives the - The fission product release levels would be quoted at S. for amount of hydrogen that would be present during the second hour the two failure modes on 1300 HWe units, and at S, and S, for the <$. after the beginning of the AB sequence on a 1300 HWe reactor, and£.failure modes, respectively, on 900 MWe units. assuming a best-estimate value of 75" for the fraction of zircaloy core inventory that could be oxidized. For clarification S^ corresponds to significantly delayed (j>i6 hours) containment failures with substantial retention of Due to convection flows in the containment, the corresponding the fission products in the containment, whereas S, - which hydrogen concentration is supposed to be homogeneous : figure 3 is smaller than Sj - is related to the normal leakage rate gives the level of such a concentration in a 1300 MWe containment of the aerial part of the containment, plus a delayed release to for AB sequence. It tends to prove - according to the SHAPIRO the atmosphere through the soil after basemat melt-through ; diagram of the flammability and the detonability limits for S. is the reference level for radioactivity release that would be hydrogen - air - steam mixtures (figure 4) - that a containment- easily handled by emergency plans for usual site features ; scale slow kinetics combustion of all the hydrogen accumulated would be conceivable, but not a detonation. b) AC scenario Independently of any particular accident sequence, the highest - The containment would fail at early times from steam possible quasi-static pressure would be obtained in combining a accumulation ; deflagration burning the totality of hydrogen possibly available from Zr-HjO reaction with the highest vapor pressure that makes a deflagration possible. A parametric study has been made to assess the temperature and d) Estimate the containment leak rate evolution as a function of quasi-static pressure in the containment, as a consequence of an the building behavior ; adiabatic combustion of a hydrogen - air - steam ternary mixture ; the rssulting containment pressures are plotted on figure 5 vs e) Assess the actual leak rate for air-steam mixtures. initial steam pressure - which is supposed to relate to saturated steam - for various hydrogen releases corresponding to a given set The accident conditions taken into account in such a of zircaloy oxidation rates. calculation are those of an ADC-type scenario, accoroing to the WASH-1400 terminology, that is a large break LOCA with the failures Two criteria have been used to identify the domains where of the emergency core cooling injection system and of the combustion is actually possible : the SHAPIRO diagram already containment spray injection system. mentioned, and a criterion based on a minimum flame temperature of 710 'C which is, according to (4), necessary to ensure a The initial conditions in the containment are a temperature of self-sustained flame propagation ; in fact the latter criterion is 140 °C and an overpressure of 3.8 (130G HWe) to 4 bar (S00 MWe) ; to be considered as pessimistic at high vapor pressure. the temperature and pressure increments curing the course of the beyond-design transient are 2 °C and .3 bar pe' hour, respectively, The issue of the study is that, for : best-estimate zircaloy which is quite moderate. oxidation rate of 75!, the maximum containnent pressure is about 7.5 bar. The temperature field in the structure is first calculated ; the displacements and the structure stresses under the above Such a pressure woulo not entail the 900 MWe containment loading are then assessed by non-Hnear mechanics calculations. integrity according to its estimated failure pressure range of 8-S bar, but would raise some questions as regsras the tightness of the The mam results to be kept in mind are the following : inner wall of the 1300 MWe containment. a) For the 900 MWe containments, the ruin of the upper part 2.3 Improved estimates of the containment failure pressures comes first, caused by the rupture of the prestressed liner on the dome ; the internal overpressure in then about 12 bar. Since the first assessments of the failure pressures of the French containments, the safety body (CEA/ÏPSN) and the utility b) For the 1300 HWe double containments, the ruin of the upper (EOF) agreed to achieve more sophisticated calculations in the part similarly comes first, but at about 8 bar internal beyond-design domain so as to : overpressure. Nevertheless some cracks propagating through the whole inner wall thickness are anticipated at an early stage a) Identifiy the parts or components of the containments which are before the ruin of the upper part of the containment. the less pressure-resistant ; Whether such cracks would induce substantial containment leakages or not is being experimentally studied ; however present b) Assess the pressure causing concrete initial cracking ; assumptions consider that such a leakage could be satisfactorily dealt with by the venting of the gap between the two walls of the 77 c) Evaluate the failure pressure and the mechanisms of containment 63 1300 MWe double containment. ruin ; 78 As a conclusion, if one supposes that hydrogen generated by d) Even if there was a coincidence between transient 64 severe core damage could accumulate without burning in the Frencn conditions allowing detonation and ignition of the mixture, a PWR containments and that the convection flows mix it fraction of the total hydrogen amount would orly be concerned homogeneously, containment-scale combustion appears possible, if in the detonation process. some delayed ignition occurs, leading to a maximum internal pressure of 