Tests 3b and 3h appear to be exceptions (ash analyses may be suspect). Test 3e is a n unqualified success. Series 4 is based upon Santowax OMP which shows the lowest film weight of the coolants tested. Test 4b was meant to be compared with test l b to see if intermediate boilers have a n effect. The possibility is that they are harmful, but more data are needed. The addition of clean, 2-micron particles of Fez03 in test 4c had almost no effect. Evidently, composite particles of organic plus inorganic material are required for severe film formation. Drastically oxidized material in test 4d produced a very heavy film which contained only 5% ash. For the first time, the test heater showed evidence of corrosion. This is an example of almost purely organic fouling, since the film had such a low ash content, much of which probably came from the corroded heater. The amount of pyrolysis of this coolant as it was being oxidized was much greater than is normally observed when the same coolant is heated in a closed capsule. Another coolant treatment which has proved most effective in reducing film forming potentials is a deep bed adsorption treatment using up-flow of coolant a t 2 to 6 feet per hour through a 2- to 3-foot-high bed of 24- to 48-mesh granular Attapulgite (Attapulgus Clay Co., Attapulgus, Ga.) maintained a t 600’ F. The results of these tests (Table 11) show that
because of changes in flow (not indicated in the table) the deep bed adsorbent is not always completely effective in lowering the ash content. However, it does remove essentially all of the fouling potential from the coolant until an effluent to adsorbent ratio of about 15 is reached, a t which time a breakthrough appears to start. Even after the breakthrough, the ash content can be very low, and the coolant will still form an appreciable amount of film in the PCFT. Also, the character of the film grades smoothly from “brown-soft’’ through “blackhard-brittle” to “black-hard-tenacious.”
Literature Cited
(1) Am. SOC.Testing Materials, Philadelphia, Pa., Comm. D-2, “ASTM Standards on Petroleum Products and Lubricants,” p. 1059, November 1957. (2) Coordinating Research Council for the Petroleum Industry, New York, N. Y., Manual No. 3, March 1957. (3) Neeley, J. H., Burggrab, F., Technical Information Service Rept. R59AGT278, March 1959. (4) Smith, A. L., Cook, W. P., Hlavin, V. F., Natl. Advisory Comm. Aeronautics RM-E56H21, November 1956.
RECEIVED for review February 10, 1961 ACCEPTEDOctober 12, 1961 Work performed for U. S. Atomic Energy Commission under the Advanced Organic Moderated Reactor Development program.
RADIOCHEMICAL BEHAVIOR OF PARTICULATE CORROSION PRODUCTS
IN THE SM-1 DURING CORE I C
.
R
.
BE R GE
N , Alco
Products, Inc., Schenectady 5 , N . .‘l
This article presents an analysis of the behavior of particulate corrosion products (crud) during Core I life of the SM-1 with reference to the long-lived gamma emitters, CoF0,Cos*, FeS9,CrS1and Mns4. The effects of power level, operating history, crud concentration, and oxygen concentration in the coolant are resolved. The resulting data provide a firm basis for predicting and comparing activity build-up of future reactors.
MAJOR PROBLEM
faced by the pressurized water reactor op-
A erator is the build-up of long-lived, induced gamma activity
on primary system surfaces ( 6 ) ,and the attendant reduction of accessibility and impairment of maintenance. The important nuclides responsible for this activity on all stainless steel system are usually considered to be Coco, C058, Fe69, Mn”, and Cr45. T h e mechanism involved in the activity build-up has not been clearly defined and, as may be expected in such a relatively new field, there is some difference of opinion as to the controlling parameters. The approach of Yerazunis is the more general and was used as a basis for this study. Figure 1 illustrates, in a simplified manner, the mechanism proposed by YeraZunis (7). The coolant serves to transfer corrosion products (crud) and attendant activity, both dissolved and particulate, about the system. This article describes a study conducted on the particulate crud filtered periodically from the SM-1 primary coolant dur10
l & E C PRODUCT RESEARCH A N D DEVELOPMENT
ing Core I life. This study is part of a continuing effort to better understand the mechanisms of activity build-up. The effects of system and chemical parameters on the behavior of the several radioactive nuclides of the crud were determined. Plant Parameters
The lifetime of the first core of the SM-1 extended from April 1957 to April 1960. The load factor average was about 55% of design (design was 10 megawatt thermal or 1.8 megawatt net electrical). In terms of neutron flux, this is 6.87 X 1OI2 neutrons per sq. cm.-sec., and 6.05 X l O I 3 neutrons per sq. cm.-sec., thermal and fast, respectively, averaged over the core. T h e primary system of the SM-1 is entirely of Type 304 stainless steel (Figure 2) including core cladding. The primary coolant is deaerated high purity water containing 20 to 40 ml. of Hz per kg. of water. It is nominally
neutral but actually is slightly basic owing to ammonia formation from traces of nitrogen in the coolant (pH ranged from about 6.4 to 8.8). Other characteristic coolant conditions are 1200-p.s.i. pressure a t 430 to 450' F., and 4000-g.p.m. flow. A bypass purification system maintains purity by treating an 0.6-g.p.m. side stream.
