ML19275A959
| ML19275A959 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 10/31/1979 |
| From: | English R PRESIDENT'S COMMISSION ON THE ACCIDENT AT THREE MILE |
| To: | |
| Shared Package | |
| ML19254E707 | List:
|
| References | |
| NUDOCS 7910300413 | |
| Download: ML19275A959 (39) | |
Text
&
TECHNICAL STAFF ANALYSIS REPORT ON CHEMISTRY TO PRESIDENT'S ConMISSION ON THE ACCIDENT AT THREE MILE ISLAND ADVANCE COPY NOT FOR PUBLIC RELEASE BEFORE AMs, WEDNESDAY, OCTOBER 1979 70)lo3co4I3 1183 041
CHEMISTRY BY ROBERT E. ENGLISH TECHNICAL ASSESSMENT TASK FORCE OCTOBER 1979 WASHINGTON, D.C.
1i83 042
TABLE OF CONTENTS
SUMMARY
AND FINDINGS..................................
4 INTRODUCTION...........................................
7 ZIRCONIUM REACTIONS...................................
7 Findings
. 10 FISSION PRODUCTS....................................... 11 Findings..........
................. 14 HYDR 0 GEN BUBBLE.
................ 15 Hydrogen Burned in Containment
................... 15 Hyd ro gen Invento ry................................ 18 Getting Rid of the Hydrogen Bubble................ 19 Hydrogen Explosion in the Reactor?............... 20 Findings
........ 23 MAXIMUM HYDROGEN EXPLOSION IN THE CONTAINMENT BUILDING. 25 Finding
.28 ifYDR0 GEN RECOMBINATION......
............ 30 Findings
... 33 TABLE 1: Fission Products in the Reactor Coolant from Samples Taken on March 29 and April 10, 1979....
24 REFERENCES
.,35
) } h; b
This docu aent is solely the work of the Commission staff and does not necessarily represent the views of the President's Commission or any member of the Commission.
This pre-publication copy is a final document and will be subject only to minor editorial changec in its published form.
1183 044
SLWiARY AND FINDINGS The following topics at TMI-2 were reviewed:
zirconium's reactions with water and the fuel UO ; the fi.. ion products released from the fuel 2
elements; and the hydrogen produced and its likelihood of axploding in the reactor vessel. The following are the findings:
Frem the measured fission products, approximately 50 percent of the core is assumed to have exceeded 4,000 F.
Ninety percent or more of the fuel rods ruptured their zirconium cladding.
There is a lir ited amount of experimental data (References 4-6 and 32) indicating that the UO fuel can dissolve in a partially-oxidized 2
liquid of Zr at about 3,450 F (1,900 C).
The significance of the data is that some liquid reactor fuel could result from temperatures well below the 5,200 F temperature for melting UO.
Chung (Reference 4) 2 further postulates that there is the possibility of forming a low-temperature-melting eutectic between UO and either unsaturated 2
-phase or phase zirconium at temperatures of 2,400-3,360 F (1,300-1,850 C).
Honekamp (References 32 and 34) shows that the total amount of liquid formed must have been small at TMI-2.
Some of the UO fuel is finely divided, and its fission products 2
are being slowly leached by the reactor coolant.
)
4
The amount of hydrogen burned in the containment building was estimated in two ways: from the height of the pressure pulse; and from the composition of the atmosphere in containment af ter the combustion.
The amounts of hydrogen burned, based on these two observations, are as follows; Pressure pulse: 294 pound-moles Gas composition:
436 1 33 pound-moles Additional amounts of hydrogen present after the comoustion were as follows:
H in c ntainment atm sphere: 79 1 4 pound-moles 2
H bubble in reactor loop:
92 1 22 pound-moles 2
H dissolved in reactor coolant:
36 pound-moles 2
The total amounts of hydrogen generated are then as follows:
500 1 22 and 642 1 40 pound-moles -- estimates with extreme ranges from 478 to 682 pound-moles. These extreme values range from 44 to 63 percent of the hydrogen capable of being produced from reacting with water and all the zirconium in the reactor.
In turn, 44 to 63 percent of the zirconium in the core has been oxidized to produce hydrogen.
Because 18 percent oxidation severely embrittler zirconium, the upper 60 to 70 percent of the fuel clad is so embrittled that it has lost its structural integrity.
1183 046 5
In pressurized water reactors, oxygen produced by radiolysis is promptly consumed by an excess of hydrogen that is deliberately dissolved in the reactor coolant.
Because of this back-reaction, there was never enough oxygen in the reactor loop for a hydrogen-oxygen explosion.
At the time of the accident, information on the use of hydrogen to suppress the accumulation of radiolytically formed oxygen was available from some staff members sf the Nuclear Regulatory Commission (NRC), from the national laboratories and the reactor manufacturers, and from textbooks on the chemistry of water for nuclear power plants.
WASH-1400 (Rasmussen Report) concludes that an explosion, or detonation, within a containment building of the type approved by NRC of all the hydrogen capable of being produced from the reactor's zirconium would not violate its ability to contain. An independent assessment by Los Alamos Scientific Laboratory (LASL) found that the containment building at TMI-2 may be marginal in its ability to withstand such a detonation of all the hydrogen produced.
The hydrogen production rate at TMI-2 was of the order of 500 times the capacity of the existing recombiner.
1i83 047 6
INTRODUCTION A number of topics in the chemistry and radiochemistry of the reactor at TMI-2 are important to understanding the accident. These are covered in this report. They are:
the reactions of the nuclear fuel's zirconium clad, with both the cooling water of the reactor and its fuel, uranium dioxide; the information that measurements of released fission products tells about damage to the fuel; the hydrogen bubble in the reactor vessel, and the likelihood that it might have exploded; the hydrogen explosions in the containment building; and the recombination of hydrogen.