7.5 bar. This pressure is not high enough, according to As regards the assessment of hydrogen spatial distribution in present calculations, to jeopardize the integrity of the large dry a PWR containment, some plans have been made at CEA to elaborate a containments of French PWRs. simplified computer code. This effort will voluntarily be limited to the level necessary to a good valorization of relevant results 2.4 Localized hydrogen detonation of the EPRI research program on hydrogen combustion and control (cf § 2-6), in which the safety body and the utility took a The potential detonation risk due to possible hydrogen participation. pocketing has also been considered. A transient .non uniform distribution of hydrogen could indeed lead to a detonation, if Therefore hydrojen detonation matters have been dealt with in ignited at concentrations higher than 19Ï in volume, provided that a conservative way. In the hypothetical case examined, the steam concentrations be lower than one bar at most (figure 4). oxidation of 100Ï of the zircaloy core inventory of a 1300 MWe reactor is assumed, and all evolving hydrogen is supposed to The identification of situations meeting these conditions accumulate under the containment done, forming a detonable requires complex studies taking into account breech location, time mixture ; ignition is then assumed on the containment axis. dependent flow rate and composition of the released gas mixture, compartment design and connections, convection flows, effect of The time dependent pressure on the containing' wall, containment spray. Several reasons however tens to indicate that a corresponding to the direct impact of the detonation wave, has been detonation hazard would be low : assessed ; it consists of two parts a) a pressure pulse - about 60 bar during some milliseconds - and b) the residual pressure roughly a) The hydrogen generation is usually preceded by an Intense corresponding to constant volume combustion. vapor release which heats up the atmosphere and structures of the compartment containing the breach, leading to high saturated steam Preliminary calculations of displacements and stresses have pressures ; been made with the PLEXUS code (5) for a section of the cylindrical part of the wall : the present issue is that the static pressure b) Hydrogen is released in a hot gas mixture, rich in vapor, has a dominant effect ana thaï the pressure pulse will just induce which tends to rise and mix with air in the upper regions of the some cracks in the concrete. containment where convection flows and possible spray operation Complementary calculations are still needed to take into account would enhance homogenization ; the cylinder heads, but our present feeling is that the above conclusion should not be drastically changed. c) Hydrogen may burn when the mixture concentrations cross the deflagration domain on figure 4 before reaching detonation Our provisional conclusion is then that hydrogen detonation is conditions ; highly unlikely, but if it should occur, the dynamic effect of the pressure pulse on .the containment walls would not impair be placed in the containment, should such a decision be made in the significantly the containment tightness. future. 2.5 Behavior of safeguard equipment The large scale demonstration tests in a 2100 mJ vessel at Nevada Test Site, for the study of equipment survivability, is Still limited data exist on the impact of an hyorogen liable to bring us valuable information for strengthening some explosion on the components of safeguard systems in the component specifications. The large-scale qualification of hydrogen contiennent. These materials should withstand thermal loads, 2nd control systems 1s also an important item to look at, so long the mechanical ones too in the case of a local detonation. eventual implementation of such systems on French PWRs will not be definitely ruled out. in fact, largt mechanical elements of tht safeguard equipments Apart from EPRI program support, efforts will be continued at CEA arc thought liable to survive high temperatures for minutes, due to on the localized detonation ease : a better characterisation of the their design and their important heat capacity. On the opposite, containment loading will be looked for, and the relevant non-metal 11c parts, especially some electric components, are a containment response will be studied, taking into account this time sourct of concern and could require local protections. the full building geometry. A '1st of small key components of safeguard systems used on French PWRs has bten stt up, Including electrically-operated gates, Conclusion pressure transmitters, connecting boxts, cables and filling resins. As the hydrogen risk analysis study is still going-on, bised These components are intended tc bt tested in « next future on developing experimental data and improved phenomenon modelling, under the conditions of an hydrogen deflagration ; an adapted the present conclusion may not Indeed be considered as final. Instrumentation should allow the tests to be performed in operating conditions. According to tht design safety objectives presented in S 1. there in two key Issues as regards hydrogen evolving and 2.6 Next sttpi of thu studies. Back-up from EPRI research subsequent burning In a PWR containment : çrecrtm on hydrogen combustion tnd control a) The eventuality of an early containment failure, resulting in In 1982. tht Hftty body (CEA/IPSN) and tht utility (EOF) largt rtlusts of rid1oact1v1ty, which could not be dtalt with dec 1fled jointly to tnttr tht EPRI rtstarch program on hyorogen by emergency planning ; combustion tnd control. This