- )

 

OAK RIDGE NATIONAL LABORATORY
OPERATED BY Internal Use Onll

 

UNION CARBIDE CORPORATION

NUCLISION 0 R N L
CENTRAL FILES NUMBER

OAK RIDGE, TENNESSEE 37831

 

65-8-32

 

 

 

DATE: August 16, 1965 ~COPY No. 3

SUBJECT: PRELIMINARY REPORT ON RESULTS OF MSRE ZERO-POWER EXPERIMENTS

T0: Distribution

FROM: P. N. Haubenreich

ABSTRACT

The MSRE first attained criticality on June 1, 1965 with a =35y
concentration within 1% of the predicted value. Initial critical con-
ditions were 1181°F, fuel circulation stopped, one control rod inserted
0.03 of its worth and 65.4 kg of “25U in the loop (plus L.2 kg in a
drain tank). Subsequently the reactor was held at 1200°F while the
_235UF4 concentration was increased by the addition of 79 enriching cap-
sules containing 6.6 kg =3°U. During these additions, experiments were
done to measure rod sensitivity and total worth, =>°U concentration coef-
ficient of reactivity, temperature coefficient, pressure coefficient, and
effect of circulation on reactivity. Dynamics experiments gave information
on system transfer functions and separated prompt (fuel) and delayed
(graphite) temperature coefficients. Nuclear power was restricted to a
nominal 10 watts, except during transients, when 10 kw was permitted. The
experiment was concluded on July 3 and the salt was drained to permit
preparations for high-power operation.

There were no mechanical difficulties of any consequence during the
experiment. Salt analyses showed practically no corrosion. Analysis of
the reactor physics data, although incomplete, has shown that all the
observed parameters are in good agreement with predicted values.

NOTICE

This document contains information of a preliminary nature and was prepared
primarily for internal use at the Oak Ridge National Laboratory. |t is subject
to revision or correction and therefore does not represent a final report. The
information is not to be abstracted, reprinted or otherwise given public dis-
semination without the approval of the ORNL patent branch, Legal and Infor-
mation Control Department. |

 
CONTENTS

.Abstract .

Introduction . . . . . . .

Nuclear Experiments.

| Initial Critical Loading.

~ Control Rod Worth . |
235 Concentration Coefficient of Reactivity.
Reactivity Effects of Circulation .
Temperature Coefficients of Reactivity. .
Effect of Pressure on Reactivity .
Dynamic Tests .
Frequency Response Measurements. .

| TTanSient Flow?Rate Tests.

- Conclusions from Dynamic Tests . . . . .

Fuel Salt Chemistry. . .
Performance of Mechanical Components .
Control Rods. .
Sampler-Enricher.
Freeze Valves

Fuel-and Coolant-System Pressure Control. . .

~Acknowledgment . . .

References . . . « «
INTRODUC TTON =

Preliminary testing of the Molten Salt Reactor Experiment began in
the fall of l96h. The salt systems were heated to 1200°F, purged with
helium to remove moisture, and 66 LiF-34 BeFs salt was charged into the
fuel and coolant systems. Salt was circulated for 1000 hr in the fuel
system and 1200 hr in the coolant system as most of the non-nuclear test-
ing was completed. This run, designated PC-1, was concluded in March,
1965. o
In the shutdown after PC-1, final preparations were made for 'zero-
power'" nuclear operation. These included installation of the sampler-
enricher, completion of the nuclear instrumentation and controls, and
operator examinations and certification.
The salt which had been circulated through the fuel system was pro-
cessed to remove oxide, then was isolated in the flush-salt tank for
future use. The basic fuel salt, lacking only the enriched uranium to .
bring it to the final composition, was charged into a drain tank. The
fuel system was heated and this "carrier" salt was circulated for 10 days
in Run PC-2. AnalysisVof the salt showed no abnormalities,»all‘equipment
operated well, and all was in readiness for the nuclear experiments.

Addition of 3°UF,-LiF to the salt began on May 24, and on June 1
criticality was first attained. Following this experiment, more =35U was
added to bring the concentration up to the level required for power opera-
tion. While the extra 35U was being added, experiments were done +to
measure the nuclear characteristics of the system. This series of experi-
ments was completed on July 3, the fuel salt was drained and the system
was flushed, concluding Run 3.

This report describeé the preliminary findings of the nuclear experi-
ments, the chemical behavior of the fuel salt and the performance of the
mechanical components.

Final preparations for high-power operation are presently being made.
These include inspection, maintenance, installation of shielding, and

sealing and testing the containment.
 

-~

NUCLEAR EXPERIMENTS

Tnitial Critical Loading

The purpose of this experiment was to prbvide'a check on the calcu-
1ations of critical concentration under the simplest conditions: the core
isothermal, rods fully withdrawn, and the fuel stationary; Tt also served
to establish the basepoint from which the =35y additiohs necessary to
reach the operating concentration could be made with confidence.

The fuel salt composition specified for power operetion is _

65 LiF¥29.2.BeF2-5'ZrF4—O.8 UF, (expressed as molar percentages). The

total uranium content is considerably above the minimum required for

criticality if highly enriched uranium were used, and was chosen for

reasons of chemistry.

The critical 25U concentration was predicted by calculations using
a multi-group, one-dimensional diffusion code, MODRIC, with thermal-group
cross sections obtained from cell calculations'by,the THERMOS code and
fast-group cross Sectiens calculated by GAM-2.1,2 The geometrical approxi-
mations were checked by using a two-group, two-dimensional code, Equi-
poise-3, with group constants for each region from MODRIC. It was predict-
ed that the reactor would be critical at 1200°F, rods out, fuel static
with 0.256 mole % Z35uF, (0.795 mole % totalVUF4); |

Instead of using 32%-enriched uranium to make'up the fuel salt, we
decided to start with depleted uranium in the salt and add the reguired
amount of 235U as highly enriched uranium (93% =2°U). This permitted
preliminary operation with uranium in the salt before the beginning of
nuclear operation and also facilitated the manufacture of most of the
uranium-bearing salt. The salt was prepared in three lots: the carrier
salt, cOntaining‘the beryllium, zirconium and most of the lithium
fluorides; T3 LiF-27 UF, eutectic containing 150 kg of depleted uranium;
and eutectic containing 90 kg of 23U in the highly enriched form.

Thirty-five cans of carrier salt and two cans of eutectic containing
the depleted uranium were blended as they were charged into a drain tank
in April. This mixture of salt was then circulated for 10 days at 1200°F

while the sampler-enricher was tested and 18 samples were analyzed TO

 
establish the initial composition. The critical experiment then consisted ~
of adding enriched uranium in increments to bring the 35U concentration |
up to the critical point. |

Nuclear instrumentation for the experiment consisted of two fidsion
chambers, two BFs chambers and an 241Am-2420m Be source, located as shown
in Fig. 1. The fuel salt itself also constltuted a neutron source, due to
reaction of =34U alphas with beryllium and fluorine.

The enriching salt was added in two ways: by transfer of moltén
salt from a heated can into a drain tank, and by lowering capsules of
frozen salt into the pump bowl via the sampler-enricher. The latter
method was limited to 85g =35U per capsule, only 0.0012 of the expected
critical loading. Therefore the bulk of the 25U vas added in 4 additions
to the drain tank. After each addition the core was filled and count-rate
data were obtained to monitor the increasing multiplicatibn.