ZIRCONIUM REACTIONS At temperatures up to about 3,000 F, the combined characteristics of zirconium, water, and UO; are well known (References 24 and 31).
Inasmuch as the Zircaloy-4 clad is almost pure zirconium (98 percent zirconium), the discussion below focuses on the zirconium, although the experimental programs investigated the alloys as well as the pure metal.
1183 048 7
Zirconium is a rare metal that has especially valuable properties as a clad for reactor fuel elements.
It readily conducts the heat from the UO int the water, and its 3,320 F melting point is about 525 F above 2
that of iron.
It has the particularly desirable quality of nc'. capturing many neutrons, thereby saving them so that they may be used to produce fission of the uranium.
At high temperatures, the zirconium also can react with, water produce hydrogen in the following way:
Zr + 2 H 0+Zr02+2H2 II) 2 Heat is also produced, and the oxidation of the zirconium can make it brittle if too much occurs. No oxygen is produced.
The way to avoid this reaction is to keep the zirconium from becoming too hot.
This problem is recognized in the design of all water-cooled commercial reactors in the United States, in part because the h3C requires that zirconium be kept below critical temperatures (Reference 1).
Therein the operating conditions of the zirconium are specified to remain within the following limits even during the design-basis accident:
(1) peak clad temperature may not exceed 2,200 F; (2) oxidation may not exceed 17 percent of cladding thickness; and (3) hydrogen generation may not exceed 1 percent of that which would be produced if all the zirconium in contact with the fuel were to react.
The ranges of operating conditions requiredtoavoiitheseproblemswerespecifiedin1962(Reference 2).
At TMI-2, all three limits were exceeded.
1183 049 s
The following examples will give some scope to the problem for the TMI-2 accident.
The Baker-Just equation (specified in Reference 1 and discussed in Reference 2) gives the oxidation rate, as follows:
du
=
0.3937 exp (- E )
(2) dt n
( RT )
where n = zirconium thickness oxidized, cm t = time, s E = activation energy, 45500 cal /g-mole R = 1.987 cal /g-mole -K T = cemperature, K A similar equation by Cathcart and Pawel uses newer values for the coefficients and is, therefore, also commonly used (Reference 4 discusses limitations of these equations).
If operated for 3 years at its normal temperature of about 650 F, zirconium would oxidize to only 1/1000 to 1/100 of the second and third limits.
If during an accident the zirconium were to reach 2,000 F, the 17 percent oxidation limit would be reached in just under an hour.
Because the zirconium-oxidation rate varies with time, even at a given temperature (equation 2), and because the afterheat from radioactive 1183 0150-9
decay of the fission products also varies with time, a direct comparison of the energies produced by the zirconium-water reaction and by the fission products requires a somewhat arbitrary selection of conditions.
For the decay heat from the fission products, the time 140 minutes after trip was chosen, this being the time that the pilot-operated relief valve (PORV) was closed and the reactor began to heat up.
In one minute at 2,000 F, the heat produced by the zirconium-water reaction would be roughly dou)le the decay heat produced in the 140th minute.
The zirconium-water reactions as tney concern the TMI-2 accident have been summarized by Paul Cohen in Reference 3.
Chung and Honekamp of Argonne National Laboratory (ANL) have summarized the literature for oxidation at 2,900 F, and above, and for clad-fuel interactions in the vicinity of 3,450 F (References 4 and 32).
FINDINGS The need to keep the zirconium below 1,000 F ir sell known and is a basic principle in reactor design.
At temperatures above 1,500 F, the zirconium oxidizes in a steam atmosphere and produces hydrogen, but no oxygen. This oxidation leads to embrittlement of the clad and clad failure.
At 3,450 F or above, partially oxidized Zr can melt.
1183 051 10
There is a limited amount of experimental data (Refcrences 4-6 and 32) indicating that the UO fuel can dissolve in this liquid, partially 2
oxidized Zr.
The significance of the data is that some liquid reactor fuel could result from a t.mperature well below the 5,200 F temperature for melting UO '
2 Chung (Reference 4) further postulates the possibility of forming a low-temperature-melting eutectic between UO and either unsaturated 2
a phase or S phase zirconiu.n at temperatures of 2,400-3,360*F (1,300-1,850 C).
Honekamp (Reference 32) shows that even at temperature; as high as 3,800 F (2,100 C) the total amount of liquid fuel mur, be small; English, however (Reference 34) shows that the amount of fuel that could be dissolved is much higher at 2,400*C.
FISSION PRODUCTS Measurements of the fission products released provide information on the extent of fuel damage during the accident. These fission products could be either gases that escaped to the atmosphere of the containment building, or substances dissolved in, or transported by, the reacter's cooling water.
In either case, the damage is assessed by comparing the measured fission products with the total amount of that species produced by the reactor.
1183 052 11
England and Wilson at LASL analyzed the operating history of the TMI-2 reactor, and f rom tha history computed the quantities of the various fission products and actinides that were generated (Reference 7).
They also determined the amounts of these various radionuclides that remained at any given time after the accident, as well as the total quantity of decay heat that resulted from their radioactive decay. The computer codes CINDER and EPRI-CINDER were used for the analysis.
Samples of the reactor coolant at TMI-2 were taken from the let-ot wn line first on March 29, and later on April 10, 1979. The first sample was sent to Bettis Laboratory for analysis of its fission products and the second was s4.nt to Savannah River, Oak Ridge National Laboratory (0RNL), Bettis Laboratory, and Babcock & Wilcox (B&W); the results (Refer ace 8) are presented in Table I.
Because the radioactivity of the various species declines with time, the decay of each individual species, from the time of shutdown on March 28, 1979, to the dates shown, is taken into accoimt in computing the fractions of core inventory in the coolant.
On March 31, a gas sample was withdrawn from the air in the contaiument building; and the radioactivity of the sample was measured by Bettis Laboratory.