program comprises several research subjtcts (6), some of them being Habit to provide a sound back-up b) The Impairing of elements of safeguard systems, not specified to the development of our risk analysis studies. tor the corresponding thermal and mechanical loads, which could lead to a more degraded situation with a potentially higher As regards the problems of hydrogen mixing and distribution in release of radioactivity. the containment, the experiments et HEDl and the relevant modelling at Battelle should help us to assess the risk of hydrogen As regards item »), pessimistic calculations of containment 79 pocketing, and better define the positions of hydrogen detectors to loading and response have been made, tending to prove that 900 MUe reactor building integrity should not be questioned. On 1300 «We 6) L. THOMPSON, "EPR] Research on Hydrogen Combustion and Control", reactors, an hydrogen explosion will possibly increase the leak Proceedings of the Second International Conference on the Impact rate through the inner wall, due to some crossing cracks in the of Hydrogen on Water Reactor Safety, October 3-7, 1982, concrete ; this point is to be further investigated, but the Albuquerque, New Mexico. present feeling is that the situation could be dealt with by the containment gap venting ; Concerning item b), the first results of equipment survivability tests from the EPRI program showed no visible or functional degradation. Further tests are needeo at a larger scale and on the specific materials used in France, so as to confirm that trend ; anyway some "strengthening" of limited materials could be easily conceived and implemented, as far as necessary. As a result of the above considerations, no P'*R design modification is presently required by the French safety authorities ana, in particular, no hydrogen control systems have yet been planned to be set up. REFERENCES 1) P. TANGUY, "French Safety Philosophy", Nuclear Safety, to be published 2) R.O. WOOTON, "BOIL 1 - A Computer Program to Calculate Core Heat-up and Meltdown in a Coolant Boiloff Accident", Battelle Colombus Laboratories (March 1975) 3) D. ROY, "Description du code PAREO 6", Note technique EDF/SEPTEN Ne TC/76-18 (Mai 1977) 4) L.E. HOCHREITER et al, "PWR Containment Atmospheric Response for Postulated Class-9 Accident", Transactions of ANS, 1980 Winter-meeting, p.3Z2 5) P. A1LLAUD, K. LEPAREUX, "Code PLEXUS ; Explosion d'hydrogène" Rapport DEMT/SMTS/LAMS/83-19 Fie i 900 MW (CPU REACTOR CONTAINMENT 'Hydrogen MCSS Of concenîrcïicn hyflrDoer- ' I V, . 1C0C 75% Zr-H.^O reaction | Time ( s I 35CC LCCO i;00 500C Fig 3 _ HYDROGEN CONCENTRATION IN THE 1300MW Fig 2 . 1300 MW ( P ', ) REACTOR C0NTAINMEN1 PI-CONTAINMENT DURING SEQUENCE AB 67 82 "Ml.2 >/ y \/ „ 100 80 60 20 H vdroger Fig. 4_ FLAMMABILITY LIMITS OF HYDROGEN- Fig 5. MAXIMUM CONTAINMENT FRESSURE AFTEr -AIR-STEAM MIXTURES (SHAPIRO) BURNING Or A HYDROGEN . AIR _ SATURATED STEiM MIXTURE ANALYSIS OF THE EFFECTS OF HYDROGEN BURNING -' . - - : : ' ' <Î -. . ^ . i •> c •"- **• W. Trier e ib one t JC Dîne Teneta AND MEASURES TAKEN FOR THEIR MITIGATION . "i'i i -r- *.." t *" -.t: • : irenai ^i:culatior. pumps. The containment 11 AT THE LOVIISA NUCLEAR POWER PLANT .3. = ~. ";-e ". • b - _ . t. ^jpre^s..;, t/pe , consisting ot prsstresbed ; ; sr. ",.1'ledi p i. Ar.*;3 nas b«en quite sat 13 f 3. B. REGNELL, S. HELYNEN • r.a: • . t. ^. «3 rsqe capacity factor nor t.^.e t". I mat ran Voima Oy, Helsinki, Finland Abstract ' * \tir\L :';i .î AND Mi ."I J AT ING The Lovusa plane units are furnished wi;r. a près s ur1 zed wa-.ei re 5etc r /, . . i :••. L . • .^ecuv .ip; i : "-ïf, t c:iat eac i 1 er assj.141r.10ns and an ice condenser type ?-*tainmen,r . The reiat 1 ,'e 1 y low ;les iqr, j e : », J:;J:I j r,t; cirabusr^on "it P. yd co^en h ad to he revised, I" pressure cf this type of containment er.nances the need fa analysis r.f ".at 'Miii wi •_ r ./. -:ut ice JC le .ne! t : nq of the fuel a large t !. .ic 3 '. ii'7 ^a^.t'E • al rouid react, ^esulti :ig in large amount s the consequences of burning of postulated large amounts of nydroger. r released into the containment. In 198 2 sucr. an analysis was urn-r take, •; ::.<-. -L- .- jSt 1 jr. jf hydro^er. Hydrogen releases for several accident scenarios we re etrimauo >s-nc e Sj.^r.y A.:' not ; ; ] cs 1 .1 c'ln 1 -via . i.e. t.:. e Insti^Jte of Hadut::r the March code .Tne response of trie cont&inmei.t to the contu-.i'.«;c L-urn; : r of the released hydrogen was calculated as 1 no the Ciasix compare. :o.je .ot ,'C i',r. I s? » , ..r»ôuid 1.1 ward s r.'.e eac 01 ' 3â0 recmrpmer.ts regard; '.: developed by Offshore Power Systems (a subsidiary of Hest.mcr.cjt-?;. T1^ yd.•-.••; en C-_-:K- • 4'- >. 'ir and : ne u 1 r ..mate pressure retiiniiig capab L! 1 ty 'I assumption of controlled barn ma JcS based on the planned install at ;o:, ie c~:r.tji.iv-.r.r Plant o»••. 1 •*r5 we:e aste*d to 3naJ /as the consuqaeTces .: "a Mi?.ierd' 1 » 1 Aii'l cu.r.oui t ion of hydrogen i.-om met al -water react 10 na of hydrogen igniters in tne conta 1r.pien.t. ;v;..:-;i:iq :0 _ .50 * ot trie oiaddir^ .nacerial in contact with fuel . Ir. ovy:rice_* i i -i I the c>:q JI: e i.er.ts »jere made ."noce demand 1 ng , me re as 1 nq r.h-= The result of this anal y s is indicated that :rie coir.ainmtr^ cc. 1 d w : t - - ;"tïcr. ..ori of ir.tcai wat yr :eacticr. te / S \. stand the pressure and temperature load ings imposed by tne Dur,;.:.,.-j hydrogen. i AC* . : . 0 <. s1 1 \\ 1 i'istrumantàt ion for :oedsur me the hydrogen Subsequently to this anal y s is work a mitigating system cons is - - - ., .;,..- . . >• ..' . ,r 1 i.?i"-;- 2 31^:11 £ 1 ;a:.c* in the case of the V": r i.. ; " .. L-;i^i..,c i' >i : • ^ ' J : .