The amount of =35y expected to make the reactor criticaliwas calcu-
lated to be 68.7 kg, using the volumetric concentration from the criti-
cality calculations and the volume of salt thought to be in the fuel loop
and drain tank. (The value of salt density’which was used to get the ~
volume from the known weight of salt is now believed to be erroneofis,

See later discussion.) |

Before the addition of enriched uranium, cbunt rateé'had been deter-
mined with barren salt at several levels in the core. Then as the core
was filled after each =35y addition;_the ratio,of count rates at each
level was used to monitor the multiplication. (Fig. 2 shows elevations;
count rates were determined with salt at 0.4, 0.6, 0.8, and 1.0 of the
graphite matrix and with the vessel full.)

Count-rate ratios with the vessel full after each of the four ma.jor
additions are shown in Fig. 3. Bach addition, fill, and drain took be-
tween one and two days, so only four major édditions had_been planned.

After the third addition, with 64.5h kg 2357 in the salt, the projected
critical loading was 70.0 * 0.5 kg ©35U. (The 1-inch BF» chamber located
in the thermal shield, whose count rates extrapolated to a higher wvalue
wa.s known to be strongly affected by neutrons coming directly from the

source.) The fourth addition was intended to bring the loading to about
ORNL~-0OWG 65-7575

REACTOR INLET
LINE 102 =

 
  
     
    
     
 
  
 

f-in.BFy CHAMBER TUBE

N

. ~*"\— REACTOR OUTLET
,\\ LINE 400
i \\ Vo 34 ’

. °
REACTOR
VESSEL

 

2-in. BF3 CHAMBER
FISSION CHAMBER

 

 

THERMAL SHIELD

INSULATION FISSION CHAMBER
NO. ¢

NUCL. INST.
PENETRATION .

SOURCE TuBE

  
  
  
  
  
    
   
  
     
  

THERMAL SHIELD
INSULATION

2-in. BF3

CHAMBER
TOP OF GRAPHITE

ELEV. 833 ¢t 54 in.

ELEV. 8301t 8in.

"REACTOR VESSEL
(CORE MIDPLANE)

ELEV. 829t 9in.

NUCL. INST.
PENETRATION
ELEV. 828 1t . 3Y2in. BOTTOM OF GRAPHITE

ELEV. 828ft OYain.

FISSION CHAMBERS

"{-in. BF3 CHAMBER
BOTTOM ELEV.
- 828 ft 3in.

ELEVATION

Fig. 1. Source and Instrumentation in Tnitial Critical Experiment.

 
ORNL-DWG 65-7573

 

 

 

 

 

 

 

 

 

 

ELEVATION (ft)

 

836 836
REACTOR OUTLET
835 ft—3%in. —- - — - -
| 834 ft—2.15n, |
{0 — —
834 834
[ op o mons *
0.9 |- 1.0 833t—-5%in. TOP OF MOST GRAPHITE
833 f1—3 in.- 51 - H _UPPER ROD LIMIT
833 —— a8 - o33
| 0.8 |- 09 - Tk
; : |
o 3 08} - 3
= 0.7 |~ 5 - O
& £ 0.7}~ - -
~ * o~ b
o X € _F w
& @ 2 L o
S ~06F & - S
O x o . =
M=% o5} & o 24 F - 831
= g o - o
O s 0.5 |- Qa - coc
™ W o 18 .
— L -
>3 ° § 2 r 2 830
S © - O
= 03} W 0.3 |~ s E
QO - .
& S C DRIVEN ROD
S ~ 0.2 — - LOWER LIMIT
o29 0.2 - § °F \ 829
o4 _a £
w
A= SCRAMMED ROD
0.1 LOWER LIMIT J
' 828 ft—0Y,in. H
828 —— oL 2 ORIZONTAL GRAPHITE 126

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig. 2. Relation of Rod Position

 

 

’and Levels in Reactor Vessel.
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Y
-DWG 65-7574
06 | | ORNL' 365-75
A 2-in. BF; CHAMBER
A14n.BF3 CHAMBER
\ e FISSION CHAMBER NO.1
\ o FISSION CHAMBER NO.?2
- 0.5 \ —
Z 0.4\
% [\
o _
o
O
o
Y
x 03 \
W
2
* Q
. \
N - A
= N |
O \ \ \
o 0.2 |
N \
\ \\ \ ]
0.4 f . \\\ ,
n\m\\
. , )

 

40 44 48 52 56 60 64 68 T2
'MASS OF 2%°U IN TOTAL SALT CHARGE ( kg)

Fig. 3. Count Rate Ratios After First Four Additions of 235U. (Vessel
full, rods at 51 in., Source at 829'-9", chamber locations as in Fig. 2.)

N

 
10

1 kg below the critical point. (This was not as bold as it has seemed to
some ; the minimum critical loading at which we were shooting was only a
checkpoint before subsequent additions. No question of safety was in-
volved if there had been an overshoot.) 4.38 kg of 35U was added and

the count rates showed the loading was within 0,8 kg of critical when

the rods were withdrawn and circulation was stopped. Preliminary esti-
‘mates of rod worth and circulation effect, based on changes in subecritical
multiplication, were!approximately the expected values.

In the final stage, enriching capsules were added through the pump
bowl to bring the loading up 85g at a time. After each addition circula-
tion was stopped, the rods were withdrawn and count rates were measured.
The external source was withdrawn for some of the measurements to see the
relative strength of the internal and external source. With the reactor
within 0.2% 8k/k of critical, slight variations in temperature caused con-
siderable changes in multiplication. (Variations in the voltage of the
area power supply change the heater inputs slightly, requiring fine ad-
Jjustments of the heater controls to keep the temperature precisely at a
specified temperature.) After T capsules, it appeared that after one
more, the reactor could be made critical. The eighth was added, circu-
lation was stopped and the rods were carefully withdrawn. At 6:00 PM,
June 1, the reactor reached the critical point, with two rods at full
withdrawal and the other inserted 0.03 of its worth.‘ Criticality was
verified by leveling the power at successively higher levels with the
same rod position. The 235y loading was 69.6 kg.

Predicted and observed =35y requirements for criticality are compared
.most logically on the basis of volumetric concentration. The required
volumetric concentration is nearly invariant with regard to the fuel-salt
density (unlike the mass concentration, which varies inversely with
density) and depends not at all on system volume or total inventory.

The "observed" 25U concentrations are on a weight basis, obtained
from either inventory records or from chemical analyses. These weight
concentrations must be converted to volumetric concentrdtion by multiply-
ing by the fuel-salt density. The amounts of 25U and salt welghed 1into
the system gave a 25U weight fraction of 1.42% at the time of the initial

criticality. The chemical analyses during the precritical operation and
11

the zero-power expériments gave uranium concentrations which were 0.985
of the "book" concentrations. (Part of this discrepancy, about half we
believe, is due to dilution of the fuel with flush salt left in freeze

valves and drain-tank heels when the fuel salt was charged. ) Applying

‘this bias to the book concentration at criticality gives an "analytical"

235 weight fraction of 1.40%. The density of the fuel salt at 1200°F we
now believe to be about 145.5 1b/ft°. This is the preliminary result of

- recent laboratory measurements of density, and it agrees with measurements

made in the reactor using the two-point level probes in the drain tanks.
Earlier measurements in the reactor, using the drain-tank weight indi-
cations and the volume of salt believed to have been transferred into
the fuel loop, gave 136.6 1b/ft?. (The amounts of =25U in the fuel loop

shown in Figs. 6 and T and the accompanying text were computed on this

 basis.)