The results are given below (Reference 8):
Isotope Concentration, u (Ci/ml)
Xe-133 676.000 Xe-133 m 16.000 Xe-135 8.100 1183 053 I-131 0.063 I-133 below 0.03 12
The following evaluati.ons of fuel damage are attributed to Bettis Laboratory (in Reference 9) following the water sample on March 29, and then the air sample on March 31:
(1) most of the volatile fission products were released to the reactor coolant, and 2 to 12 percent of the fuel reached 3,000-4,000 F.
Based on this, and the amounts of strontium, barium, and uranium present, it was determined that little, if any, of the fuel melted.
(2) About 90 percent of the 36,816 fuel rods burst their clads, and about 30 percent of the reactor fu;l exceeded 3,500 F.
Again, little, if any, of the fuel melted.
Cohen (Reference 10) concludes that 57 percent of the xenon was released, and that some of *.he fuel is probably in a finely divided form from which fission products are slowly being leached by the reactor's cooling water.
J. Rest of ANL (Reference 11) points out that the release of fission products from the fuel depends on the degree of grain-boundary separation which, in turn, depends on the fuel's operating history.
R. Lorenz of ORNL (Reference 12) concludes that the sizeable release of gaseous fission products could be produced from having 40 percent of the fuel at 4,350 F (2,400'C) and the remainder at lower temperatures.
Another interpretation is that the 57 percent release of xenon, all from the upper, hot two-thirds of the core, requires temperatures over 4,000 F throughout this region.
None of these temperature estimates is so high that the UO itself 3
would melt (5,200 F), but the estimates are generally high enough for formation of a ternary liquid of Zr-Zr0
-U0.
2 1183 054 13
FINDINGS approximately 50 percent of the core exceeded 4,000 F; ninety percent or more of the fuel rods ruptured their zirconium clad; and some of the UO fuel is in a finely divided form from which 2
fission products are being slowly leached by the reactor coolant.
1183 055 14
HYDROGEN BUBBLE Clearly, large amounts of hydrogen were produced by the reaction in equation one.
This hydrogen created a bubble, or bubbles, of noncondensible gas in the reactor loop, that dissolved in the teactor coolant, and escaped to the containment building, where i*., in part, exploded. An inventory of all this hydrogen will be eful in assessing damage to the zirconium cladding of the UO fuel fr m
.s reaction with water.
2 HYDROGEN BURNED IN CONTAINMENT In constructing the inventory, a set of simultaneous measurements would be valuable, but these ar. lacking. The hydrogen explosion in the containment building occurred on March 28, 1979. About 9 hours1.041667e-4 days <br />0.0025 hours <br />1.488095e-5 weeks <br />3.4245e-6 months <br /> and 50 minutes after the accident began, pressure in the centainment building rose above the setpoint of 28 pounds i f. r square inch gauge (psig). That pressure initiated the water spray into the building, but it did not indicate the actual height of the pressure oulse. On the other hand, the reactimeter recorded data on the pressures in the steam generators, these pres ures being the difference between steam pressure and building pressure; they were recorded every 3 seconds.
The Electric Power Research Institute (EPRI) analysis of these data (Reference 13) shows the following:
(1) The magnitude of the pressure pulse was not greater than 28 psi; and (2) the time for the pressure to rise to its peak was between 6 and 9 seconds after the rise began. Overall, the pressure pulse was a relatively slow, gradual process and, at least in the vic.nic of the pressure sensors on the steam generators, no detonation was apparent.
1183 056 15
The amount of hydrogen burned war estimated by the staff in two ways first, from the size of the 28 psi pressure pulse; and second, from the depletion of oxygen in the air within the containment building.
The containment building is described in several places in the Final Safety Analysis Report (FSAR) for TMI-2.
In paragraphs 3.8.1.1 and 6.1.1.1, the building is described as a vertical cylinder resting on a flat floor and topped by a dome that is a portion of a sphere. The inside diameter of the cylinder is 130 feet; the height of the cylindrical wall is 157 feet; and the radius of the spherical dome as 110 feet.
In Table 6.2-1 and in paragraph 6.2.5.1, the free volume is given as ' ! million cubic feet; in Table 6.2-9A, the free volume is stated to be 2,116,000 cub'ic feet.
The initial atmospheric conditions were taken as dry air mixed with hydrogen at 120 F and one atmosphere of pressure; final pressure was taken as 28 psig. The combustion was assumed to be adiabatic, to burn all the hydrogen, but to burn nothing else. The gas properties used (References 14 and 15) incorporate the changes in molecular species as well as the variations in their thermodynamic properties with temperature.
The results showed that the containment atmosphere contained 294 pound-moles of hydrogen, or 5.9 percent hydrogen by volume. Air temperature at the end of this combustion was 1,277 F.
If the rixture before combustion was not uniform (as assumed), the quantity of hydrogen burned would be 1183 057 16
unchanged, but the temperatures af ter combustion would vary with the local mixture and would be both higher and lower than 1,277 F.
Although the air in containment was assumed to be dry initially, humidity has only a minor effect on the amount of hydrogen needed to reach 28 pstg.
Measurements of gas composition within the containment building also indicate the quantity of hydrogen burned because the hydrogen's combustion depletes the oxygen. On March 31, 1979, the first samples of the containment atmosphere were obtained; these two samples contained 1.7 and 1.9 percent hydrogen (Reference 9).
The oxygen levels were 15.7 and 16.5 percent (Reference 13).
Later measurements indicated both higher and lower oxygen concentrations. Why the oxygen level rose and fell with time is not clear, apart from experimental error. Some indication of the experimental errors is given by the April 1,1979, data from Cohen (Reference 9); Cohen indicates that the hydrogen concentration was measured eight times between 7:00 a.m. and 10:00 p.m.
and that the concentration was 2.3 + 0.3 percent hydrogen -- but he does not state whether the + 0.3 is the standard deviation or the extreme range of the data.