-,me iL s JC» inerted , and hydrogen burns INTRODUCTION > .->•' J-.-.-'. i - a; -; L." •--;:•;••:. The 5 : "'i^t . on \n tne Lov n sa plant was : ' •-•• r •_• i! , ..(..wj .••::. , ;^. „* • ) tr.fi '.ibe of the L.*e condenser containment. Presently, four nuclear power plant un 11^ are it* cortnitc ^ c. -, 1 "pe.^L Finland, at two different sites and ope rat ed oy two uuliues. we ex 1,iei.se 1 is the relatively i:i er, ' urns cut to be a drawDack frcu The state-owned Imatran Voima Oy 1 i **c ^ ooeratts n 1 UJClear p The containment strength calculations were performed D> T"'\. ' S own experts. As a result of tne se calculation we coula witr. reason aal e confidence state that the conta- nment would n jt f 3 -1 jr-1 e ", s r^n- over- pressure exceeds 2.25 bar. This is to De compared witr " The results show that hydrogen was ignited in the lower and upper The igniter system consists of about 70 glow plug type igniters which plenum of ice condenser, in the lower compartment and in separate rooms are divided in two separate independent groups. Igniters are located in located near the lower compartment (dead ended volumes). No burns took every separate room in the containment where hydrogen accumulation can place in the upper compartment. In every case the pressure increase is be expected, in order to avoid developing of high local hydrogen well below containment failure pressure (325 JcPa) and only in the most concentrations and uncontrolled burning. The igniter Installation, commissioning and testing In the first application of igniters for controlled hydrogen burning Installation of the igniter system was for the most part carried out at ordinary glov plugs intended for use m diesel engines were used both units during the annual refuelling outages at che end of >982. As 1/5/). Because these were low voltage devices, transformers had to be part of the commissioning tests all igniters were energized and the used between the electric power supply network and the igniters. These surface temperature of the heater element was measured using an optical transformers had to be qualified for use within the containment in Fyrometer. accident conditions. The transformers could be avoided by using ig- niters designed for operation at a higher voltage. In-service inspection and testing requirements are as follows: The igniters used at the Loviisa plant are manufactured by Tayco The surface temperature of the igniters shall be measured once Engineering Inc. USA, especially for this application. Extensive in five years testing, e.g. at the Whiteshell laboratories in Canada, have been carried out in order to ensure the capability of the igniters to re- The power consumption of every subgroup shall be measured once liably ignite hydrogen in various conditions. Also, the igniters have a year been environmentally qualified. The operating voltage of the Tayco igniters is 120...132 V AC and the surface temperature is about 870'C The resistance and the insulation resistance of all subgroups shall be measured every three months. System design The function of the remote controlled system operation from the control room shall be tested every three months. Igniters in both redundant groups are connected to the reliable power supply, i.e. they will oe fed from the dieselgenerators if necessary. By limiting the most frequent tests to resistance measurements, the Each group includes eight subgroups where from one to six igniters are number of thermal cycles imposed on the igniters is reduced. connected in parallel. The distance between igniter locations was selected to be not iwt= thar. stout 10 m, excluding the upper compart- ment, where no burns were predicted to occur. The igniters are actuated CONCLUSIONS manually when a loss of coolant accident is verified or when a contain- ment high pressure signal is received, intentional actuation of the Controlled burning of hydrogen using igniters is the mitigating method system does not cause any harmful consequences. selected for a number of ice condenser plants and for some boiling water reactors with non-inerted containments. Tests and analyses carried out indicate strongly that this approach is a viable one. In a recent paper R.E. Henry concludes that "intentional ignition with Environmental qualification numerous sources (igniters) is a very effective means of preventing hydrogen accumulation". The hydrogen mitigation system has to perform its function in the conditions created by the controlled hydrogen burns. The qualification Even so, uncertainties remain, e.g. regarding containment integrity, of the igniter themselves was already mentioned. The power supply the conditions for detonation, distribution of hydrogen in the contain- cables, connectors, etc have also to be qualified for this appli- ment and the understanding of the combustion process. Furthermore, it cation. In the Loviisa plant two different cables were used (Nokia, is of great importance that the survivability of components important type LJNSM and ACOME PLAHTROL, made in Finland and France, resp.), both to safety be ensured in hydrogen burn conditions. capable of withstanding high temperature conditions according to their technical specifications. Thus, both theoretical and experimental work remains to done before the hydrogen problem can be considered as solved. Testing at the peak temperatures reached in the hydrogen burns had not been done, however. Although the duration of these peaks is short, at most a few minutes, it was felt appropriate to carry out some ad- ACKNOWLEDGEMENTS ditional testing. At IVO'S own laboratories a