Corrections must be applied because the initial critical cdnditions
were not exactly»thésame as those assumed in the predictidns. Thé core
temperature was 1181°F instead'of 1200°F and the control rods were
poisoning 0.184% Sk/k instead of none. (Two rods were at maximum with-
drawal, 51 inches, and one was at 46.6 in.) The predicted 235U concen-
tration for criticality at»the reference condition was 32.87 g/Zj
corrected to the actual conditions, it is 33.06 g/z, This predicted

value is compared with "observed" concentrations in Table 1.

| -Table 1 |
COMPARISON OF CRITICAL #35U CONCENTRATIONS
1181°F, PUMP OFF, 0.18% &k/k ROD POISONING

 

£35U Conc. Fuel Density 2357 Conc.

 

 

. (we. %) (16/£%) (e/2)
Predicted -——— ee--- 33.06
Book 1.k 145.5 33.10

| 136.6 31.07
Analytical 1.40 145.5 32.60
136.6 30.60

 
12

If, as we estlmate, the true concentratlon was about half way be-
tween the book and the analytlcal and the den51ty is about th 5 lb/ft3
the actual concentratlon was extremely close to the predlctlon

Laboratory'measurements now in progress should conflrm the value of
fuel-salt den81ty. The uncertalnty between book and analytlcal concen-
trations will also be reduced as a result of the next startup, when
analysis of the fuel after another flushlng, draining, and refllllng

will help us evaluate the dilution effect

Control Rod Worth

The addition of 2357 beyond the minimum critical loading had a two-
fold objective: to end with enough excess reactivity to permit operation
at full power‘and in the procees to make measurements which could be
analyzed to give control-rod worth and various reactivity coefficients.
The final emount of 235U was to be enough to be critical at 1200°F with
the fuel statlonary and one rod fully 1nserted The method was to add
85 g 2357 at a time through the sampler- enrlcher, after each capsule
determine the critical rod position, and at longer intervals do other
experiments. | - | |

Experlments speclflcally aimed at rod worth were stable period
measurements and rodfdrop experlments. These were done with the fuel
static and with it circulating. The results will give, as accurately as
possible, the total and differential worths of the regulating rod (rod 1)
over its entire travel with the other two rods fully withdrawn. In
addition, worth values will be‘obtained for each of the three rods with
the other two withdrawn and at intermediate positions. These will lead
to evaluations of rod shadowing and '"ganged" rod worth. So far we have
finished analyzing only the period measurements with stationary fuel for
rod 1. |

After the initial critical experiment, another 8 capsules were re-
quired before the reactor could be ma.de crltlcal at 1200°F with the fuel
pump running (a consequence of the loss of delayed neutrons during circu-
lation). Thereafter we measured the critical rod position with the pump

running after each capsule. At intervals of U4 capsules, we made period
~N

13

measurements with the pump running then turned it off, determined the new
critical rod position, and made more period measurements. This went on
until a total of 87 capsules had been added. Three times during this ex-
periment (after 30, 65, and 87 capsules) we observed rod-drop effects.

Period measurements were usually made in pairs. The rod whose sensi-
tivity was to be measured was adjusted to make the reactor just critical
at approximately 10 watts, then it was pulled a prescribed distance and
held there until the power increased about 2 decades. The rod was then
quickly inserted to bring the power back to 10 watts and the measurement
was repeated at a shorter stable period. Periods were generally in the
ranges from 4O to 50 sec and from 70 to 120 sec.

For the measurements with the pump off, the usual inhour relation

was used to calculate excess reactivity from the observed stable period.

) o

4 T Pi

PTrrT T+ _E: T+ 3.1
) L TF

 

The delayed_neutron fractions Bi, used in the stationary-fuel cases were
accepted values, corrected for the increased importance of the lower-

energy neutrons in the MSRE.?!
- Table 2

DEIAYED NEUTRON FRACTIONS IN MSRE

 

 

104 x fraction (h/n)

 

Half-Life | ~ Effective

Group (sec) ~ Actual (Static Fuel)
1 55,7 2.11 2,23
o 22,7 14.02 14.57

3 6.22 12.54 13.07

L 2.30 25.208 | 26.28

5 0.61 7.4 T.66

§

0.23 2.70 - - 2.80

 

 
1k

Two fission chambers driving log-count-rate meters and a two-pen recorder
were used to measure the periods. (The stable period was taken to be the
average of their two readings, which usually agreed within about 2%.) We
watched a linear recorder, driven by a compensated ion chamber, while we
leveled the power at the beginning of a measurement. In the analysis, we
computed the initial reactivity (usually less than 0.003%) from this
recorder chart if there had been a slow drift. The difference between the
reactivity during the transient and the initial reactivity was. divided by
the rod movement and this sensitivity was ascribed to the rod at the

mean position. | | |

We obtained the total rod worth by integrating the curve of rod
sensitivity vs position shown in Fig. 4. Because rod worth is affected
by the 23U concentration in the core, it was necessary to apply theo-
retical corrections to the measured sensitivities to put them all on the
basis of one concentration. The points in Fig. 4 were corrected to the
initial critical concentration, where the sensitivity 1s the highest.

The correction factors which were applied incréase linearly with <3°5U
concentration to a maximum of 1.03 at the final concentration (the points
between 1 and 2 inch). Had the points been corrected to the final concen-
tration, the curve would have been lower by three percent.

Figure 5 shows a curve of rod effect vs position at the initial con-
centration which is the integral of the differential-worth curve in Fig. L.
The curve for the final concentration is simply the first curve reduced by
a factor of 1.03. | | o

The theoretical predictions of rod worth were based on four-group,
two-dimensional diffusion Calculations made with the Exterminator program.
The change in the static multiplication constant with all three rods in-
serted all the.way through the MSRE core, éhd with the fuei composition
practically the same as the initial experimental composition, was
—5.51% in ke. This is equivalent to a static reactivity of —5.79% Ak_/ke.
For these same conditions, this theoretical model gave a static reactivity
of —2.28% Ake/ke for rod 2 with rods 1 and 3 completely removed. Although

the actual rod calibrated was rod 1, complete calculations were made with
15

ORNL-DWG 65-8033

 

0.07

 

 

o
o
o0
®
°
e L

 

£ |
N .

s | * N .
s 005 £ ™

" SN

o

o

b

p
o

 

 

oa . ? L \ \\

003 — » - _\\

0.02 ' — , , 1 |
7 ‘ . w \\

0.0 . — 4 _ 4 , 1 |
ol | ' o ‘ | ]
0 4 8 12 6 20 24 28 32 36 40 44 48 52
| | ROD POSITION (in.)

 

DIFFERENTIAL WORTH

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig. 4. Differential Worth of Control Rod No. 1. (Adjusted to initial
critical loading.) |

 
2.2

REACTIVITY WORTH (% 34/k)

16

ORNL-DWG 65-8034

 

7

/

"

 

2.0

7

 

1.8

 

1.6

1.4

 

 

 

CRITICAL LOADING

|z
2 |

 

(61.52 kg 235U IN LOOP) /4

 

.’

 

FINAL LOADING |
(67.98 kg 223y IN LOOP)

 

 

 

 

 

t.2

 

 

1.0

 

0.8

 

0.6

 

0.4

 

0.2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

12

16

20

24

28

32

ROD POSITION (in)

36

40

44

Fig. 5. Integrated Worth of Control Rod No. 1.

48 82

13
17

this model only for rod 2, since this rod lies diagonally opposite and
symmetrically with respect to the graphite sample holder.

The worth of the partially inserted, banked rods was calculated for
a cylindrical annular model of the rod bank, using the two-group, two-
| dimensional program, Equipoise-3. These results indicated that the
available travel of 51 in. covered 0.927 of the worth of rods all the
'way through the core. (The effective height of the core, extending into
the upper and lower heads, is 78.2 in.) Thus the effective static reac-
tivity for rod 2 was —2.28 x .927 = 2.11%. Earlier calculations made for
different fuel salt compositions had indicated that the worth of either
rod 1 or rod 3 is about 1.03 times that for rod 2, due to the positions
of the rods relative to the graphite sample holder. Thus the predicted
worth of rod 1 would be about 2.2%, in favorable agreement with the
experimental measurements.