The average values of the data obtained on March 31 were used in this report.
The averages were used because the variations between the measurements on March 31 are probably not significant, the experimental errors shown above being considered. March 31 was also the first day that the bubble volume in the reactor loop and the compcsition of the containment atmosphere were both measured; the near simultaneous taking of these measurements is important in constructing a hydrogen inventory.
The following composition is thus assumed for the containment atmosphere 1183 058 17
on March 31 on a dry gas basis:
1.8 1 0.1 percent hydrogen, 16.1 1 0.4 percent oxygen, 82.1 1
.5 percent remainder (81.06 percent nitrogen plus 1.04 percent argon).
At the time of the explosion, much water vapor had been added to the containment atmosphere, so 100 percent relative humidity is assumed.
Also, air temperature is taken to be 120*F, close to the 117 F on March 29 (Reference 9).
The result is that 436 1 33 pound-moles of hydrogen are required in order to produce the measured depletion of orygen, 148 percent of the 294 pound-moles computed to produce the 28 psi pressure spike.
Thus, the hydrogen burned is taken to range from 294 to 469 pound-moles.
HYDROGEN INVENTORY To these must be added the 1.8 1 0.1 percent hydrogen present in the containment building's atmosphere on March 31, 1979, or 79 1 4 pound-moles.
On March 31 the hydrogen bubble was described as follows (Reference 9):
"The calculated volume of noncondensible gas was approximately 823 cubic feet at a reference pressure of 875 psia"; the uncertainty in this value is given as 1 200 cubic feet.
For the purposes of this report, the bubble is taken as 823 1 200 cubic feet of hydrogen at 875 psia at the loop temperature of 280 F; the resulting quantity of hydrogen is 91 1 22 pound-moles.
In addition, 36 pound-moles of hydrogen were dissolved in the reactor coolant.
The amount dissolved in the water within the containment building is negligible.
I183 059 18
The sum of all these quantities of hydrogen ranges from 500 + 22 to 642 + 40, or 478 - 682 pound-moles of hydrogen.
The values herein differ significantly from the values presented in the NRC's "The Evaluation of Long-Term Post-Accident Core Cooling of TMI (Reference 16 at A-12); NRC values are attributed to W. R.
Butler (Reference 17). The principal difference is in the hydrogen burned in the containment building.
G. Marino (Reference 25) estimates the total amount of zirconium in the reactor at 49,711 pounds.
Because reacting all this zirconium with water could produce 1090 pound-moles of hydrogen, the inventory totals above amount to 44 to 63 percent of the amount theoretically possible.
The proportion of zirconium embrittled severely by oxidation exceeds these proportions because even 18 percent oxidation produces severe embrittlement (Reference 31).
GETTING RID OF THE HYDROGEN BUBBLE Cohen (Reference 19) and Jenks (Reference 20) appraised the means by which the hydrogen bubble was removed from the reactor loop.
Each of these references estimates that the differential solubility of hydrogen in water is not sufficient by itself to explain the rate at which the bubble disappeared, although this mechanism was a big contributor.
Cohen postulates that a significant amount of gas may have leaked past the 0-ring seal between the reactor vessel and its head.
During the
,,n-C40 l1C>
19
period in which considerable hydrogen was trapped in the reactor vessel, the gas had access to this seal and could have escaped through any leak there. When the water level rose and covered this seal, this gas leakage would no longer have been possible, but it could have been a significant contributor for 2 days or so.
Another possibility is that the high temperatures reached at the reactor outlet may have overheated and damaged the synthetic-rubber 0-rings that seal the top of the control-rod-drive mechanisms.
HYDROGEN EXPLOSION IN THE REACTOR?
The complete chronology concerning the views and actions concerning the possibility of a hydrogen explosion in the reactor vessel was reviewed and assessed.
Included in this review were chronologies from the NRC (References 18, 27, and 33). This material was critiqued by a group at ANL led by Closs (Reference 21). A second group at ANL led by Honekamp prepared a supporting technical analysis of the hydrogen bubble (Reference 22).
In Paul Cohen (Reference 23) addressed various mechanisms by which oxygen could be produced within the reactor loop or transported into that loop from outside, by being dissolved in the water of the emergency core cooling system (ECCS), for example.
Jenks (Reference 20) also investigated radiolytic production of oxygen within the reactor loop and describes his contribution to the NRC's evolving view of the potential for explosion.
))83 061 20
Basically, radiolytic decomposition of water always occurs in water-cooled nuclear reactors, both while they are operating, and after they have been shut down. Knowledge of this phenomenon and of how to deal with it was evolved long ago and is discussed in considerable detail by Cohen (Reference 24). The usual method of dealing with this oxygen (and the one used at TMI-2) is to add hydrogen gas to the make-up supply of water. This is acccmplished by merely keeping hydrogen gas above the water in the make-up tank; at a pressure of 4-5 psi, enough hydrogen (20-2c Scc of hydrogen per kilogram of water) will dissolve in this water to suppress oxygen formation within the reactor loop; 0.1 Scc /kg will produce sufficient recombination (Reference 18).
In fact, the same radiolysis that produces oxygen also stimulates it to recombine with the excess of hydrogen that is present in the water, once more forming water.
When boiling occurs in the reactor, as it did at TMI-2, some of the radiolytically formed oxygen can (before recombination) escape from the liquid into the steam bubbles and be carried out of the liquid into the bubble above.
Cohen (Reference 23) points out that only about 5 percent.
of the decay heat is useful in radiolytically decomposing the water and that only 0.225 molecule of oxygen is produced for each 100 eV of radiolytic energy deposited in the water.