fairly large electrical eguipment qualification program has been under way for several years. This work would not have been possible without the assistance of many New tests were devised for the simulation of hydrogen burn conditions, people. Our sincere gratitude goes to N. Liparulo and his group at the e.g. the cables were exposed to gas flames. Westinghouse Nuclear Coûter and to Dr Martin Fuis and his staff at the Offshore Power Systems. The fact that the igniter systems for the Lo- As a conséquence of the requirement to assume nydrogen burn in the viisa units could be designed, installed and commissioned in a very containment, qualification of safety related equipment besides the short time is the result o5 dedicated work by many people at IVO, igniter system has to be re-evaluated. This work is still in progress. T. Juntunen and K. Kontteli at IVO's Electrical Department deserving special thanks. REFERENCES 1 silvennoinen, P., EeriJcàinen, L. Research Report., Tecr .: :;31. Research Center. In Finnish, not published 2 Eerikàinen, L., Private communication 3 wooton, R. 0., Avci, H. I., March, Code Description and Us^r' Manual NUREC/CR-1711, BMI-2064, 1980 4 Fuis, G. M., The Clasix Computer Program for the Analysis .•••• Reactor Plant Containment Response to Hygrogen Release and Deflagration, offshore Power Syscems, OPS-36A31, '9s 1 5 DuKe Power Company, An Analysis of Hydrogen Control Measures at McGuire Nuclear Station, 1981 6 Henry, R. E., Modelling of Incomplete Burns for Containment Analysis. Proceedings, International Meeting on Light water Reactor Severe Accident Evaluation, Cambridge, Massachusetts, Aug. 28 to Sept. 1, 1983 if 1 87 HYOROGEN RELEASE TO CONTAINMENT IKG/S ) " § o z a TOTAL PRESSURE IPSIA) LC lJL< O "o o o O O o o" o o oo •— o1/1 Z S \ UJ g" O vV C-4 CO 30 (VISdl 3df1SS3ad 1V101 > \ —1 § \ O fa \ o O 8 o" 8 o O 31 (Jl 3UniVtJ3dH31 90 THE BASIS FOR SAFETY STANDARDS AIMED AT 76 AVERTING FIRE AND EXPLOSION HAZARDS DURING WORK INVOLVING HYDROGEN isu.g A.N. BARATOV All-Union Fire Fighting Scientific Research Institute, Moscow, i 140.0 _4 Union of Soviet Socialist Republics Translated from Russian 130,0 Abstract J Safety standards, [or .iv»rting tire ami explosion hazards due Lo hydrogen require ,i sound knowledge ot hydrogen propvrt les at various conditions. Re 1 e v.ini re;>e j rch work is referred to, re 1 Hydrogen is widely used nowadays in various branches of science and 100.0 technology, and it is also formed in some sources ar atomic eneiqy. 0,0 2000,0 M)O0,0 6000,0 8000,0 Hydrogen's unique combu- - • >r r>moerties are such as to call for thorough TIME (SECONDS) investigation of its explosive propertles and of the conditions under which it forms explosive mixtures with air and to demand the development of methods and equipment for prevent ing ignition of air-hydrogen mixtures and for FIG. 6. extinguishing any fires which occur. Only such informa**.on and equipment can provide a sound basis for measures to ensure the safety of work involving hydrogen. Numerous sources quote the following data with regard to the explosive properties of air-hydrogen mixtures.* Area Tne greatest hazard is presented by accidents in which hydrogen leaks :r"u the building and forms mixtures capable of detonation inside items of Combust ion region 4.12 - ""5. 0 vol. ft equipment. Gaseous or liquid hydrogen leaking or escaping as a result of an Minimum îqnit ion energy ~ 0.02 mJ accident forms potentially explosive media. Research carrled out m our Se lf-iqntion temperature - ?33 K institute I z-] has demons tr a ted tha*", depending on the cooling conditions at standard flame propagation velocity - 2.7 m/s -3 tht site of evaporation, the rate of evaporation from various surfaces Critical diameter - 0.6 x 10 (•?:ave 1, soi I, etc ranges from 0.014 to 0.44 kg* m -s . Gaseous Minimum explosive oxygen content. - 5 vol. S ny-iroger. mixes with .air ir. a turbulent ptocess, and less than 50% of th» The most important indices cequ l red £or devising prevent ive measures are j eaker! hydrogen contributes towards the format ion of an explosive medium. The -r~i -anit ion limits» which provide a means ot evaluating explosion Hazards, hclri :f -.he hydroqen is ô isyersed in the air without forming combustible ietemi.-.e the maximum permissible oxygen content or mixtures, and so zrt. fixtures. In tnis connection, the special features of tne combustion of lean Investigate -sn of the combustion of hydrogen clouds in open spaces [6] has Tear-limit hydrogen-air mixtures should be mentioned. It is well virtue -»£ the fire hazard properties and quantities ot tr:e 5^35 tar.ee ^ is the leakage factor for the buildinq (in m ); :nvolved. In particular, dangerous {category A) facilities ara tnose m ^hicn explosive mixtures may be formed containing substances whc-:~ . -.v.e; .jr.1 :,.-- .Î t: f siardard flame velocity (m m s ") ; and limit is below 10 vol, % (e.g. hydrogen) and whose volume excels >% ct :r.a-. . o tne j^qree of expansion Jurinq r.ydrogen combustion, 1 of the building. For buildings assigned to this category -.p. * standards ,", îssjn1 d to be equa 1 to 5. provide for : limits on the number of storeys and the WO:<:.-.G jr-jas. jev^ee: fire walls depending on the fire-resistance of the bu:.c inq ; requ.. cements relating to lifts and their location and to évacuât: 3- of tr.e au^lairç ; Whe'p T - T (m bpinq the possible hydrogen release, in kg) the buildirq erection of easily removable structures covet ing -LGAJ calculated -r. is considered to present an explosion hazard and is assigned to category A. accordance with SN 502-7 7, but amounting to at Isas* :, the premises; and fire-proof ing of load-bearing structures In .nàustnai facilities the source of ignition is commonly electr ica 1 eauipne.