We are working on the analysis of the period measurements with the
fuel circulating. As will be discussed later, a new mathematical treat-
ment of delayed-neutron effects in the MSRE is being programmed; this
will be used to relate period to reactivity. The sensitivities determined
in this way should, of course, coincide with the results of the static-
fuel measurements. Because of the difficulty of accurately treating the
complex pattern of precursor distribution in the circulating fuel, the
values obtained withvthe fuel static should be more reliable.

The rod-drop experiments are also béing analyzed.‘ They will give
independent values of rod worth and may show the effect of £35U concen-
tration on rod worth. This latter effect is also the subject of multi-
group calculations at the concentrations observed in the experiment.
(The correction factors used in the preliminary analysis were interpo-

lations of results at higher and lower concentrations.)
18 . s

N’
235U Concentration Coefficient of Reactivity

The 235U concentration coefficient of reactivity is given by the
ratio of the change in reactivity to the fractional change in €35y concen-
tration (or circulating mass) as a result of a small addition. The effect
of each capsule addition (after initial criticality) on the critical po-
sition of the control rod was determined with the pump running. The
critical position with the pump off was measured after every fourth ad-
dition. Rod positions were converted to reactivity, using Fig. 5 and
correcting for the concentration effect on rod worth. Results are shown
in Fig. 6.

The slope of a curve in Fig. 6 at any concentration multiplied by
that concentration is the desired concentration coefficient of reactivity,
(8k/k)/(dm/m). The coefficients obtained in this way range from 0.22 at
the final concentration to 0.23 at the lowest concentration. The coef-
ficient predicted from multigroup criticality searches about the minimum >

critical concentration was 0.248. (ref. 2)

Reactivity Effect of Circulation

The reactivity effect of circulation, given by the difference of the
two curves in Fig. 6, is —(0.212 + 0.004)% &k/k. The effect of changes
in delayed-neutron precursor distributions with circulation had been pre-
dicted to be —0.30% 8k/k.1»3 Another —0.2% 8k/k was expected because of
entrained bubbles of helium in the circulating salt. ‘As will be discussed
later, the evidence shows there were practically no circulating gas bubbles
excépt for a brief‘period when the fuel level 1in the pump bowl was lowered.
Thereforé~thé gas effect attending circulation was practically nil.

The difference between the predicted and observed delayed-néutron
losses we bzlieve is due to inadequate accounting for delayed neutrons
emitted just dutside the graphite region of the core in the upper and
lower heads. A more realistic model has been developed and 1is now being ,
programmed. The program will compute precursor distributions under both

steady-state and transient conditions, taking into account mixing,
8h/k (%)

19

ORNL - DWG 6.5-0031 o

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.4
7
2.0 e
A& ’
1 | FUEL NOT CIRCULATING ¢~ |s*
1.6 'l/
- Sl /O’ g
d
i A .
'./‘/,-f’;’u& CIRCULATING
o8 o
e ,' .'
2
o
. '/'
0.4 2 pd
R ~ _....v |
| | P,,/’ e
6t 62 63 64 65 66 67 69

Fig. 6. Effect of 235U Mass on System Reactivity.

MASS OF 233U IN FUEL LOOP (kg)
20

velocities, volume fractions, and flux distributions in each of the prin- N~
cipal regions. The weighted contributions of the delayed neutrons from
each group will be computed, taking into account the initial energies of
the neutrons and the nuclear importance of each region. The result will
be a "circulating-fuel inhour equation' for the MSRE whose uses will in-
clude the analysis of the circulating-fuel period measurements and, as a

special case, the steady-state effect of circulation.
Temperature Coefficients of Reactivity

We measured the effect of temperature on reactivity by adjusting
the electric heaters to change the system temperature slowly (about 15°F
per hour) while we observed the critical position of the regulating rod.
This experiment gave the overall temperature coefficient, i.e., the sum
of the fuel and graphite coefficients. We also attempted to separate the
fuel (rapid) and graphite (sluggish) coefficients by an experiment in
which the coolant system was used to increase the fuel-salt temperature -
rather abruptly.

Figure 7 shows results of the three experiments involving slow
changes in temperature. The first experiment, with 64 kg 3°U in circu-
lation, gave a line whose slope ranges from —(6.6 to 8.3) x 107> °F~1. At
66 kg the experiment gave a straight line with a slope of —7.2k x 107> °F ™.
In the last experiment, at 68 kg, the slope of the curve above about 11L4O°F
is =7.3 X 10”5 °F"1. A value of —7.0 X 107° °F~! had been predicted:
—3.3 x 1075 °F-1 for the fuel and —3.7 x 1075 °F ! for the graphite. The
fuel-salt coefficient of thermal expansion used in this calculation was
obtained by an empirical correlation of density and composition of salts
other than our fuel salt. Recent measurements of fuel-salt density gave
a higher thermal-expansion coefficient, leading to a calculated fuel
temperature coefficient of reactivity of —5.6 x 107> °F~1 (ref. L4)~The new
value for overall temperature coefficient is —9.6 X 1075 °F *. The ob-
served coefficient is in better agreement with the earlier prediction
(on which the safety analysis of the MSRE was based). The density measure-
ments are being reviewed and they will be checked by an independent

method of density determination.
 

"

18

2l

ORNL-DWG 65-8032

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

16 , e /'
64.14 kg 233U IN LOOP 4
1.4 —p
& . .
& |
S os 4 L L,
. 67.98 kg 233U IN LOOP »/
0.6 > 4
+7
04 /l-/ 76643 kg U N LOOP
0.2
1000 1100 1150 - 1200 1250

FUEL TEMPERATURE (°F)

Fig. 7. Effect of Fuel Temperature on Reactivity.

 
22

The experiment at 68 kg =35U shows a lower slope below about 11L4O°F.
We do not believe that the temperature coefficient is lower in this range
but that another phenomenon became significant during this part of the
experiment. That is the appearance of an increasing amount of helium
bubbles in the circulating salt as the temperature was lowered. The evi-
dence for this is discussed in the next section on pressure effects. The
effect so far as the temperature experiment is concerned was that the bub-
bles tended to reduce the amount of fuel salt in the core, compensating
to some extent for the increase in density of the salt 1tself as the
temperature was lowered. Thus the slope of the lower part of the curve
-~ cannot be interpreted as a temperature coefficient of reactivity in the
usual sense, |

The hot-slug transient was done by stopping the fuel pump, raising
the temperature of the circulating cdolant salt and the stagnant fuel in
the heat exchanger, then restarting the fuel pump to pass the hotter fuel
salt through the core. The output of a thermocouple in the reactor-vessel
outlet, logged digitally at 1/k-sec intervals, showed a brief increase of
5 to 6°F as the hot salt first passed. It then léveled at about 3.5°F
for a few loop transit times before decreasing gradually. - The noise in
the analog-to-digital conversion (* 1°F) limited the accuracy of the
measurement, but by taking an average of 50 points during the level period
after mixing and befbre the graphite temperature had time to change signifi-
cantly, a value was obtained for the step in fuel temperature. Reactivity
change was obtained from the change in rod position, corrected for the
decrease due‘to_circulation,,and ascribed to the fuel- temperature increase.
The result was a change of —(L4.9 * 2.3) x 107> °F~'. Predicted values of
the fuel- temperature coefficient lie in this range. We propose to repeat
this test with the thermocouple signals biased and amplified to reduce
the effect of noise in the analog-to-digital conversion. More precise

results can be expected in this case.
.23
Effect of Pressure on Reactivity

We performed three tests to explore the_effect on reactiVity of chang-
ing system overpressure. Theoretical considerations had indicatedthat for
slow changes a very small, possibly negative, pressure eoefficient of re-
activity could be expected but for rapid changes the coefficient would be
positive. The existence of any pressure coefficient was based on the
assumption that undlssolved helium would be entrained in the c1rculat1ng
fuel. In each of the three tests the loop OVErpressure was slowly in-
creased from the normal 5 psig to 10 to 15 psig and then qulckly‘relreved,
through a bypass valve, to a drain tank'that had been previously vented to
atmospheric pressure. |