Table 1.
Nuclide COB0
cos Fe59
MnS4 Cr6'
Radioactive Properties Beta Energy, Gamma Energy, Half Life MeV. MeV. 1.33; 1.17 5.27 yr. 0.31 71 days 0.48 ( p + ) 0.81 45.1 days 0.46 1 . 2 9 ; 1.10 280 days ... 0.84
...
27.8 days
0.32
Methods
Particulate crud is collected a t a point just downstream from the purification system cooler. T h e crud filter is connected to the 1-inch purification system piping by l/r-inch stainless steel pipes. T h e coolant traverses about 15 feet of '/(-inch line before it reaches the filter. The full purification system flow of 0.6 g.p.m. is put through the filter. A 75-mm. Millipore Microweb-type WH filter is currently used. This filter is reported to retain all particles larger than 0.45 micron. Prior to October 1958, a sintered stainless steel filter was used, but this method was discarded because of the difficulty in weighing the collected crud. The sintered filters were rated at 2-micron pore si7e. The collected crud is ignited a t 800 to 900' C. and crud level is determined from the ignited weight. The ignited crud is dissolved by treating with mixed nitric, perchloric, and hydrofluoric acids. Aliquots of the solution are taken for radiochemical assay of Cow, Coj*, Fe59, Mnj', and Cr5l. Appropriate carriers are added and standard wet chemical techniques are used to separate the active elements for counting (Table I). The activity of the nuclides are measured on a Baird single channel pulse height gamma ray analyzer. All results are normalized to the date of sampling.
_I
COOLANT
PRIMARY
SYSTEM CORROSION
CORE CORROSION SYSTEM SURFACES CORE DEPOSITS
I
CRUD TRAPS
__
CRUD TRAPS I
I
-
I N FLUX
Figure 1.
OUT-OF-FLUX
Mechanism of activity build-up flow diagram
Method for Separation of Effects of Parameters
The specific activities of about 60 crud samples were determined. Typical results are shown in Table 11. The parameters which may be expected to show measurable effects are system age, average power level, relative power factor, crud level, and recent thermal history. The average power levt-1 varied with the age of the SM-1 as shown in Figure 3. Because system age (elapsed calendar time) and average power level are related, it was unnecessary to determine the effect of both variables. The correlation with system age may be expected to show equilibrium activities for those nuclides which would reach equilibrium during Core I. Since equilibrium depends on flux levels (average power factor), the shorterlived nuclides, Fe59, Cob*, Mnu, and Cr5I, would be expected to show a specific activity varying with the reactor age in a manner similar to that of the average power factor. Since the SM-1 is a training reactor, the day-to-day power release was quite variable. I n order to determine the long term (equilibrium) correlation with system age, it was necessary to correct for the short term variation in specific activity due to the day-to-day variation in power release. This correction was termed the "relative power factor" (R.P.F.) which is defined as the power release rate during the half life preceding the sampling date for the nuclide divided by the average power release rate during the preceding six half lives. For this analysis, the time periods were taken as 1000 and 6000 hours, respectively, rather than the actual half life of the various nuclides. If the crud is in equilibrium with in-core material, the effect of relative power level may be predicted. I t may be shown that the following relationship holds for an in-core nuclide, and for material in equilibrium with the in-core nuclide: v/y' = 0.5(+/+')
4- 0.5
(1)
STEAM GENE_RATOR
PRESSURIZER
I I
F? 1
I
PUMP
PRESSURE VESSEL
I
a I
COOLER
LEAK- OFF COLLECTION
MAKE-UP
MAKE-UP CONDENSATE DEJINERALIZERS
PUMPS FILTER MAKE-UP TANK
Figure 2. system
SM- 1 primary system including purification
VOL 1
NO. 1
MARCH 1962
11
~~
~
Table II.