-. t. in trie JSSH a het of regulations 19] has been arawn up to prevent We have proposed [8] the following equa z\jzh ace lient s. Under these régulât ions exp losi ve xixtjreâ arç char acte r 1 zed 5f explosion hazard presented by a building; -.v 3 cri^cal -1 jr.-t.er (category) and sel'"-1 "nit ion tempt'i3i;jr A ^omrcr r.i-*- .-jtxJ '-• f ocev^nttng the forma" ion of an *?xplosvie medium is that where: m is the maximum permissible quantify of hydrogen -i" kç ,f erne rqency ';?ni:lat:on wnich is automatirally activated by stationary gas may be produced without constituting an explooior, hazar i nalyser * «hen e !".y dr oqpp concentration of 2% is ceacned. Where the quant it 7 r / 1 : ijqe ri r •ïlca^'io : :. 70ns iderable, howeve r, ventilât ion may pr<-v.* is the heat capacity of the air (m kczi -,q ~-Ut»g/ ; . r e : t •ect : ve . ;.- -^ j.jn -3 se s r.nt» stabilizat .en net hod 1 s recommended , whereoy 3L.rjsrjr.ce s which inhioit flame propagot ion in hydrogen-air mixtures a r*i is the air density (m kg-m ); • :i t rodu (-•»(' mtu the space r.o De protected. Informât ion on such stabilization is the minimum combustion heat (in kcai-ng '; ; • v '. • o'J -. ' î j 1 ven be 1." w. As mentioned above, in order to mitiqate cne consequences of explosions Table 1) which involves degenerate branching by way of H.,0., the relevant standards (71 provide for the erection of easily removal * «compos it. ion. With this assumption, the promot ion of hydrogen combustion Dy structures. However, the area prescr ibed for such structures ;nder ""•;**? Ha Ions i standards (10) is based on flame propagat ion under normal '.jnperVjc^ i) and 3r laminar conditions. Moreover, to repeat a point -nade earlier, a f Lame pass ir-.a the formation of hydrogen peroxide, the latter break ing down readily into throuqh an obstacle may speed up dramaticaily • In Ref . ill], for example, it active hydroxyl radicals. Apparently a straight-chain mechanism is is shown that when a flame passes a series of obstacles (such as a row of characteristic of lean hydrogen-air mixtures, which may explain why Halons machines, central supporting columns, girders, partitions, assemci/ platforms, lave only a slight effect on their combustion (lower branches of the cranes, etc.) its velocity can increase by an order o£ magnitude or more stabilization curves). It has therefore been suggested [12 1 that tne (Fig. 6). effectiveness of Halons could be increased by reducing the oxygen content in Undet such conditions the easily removaole structures cover inq "he area thç combustible mixture concurrently with tr.e application in inhibitors (i.e. prescribed by (10] may prove to be useless or inadequate tc or event the enriching the mixtures in the fuel component). The results of experiments on destruction of the building. It is therefore necessary to revise the metfed stai: ilizmg hydrogen-air mixtures while siirultjr.eously adding an inert diluent of calculating their area taking account of possible acceleration when the *fitrc«gen or carbon dioxide) are set out in Figs 11 and 12. They show that as flame meets an obstacle. tht* T devised for the specific conditions of a peninsula of Eire. *t higher In conclusion, data are g iven regarding the stabilization and pressures the mechanism is different. This is confirmed, for example, by data extinguishing of a hydrogen diffusion flame by introducing nitrogen into the regarding the effect of C2?4Br2 on the kinetics and temperature of hydrogen-filled pipe. At escape rates of up to lOm-s {flame-out was hydrogen self-ignition, the dependence of which on composJtion is shown in observed at 14 0 m-s"1) the flame was extinguished with tenfold dilution (in Figs 8-10. Although the Halon may inhibit hydrogen oxidation in the theory 18-Eold dilution should be necessary). peninsular region, it can either inhibit or promote combustion a- the third limit. In view of these data we propose a hydrogen oxidation mechanism (se- 93 79 REFERENCES 94 (1] KRIVULIN, V.N. et al., Dokl. Akad. Nauk SSSR 247 5 (1.379) 1184. ird Z.Tmouse ion l.iniDlcun [21 FURNO, A.L., et al., 13th Symposium (International) on ComDustion (1971} E + 0 -*- E0 + K 593. 2 2 2 il + On + J JlUo + -'I [3] KRIVUL-N, V.N., SHEBEKO, ÏU. N. , KjUR'.'-V.'TSKV, C.A., PAVLOVA, V.I,., BARATOV, A.N. , Khimicheskaya Fizika [Chem-.cal Physics! 8 11963). [4' KUDRYAVTSEV, E.A. , et al., in: The fire and explosion -.izards presented H202 + M — 2CI-I + H by substances and materials Un Russian), VNIIPO, Moscow (1982). H02 + E -*• 20E [5] MAKEEV, V.I., et al., in: Fl re-r ignting science and technology (m OH + Eg —*• S20 + E Russian) VNIIPO, Moscow (1977) 119. H + 02 —> OH + 0 [6] MAKEEV, V.I., et al., in ?roc. 7tn All-union Practical Scientific Conference on Combustion ana Problems of Fire-flghting (in Russian), 0+H2 -* OH + H VNIIPO, Moscow (1981) 17. + H02 + H -*• 'h °2 [7] Stroitel'nye normy i pravila [Construction standards and regulations!, H02 + H -> H9O + 0 SNiP P-90-81, Strojizdat, Moscow, p. 13. H + RX —*- [81 SARATOV, A.N., Zh.Vses.Khim. O-va D.I. Mendeleeva, 2_5 1 (1982) 22. HX + p- " . V HX + H 5=fc -lo + A. (9) Pravila ustrojstva ehlektroustanovok [Regulations £or constructing electrical equipment], PUEh-76, Section VII, Atomizdat, toscow il980) 19. OH + HX -> E^ + X [101 Instruktsiya po raschetu ploshchade;) legkosbrasyvaemykh kcr.strukt s i j EOo + X 5> HI + 02 [Instructions for calculating the area of easily removable structures!. X + X + M — »• I2 +• U SN 502-77, MISI, Moscow (1977). K02 + HX —+ H202 + X [Ill AGAFONOV, V.V. , ABDURAGMOV, I.M., BARATOV, A.N., RUMYANTSE\', V.S., Fiz. X + H ^ HX + X Goreniya Vzryva J^9 4 (1983) . 