The first two tests were carried out at normal system temperature with
the normal operating level of salt in the fuel-pump bowl. ‘No change in
control-rod position was required to maintain criticality and no signifi-
cant change in pump-bowl level was obserred &uring either of the tests.
These indicated that the pressure coefficient was negllglbly small and
that essentially no hellum bubbles were clrculating with the salt. Further
evidence of the lack of circulating voids was obtained from a gamma-ray
densitometer on the reactor inlet line; this instrument showed no change
in mean salt density during the tests. a

The third test was'performed at an abnormally'lOW'pump—bowl level
which was obtained by lowering the operating temperature to 1050°F. Figure
8 shows the pressure transient and the respenses of the regulating control
rod, densitameter; and fuel-pump level during the rapid pressure release.
Independent evaluations of-the void fraction from these three parameters
all gave valnes between 2 and 3 percent by volume. The frequency response
characteristics of the effects of pressure on reactivity were calculated
from the pressure end rod motion and are shown in Fig. 9 along with the
predicted high-frequency respense'for a void fraction of 1.2%,_ Extrapo-
lation to the observed curve gives a void fraction of 2 to 2-1/2 percent.
The low- and high-frequency pressure coefficients were + 0.0003 and

+ 0.014(% 8k/k)/psi, respectively, for this particular condition.

 
24

ORNL-DWG 65-8074

E P LEVEL '(70)

OVERPRESSURE (psig)

 

DENSITOMETER

 

CONTROL ROO POSITION (in.)

0 4 8 12 _16 20 24 28
TIME (min)

Fig. 8. Conditions During Rapid Pressure Release While Circulating
Helium Bubbles. |
 

e | R ORNL DWG. 65-8304
1070 - |

Hithréquenéy Value
Predicted for

1.2% Void Volume

 

10 -

1 1.0 | 10.

Frequency w, rad/min - « | o

"Fig. 9. Reactivity-Pressure Transfer Function with 2 to 3% Void Volume in Fuel. (Calculated
from pressure release experiment using Samulons method with 0.2 min sampling interval.

W

¢z
26 .

Dynamic Tests

We performed a variety of dynamic tests during the operation at zero
power. These tests were the start of an extensive program to evaluate
experimentally the inherent nuclear stability of the MSRE at all power
levels. The reactor system has been analyzed on a theoretical basis®
and the tests are designed not only to characterize the present system
but also‘to evaluate the techniques and mathematical models used in the
theoretical analysis. Preliminary results from the analysis of the zero-

power tests are presented below.

Frequency Response Measurements |

A series of tests was run to determine the frequency response of neu-
tron level to reactivity perturbations. These experiments included pulse
tests, pseudo-random binary reactivity-perturbation tests, and measurements
of the inherent noise in the flux signal. Tests were run with the fuel
pump on.and with it off. DNoise measurements were also made during a
special run with a low pump-bowl level where there were entrained bubbles
in the core.

The frequency response is a convenient measure of the dynamic charac-
teristics of a reactor system. Classically the frequency response is ob-
tained by disturbing the reactor with a sinusoidal reactivity perturbation,
and observing the resulting sinusoidal neutron-level variations. The mag-
nitude ratio is defined as the ratio of the amplitude of the output sinu-
soid to that of the input sinusoid. The phase angle is defined as the
phase difference between the output sinusoid and the input sinusoid.

Other procedures, such as those described in this report, can be used to
yield the same results as the classical methodgéfith less experimental
effort. - / ) |
| The zero-power frequency response tests serve to check the theoreti-
cal, zero-power frequency response predictions, but they do not furnish
direct information on the stability of the power-producing reactor. The
zero-power tests, however, do serve as an indirect, partial check on the
at-power predictions because the_dynamic behavior at power is simply the

zero-power case with the addition of feedback from the system. Thus,

 
27

verification of the zero-power kinetics predictions lends some support

to the predictions regarding power operation.

 

Procedures -- In the pulse tests a control rod was withdrawn 1/2 in.,
held there for 3-1/2 or T sec., then returned to its original position.
The rod was placed so that this rod motion caused a change in keff of
about 0.03%. The rod-position signal and flux signal were recorded digi-
tally at 0.25-sec intervals using the MSRE on-line digital computer
(referred to in Figs. 16 and 17 as "logger"). The frequency response

was obtained by numerlcally Fourier transformlng the input and output

R

signals. . 7Y Hrid A ,
The pseudo- random binary tests con51sted of a spe01ally selected
series of p051tive‘and_negat1ve pulses. The series contained 19 bits
each with a duration of 7 sec., and was repeated several times so that
the initial transients could die out. The rod motion about the average

position corresponded to a Ak of about +O'DQ015?{ The rod—position

signal and flux-level 31gnal weie recorded digitally at 0.25-sec intervals
using the MSRE computer. The frequency response was obtained by two
different methods. The direct method used a digital filtering'technique
to obtain the spectral density of the input and the cross spectral density
| of the input and the output. The frequency response of the system is the
ratio of the cross fio input power spectral densities. The indirect
method invelved calculation of correlation functions and subsequent numeri-
cal Fourier transformation. Both methods gave essentially the same
results. |

The noise measurements and»analyses used direct analog filtering of
an analog tape record of the inherent noise in the fluxélevel signal. The
tests required one hour of data recording to give statistically accurate
results. These tests were hampered by the unfavorable location of the de-
tector and the resulting low flux-signal level at low power (~ 10 watts).

Results -- Preliminary results of the frequency response measurements
are shown in Figs. 10 through 14 along with theoretical predictions for
the runs with the fuel pump on and with the pump off. As shown on the
legend in the figures, several different procedures were used to obtain

these results, but this represents only a partial analysis of the available

 
ORNL DWG. 65-8905

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

loh |
8 ® Binary Test - Indirect Analysis
6 A Binary Test - Direct Analysis .
N : fi Noise Analysis AT e T TF : £
2 \‘\
100 | ,
8 o
U e,
e © 04805 )
/K| 2 g 4$-.§’“%¢4&;;\\\4§
N A
2 ~ .
10 N 1
8 . N >
6
L N
2
10 . ; - - -
_ ] _ ] ]
2 4 6 8 2 Ly 6 8 2 L 6 QLO 2 4 6 suxo 2 L 6 %Dox)

.001 . .01 v 1
| Frequency w, rad/sec

1\1/1\1o
K/KO

 

Fig. 10. Frequency Response - Magnitude Ratio of with Fuel Circulating. Results of binary

and noise tests.

8¢
ORNL DWG. 65-8906

 

 

 

 

 

 

 

 

 

 

 

 

 

th
| O Test No. 1
o 0 Test No. 2
(
O
lO3
8N/No
8(71{0
102
lOl .
.001 | 01 Jd 1.0 10.0 100.