Because of these factors, 20,000 times as many water molecules are released (or boiled) into the bubble as are oxygen molecules released by radiolysis, even if every oxygen molecule formed by radiolysis (that is, zero recombination) could escape into the bubble. With such a dilute concentration of oxygen, no combustion is possible (Reference 30).
In any real case, the amount of oxygen released into the bubble would be significantly smaller than the pessimistic case considered here.
1183 062
From these references, the following overall judgments with respect to oxygen formation or explosion in the reactor vessel can be drawn.
No explosion within the reactor vessel was possible at any time.
The largest proportion of oxygen was released to the bubble when boiling first began. Even then, the concentration of oxygen in the bubble was far below any combustible limit.
~
As boiling continued on March 28 and the reactor heated up, hydrogen was formed according to equation one, and this further diluted the oxygen and prevented any combustion within the reactor loop.
Sixteen hours after the accident began, no additional boiling occurred within the reactor loop, preventing the release of any more oxygen (Reference 21). Any oxygen present in the water would have been completely recombined with hydrogen in less than 5 minutes. Any oxygen in the bubble at that time gradually dissolved in the water and there disappeared by recombination.
In spita of the impossibility of a hydrogen explosion within the reactor vessel, the NRC was greatly concerned about such an explosion from March 30 until April 2.
ANL reviewed the chronology and the judgments concerning the handling of the hydrogen bubble in the reactor vessel at TMI-2 (Reference 21), and they reached the following conclusion:
1183 063 22
It is clear that the erroneous conclusions about dangerous concentrations of 0, in the H, bubble originated from a number of calculations neglecting the iniportant back reaction.
Since the radiolysis of water has been studied for decades by radiation chemists, it is hard to understand why none of this country's outstanding radiation chemists were contacted, or as in the case of KAPL and Bettis, were asked so late in the incident.
Expertise in radiation chemistry is available at each of the National Laboratories....
Certainly, there was nothing in the TMI - bubble incident for which the fundamental science was not well known.
For example, the all-important H,-0 back reaction, which was left out of the NRC estimates 2
on oxygen formation, is the basis for adding H, to the primary cooling system under normal operating conditions.
Findings From assessments of the amount of hydrogen burned in the containment building, the inventory of all the hydrogen produced at TMI-2, and the handling of the hydrogen bubble, came the following findings:
From the magnitude of the 28 psi pressure pulse, the amount of hydrogen burned in the containment building was computed to be 294 pound-moles.
However, combustion of 436 33 pound-moles is required to account for the measured deficit in oxygen in the atmosphere within the containment building.
Additional amounts of hydrogen present after the combustion were as follows:
H in c ntainment atm sphere: 79 + 4 pound-moles 2
1183 064 23
H bubble in reactor loop:
91 + 22 pound-moles 2
H dissolved in reactor coolant:
36 pound-moles 2
The total amounts of hydrogen generated are then as follows:
500 1 22 and 642 40 pound-moles, estimates with extreme ranges from 478 to 682 pound-moles. These extreme values range from 44 to 63 percent of hydrogen capable of being produced from reacting with water all the zirconium in the reactor.
In turn, 44 to 63 percent of the zirconium in the core has been oxidized to produce hydrogen.
Because 18 percent oxidation severely embrittles zirconium, the upper 60-70 percent of the fuel clad is so embrittled that it has lost its structural integrity.
In pressurized-water reactors, oxygen produced by radiolysis is promptly consumed by an excess of hydrogen that is deliberately dissolved in the reactory coolant.
Because of this back-reaction, there was never enough oxygen in the reactor loop for a hydrogen-oxygen explosion.
Information on the use of hydrogen to suppress the accumulation of radiolytically formed oxygen is available from members of NRC, from the national laboratories and the reactor manufacturers, and from textbooks on the chemistry of water for nuclear power plants.
1183 065 24
MAXIMUM HYDROGEN EXPLOSION IN THE CONTAINMENT BUII.DIhG
'Because only about half the zirconium in the reactor was reacted with water to produce hydrogen, one might ask if the containment building is strong enough to withstand a more severe hydrogen explosion. Marino estimate; that the reactor contained 49,711 pounds of zirconium (Reference 25).
Complete reaction of this much zirconium with water would produce 1,090 pound-moles of hydrogen.
As an extreme case, consider that all this hydrogen was released to the containment building, uniformly mixed with the atmosphere there, and then ignited. Gordon (Reference 26) computed the pressure from a combustion for two cases:
(1) thermodynamic equilibrium after a constant-volume adiabatic combustion; and (2) a one-dimensional Chapman-Jouguet detonation.
In each case the initial conditions postulated were 120 F, 100 percent relative humidity, and pressure of one atmosphere.
The overall results are as follows:
equilibrium case:
final pressure, psig = 79 final temperatute, 'F = 3,668 detonation case:
final pressure, psig = 166 final temperature, F = 4,042 i183 066 25
The containment building at TMI-2 was designed for an internal pressure of 60 psig and has bee proof-tested at 69 psig. With its safety factor of 1.5, the building should withstand 90 psig without loss of its containment capability. The concrete snell might develop visible cracks, but the reinforcing steel should maintain the building's structural integrity. The steel plate that lines the inner surface of the walls and dome of the building should fulfill its role as a membrane that would prevent leakage of fission products even if the concrete were to crack. All this indicates that the building would successfully withstand the 79 psig gas pressure that would load the building shell for perhaps 5 seconds and then gradually decline when the water sprays inside the building cool the air.
The detonation case presents a more difficult problem because of the dynamic interaction between the detonation's impulsive load and the elasticity of the building. WASH-1400 (Reference 28) concludes that the containment buildings of the type approved by NBC should withstand such a detonation.
Los Alamos also evaluated this structural problem for TMI-2 by drawing on their background in explosions derived from the weapons program (Reference 29).