2 [12] BARATOV, A.N. , IVANOV, E.N., Fire protection in the Chemical petrochemical and petroleum-refir.ing industries tin Russian), Khimiya, Moscow (19?9) . X = ha loger». [13] MYSHAK, Yu.A., et al.. Instructions for training fire-c îght inq personnel and conveying them to the site of the tire (in Hussiani , VHII?O, Moscow (19 82) . [14] MAKEEV, V.I., GOLINE7ICH, G.E. , CHUGUEV, A.p., in: Combustibility of substances and chemical £ ire--f îght mg agents Un Russian), VNIIPO, Moscow (19 76) . Fiant» fnnc :r.insi[un v«loci:v in Tiixc.re-; ."U nvdrogen ind pr Smothering sy me^ns of 3 ^omb 1 neo igent "< -- . F .!r . Combined agent Extinguishing JV -neans C_r 3r^, vol.% Mixture, vol. Hydrogen 0.53-1-1 33.7-3Q.3 Petroleum products 0.23-û.^O 2.5-:.C CL2 0j 04 OS 13 f2 1,4 f Dependence ^t .lp and T on che eomoos;:::n of ihe H,-air mixture: r no-" i am-- sgljn ; EP f L.itne propane ion ; txpt? ri::ienta L data i. tp I ; . -i Leu Laced /a Lue of up ; B c a l eu Lar ?d v.i Lue of T. Hydrogen flame in cube 5.5 cm in diameter and 1.5m long (ignicion in lower pare of Lube): (a) 7.5% H2 - 5% C * 10"'. C02 - 77.5". air; (b) 8.07. H2 - 57. C - 10% CO2 - 77"'. air; (c) 9.07. H2 - 5% C » 10''. CO, . 7 (d) 11.0% H2 - 5", 2 - 10% C02 - 74% air. 95 120 2.0 0.02 Û,ûf OM S£8 ; 1 ' S i risv ax Additive, % Dependence of. Élame radius r and apparent comDuscion velocity -J Stabilization of propane-air mixture on tîme C . (1) curve r(t) plotted from the data for a mixture of volume A, a - standard tube; t, 0 - new system; "T= 32 ,-n and hvdrogen concentration C = 3^ vol.',; ( (1,2) - C^F ,3^.,; 3,^*) - nitrogen. '2) curve •«(, c ) , which LS the derivative of curve 1 ; (3 ^ curve r(c), plocced from the data for a mixture of volume "V= 32 m and hydrogen concentration C = 35 voL.1». H M Fit;. 5. Dependence of pressure Up in shock wave on discance from explosion /?,, '•ol.", centre R. curve plotted from the calculated data for an explosion produced b1-' F \ j). 7 . Stabilization ot propane and hydrogen a TNT charge of mass 0.02 kg; l ) C ,FAa (2 i CF^Br. o - experimental points for the explosion of a gaseous nntur- of volume 3* m with 35 vol.* hydrogen concentration. m SSÛ T.K 3. Inhibition of H^ oxidation through the addition of C-F.Brj in a peninsular fire region Fie. 10. Effect of C^F^Br on the seIf-tgnltiûn vemperatur» of a stoichiometric H^-air mixture UT 59 -'.J -a 4 s i to a H <* .Additive, '/.. 'fit Fig. 11. Stabilization of H2-air mixture by means of a combination of Fis. 9. Effect of C.F.Br, and C,H,Br. on H, oxidation at the third limit. C2F4Br2 and Nj. —a 2 4 2 2 4 2 2 Mixture containing 20 vol.% H- in air. Substituted by (1) Without additives; (2) 1% CjHjBr; (3) 0.5% CjHjBr; (1) 40% air; (2) 30% air; (3) 20% air; (4) 10% air; (5) 5% air; (6) undiluted. (4) 1% C.F.Br.; (5) 0.5% C,F.Br,. STATEMENT ON CURRENT AND PROPOSED ACTIVITIES OF DG XII OF THE COMMISSION OF THE EUROPEAN COMMUNITIES* B. TOLLEY, Commission of the European Communities, Brussels The Commission of the European Communities (CEO formed in 1972 a working Group N° 2 on wate- Reactor Safety Research. It is composed of representatives from Member states who are appointed by various bodies responsible for reactor safety, bodies responsible for publicly funded safety research programmes, electricity producing bodies and manufacturers. In addition to acting as a forum on national safety research programmes, it assists the CGC to identify reactor safety topics where further information is considered necessary from o i t i 4 s i r 's Additive, %. a European viewpoint and to propose how possible relevant European collaborative work could be undertaken. 12. Stabilization of H^-air mixture by means of a combination of CjF^r and CO,. One such topic is that of hydrogen phenomena during severe accident situations. Substituted by CO,! (1) 307. air; (2) 207. air; (3) 10% air. Accordingly in 1980 Working Group N° 2 formed an ad-hoc expert subgroup to assess the current situation regarding research in this area, to indicate where further information is considered necessary, and to advise on possible courses of collaborative action to obtain suih information. * Presented by J. Duco (France) This subgroup has met on various occasions during 1980/81 and Considered to be of particular priority is rapid and reliable produced a State af the Art Report (August 1981). This considers detection of hydrogen and other gases within the containment the situation under the following headings: under severe fault conditions. - Hydrogen Generation °rocedures are now underway to obtain the approval of this - Hydrogen Distribution programme by the Council of Ministers such that invitations - Hydrogen Combustion Phenomena to tender can be issued as early as possible in 1984. - Technical Control. Results of such a orogramme will be fully distributed within various recommendations were produced for future European the European Community. Distribution outside the Community collaborative activities. will depend on the approval of member states. These form the basis of a proposed safety research activity in this area which has now been accepted by the CEC as part of its proposed overall research programme for 1984-87. The work on hydrogen phenomena is planned to be undertaken by" contracts on a cost-shared basis at various institutes in Europe/ also possibly at the Joint Research Centre, Ispra and to connect up with various other programmes outside the European Community. 