Frequency w, rad/sec

6N/NO
6K/KO

 

Fig. 11. Frequency Response - Magnitude Ratio of with Fuel Circulating. Results of two

7-sec pulse tests.

6¢C
Phase, deg

ORNL DWG. 65-8907

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20

10

0
-10 . ‘

O 7-Sec Pulse, Test No. 1 TA‘

0| 9 T-Sec Pulse, Test No. 2 .

- ® Binary - Indirect Analysis » AA “\
4 Binary - Direct Analysis 2 \
-30 y :A .
a A ado3/0
=40 .D ol
o /
% ___-———'J
=50 ‘ . 7’
' o
o | //
/ °
(o) )V
o

-&) /J / ?
-90
-100 ' - ‘

2 L 6 8 2 L 6 8 2 L 6 8 2 L 6 8

.001 .01 Jd 1. | - 10.

Frequency w, rad/sec

SN /N
Q

Fig. 12. Frequency Response - Phase of 6K/K with Fuel Circulating. Results of pulse and
binary tests. O

O¢
 

ORNL DWG. 65-8308

 

 

T-Sec Pulse, Test No. 1
'7-S¢c Pulse, Test No. 2
3-1/2-Sec Pulse, Test No. 3
3-1/2-Sec Pulse, Test No. L -
Noise Analysis / My a 4 m

 

 

 

mOvcQOe

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2
103 é N)\
8 g
° 4
og
L i3
| U3
&N/No
/K,
‘;ngYY ‘ir-——_qiqfl;fl;g
4 | Qg O %\J\
10°
8
6
L N
2
10t 1
2 L, 6 8 2 4 6 8 2 Yy 6 8 2 y 6 8 2 L 6 8

 

 

.001 .01 | 1 1.0 10. 100.
Frequency u, rad/sec | |

N/N

Fig. 13. Frequency Response - Magnitude Ratio Of‘K/K

noise tests.

with Fuel Static. Results of pulse and
o

e

* “ A,::‘,.hg“m.' ,;‘i;"
L FVe
Phase, deg

ORNL DWG. 65-8909

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20
10

0
-10 A1t

P E—p—"'—- ~——
0O f o \
-0 ) N
b B
-30 - 5
o9 Popa B
-40
'f "”4-"
-60 5 '
| ® 7-Sec Pulse, Test No. 1
. n | 0 7-Sec Pulse, Test No. 2
"1 (, A 3-1/2-Sec Pulse, Test No. 3
n Y //‘ | | © 3-1/2-Sec Pulse, Test No. k
-80 J i//b
o
et

-90

-100 ,
001 2 L 6 8.01 2 k4 6 8 3 2 L 6 81.0 2 L 6 8 10.

Frequency w, rad/sec

SN /N |
O
Fig. 1l4. Frequency Response - Phase of 6K/K with Fuel Static. Results of pulse tests.
- o

ce
33

data. All of the experimental points except for the noise-analysis re-
sults are in absolute units (fractional change in neutron flux per change
in keff)' The noise'analysis results are based on the assumption of a
white-noise input and contain an unobtainable proportionality constant.
Thus, the noise-analysis data were arbitrarily multiplied by the factor
required to give agreement with theoretical results at a'frequency of
9 radians/sec. |

Figure 15 shows the noise analysis results for operation at a reduced
pump-boWl level, which caused increased bubble entrainment in the fuel
salt. Comparison of the noise spectrum obtained with bubbles circulating
(Fig.’15) and the noise spectrum obtained with no bubbles circulating
(Fig. 10) indicates that the bubbles increased the amplitude of the power
spectral density significantly in the I- to lO-rad/sec region. Previous
experiments with the MSRE-core hydraulic mockup indicated that random,
hydraulically-induced pressure fluctuations in the core would probably
cause a significant modulation of the core void fraction, thus causing
reactivity fluctuations. Hence, additional flux noise in this frequency
range was expected, although it was not possible to predict the "shape"
or characteristics of this spectrum.

Unfortunately, the noise measurement was the only type of frequency-

response measurement made during operation with a reduced pump-bowl level.

We propose to operate again with a reduced pump-bowl level early in the §f¢,j;

é
next critical run, repeating the noise tests and also doing pulse and %
pseudo-random binary tests. This should adeguately determine the influence\

| BT

of bubbles on dynamic behavior.

Evaluation of Results -- Significant results from the zero-power

 

pulse and binary noise tests were expected only if the.temperature drifts
during the tests weré held to essentially zero, and the control rod could
be positioned with accuracies better than 0.0l inch.. In spite of these
stringent requirements, results were obtained which agree favorably with
the predictions. Reasons for the consistently low values of magnitude
ratio are currently being studied to determine if they are due to inac-

curacies in the measurements or to. imperfect data

or models used in

-~

-t

  

theoretical calculations. ————

 
'ORNL DWG. 65-8910

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10
8 —
*The Frequency Response X is Proportional to
. o ‘
¢l VPSD if the Disturbance in 9K is white noise
o o o ,
O
4 O
a
&
' .0
2
o .
o o :
°l o o
ma,L | | -
1.0 - 2 4 6 8 2 4 6 8

Frequencyw , rad/sec |

Fig. 15. Frequency Response - SQuare Root of Power Spectral Density (PSD) of Flux Signal* -

Fuel Circulating with Bubbles Results of Noise Analysis.

e
35

It was encouraging that the necessary data could be obtained in the
important frequency range for the MSRE without the purchase of spec1al- |
purpose equipment However, some lessons were learned that will lead to
slight modifications in subsequent tests.

The standard analog to- digital conversion of the required 51gnals
'by the computer is not accurate enough. The requirement of operation
over & wide range precludes the possibility of accurate recording for the
relatively small incremental changes used in these tests. It is eXpected
that this can be corrected by use of an appropriate bias and additional
electronic amplification of the signals. Future tests will also use'anaQ
log-tape recording of the data in parallel with the recording by the MSRfififfiw
computer. This will serve to check the computer_results and will alsg,j
permit testlng of other analys1s procedures —~

The success with the pseudo-random binary tests will lead us to |
expand the use of this test. In particular, the lengthqof each bit in 'gfi;'iflu
the l9—bit sequence will be increased in somevof the tests, and sequences%”b”i‘

| = - | T
with more bits will be used with the rod motion controlled by the MSRE L

1‘ e

computer. These modifications will give the resolution needed to detectqfifl}aq,
the resonance peaks expected at low power in the frequency-response magnij
tude-ratio curves. '_ .
The results shown in Figs. 10 through 14 were obtained by straight-
forward application of basic procedures. The pulse test results were
filtered prior to analysis, but the‘binary test results were used in un-
modified form. These results will be further refined by application of
additional data-handling techniques such as smoothing and automatic trend
removal. However, early results were good enough that these modifications
should not produce much improvement. The noise-test results were in-
fluenced by the tapeérecorder characteristics. This instrument is cur-
rently being calibrated and the results will he.corrected. This correction
should be important only below 6 radians/sec.
Transient Flow Rate Tests .

The purposes of the transient flow-rate tests'were: 1) to obtain

 

startup and coastdown characteristics for fuel- and coolant- pump speeds

and for coolant-salt flowrate; 2) to infer fuel-salt flow-rate transients
36

from the results of (1); and 3) to determine the transient effects of
flow'changes on reactivity and void fraction.