This evaluation showed thst the force from the detonation would be imposed for a period much shorter than the building's periods of natural oscillations. As a result, t.he building's inertia as well as its strength would be called upon to resist the detonation; and the maximum load, as computed in Reference 29, for the structure at TMI-2 would be below, but close to the building's structural limit.
Inasmuch as the analyses of the detonation and of the structural dynamics were each on a somewhat simplified basis, additional study is 1183 067-26
required before one could conclude with confidence that the containment building at TMI-2 could withstand such a detonation.
In WASH-1400, the impulsive load from the shock was found to be less than 2.5 percent of the strength of the structure.
The key reason why the result in Reference 29 is so different from that of WASH-1400, is that the duration of the imposed shock loads are so different, 8 milliseconds Reference 29, and 10 microseconds in WASH-1400.
The pre;sure load from the detonation wave would be imposed upon the wall in the following way.
The shack front that is the detonation wave would strike the wall and be reflected, thereby producing a sudden rise in pressure on the wall. This increased pressure level would be sustained until rarefaction waves following the shock reached the wall and lowered the pressure there. Thus, the ducation of this p essure loading of the wall would be a fraction of the following time.
the radius of the building divided by the speed of sound in the gas within the building.
The speed of sound in the product gases is 5,860 feet per second, and the radius of the containment building at IdI-2 is 65 feet. The duration of the pressure-pulse on the wall would therefore be of the order of a few millisoconds.
The pulse duration ot 10 microseconds in WASH-1400 (Appendix VIII at 123) appears to be based on the thickress of the shock front, rather that on the time interval between the shock wave's and the rarefaction wave's reaching the wall.
For this reason, WASH-1400 is probably in error.
1183 068 27
Even if this qua: ity of h3 ogen were to be released to the containment building, explosions of either of these magnitudes appear extremely unlikely because of the likelihood that the hydrogen will be ignited before all of it enters the containment 'ouilding.
As TMI-2 demonstrates, the t tilding does contain ignition sources, such as a switch that arcs. The peak pressure and the potential for damage would be reduced if the same amount of hydrogen were burned in several bursts, each individually smaller than the ultimate explosion. Perhaps deliberate introduction of an ignition source for this purpose would be prudent and installation of recombiners, as discussed below, might entirely avoid the problem, in most cases.
Qndings At TMI-2, combustion of al the hydrogea producible would not exceed the strength of the containment building.
WASH-1400 (Rasmussen Report) concludes that detonation within containment buildings that meet NRC design criteria of all the hydrogen producible from the reactor's zirconium will not violate its ability to contain.
In an independent assessment by the LASL the strength of the TMI-2 containment building was found to be somewhat above the loads conceivably imposed by detonation of hydrogen within the building; however, the strength margin was less than the errors in the approximate analysis. Additional study is required to established with confidence that the building will withstand the detonation.
83 08 28
In WASH-1400, the method for analyzing containment-building tolerance of hydrogen detonations is probably in error.
1183 070 29
HYDROGEN RECOMBINATION The FSAR specifies that a hydrogen recombiner be available for connection to containment and for peacefully reacting any hydrogen in the containment building. This recombiner is sized to remove hydrogen at the rate at which it would be formed by radiolysis from water in the sump of the containment building -- soecifically, at the rate of 0.7 pound per hour (FSAR, paragraph 6.2.5.)
In addition, hydrogen could be produced by the zirconium-water reaction discussed above; the maximum amount of hydrogen allowed by NRC regulaticas (Reference 1) is that resulting from reacting 1 percent of the reactor's zirconium.
A recombiner this low in capacity would require a long time to remove all the hydrogen raleased to containment at TMI-2.
The hydrogen inventory herein treated this amount of hydrogen in two segments:
the amount burned and the amcunt in the atmosphere on March 31.
The oxygen deficit in gas analysis gave the larger value for the amount burned (436 pound-moles of hydrogen), and that value is used here. The hydrogen measured in the gas analysis was 79 pound-moles, for a total of 515 pound-moles. At 0.7 pound reacted per hour, the existing recombiner would require 6 weeks to reduce the hydrogen content of containment to 3 percent, just under the 4 percent limit discussed by Rose (Reference 29). Expressed differently, the rate at which hydrogen was produced at TMI-2 exceeded the capacity of the recombiner by a factor of about 500.
})0 30
Inassmch as TMI-2 exceeded the capacity of the existing recombiner.
some consideration was given during the acci.~ent investigation to reacting this hydrogen as rapidly as it was formed. The basic concept was to react the hydrogen with containment air (burn it) in a controlled 2aaner and to vent the combustion products (water vapor, unburned air, and fission products) to the containment atmosphere. The combustior. concepts were discussed with Larry Diehl, head of the Combustor Fundamentals Section, and Robert Jones, head of the Combustion Technology Section, both of NASA's Lewis Research Center where considerable research on hydrogen combustion has been conducted.
Two combustor concepts were selected (1) a catalytic reactor; and (2) a conventional ficme-type combustor.
Catalytic reactors can readily react hydrogen with air over a wide range of flows and velocities up to a rated inlet-air velocity of 50-75 feet per second. This reaction would, for all practical purposes, consume all the hydrogen and release only a negligible unburned fraction to the containment atmosphere.
The catalytic reaction of the hydrogen could be initiated by an electrically heated platinum grid on the entrance face of the catalyst bed.
In concept, such a catalyst bed 13 inches in diameter could react 220 pounds of hydrogen per hour -- about 300 times the capacit" of the present recombiner.
Conceivably, all the zirconium in the reactor could react with water to produce hydrogen -- about 2,200 pounds of hydrogen.
Discharging this hydrogen through the catalytic reactor would permit recombination of up to 10 percent of this aroaunt each hour or 100 percent in 10 hours1.157407e-4 days <br />0.00278 hours <br />1.653439e-5 weeks <br />3.805e-6 months <br />.