8y this means overall objectives of the CEC to stimulate collaboration between Member State institutes and to use more efficiently resources in such institutes will be satisfied. Specific detailed items of such study in the proposed research programme on hydrogen phenomena are to be within the following areas: - - Distribution of hydrogen and other evolved gases within realistic, compartmented containments. - Combustion phenomena, with the aim of better definition of the possibility of detonation and consequent effects on containment and safety related equipment. 99 - Technical control methods. LIST OF PARTICIPANTS HUM.'AKY Mr. Gy. Kapocs Paks Nuclear Power Station P.O.box 71 CZECHOSLOVAKIA P.IKS Mr. L. Tomtk Nuclear Power Plane Dlstr. Trnava Mr. K. Kovacs Design Institute for Power Grid 919 00 Jaslovske Uonunice and Stations (EROETtKV) Szichenyi ut 3 Budapest S FIMLAND Ms. M. Roder Central Research Institute for Physics Mr. B. Regnell Imatran Volma Oy P.O.Box 49 P.O.Box 138 1525 Budapest 114 00X01 Helsinki 10 Mr. K. Eerikainen Technical Research Centre of Finland INDIA Vuorimiehentfe 5 02150 Espoo Mr. P.K. Das Tarapur Atomic Power Station Tarapur FRANCE UNION OF SOVIET SOCIALIST Mr. J. Duco Commissariat â l'énergie atomique REPUBLICS IPSN B.P. fa Mr. A.N. Baratov All-Unlon Fire Fighting Scientific F-92260 Fontenay-aux-Roses Research Institute Moscow Mr. P. Cassette Commissariat à l'énergie atomique IPSN Mr. A.V. Dubrovin The Kurchatov Atomic Energy Insritute B.P. 6 Mr. V.A. Ermakov Moscow F-92260 Fontenay-aux-Roses Mr. B. Gel'fand Institute of Chemical Physics of the Academy of Sciences of the USSR GERMANY, Federal Republic of Moscow Mr. P. Papadlmitrlou Gesellschaft. fur Reaktorsicherheit m-b.H. Forschungsgelande Mr. A. V. Ivanov All-Union Fire Fighting Scientific D-8046 Garchlng Research Institute Moscow UNION OF SOVIET SOCIALIST Mr. L. Turerskii The Dzerzhinskli All-Union Thermo- REPUBLICS (continued) Technlcal Institute Moscow Mr. S.A. Kabakchl Institute of Physical Chemistry of the Academy of Sciences of the USSR Mr. Yu. N. Shebeko All-Union Fire Fighting Scientific Moscow Research Institute, Moscow Ms. E. Klementieva The Dzerzhlnskli All Union Thermo- Technlcal Institute Moscow INTERNATIONAL ATOMIC ENERGY AGENCY Mr. S.M. Kogarko Institute of Chemical Physics of the Academy of Sciences of the USSR Mr. H. Andres Moscow (Scientific Secretary) International Atomic Energy Agency Division of Nuclear Safety Vienna, Austria Mr. 0. Kovalevlch The Kurchatov Atomic Energy Institute Mr. V. Osmachkin Moscow Mr. I. Koukhtevish Institute Atom Teploelectro-Project USSR STATE COMMITTEE FOR Leningrad UTILIZATION OF ATOMIC ENERGY Ms. D. Khoklova Department for Internationa1. Relations Mr. V « Krasnoshtanov The Kurchatov Atomic Energy Institute Staromonetny Pereulok 26 Moscow Mr. M. Nikirln Moscow ZH 18U Ms. I. E. Lebedeva Institute of Physical Chemistry of the Academy of Sciences of the USSR Moscow Mr. S. Lesnoi The All-Union Institute of Atomic Power Plants Moscow Mr. A.G. Lyamin Institute of Chemical Physics of the Academy of Sciences of the USSR Moscow Mr. B. Samoilov Novovoronezh Atomic Power plant Novovoronezh Mr. S.A. Tsyganov Institute of Chemical Physics of the Academy of Sciences of the USSR Moscow HOW TO ORDER IAEA PUBLICATIONS An exclusive sales agent for IAEA publications, to whom all orders and inquiries should be addressed, has been appointed in the following country: UNITED STATES OF AMERICA UNIPUB, P.O. Box 433, Murray Hill Station, New York, NY 10157 In the following countries IAEA publications may be purchased from the sales agents or booksellers listed or through your major local booksellers. Payment can be made in local currency or with UNESCO coupons. ARGENTINA Cornisiôn Nacional de Energia Atomica, Avenida'del Libertador 8250, RA-1429 Buenos Aires AUSTRALIA Hunter Publications, 58 A Gipps Street, Collmgwood, Victoria 3066 BELGIUM Service Courrier UNESCO, 202, Avenue du Roi, B-1060 Brussels CZECHOSLOVAKIA S.N.T.L., Spàlenà 51. CS-113 02 Prague 1 Alfa, Publishers, Hurbanovo némestie 6, CS-893 31 Bratislava FRANCE Office International de Documentation et Librairie, 48, rue Gay-Lussac. F-75240 Paris Cedex 05 HUNGARY Kultura, Hungarian Foreign Trading Company P O Box 149, H-1389 Budapest 62 INDIA Oxford Book and Stationery Co.. 17, Park Street, Calcutta-700 016 Oxford Book and Stationery Co., Scindia House. New Delhi-110 001 ISRAEL Heiliger and Co., Ltd., Scientific and Medical Books, 3, Nathan Strauss Street, Jerusalem 94227 ITALY Libreria Scientifica, Dott. Lucio de Biasio "aeiou". Via Meravig!: 16, 1-20123 Milan JAPAN Maruzen Company, Ltd., P.O. Box 5050, 100-31 Tokyo International NETHERLANDS Martinus Nijhoff B.V., Booksellers, Lange Voorkout 9-11, P.O. Box 269, NL-2501 The Hague PAKISTAN Mirza Book Agency, 65, Shahrah Quaid-e-Azam, P.O. Box 729. Lahore 3 POLAND Ars Polona-Ruch, Centrala Handlu Zagranicznego, Krakowskie Przedmiescie 7, PL00-068 Warsaw ROMANIA llexim, P.O. Box 136-137, Bucarest SOUTH AFRICA Van Schaik's Bookstore IPty) Ltd., Libri Building, Church Street, P.O. Box 724, Pretoria 0001 SPAIN Diaz de Santos, Lagasca 95, Madrid-6 Diaz de Santos, Balmes 417, Barcelona-6 SWEDEN AB CE. Fritzes Kungl. Hovbokhandel, Fredsgatan 2, P.O. Box 16356, S-103 27 Stockholm UNITED KINGDOM Her Majesty's Stationery Office, Publications Centre P.O. Box 276, London SW8 5DR U.S.S.R. Mezhdunarodnaya Kniga, Smolenskaya-Sennaya 32-34, Moscow G-200 YUGOSLAVIA Jugoslovenska Knjiga, Terazije 27, P.O. Box 36, YU-11001 Belgrade Orders from countries where sales agents have not yet been appointed and requests for information should be addressed directly to: Division of Publications International Atomic Energy Agency Wagramerstrasse 5, P.O. Box 100, A-1400 Vienna, Austria