Figures 16 and 17 show the fuel-pump speed, coolant-pump speed,
and coolant-salt flow rate vs time for pump staftup and coastdown. Data
were taken with the computer and with a Sanborn oscillograph. The output
of a differential-pressure cell acrossythe coolant-salt venturli was re-
corded directly on the oscillograph, and the square root of that signal
was taken as flow rate. The lag in the response of the computer flow
signal is due to the responsé characteristics of the EMF-to-current con-
verter and the square-root converter between the differential—pressure
transmitter and the computer input. |

It was hoped that the coolant-pump speed and coolant flow rate would
coast down in unison so that the fuel flow rate coastdown'couid be in-
ferred directly from the fuel-pump speed coastdown curve. This was not
the éase, however. Othegmmethgfigwg£/aflaiysié“willbew@ttemptedwlatéf:

Reactivity effects of fuel flow-rate transients were measured by
letting the flux servb controller hold the reactor critical during the
transients; the reactivity added by the rod is then equal (and opposite)
to the reactivity change due to the flow perturbations. The data for
the pump startup were taken on the computer but were inadvertently erased;
the reactivity transient for the pump coastdown was recordéd. Due to the
absence of voids in the fuel loop during normal operation, this transient
is due entirely to flow effects. Further analysis of this curve will be
made to try to determine the flow coastdown transient and to check on the
model used to represent the circulating delayed-neutron precursors.
Conclusions from Dynamic Tests |

The two most significant conclusions to be obtained from the dynamic
tests are 1) the information obtained gives no indication of the existence
of inherent characteristics that might lead to operating difficulties in
the low-power runs, and 2) the selected tests were, on the whole, quite
satisfactory. These tests gave results which show good agreement with
theoretical predictions giving increased confidence in the theoretical
model and in the predictions for stable powér operation. Since the
ZETrO-power tésts of this type are always mofe‘difficult than power-level

tests, very good results are expected from later tests.
 

% Full Scale

120

CRNL DWG. 65-8911

 

 

 

 

  
 

 

 

 

 

 

 

Fuel .
0l e ' 55— 5——0—0——0—0—0—
Speed _
Coolant
Pump
- Speed
80 o
o
60
O Logger
® Oscillograph

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

&P

5 6 T 8 9

Time, sec

Fig. 16. Pump Speed and Flow Startup Transients.

10

LE
% Full Scale

ORNL DWG, 65-8912

 

O Logger

@® Oscillograph

 

 

 

 

 

\\ \

8¢

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20 V
Coolant Pump Speed |
o “N | |
O ) N ooy [ Jbemy
0 12 1k 16 18

Time, sec

Fig. 17. Pump Speed and Flow Coastdown Transients.
39
FUEL SALT CHEMISTRY®

Fuel samples were removed daily from the pump bowl during the last
precritical operation (Run PC-2) and the zero-power experiments (Run 3).
Chemical composition, contaminant levels and isotopic analyses were de-
termined on a regular ba81s Changes in composition during the operation
were only such as could be explained by the addltlons of enriching salt.
The analyses support the conclusions that the salt was given excellent
probection against moisture or oxygen, that no uranfum precipitated, and
that essentially no corrosion of the INOR-8 in the fuel system occurred

during the period of approximately 1100 hours.
PERFORMANCE OF MECHANIGAL COMPONENTS

Insgeneral, the performance of the many mechanical components of the
system was highly satisfactory throughout the run. This is particularly
true in view of the fact that some of the items were being integrated
into the system operation for the first time. Some difficulties were
encountered which caused temporary inconvenience but no program delays
resulted and no extensive modifications will be required to improve future
performance. This section deals only with the difficulties that were
experlenced their actual and potentlal effects, and the changes whlch

they incurred. |

,Control Rods o | | | |

Two of the control rods and drlves were 1nstalled durlng run PC-1
and were used in simulator tralnlng Before PC-2 all three rods were
1nstalled and subgected to a test cons1st1ng of 100 cycles of full w1th-
drawal and scram. The rods operated freely and never falled to scram,
but occas1onally the lower limit switches falled to clear properly as the
rods were withdrawn. We found the cause to be galllng in the cam actu-
ator for the.switch After we 1nstalled stelllte bearlng strlps to
remedy this problem, each rod wa.s successfully ralsed and scrammed 30
times w1thout any malfunction. (The lower 1imit sw1tch on Rod 2 at

first stuck as before, and we found that a shim had been left out of the
Lo
switch-actuator assembly. After the shim was replaced there was no
further trouble.) Trouble-free operation continued throughout Run 3.
Rod drop times were measured in the tests in PC-2 and in a series of
LO scrams at the end of Run 3. The results (Table 3) show that the drop
times became slightly shorter and more uniform. This is consistent with
development experience in breaking in new flexible rods.

Table 3, Observed Drop Times of Control Rods

 

Number of | | ___Drop Times?® (msec)
| | _Timed Drops Average ___Standard Deviation
~ Test Series Rod 1 Rod 2 Rod 3 Rod 1l Rod 2 Rod 3 Rod 1 Rod 2 Rod 3

Original L5 6L L6 820 818 870 17 25 30
- At end of | - |
~ Run 3 Lo L L1 793 TT5 T92 4.0 L.1 3.7

 

Nomgs—

aTime from actuation of the scram switch (with rods at 51-in. withdrawal)
until actuation of lower limit switch (0 inch).

_Measurements of indicated rod positions with the lower end of the
poison at the fiducial zero position were made after installation and at
intefvals during Run 3. The data did not indicate any stretching of the

rods.

 

| SamPler*Enricher‘
| The installation of the fuel-system Sampler-Enricher was completed
during the shut-down period before Run PC-2. Since that time, 54 samples
werewithdrawn and 87 enrichingcapsules were added. Although séveral
minor problems occurred during the reactor operation, none of these prob-
lems prevented the sampler from being operational. In some cases delays
were incurred during sampling or enriching, but the overall schedule was
not signifiCahtly'affected. ‘Specific problems that were encountered are
listed below. | | |
1. Both the operational and the maintenance gate valves developed
| vleaké through one of the two seats of each valve. (The space
between the seats of each valve is helium-buffered and monitored
continuously when it is closed.)
2. The removal valve leaked and required a greater-than-normal
helium flow for buffering. | |
3. A solenoid valve ohthe removal—ValVe actuator failed and was

replaced.

 
/’\‘

b1

k.  The removal valve occasionally failed to close completely and
had to be closed manuvally. |

5. The access port periodically failed to close properly because
of faulty operation of the clamps on the clamp actuators.

6. One sample capsule was accidently dropped down the sample
tube to the operational valve gate. The capsule was recovered
and removed without significant difficulty.

T The drive motor stopped once during the removal of an empty
capsule. The capsule was:inserted'about 12 inches and was then

- successfully withdrawn. The reason for this stoppage:is'un-
‘known.

8. There were 3 boot failures on the manipulator with one of these
involving both boots. The inner boot was damaged twice by the
operation of the manipulator. Both boots were blown off by

- excessive pressure difference which resulted from an operational
error. No detectable release of contamination occurred with
any of the failures.

9. The manipulator arm and fingers were bent causing some diffi-
culty in gripping the latch cable and moving fhe manipulator

arm. | | | | | |

Most of thé troubles.resulted f'rom this period being one of testing

of equipment and training of operators. A mock-up had been thbroughly
tested but the installation of the samplerfenricher on the reactor was
completed just béforé,the ¢ritical experiments wére begun._»The»operators
are now well trained and some minor chénges are being made to improve the
reliability of fihe equipménfl and thé Safety of the bperation. The device

is expected to perform satisfactorily during power operation.