1183 072 31
A conventional flame-type combustor could handle higher flow rates.
Although such a combustor cannot successfully consume all the hydrogen at the very lowest flows (which the catalytic reactor can handle very well), it can react larger flows than the catalytic type inasmuch as an inlet vslocity of 150 feet per second is acceptable, and it could tolerate the stoichiometric mixture as well.
Such a combustor, 24 inches in diameter, could recombine 2,200 pounds of hydrogen, or all the hydrogen that could conceivably be produced, in one hour's time. Water sprays totaling 300,000 gallons would be sufficient to cool these combustion products and to condense the water vapor produced.
Although neither of these concepts was carried to the design stage, they show that devices for reacting all the hydrogen from the reactor loop can be small in relation to the other component.s of the power plant. Through their use, the actual explosion in the containment building at TMI-2 as well as the largest hydrogen explosion possible at TMI-2 could have been avoided.
For both the catalytic and flame-types of combustor, the discharge of hydrogen from the reactor loop was assumed to pass directly through the combustor. This presumes that the combustors, or recombiners, would all be connected to the reactor coolant vent system and ready to receive the hydrogen.
(This contrasts with TMI-2 where several_ days were required to put the recombiners into service -- according to PNO-79-67F and
-67K.)
It also presumes that the hydrogen would be vented as designed tor.
In case of an accident, this might not occur, and the hydrogen 1183 073 32
could conceivably be vented directly to the containment atmosphere through, say, a ruptured pipe.
In that event, the catalytic reactor described above could remove the hydrogen from the containment atmosphere by processing 70 cubic feet of that atmosphere each second.
In 8.33 hours3.819444e-4 days <br />0.00917 hours <br />5.456349e-5 weeks <br />1.25565e-5 months <br />, a volume equal to the entire containment atmosphere would pass through the catalyst bed, and the concentration would be diminished by 63 percent.
The concentration would be diminished by 90 percent in 19.2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> and by a further 90 percent in each successive 19.2-hour period of operation.
The hydrogen released to containment at TMI-2 could be recombined to below the flammable limit in 10.3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> rather than the 6 weeks for the existing recombiner at TMI-2.
Findings The hydrogen production rate at TMI-2 exceeded the capacity of the existing recombiner by a factor of about 500.
Alternate concepts assessed only superficially make increasing the recombiner capacity appear feasible.
1183 074 33
TABLE 1:
Fission Products in the Reactor Coolant from Samples Taken on March 29 and April 10, 1979 let l*:Intry colant 2rst Prim 1ry Qx>lant S.mg >1e Armityses Cur rmtel to Analyse a u>rrectal to 4/11/79 3/30/79 Ilettis Sit!.
Oi4R.
Ihttis inW ii31. ant nue-FracUn7e Ouolant cort-Fractiosi*
Qulant unt-Fractim*
bolant unt-
'Fr action
- bolant axe-E action
- mnt rat ion of Qire in centration of Que in amt ra tion af Core in nsitration of Core in mnt ratice of 03re in thiclisl e T I/2 (p Ci/u-)
Primary
( p ci/cc)
Primsay (p C1/m)
Primary (HC1/cc)
Pa lmary
( U Ci/n:)
Primury Cantant omlant Onlant ruolant O nlant I-Ill Ikl 1.4 x 10' O.095 4.5 x 10' O.006 8.2 x 105 0.155 8.5 x 108 0.16 6.7 x 105 1 11T"
- 20 Bli E 3 d '10'-~
0:083 -
0.13 d'i34 '~
-~ 2Y-'
~T6.1 x 16'-~
- 0. 0,a TTx 105-o.006 8.2 x TOT-i)T091 1.Tx 15r o,og5-
,3,3,Igi-0.082 6-'ill
~ks~
~'1.8 x 15' - ~ 5.'15----
1 Tx 10-5.12 1;T T 10' 5~12
~1:1 T 15 r 7 :12 9.57 IDr T1o bi:11 F 1 11 -
O.13 3.3 x 103 P.13 3.4 x 10I
- 2. 7 T56' '
O.11 1 2 x 107~~-'
0,13 2.8 x 10' O.11 Er-90 _
__ 553 _[
~5.~4 0.0000lf 4.5 x 10 0.009 6,Q..xj03 LOQ42 7.3 x 10I ~0.604 F
Ss-89 29Y 5.0 x 10' O.022
~.6 x 15;
O.000019 E 106 ~
~ l6M ~
3 I-O J_K)ll-1.5 x 161-0.00691 -
T-L O0l6 2,2 x ID ILUl40
~ 12.M ~'}.fx 10'-~ ~6 6065/3 1,7 x 167-0.0010 2,1 x_IO I
ta-140 4oh 1.4 x 10 0.00075 1.6 x 108 0.00006 1.4 x 10 o,00075 Ft >-99 tih 1.3 x 10I 0.012 1,8 x 10 0.017
~~1.3 x 10:
~0.012 3
7-Eteli ]2d]
~ ~ ~ ' ~
~ ~ ~ ~0.0012 2.0 x 10 Te-lie 78h
~5 057 56 I 0.00051_ _
s 0.35s Its -
doisjh-9.0 x 10' O.61 l bn ter of Q
co-136 Cr oss g
3.6 x 10 '
1 x 10
<1 x 10-3
<1 x 10-8 4.5 x 10
<4.6 x 10-8 1.3 x Ig-s 1.3 x 10-s Q_.W >
5 u
< l f ta, C.,3 gr__
_ 12y--
76 93 U
B3 1.2 0.1 N
W
- lhnel on a 3.a inury u nila.it vulisie of 3.0 x 10' mil.
If tie nuke-tip water (9 x 10' atl) frun tie InCT is irc1talet anal axml&rnt to le at tie r zie un axmt s at it es, tie traction of tie u>ae in tJe unlant wuld le almut a factor of 3 tilgter.