~~
~
~
Experimental Specific Activity
Typical Results for Filterable Crud during Core I of SM-1
Crud Hr.. Since (Cos0 X 706) Level, Last D.P.M.b/ Date P.P.M. Start-up Mg. 1.46 0.20 11-23-57 100 0.10 12-24-57 1.98 250 0.45 1-31-58 4.96 425 0.07 2-17-58 1.54 200 4.34 3-27-58 0.13 4-10-58 0.12 350 2.30 1.31 100 5- 2-58 6.79 0.43 425 6- 5-58 5 .OO 0.59 9- 1-58 50 2.79 7.92 10-11-58 4.50 50 10-26-58 0.31 325 9.79 0.25 125 1.37 12-19-58 0.04 200 12-28-58 0.68 400 1- 6-59 6 .O 225 3- 7-59 0.03 7.8 0 0.75 10.2 5-1 6-59 50 13.0 1400 7-1 1-59 1.8 0.07 0.77 0.14 14.0 9-23-59 500 8.6 0.59 0.08 11-30-59 75 12- 1-59 0.60 0.11 13.0 100 1 . 5 0.05 150 13.0 1-28-60 1.6 0.60 2-25-60 750 13.0 5 Shut dowm for at least 8 hours only. b Disintegrationsper minute. ~~
w
~~
~~
I
80,
I-4
aL W W
Age of System, Hr. 4000 4600 5500 5900 6900 7300 7800 8700 10750 11850 12170 13424 I3495 13800 15374 17115 18400 20203
Rel. P.F. 0.83 1.03 1.51 1.30 1.40 1.30 1.10 1.37 0.72 0.675 1.79 0.58 0.71 0.85 0.75
I
I 1
I
60
-FJ
I I
/
\
I
I
4000
8000
I
I
1
12000 16000 20000 AGE OF SYSTEM HOURS
I
I
24000
2’5000
~
Figure 3. SM-1 average power release rate computed for each 6000 hours during Core I
20 a~
E
7
D.P.M.b/ MS. 1.10 3.0 3.4 1.5 3.6 1.7 4.5 3.7 1.4 1.5 6.5 2.2 3.6
5.5 7.5 5.1 13.0 12.3 6.8 10.9 13.0 12.0
0 2
(P.P.B.) 10 31 14 10
10
4.2 9.9 . .. 8.2 0.27 5.2 9.5 6.4 1.3 2.6 8.2 1.1 5.7 8.5 3.8 6.9 13.2 2.0
io
10 10 10 10
418
10
52 70 230 64 46 16 37 10
2.1 1.4 5.7 5.1 3.2 3.8 5.6 7.0
10
y represents the specific activity after one half life of operation
\
40
(Cr61 x 705) D.P. M.b/ Mg. 2.9 1.8 4.7 2.8
70‘3) (Re69 X 700)
D.P.M.b/ Mg. 6.0 7.8 15.3 5.2 13.0 6.6 17.2 14.6 5.4 5.9 17.5 1.3
where 4’ = the long time flux average y’ = equilibrium specific activity
1
I
I
(C068X
at a different flux 6. Since the relative power factor approximates the ratio $I/+’,the effect of the relative power factor may be expected to show if the crud is in equilibrium with the in-core material. Crud level is defined as the amount of filterable crud found in the coolant expressed as parts per million by weight.
7E==Ez=l
.i
74
i
79
W
7.0
0 0
16
I2
= .ai. . e z 4 ws 0 W
66
4000
8000
12000
16000
20000
24000
AGE OF REACTOR SYSTEM -HOURS
0
%: 0.2 AGE I
I
OF I
REACTOR-HOURS I
I
I
1
W
3 -0.2
I
-0.4
a ‘I
5
0.4
U
2- 0.2 u w
s
o
-0.2 -0.4
b Figure 5. 12
iaEc
Correlation of C0ESspecific activity
PRODUCT RESEARCH AND DEVELOPMENT
-0.6
0.5
1.0
1.5
1
1
1
2.0
2.5
3.0
R E L A T I V E POWER FACTOR
I
T h e specific activities were also shown to be affected by shut downs which involved cooling of the primary system. More precisely, the activity varied as a function of operating time following such shut downs. This parameter is termed recent thermal history. Two other parameters were found to have important effects on CrS1 activity-namely, the dissolved oxygen concentration in the coolant and shut downs of the reactor which lasted more than 8 hours. The effects o n Cr5l are discussed in a later section. I n an effort to determine the relative effects of system parameters the data were analyzed using the multivarient correlation technique as described by Ezekiel ( 5 ) . I n this correlation, y, the specific activity, was considered empirically to be the sum of the contributions of the various parameters x1, x 2 , etc.