Freeze Valves

 

Several problems occurred with the freeze valves, but most of these
were corrected prior to Run PC-2. During critical operation the only
important requirement is that of ensuring a sufficiently short thaw time
for the drain valve, FV-103. A proportional controller was added to the
FV-103 cooling-air supply to maintain the freeze valve at a preselected

temperature which would result in a suitable thaw time. The temperature

 
L2

control and the valve temperature distribution were controlled satis-
factorily at any steady temperature level but it was necessary to change
the controller set points whenever the fuel-system temperature was changed
appreciably. The following thaw times were recorded for the drains after
Runs PC-2 and 3. |
Run PC-2: Carrier-Salt Drain - 16 1/2 min (system cooled to
| 1100°F)
Run  3: Fuel-Salt Drain - 10 min
Run 3: Flush-Salt Drain - 18 min
An incident occurred during Run 3 which could have delayed an emer-
géncy drain of the fuel system had it been required.v Electrical power
to the freeze-valve control modules was lost for a period of about 10 min
because the terminals of a temporary recorder were accidently shorted.
The loss of module power turned blast air onto the freeze valves. Normal
operation of the valves was restored after module power was regained.
Control circuit revisions will be made to prevent the recurrence

of this problem.

Fuel and Coolant System Pressure Control

Difficulties were experienced in controlling the fuel and coolant
system pressures within close limits because of accumulation of solids in
the offgas throttling valve used for fuel-system pressure control and in
the filter just upstream of the coolant-system pressure control valve.

During Run PC-1 (January - March, 1965), when the coolant salt was
circulated for 1200 hours, the coolant offgas filter plugged and was re-
placed twice. When the filters plugged, pressure was controlled at
5 = 2 psig by manual venting through a larger bypass valve.) Inspection
showed that the filter was covered with amorphous carbon containing traces
of the constituents of the coolant salt and INOR-8. The filter was re-
placed before PC-2 with one having 35 times the surface area. Coolant
salt was not circulated again until near the end of Run 3 and then for
only 118 hours. During this time the pressure control again became
erratic, indicating obstruction of either the filter or the valve. These

will be inspected and cleaned or replaced.
43

The fuel-system pressure COhtrol became erratic near the end of
Run PC-1 and at the conclusion of that run the offgas filter was removed.
Tt proved to be clean, so the pressure control valve was removed and found
to be partially plugged. The obstruction was blown out with gas and the
valve was washed out with acetone. The valve then performed normally and
was reinstalled. The acetone rinse was darkened and contained small
(1 — 5u) beads of a glassy substance. | -

Fuel-system pressure control was éatiSfaCtory'at'the beginning of
Run PC-2, but within a week the valve began sticking again. This time
the valve was replaced with one having a larger C_ (0.0T7 instead of 0.02).
The original valve was cut open for inspection and a black deposit was
found partially covering_the tapered stem. The deposit was about 20%
amorphous carbon and the femainder was the 1 — 5u glassy beads, which
proved to have the Qompositioh of the flush salt. | )

The larger repiacement valve in the fuél offgasline:gave,adequaté
pressure cbntrol until near the end of Run 3, after about 30 days of salt
circulation, when it appeared to be sticking occaSionally. When this
happened, the pressure built up slowly to about 6 psig before the valve
opened to drop it back to the normal 5 psig.

| The cause of the solids in the offgas lines is not yet known. There
is reason to believe that some carbon may have been introduced to the
reactor with the salt, accumulated on the surface in the pump bowl and
been carried into the offgas line as a dust. O0il contamination of the
salt system has also been suggested. The glassy salt beads in the fuel
offgas line are probably frozen droplets of mist caused by the stripper

spray, but we do not know whether these are carried into the line con-

- tinuously during operation or were swept out of the pump bowl by sudden

venting.

A filter, capable of removing l-y particles, 1s to be installed in
the fuel system. (The pore size of the original filter was about 25pu.)
Presumably this will protect the valve from further accumulations.
Measures to remedy the coolant-system problem will depend on the results
of the inspection of the filter. In any event, the coolant offgas filter
and pressure-control valve can be maintained directly after power opera-

tion should this prove necessary.
Ly

Failure of the controllers to maintain absolutely constant system
pressures will have very little effect on either the operation or the
safety of the system. The pressure variations that have been observed
have been both slow and small. Furthermore, we will normally operate
with almost no entrained gas in the fuel loop so that the pressure
coefficient of reactivity will be very small. As a result, the reac-
tivity effects associated with pressure variations, if they are observable

at all, will be easily handled by the servo control system.
ACKNOWLEDGMENT

This report is the work of several people'who were responsible for
the experiments ahd who contributed substantially to this summary . They
are J. R. Engel, B. E. Prince, S. J. Ball, T. W. Kerlin, R. E. Thoma, and
C. H. Gabbard. The experimental operation was by the entire staff of the
MSRE, with substantial participation and support by many in the Reactor
Division Development Sections, Reactor Chemistry Division, Instruments

and Controls Division, and Plant and Equipment Division.

r.
N

L5
REFERENCES
P. N. Haubenreich, et al., MSRE Design and Operations Report,

Part TTIT, Nuclear Analysis, USAEC Report ORNL TM-730, Oak Ridge
National Laboratory, February 3, 1964,

 

B. E. Prince, MSRP Semiannual Progress Report, January 31 l96h
USAEC Report ORNL-3626, pp 53-5L4.

P. N. Haubenreich, Prediction of Effective Yields of Delayed Neutrons
in ‘the MSRE, USAEC Report ORNL TM-380, Oak Ridge National Iaboratory,
October 13, 1962. |

B. E. Prince, Comparison of Calculations and Uncertainties in the

Temperature Coefficients of Reactivity in the MSRE, USAEC Report

 

 

 

 

ORNL CF 63-3-1, Oak Ridge National Laboratory, March 3, 1965.

S. J. Ball and T. W. Kerlin, MSRE Stability Analysis, USAEC Report
ORNL TM-1070, Oak Ridge National ILaboratory (in press).

R. E. Thoma, MSRE Salt Chemistry During Precritical and Zero-Power
Experiments, Internal Memo MSR-65-40, August 2, 1965.

 
7N\

Distribution

MSRP Director's Office, L6,

Rm. 219, 920L4-1 L.
G. M. Adamson L8,
S. J. Ball L9.
S. E. Beall 50.
E. S. Bettis 51.
F. F. Blankenship 52.
W. H. Cook 53.
J. L. Crowley 54,
5. J. Ditto 55.
J. R. Engel 56.
E. P. Epler 57.
C. H. Gabbard 58.
R. B. Gallaher 59.
W. R. Grimes 60.
R. H. Guymon 61.
P. H. Harley 62.
P. N. Haubenreich 63.
V. D. Holt 6L .
A. Houtzeel 65-66.
T. L. Hudson 67-68.
P. R. Kasten
R. J. Kedl 69-70
T. W. Kerlin T1-73.
A. T. Krakoviak Th.
R. B. Lindauer
C. D. Martin
C. E. Mathews
H. G. MacPherson
W. B. McDonald
H. F. McDuffie
C. K. MecGlothlan
R. L. Moore
H. R. Payne
A. M. Perry
H. B. Piper
B. E. Prince

External

H. M.

4

 

ORNL-CF-65-8-32

I,. Redford
Richardson

H. Shaffer
J . Skinner
N. Smith
G. Smith
F. Spencer

R. Tallackson
E. Thoma
C. Ulrich
. H. Webster
M. Weinberg
E. Whatley
. White
Central Research Library
Document Reference
Section - Y-12
Reactor Division Library
ILaboratory Records
Laboratory Records - RC

SH.ZFw.Z!?U‘:—lb':UHS*'UlDZCflUmZC-*
N
o
H
%
W

Q

Roth, Division of Research and Development, AEC-ORO