La V
e
REFERENCES 1.
Rules and Regulations of the NRC, Title 10, Code of Federal Regulations
- Energy, Part 50 at paragraph 50.46 and Appendix K.
2.
L. Baker, Jr., and L. Just, Studies of Metal-Water Reactions at High Temperatures. III - Experimental and Theoretical Studies of Zirconium - Water Reaction. ANL-6548, 1962.
' 3.
Paul Cohen, " Zirconium-Water Reactions - Application to Three Mile Island-2 Accident," July 16, 1979.
4.
H.M. Chung "A Discussion on Reactions and Damage in Fuel Rods of the Three Mile Island-2 Reactor," August 1979.
5.
P. Hofmann and C. Politis, " Chemical Interaction Between UO and 2
Zircaloy-4 in the Temperature Range Between 900 and 1500 C," Zirconium in the Nuclear Industry (Fourth Conference), ASTM STP 681, American Society for Testing and Materials, 1979, in press.
6.
S. Hagen, A. Grunhagen, H. Malauschek, H. Schulken, and K. Wallenfells, "Experimentelle Untersuchung der Abschmelzphase von UO -Zi#C#1 Y 2
Brennelementen bei versagender Notkuhlung," Projekt Nucleare Sicherheit Halbjahresbericht, 1977-2, KfK2600, Kernforschungszentrum, Karlsruhe Germany, at 416-428.
1183 076.
35
7.
T. England and W. Wilson, "TMI-2 Decay Power: LASL Fission-Product and Actinide Decay Power Calculations for the President's Commission on the Accident at Three Mile Island." LASL LA-8041-MS, 1979.
8.
F. Miraglia, "Lettet from NRC to Robert English, staff of President's Commission on TMI J," July 5, 1979.
9.
Anonymous, " Preliminary Annotated Sequence of Events, March 29, 1979 thru April 30, 1979." Metropolitan Edison Company, July 20, 1979.
10.
Paul Cohen, " Fission Product Release from the Core, Three Mile Island-2," July 20, 1979.
11.
J. Rest, " Letter from ANL to G. Marino of NRC," April 16, 1979.
12.
R. Lorenz, " Letter from ORNL co Robert English, staff of President's Commission on TMI-2," Aug. 14, 1979.
13.
Personal communication from Allen Miller, EPRI, undated.
14.
Robert English and William Wachtl, " Charts of Thermodynamic Properties of Air and Combustion Products from 300 to 3500 R."
NACA TN 2071, 1950.
15.
Rc.bert English and C. Hauser, " Thermodynamic Properties of Products of Combustion of Hydrogen with Air for Temperatures of 600 to 4400 R."
NACA RM E56G03, Oct. 12, 1956.
1183 077 36
16.
.. valuation of Long-Term Post-Accident Core Cooling of Three Mile Island Unit-2,"
NRC NUREG-0557, May 1979.
17.
W. Butler, " Memo to R.L. Tedesco," NRC, April 25, 1979.
18.
W. Milstead, " Memo to R.J. Mattson," NRC, July 9, 1979.
19.
Paul Cohen, "Three Mile Island-2 Accident of March 28, 1979 - Bubble Dynamics," July 18, 1979.
20.
Glenn Jenks, " Letter from ORNL to Robert English, staff of President's Commission on TMI-2," Aug. 12, 1979.
21 G. Closs, S. Gordon, W. Mulac, K. Schmidt, and J. Sullivan,
" Report by tDe Ad Hoc Committee of the Radiation Chemistry Group of Argonne National Laboratory to the President's Commission on the Accident at Three Mile Island," undated.
22.
J. Honekamp, S. Gordon, K. Schmidt, and D. Malloy, "An Analysis of the Hydrogen Bubble Concerns in the Three-Mile Island Unit-2 Reactor Vessel," undated.
23.
Paul Cohen, "0xygen Generation and Gas Composition of Bubble,"
June 24, 1979.
1183 OL78 37
24.
Paul Cohen, " Water Coolant Technology of Power Reactors," Gordon and Breach, New York, 1969.
25.
G. Marino, " Memo for Files," NRC, April 25, 1979.
26.
Sanford Gordon, " Memo to Deputy Director, Technology," NASA, Lewis Research Center, Aug. 22, 1979.
27.
Roger Mattson, "NRC letter, to Stanley Gorinson, Chief Counsel, President's Commission on the Accident at Three Mile Island,"
Sept. 6, 1979.
Principal enclosure:
" Chronology of TMI-2, Hydrogen Bubble Concerns."
28.
N. Rasmussen, " Reactor Safety Study - An Assessment of Accident Risks in U.S. Commercial Nuclear Power Plants," NRC WASH-1400 (YUREG-75/014), Oct. 1975, App. VIII.
29.
Donald Rose, " Letters from LASL to Robert English, staff of President's Commission on TMI-2," Oct. 2 and 15, 1979.
30.
I. Drell and F. Belles, " Survey of Hydrogen Combustion Properties,"
NACA Rep. 1383, 1958.
31.
J. Hesson, R. Ivins et alia, " Laboratory Simulations of Claddiag-Steam Reactions Following Loss-of-Coolant Accidents in Water-Cooled Power Reactors," ANL -7609, Jan. 1970.
i183 079 38
32.
J. Honekamp, "Some Considerations Related to the Potential for Fuel Clad Eutectic Formation in the Core of TMI-2,"
Argonne National Laboratory, September 1979.
33.
Roger Mattson, "NRC letter to William Stratton, staff of President's Commission on the Accident at Three Mile Island," July 30, 1979.
34.
Robert English, Dissolving the Reactor Fuel--An Assessment of Honekamp's Analysis, Staff report for President's Commission on Three Mile Island, Oct. 17, 1979.
))0b 00 39