7.0 I
=
6.8
2
6.6
g
6.4
a -I
6.2
I
Fc
’
0.6 I
” -0.2 L i
-0.4
I
I
I
I
1
6’oL
3
I
I
‘
7
CORRELATION WITH SYSTEM AGE
I
4dOO
Sob0
I
I
12000 I 6 k O 20b00 AGE OF REACTOR -HOURS I
I
I
24b00 I
1
I Fe5’, VARIATION WITH CRUD :LEVEL
wheref(x) = function of the parameter = d.p.m./mg.; or log d.p.m./mg. y
If only one parameter were varying, the effects of the other parameters would be lumped together in a constant term. T h e correlation of the variable parameter would then consist of a least mean square curve fitting. The technique used was a n extension of the mean square method to multiple parameter systems. The correlation was made graphically because of the discontinuous nature of the reactor operating history. The object of the correlation was to determine the effects of the various parameters, which consists then of defining ~ ( x J , f ( x d . etc. The effects of the individual parameters may be expressed by rearrangement of Equation 2.
f(%)
= Wx,)
- y
(3)
where x I represents all parameters except x i . The standard error is a measure of the reliability of the total correlation. It is obtained by comparing each. experimental specific activity with that predicted by the total correlation Equation 1 ( 5 ) . Thus it represents the possible total deviation of the over-all correlation from the true behavior of the nuclides. It is used in this article as a measure of the correlations for each nuclide. Results. T h e graphs used below to illustrate the parameter correlations are plots off(xt) us. x i . I n this study both specific activity and the log specific activity were correlated. Only the result which provided the smallest standard error is reported. Corn Behavior. Not all parameters affect each nuclide. Observable short term variation of Cow due to the relative power factor is precluded since Core I lifetime was less than one half life for Corn. Thus Cow could not reach equilibrium levels during Core I. T h e specific activity would be expected to increase throughout the study period. Figure 4 shows this to be the case. Figure 4 also shows that Cow specific activity increases with the crud level over the range of normal crud levels in the SM-1. At high crud levels there is little effect attributable to crud level. The correlation with thermal history involves a “correction factor” in that the specific activity during 200 hours following a cold system start-up is found to be reduced by about 3.8 X 106 d.p.m./mg. It appears that a n unusual amount of crud from a source of low activity is released to the coolant shortly after such a start-up thereby diluting the Cow. Because the probable source of crud with such reduced activity is crud traps, the effect is labeled the “crud trap effect.” The crud trap effect is based on only six observations and should be regarded only as a tentative finding.
T h e standard error of the CoM specific activity predicted from the variance between the calculated and experimental values is i15%. Co” Behavior. Because better correlation was obtained, the logarithm of the Cos8 specific activity was used. T h e experimental specific activity of C058 was correlated with relative power factor as shown in Figure 5 and the resulting values of the specific activity were correlated with system age, crud level, and thermal history. As in the case of Corn, the behavior with system age (Figure 5 ) is apparently in agreement with predictions from power release rate. More will be said about this behavior. Figure 5 also shows the effect of crud level. Figure 5 shows that Cob8 specific activity varies with crud levels approximately as does Cow. An apparent release of material from crud traps (crud trap effect) is again observed following a cold shut down which results in a 6070 reduction of activity. The standard error for the C058 correlation is rt20oJ,. Mn54 Behavior. T h e manganese in the SM-1 is mainly in the nonfilterable form (4). Because of mechanical difficulties the crud samples were not washed. Thus the crud contained coolant, along with any materials dissolved therein, For the other nuclides herein discussed, the error due to the dissolved material is negligible because of low solubility. Attempts to correct for the soluble Mn” contamination of the filterables reveal that the correction is approximately equal to the resulting measured value. Thus the experimental error is large and any correlation for MnM specific activity would be open to question. I t may be stated that the soluble Mn” comprises more than the 95% of the total Mn51previously reported (2). Fe59 Behavior. As with C058, a better over-all correlation was obtained with the logarithm of the Fe59 specific activity. Correlation with the relative power factor agrees well with theory in the case of Fe59 because the half life corresponds closely to the 1000 hours used to obtain the relative power factor. The data were also correlated with crud level and age. The results are shown in Figure 6. The general relationships agree with those found for C058, including a 60% reduction because of the “crud trap effect.” One difference is seen in the behavior with system age. No dip is found corresponding to the period of low power release after 12,000 hours. T h e standard error of the correlation was less than 1.5%. VOL. 1
NO. 1
MARCH 1962
13
= z
I
4
9.2
-11
I
I
I
I
I
Cr" CORRELATION WITH S W E Y AGE] n
i,
o
